STATE  OF THE ART:   1971
INSTRUMENTATION  FOR  MEASUREMENT
       OF  PARTICULATE  EMISSIONS
      FROM COMBUSTION  SOURCES
      VOLUME  III:   PARTICLE SIZE
                      by
             Gilmore J. Sem, John A. Borgos,
           Kenneth T. Whitby and Benjamin Y.H. Liu

                 Thermo-Systems Inc.
               2500 North Cleveland Avenue
                St. Paul, Minnesota 55113
                Contract No. CPA 70-23
               Program Element No. 1AA010
            EPA Project Officer: John O. Burckle

             Chemistry and Physics Laboratory
           National Environmental Research Center
            Research Triangle Park, N. C. 27711
                   Prepared for

           OFFICE OF RESEARCH AND DEVELOPMENT
          U.S. ENVIRONMENTAL PROTECTION AGENCY
               WASHINGTON, D.C. 20460

                    July 1972

-------
This report has been reviewed by the Environmental Protection Agency
and approved for publication.  Approval does not signify that the
contents necessarily reflect the views and policies of the Agency,
nor does mention of trade names or commercial products constitute
endorsement or recommendation for use.
                                  11

-------
                        TABLE OF CONTENTS




                            VOLUME III





                                                                 Page




FOREWORD 	  1




ABSTRACT AND CONCLUSION 	  2




A. INTRODUCTION	4




B. TECHNIQUES WITH SEPARABLE CLASSIFICATION AND SENSING ...   .13




   1. Particle Classification Techniques 	 13




   2. Particle Sensing Techniques 	 44




   3. Combinations of Classifiers and Sensors 	 44




C. TECHNIQUES WITH INSEPARABLE CLASSIFICATION AND SENSING .  .   .54




   1. Optical Techniques 	 54




   2. Impact and Momentum Sensors	61




   3. Piezoelectric Single Particle Counter 	 62




D. LABORATORY POWDER SIZING TECHNIQUES 	 63




E. A DIFFERENT SIZING CONCEPT: PARAMETRIC MEASUREMENT 	 65




F. SUMMARY AND CONCLUSIONS	69




G. REFERENCES	72
                              111

-------
                                    FOREWORD
     The compilation of the information contained in this publication was per-
formed pursuant to Contract CPA 70-23 for the Environmental Protection Agency.
The work was sponsored by the National Environmental Research Center at Research
Triangle Park for the purpose of conception and evaluation of instrument systems
for particle size distribution applicable to the measurement of emissions from
stationary sources.

     This report was prepared during the period from January to August, 1971, and
is the third of a four volume series.  These volumes contain the following:

Volume I of this report is written for the engineer or planner who needs to know
a few basic facts about a particulate mass measurement technique and wishes to
minimize the time required to obtain this information.  Volume I is intended for
use as a quick reference guide.

Volume II of this report is designed as a detailed in-depth report on operating
principles, techniques, historical data, and discussion of the more viable tech-
niques for particulate mass monitoring.  Volume II is designed for the plant
engineer, abatement and control officials, and others who may not be familiar
with the detailed technology of these areas.  Included are sections on power
plant emissions properties and extraction sampling probes.

Volume III of this report is a comprehensive survey of particle sizing techniques
which may be used by the plant engineer, abatement and control officials, and
others as a quick reference guide or as a source of more detailed information,
including references to original work.

Volume IV of this report describes an experimental evaluation of the beta radia-
tion attenuation technique for mass concentration measurements on a coal-fired
power generating plant.  Problem areas requiring further developments are iden-
tified for personnel concerned with improving the techniques.

These reports have been issued as they were completed to make them available
to the public on a timely basis.  Volume I and II were issued in September 1971.
Volume IV will be issued late this Fall.

-------
                            ABSTRACT AND CONCLUSION


     Volume III (this volume) discusses candidate techniques for automatic
or semi-automatic measurement of particle size distribution in combustion source
effluents.  Automatic or semi-automatic particle size measuring instruments do
not yet exist for this application.  This report considers the application to
effluent streams of particle size measuring instruments used in other fields.
The discussions emphasize the particulate concentration parameter (mass, number,
surface area, etc.) which each technique senses as well as the method of classify-
ing particles into size ranges (aerodynamically, electrostatically, optically,
etc.)  Included are descriptions of the basic operation of each technique,
discussions of limitations of each technique, suggestions of possible major
problems in applying each technique to effluent streams and an overall evaluation
of each technique relative to others.

     The most promising approach for detailed size analysis on a routine basis
at this time is an aerodynamic particle size classifier combined with either
beta radiation attenuation, piezoelectric microbalance, or photometric concentration
sensors.  The impactor classifies particles aerodynamically, the most useful method
for most air pollution applications.  The beta and piezoelectric sensors (see
Volume I and II of this report for discussion of sensors) can detect the mass con-
centration, while a correctly-designed photometric sensor detects a parameter
(related to surface area) which could be used as an indicator of the effect of the
emissions on visibility.  Although the more promising aerodynamic  techniques can
classify particles in an approximate range from 0.2 to 30 microns, the particles
above and below these limits could be lumped into separate size categories, per-
mitting a gravimetric size measurement covering nearly the entire range of particles
found in effluent streams (from about 0.001 to above 100 microns).

     Several techniques appear applicable to special sizing problems in effluent
streams. Cyclone classifiers can be used to separate the "respirable fraction"
of particles.  Electrostatic and diffusion classifiers can possibly be used to
measure detailed size distributions in the range from 0.001 to 0.6 microns.
Modified optical particle counters may be useful for some research applications.
Holography offers the ability to photograph (in 3-dimension) the effluent par-
ticles in the stream without disturbing them in any way, but appears limited for
most sizing applications by its lower particle size resolution when operated across
stack distances, by its cost, and by its complexity.

     Another concept looks promising for use as a continuous, routine monitoring
technique for effluent streams.  The technique uses the size limitations of at
least three particle concentration monitors to measure particle concentration in
three or more size ranges.  Three candidate concentration monitors measure:  (1) the
mass concentration (sensitive primarily to 1 - 100 micron particles), (2) the opacity
of the effluent stream (sensitive primarily to 0.1 - 10 micron particles), and  (3)
the number concentration (sensitive primarily to 0.001 - 1.0 micron particles).

-------
Respirable mass concentration is another promising candidate.  Instruments which
measure most of these parameters have already been used in effluents so only a
moderate amount of hardware development appears needed.  Simple analysis of such
measurements made simultaneously may provide sufficient particle size and con-
centration information for most routine monitoring applications.

     Although not covered in this report, considerable research and development
must be done to develop sampling systems which can deliver truly representative
particle samples to tlie sizing instruments.  No practical candidate for automated
sizing exists which has sufficient resolution to cover the most important effluent
size range and which does not require extraction of a sample from the effluent
stream.

-------
                              INSTRUMENTATION FOR

                       MEASUREMENT OF PARTICULATE EMISSIONS

                             FROM COMBUSTION SOURCES

                           VOLUME III:  PARTICLE SIZE
A. INTRODUCTION

     The concentration, size, and chemical composition of airborne particles
are the three most important properties defining the potential effects, harm-
ful or otherwise, of most particulate dispersions.  Automatic measurement of
the concentration, specifically the mass concentration, of airborne particles
is the subject of Volumes I and II of this report.  This volume discusses the
automatic measurement of size and size distribution of airborne effluent par-
ticles from large coal and oil combustion sources.  Chemical composition and
its measurement is not a subject of these reports.

     The size of an airborne effluent particle plays a very important part in
determining its future as an air pollutant.  One of the primary purposes for
the measurement of particulate effluents is to monitor the potential harmful-
ness of the emissions so that the degree of control can be evaluated.  A con-
tinuous measurement of particle concentration or emissions rate is usually not
a sufficient indication of the potential harmfulness of particulate emissions.
Measurement of the size of the particle greatly improves the estimate of the
potential harmfulness.

     To comprehend the magnitude of the effect of particle size, consider one
microgram of particles that is made up of a single 100 micron (ym) diameter
particle with a density of about 2.0 grams per cubic centimeter.  Nearly all
100 urn particles are collected by most control equipment on effluent sources.
The few 100 ym particles that are emitted to become air pollution settle to
the ground quickly, usually within a mile or two of the source.  Their settling
velocity is about 50 centimeters per second.  The settled particles result in
such harmful effects as dirty cars, houses, and streets, contamination of soil
resulting in harm to nearby farm crops, contamination of nearby lakes and rivers,
and the aesthetic degradation of the community.  There is almost no direct health
hazard to humans from inhalation of such particles because very few reach the
person, very few of those that do reach him enter his respiratory system, and
nearly all which enter the respiratory system are collected by the entrance nasal
passage.  The chemically-reactive surface area of 100 ym particles is relatively
small compared to the same mass of smaller particles.

     If that same microgram of particles is made up of particles 1 ym in diameter,
there would be about 1,000,000 of them.  Most particulate control equipment in use
today probably collects only about half of the 1 ym particles in an effluent stream.
Most cyclone collectors collect almost none of these particles while high efficiency

-------
electrostatic precipitators can be designed to collect most of them.  The
settling rate of 1 ym particles is about 0.05 centimeters per second, almost
negligible in many cases.  A 1 ym particle does not grow very rapidly by
agglomeration with other particles or by condensation of liquids on its surface.
Thus, these particles will travel long distances from the effluent source before
settling to the ground or depositing onto a surface.  A large portion of such
particles emitted by effluent sources in industrial areas reach heavily-populated
centers where they scatter large amounts of light resulting in reduced visibility
and less sunlight reaching the ground, where they coat surfaces with grimy films,
and where people inhale them.  One ym particles penetrate more deeply into the
respiratory system than nearly any other size.  The chemically-reactive surface
area of 1 ym particles is about 10,000 times greater than an equal mass of 100 ym
particles.  One ym particles, or slightly smaller, penetrate even the highest
efficiency fiber and membrane filters more easily than any other size.  Thus, 1 ym
particles are a different type of air pollutant than 100 ym particles, and should
not be lumped with 100 ym particles for measurement purposes.

     Now, if that microgram of particles is made up of 0.01 ym particles, there
would be 10-J-2 of them.  Although no comprehensive study has been made, even the
highest efficiency electrostatic precipitator control equipment in use today
probably does not collect a significant fraction of these particles.  It appears
that only high efficiency filters can collect them.  Their gravitational settling
velocity is about 10"-' centimeters per second, completely insignificant when com-
pared with motions caused by forces such as wind, Brownian diffusion (random
molecular bombardment), temperature gradients, electrostatic fields and charges,
and Van der Waals molecular attractive forces.  These particles can grow rapidly
by condensation of vapors, resulting in a higher liquid content of the particle.
They can also grow by agglomeration with other particles resulting in a lesser
number of larger particles.  The size distribution of particles in this size range
often changes rapidly, making measurements difficult, and resulting in larger
particles (0.1 - 1.0 ym) after a period of time.  In normal city atmospheres, this
growth process requires a few hours.  Particles in the size range from 0.01 - 1 ym
make up most of the photochemical smog so prevalent in cities like Los Angeles.
Brownian diffusion causes most of the 0.01 ym particles to deposit in the upper and
middle portions of the respiratory system, causing them to penetrate less deeply
than 1 ym particles.   These particles are highly reactive, having a surface area
about 100 million times greater than an equal mass of 100 ym particles.  Particles
in the 0.01 - 1 ym range are often the nuclei for raindrops and ice crystal formation.
The effects of these small particles has little in common with the effects of 1 or
100 ym particles, and the concentration of such particles should be measured
separately.


     Combustion effluent sources emit particles over an even broader size range:
from about 0.001 to over 100 ym.  All portions of the size range are important in
defining the pollutant potential of particulate emissions.  Prediction and con-
trol of photochemical smog formation requires information about the very small

-------
particles (from about 0.001 to 1.0 ym) .   The study and prediction of mete-
orological effects such as modified raindrop production also need measurements
in this size range.  Control of respirable particles requires measurement of
particles in the 0.1 - 10 ym range.  The range from 0.1 - 10 ym affects visibility
most strongly.  Control of dust fall near an effluent source requires measurement
of the particles larger than about 10 ym.  Thus, the need .for measurements of
particle size over the entire range is established.

     The size of a particle is usually defined in terms of a diameter or radius.
However, anyone looking at a sample of effluent particles with a microscope
realizes that most particles are not spherical and, hence, do not have a well
defined diameter.  It is difficult, for example, to assign a single character-
istic dimension to a rod- or flake-like particle.  Many conventions have been
used for classifying particles by size with a microscope, such as using one of
the following as the characteristic size:  the longest dimension, the diameter
of a circle with cross-sectional area equal to the particle in question, or the
diameter of a sphere with volume equal to the particle in question.

     However, the most useful method of classifying particles by size in most
air pollution systems is based on how rapidly the particle settles out of the
atmosphere.   The relative harmfulness of a particle depends on several factors,
one of the most important being whether the particle remains airborne long
enough to reach high density human populations.  Thus, the most reasonable size
classifying system places particles in categories depending on their relative
settling velocities.  Particle size measured in this way is called the aero-
dynamic size, and is usually referenced to the settling velocity of a sphere
with a specific density of 1.0 gram per cubic centimeter.  Thus, a particle
which settles at the same velocity as a 10 ym diameter spherical particle with
specific density of 1.0 gram per cubic centimeter is said to have an equivalent
aerodynamic size of 10 ym.

     Particles with the same equivalent aerodynamic size, even though with
different specific densities and shapes, will have an equal chance of settling
to the ground within a given time period.  For example, a 6.3 ym spherical
particle with specific density of 2.5 grams per cubic centimeter has an equivalent
aerodynamic size of 10 ym.

     Equivalent aerodynamic size also characterizes the ability of a particle to
penetrate into the deepest portions of the human respiratory system after entering
the nose.  Many of the dynamic properties of airborne particles, such as agglomeratior
collision, and reaction rates between particles, depend upon equivalent aerodynamic
size.

     From the time the particle enters the atmosphere until It becomes firmly
attached to a surface, i.e., during its entire life as an airborne pollutant,
equivalent aerodynamic size is usually the most useful characteristic size.  It
is our opinion that, although other methods of classification may be useful in
specific cases, for most pollution measurement applications, size classification by
aerodynamic methods is preferred.

-------
     A  look at  the equations defining equivalent aerodynamic size will show
 the important parameters which affect it.  The gravitational settling velocity
 v  of a spherical particle in air is expressed in terms of particle and air
 properties by the well-known Stokes equation  (e.g., see Ref. 1333, p. 23):

                v  = T-m|-                                  '           (A.I)
                s   3irnD
                        P
 where
                m = mass of the particle

                g = gravitational constant

                n = viscosity of the air

                D  = diameter of the particle.
                P
 Equation  (A.I)  is valid only at atmospheric pressure and for particles having an
 aerodynamic diameter larger  than about 1 ym  diameter.  In the molecular slip flow
 region below 1  ym, the Cunningham slip correction C to the Stokes equation applies.
 The Cunningham  slip correction takes account  of the discontinuous nature of particle-
 molecule  collisions when the particle is comparable to the mean free path in size.
 (See Fuchs^--" , p. 25, for detailed discussion.)  Equation (A.I) then becomes:

                                                                      (A.2)


 With particle mass m expressed in terms of particle volume and density y >
 Equation  (A.2)  becomes:

                     D^gC
                                                                      (A.3)
To calculate the equivalent aerodynamic diameter of a particle which settles with
velocity vs, solve Equation (A.3) for D , which then becomes equivalent aerodynamic
diameter referenced to the particle density y  •  Thus, Equation  (A.3) defines
equivalent aerodynamic size.

     We have now briefly discussed some of the various ways of defining particle size.
An equally important factor in expressing particle size distributions is how the
amount of particles in each size category is expressed.  This is called the weighting
of the size distribution.

     There are many ways to express the amount of particles in each size category.
One way often used in expressing atmospheric particle size distributions is the number
of particles per unit air volume within each size range.  This is called the number
concentration size distribution.  Other commonly used particle concentration

-------
                                                                          8
weightings include particle mass, particle surface area, and particle volume.
Rather than expressing the actual particle concentration in numbers, milligrams,
square centimeters or cubic centimeters of particles per unit air volume, the
concentration weighting can be expressed as a percentage of the total which is
contained within each size category.  For example, in the case of a particle mass
percentage weighting, the result would be a graph showing the percentage of the
mass of particles contained within each size category, or-more commonly, the per-
centage of the mass of particles contained by particles larger than (or smaller
than) a given size.

     Figure 1 illustrates several of the more common ways of expressing a particle
size distribution in graph form.  All nine graphs in Figure 1 represent the same
particle size distribution.  Notice that each curve has a different shape or slope,
depending on which particulate parameter is represented and on how the investigator
chooses to represent it.  There are many other ways to plot particle size distri-
bution, each resulting in a curve with a unique shape.

     It is not always clear which way of expressing size distributions is most
advantageous in any given situation.  Each has its limitations.  A mass distri-
bution generally emphasizes large particles and deemphasizes or ignores small
particles which are important but do not contain significant mass compared to
the large particles.  A number distribution, on the other hand, emphasizes small
particles which are present in large numbers.  A number  distribution ignores
the one or two large particles in a cloud of millions of small ones, even though
the one or two large particles may weigh more than the millions of small particles.
A concentration distribution shows the actual particle concentration level, an
important factor determining particle interaction rates and as essential factor
for pollution control monitoring.  The percentage distribution, as well as several
other distributions, shows quickly, clearly, and without ambiguity, exactly which
size range contains the majority of the particles.  Although one can usually con-
vert mathematically from one distribution to another, any measurement error becomes
magnified with each conversion.  One cannot convert from a percentage distribution
to a concentration distribution without  additionally knowing the total concentration
of the particle sample.

     Thus, it is important to decide which method of expressing size distribution
is most useful for a given application before choosing the size measurement
technique.  The investigator will then choose a size measuring technique which
yields the desired size distribution as directly as possible.  Although a strong
case can be made for aerodynamic size as the most useful for most air pollution
measurements, the choice of the method of expressing particle concentration is not
as simple.  One must be very careful not to ignore an important range of particle
sizes simply because of the choice of particle concentration weighting.

-------
                                                                    c
                                                                    o
                                                                   cr
                                                                   <
                                                                   o_
      SSVIfl                   V3dV

               NOI1VU1N30NOO  310llHVd
Figure 1. Nine ways of presenting the same particle size distribution.

        The distribution shown is not necessarily realistic, but is

        made up entirely of particles in the 0.1 to 100 micron range.

-------
                                                                              10
     The discussion above is concerned primarily with complete size distri-
butions and methods of presenting them.  There are ways of expressing important
features of a distribution as single numbers.  A common method is to express an
average size (usually diameter) accompanied by a number representing the spread
in the distribution (usually standard deviation or geometric standard deviation).
The range of particle size, meaning the size of the largest and smallest particles
in the distribution, is another measure of the spread of a distribution.  Other
"averages" include the mean, median, and mode of the distribution.  The number
median and mass median diameters are commonly used.  The mass median diameter is
the diameter which breaks the distribution into two size ranges, each containing
50% of the total mass.

     This discussion has included only a brief summary of the most important aspects
of the definition of particle size and size distribution.  Any reader contemplating
work in this area is strongly urged to refer to a text such as Herdan (Ref. 1357).
Many other aspects are covered in Herdan, including details on converting from one
distribution to another, how to choose the optimum number of size ranges to cover
a size distribution, other ways of presenting size distributions, hints on in-
terpreting a size distribution, etc.

     At present, no instrument is available that can measure the complete range of
particle sizes in an effluent source.  One reason is the inherent problem of the
choice of particle weighting discussed earlier.  Thus, if the mass size distribution
is measured, small particles are ignored, etc.  However, every instrument also has
practical limitations of size range.  Few instruments can cover more than a range
of one-two decades in particle size.  No instrument with potential for automation
can accurately cover more than 2 size decades.  Since combustion effluents cover
at least 5 size decades (0.001 - 100 ym) , either several different instruments
must be combined to cover the complete size distribution, or one must choose the
size range of primary interest and find a suitable instrument for that range.

     In the past, the range of sizes measured in combustion sources was dictated
by the available technique.  Thus, almost all measurements were of mass concentration
in each of several size ranges from about 2 ym to about 100 ym.  Particles below
2 ym could not be size classified by the technique and their mass could not be
detected in the presence of the larger particles.  This has led to the conclusion
that there probably is no contribution to the mass emissions by particles below about
1 ym in combustion effluents; however, this has not been documented by direct measure-
ments.  In the case of particles above 100 ym, most such particles settled out of
the sampling line before reaching the sizing instrument.  Thus, most existing data
on particles above 100 ym also appears to be inaccurate.  Most of the existing
particle size data from combustion sources is highly questionable for these reasons.

-------
                                                                            11
     Sampling statistics also often limit the validity of particle size data,
especially number count data.  Much of the existing microscopic count size data
was obtained using a single microscopic magnification.  With most atmospheric
and combustion-source aerosols, there are so many small particles that the likely-
hood of counting a large particle with this method is small, even if 1,000 - 10,000
particles are counted for a single size distribution.  For example, there are usually
1,000 - 10,000 times as many 1.0 pm particles in atmospheric aerosols as there are
10 pin particles.  Thus, for a reasonable statistical chance of obtaining an accurate
distribution covering only a single order of magnitude of size using a single
magnification requires the counting of about 100,000 particles, a formidable task
One method of avoiding this problem is to use several magnifications covering the
size range of interest.  Each magnification should cover no more than a factor of
5 in particle size.  Each magnification can then be chosen such that the smallest
size of interest is easily detectable and yet, the largest size of interest will be
counted enough times to result in a statistically valid distribution.  This same
statistical limitation also applies to any instrument which counts the number of
particles in a distribution.  The solution for such instruments is usually to count
enough particles to overcome the limitation.

     Better sizing instruments must be made available so that measurements will
include the entire size range of interest.  The increasingly recognized importance
of micron- and submicron-sized particles places special emphasis on the small particle
No single sizing technique can cover the entire size range.  Thus, several techniques
will have to be used.  The interpretation of the data obtained from several different
measuring techniques will be a problem in some cases.  However, no method is foreseen
that will measure detailed distributions over the entire size range with a single
measuring technique.

     The remainder of this report discusses specific particle size measuring
techniques in detail.  The discussions are limited primarily to a basic
evaluation of each technique rather than a discussion of exact design details.
The reason for this is that most techniques are not developed for application
to measurements of combustion effluents, making a discussion of design details
rather speculative.

     All particle sizing instruments must perform two distinct functions: (1)
classify the particles by size and (2) sense the amount of particles within each
size range.  In some measurement techniques, the two functions can be considered
separately and developed hardware may consist of the two distinct components.
Section B discusses first classifiers, then sensors, and last, combinations of the
two components.   It will be seen that most classifiers can be used with most sensors,
some combinations having better features than others.  Section C discusses sizing
techniques which, while performing both functions, does not allow separation of
hardware for each function.  Throughout Sections B and C, emphasis is placed on the
more feasible techniques.

-------
     A number of particle sizing techniques exist for powders and liquid
slurries.  Section D discusses possible application of these techniques to
combustion source aerosols.  These techniques are deemphasized because of
the difficulty in relating the size of the slurry particles to the particles
in their airborne state.

     Section E discusses a different concept of size monitoring for use in
combustion source effluents as well as elsewhere.  This concept has strong,
practical appeal, but requires more study and a moderate amount of develop-
ment and testing.

     Finally, Section F lists the most important conclusions and summarizes
this study.

-------
                                                                            13
B. TECHNIQUES WITH SEPARABLE CLASSIFICATION AND SENSING

     Many instruments for measuring the size distribution of an aerosol can be
broken into two independent components:

     1. A classifier which separates the particles into several
        size fractions, and

     2. A sensor which measures the amount of particles within
        each size range.

Conversely, practical sizing instruments can often be assembled by connecting
various classifiers upstream of various sensors.  This discussion first con-
siders particle classification techniques, then considers particle sensors, and
finally considers the most reasonable combinations of the two.

1. Particle Classification Techniques

     This chapter discusses all known particle classification techniques and
all known .forces which can act on airborne particles.  The classification
techniques  discussed below are:

          a. Aerodynamic Classification
               i.  Gravitational Sedimentation
              ii.  Gravitational Elutriation
             iii.  Inertial Impaction
              iv.  Centrifugal Spectrometry
               v.  Cyclone Separation
              vi.  Centrifugal Elutriation

         b. Electrostatic Classification

         c. Sieve Classification

         d. Filtration

         e. Brownian Diffusion

         f. Other Forces

     a.  Aerodynamic Classification

             When  a particle moves with respect to the surrounding air, the
        force which resists that motion is called aerodynamic drag force.
        For a spherical particle larger than 1.0 ym diameter moving with a
        slow,  constant velocity through air at atmospheric pressure, aero-
        dynamic drag force F,  can be expressed by the Stokes equation:

                          v                                           (B.I)

-------
                                                                  14
where:

       n = viscosity of  the gas,

      D  = particle diameter, and
       P
       v = particle velocity with respect to the gas.

The negative sign means  that the force acts in a direction opposing
the motion of the particle with respect to the gas.   For particles
between 0.1 and 1.0 pm (i.e., about the same size as the molecular
mean free path of standard air) and with all other conditions equal
to those for Equation (B.I), F, can be expressed by the Stokes -
Cunningham equation:
       F, = - -^                                          CB.2)

where:

       C = Cunningham slip-flow correction.

As D  becomes large compared to the molecular mean free path, C->1.0
and PEquation (B.2) becomes Equation (B.I).  Equation CB.2) is valid
with reasonable accuracy for particles with 0.1 ym
-------
                                                                               15
              Expressing m in terms of particle volume and density, Equation
              (B.3) becomes:
                         P P
              where:
                          18n
                       density of the particle.
                                                                         (B.4)
              Equation  (B.4) shows that, for particles of constant density, the
              settling velocity increases as the square of the particle size.  Thus,
              the length of time t^ it takes a particle to settle a given distance I
              decreases with an increase in the particle size:
£_
V
                                                                         (B.5)
                   In terms of hardware, a sedimentation size classifier is shown
              schematically in Figure 2.  The particles settle through the laminar
              clean air streamlines and deposit onto the bottom of the chamber.  The
              sharpness of the size classification depends on the height of the
              aerosol entrance duct with respect to the average settling distance, H,
              and on the velocity profile of the aerosol and clean air flow.
                                                               Fine
                                                               Screeny
Aerosol
  In
                      Trajectory of
                     •non-deposited particle
 Clean
  Air
  In
 Trajectory of
 deposited particle-
                Filter
                      Figure 2.  Schematic sedimentation size classifier with
                                horizontal air flow.

-------
                                                                      16
         If operated upstream of a concentration sensor, the sedimentation
    size classifier would collect particles larger than a given size,  and
    particles smaller than that size would pass through to the sensor.
    The particle size cutoff could be decreased by reducing the settling
    height £ or by decreasing the horizontal flow velocity through the
    classifier.  The former results in a less sharp size cutoff and the
    latter results in greater interference by convective currents.  Thus,
    any classifier of this type can be designed for only a narrow range
    of adjustable particle size cutoffs, probably one order-of-magnitude
    at most.  The overall practical size range for sedimentation classifiei
    is from about 1.0 |jm to about 50 pm.  The lower size limit is set  by
    the practical limitations on the smallest values of £ and air flow rate
    as well as by interference by convective currents.  The upper size
    limit is set by the difficulties in introducing large particles into
    the entrance without losing them.

         The sedimentation classifier dilutes the aerosol which passes
    through by the ratio of clean air to entering aerosol.  The upper  con-
    centration of entering aerosol is not a limiting factor for effluent
    aerosols.

         The Hexhlet dust sampler manufactured by

              C. F. Cassella & Co., Ltd.
              Regent House, Fitzroy Square
              London Wl, England

    uses a simplified sedimentation classifier which consists of parallel
    plates through which the aerosol passes.  There is no clean air sheath,
    making the classification considerably less sharp, but making operation
    easier.  Other authors report other models of sedimentation classifiers
    often called horizontal elutriators.274,834,1299,638,833,912

ii. Gravitational Elutriation

         Elutriation is similar, in many respects, to sedimentation.  The
    principle of a gravitational elutriator is shown schematically in
    Figure 3.  Laminar air flows vertically within the elutriation tube.
    Aerosol is introduced at the bottom of the tube.  Large particles, i.e.
    those with settling velocities (Equation B.4) greater than the vertical
    air flow velocity, will be carried out with the exhaust.  The particle
    cutoff size is determined by the vertical air velocity in the tube.

         The major factor causing degradation of the sharpness of the
    particle size cutoff is the shape of the air-velocity profile within
    the tube.  Ideally, the air velocity profile would be flat at all
    points.  In practive, however, a boundary layer forms at the tube
    wall, allowing small particles to settle to the bottom of the tube

-------
                                                                      17
                      Air and
                 Fine Particles out
                    t    t    t
                                     Air Distribution Screen
                                     (large enough  to pass
                                      nearly  all particles)
                         t
Large
Particles
out
                     Aerosol  in
        Figure  3.  Principle  of  a gravitational  elutriator.
through the boundary layer.  Convective currents caused by thermal
gradients within the elutriator also modify the velocity profile.
The ability to overcome these two related problems determines the
ultimate sharpness of the particle size cutoff in practive.  The
concentration levels within the elutriator must be low enough so that
large particles settling downward do not intercept appreciable numbers
of smaller particles.  The applicable size range of elutriation is
about 1 to 100 ym.  Other limitations are similar to those with the
sedimentation classifier.

-------
                                                                       18
          Several authors have considered gravitational elutriation in
     more detail. 638, 833, 912,1069, 1333(p. 42,43).

          A commercial gravitational elutriator for size classification
     of powders is manufactured by:

              Geoscience Instruments Corporation
              435 East 3rd Street
              Mount Vernon, New York 10553.

     The advertised size range is 5 to 100 ym.  This unit would not lend
     itself to the sizing of stack effluent particles on a continuous
     basis without considerable modification.

          Gravitational elutriation size classification is a definite
     candidate for application to effluent aerosols.  Gravitational
     elutriators operate over the approximate size range of 1 to 100 ym,
     and may have the ability to classify even larger particles because
     higher velocity in the entrance tube allows the larger particles
     to enter the sedimentation classifier.  Several elutriators designed
     for different size cutoffs would probably be necessary to cover the
     entire 1 to 100 ym range.  A strong feature of this classifier is
     that it results in an aerosol classified by aerodynamic equivalent
     size.

iii. Inertial Impaction

          A third method of using aerodynamic drag to classify aerosol
     particles is inertial impaction.  Figure 4-L340 shows an impactor
     schematically.   A flat plate is located at the exit of a nozzle,
     perpendicular to the air jet.   Air is sucked through the nozzle and
     must turn a sharp corner to pass around the flat plate.  Because of
     their inertia,  sufficiently large particles cannot turn the corner,
     and therefore hit the plate.  Smaller particles do not have enough
     inertia to cross the air streamlines and are carried along with the
     air stream.

          As particles cross the air streamlines in an impactor, the force
     which resists that motion is aerodynamic drag force expressed in Equatio
     (B.2)  with v being the component of the particle velocity perpendicular
     to the streamlines.

          When discussing impactors, it is convenient to define a parameter
     known as Stokes number:

                Y Cv D2

                                                              '  (B-6)
     where:

          Stk = Stokes number,

           v  = average air jet velocity, and

            W = jet width (rectangular jet) or diameter (round let).

-------
                                                            19
STREAMLINES
IMPACTION
PLATE
                        K
       w
r
T
L
A
)
L-y
\
1
1
/ /

-r-
1
1
1
1
I
\\
A
\
\
\
\
K
                                                  JEf
                         JET EXIT
              OF
IMPACTED  PARTICLE
TRAJECTORY OF
PARTICLE  TOO
SMALL  TO
IMPACT
  Figure 4. Schematic  of an impactor showing particle trajectories.1340

   A dimensionless particle size is defined as (Stk)1/2.  Particles with
   the same Stokes number have an equal chance of being collected by a
   given impactor.

       Ideally,  an impactor should collect all particles larger than
   a certain cutoff size and all smaller particles should escape as
   shown by the dotted line in Figure 5 "AO  In practice, however,
   impactors have a characteristic "S" - shaped efficiency curve.
                1.0
             ui
             O
             u.
             u.
                                       ACTUAL
              IDEAL
 Figure 5. Typical ideal and experimental collection
          efficiency  curves of an impactor.13^0

-------
                                                                20
There are several reasons for this nonideal operation.  These
reasons have been investigated in considerable detail in the
literature (see especially, References 1333 (pp. 151-159) and
1340), including:

     a) the velocity profile at the jet exit is not flat
        giving particles near the center of the jet greater
        inertia than particles near the jet walls,

     b) a boundary layer forms along the plate deflecting
        particles which are near the edge of the jet more
        than particles which are near the center line,

     c) the high jet velocities needed to collect small
        particles blow some of the larger particles off the
        plates or cause them to bounce off,

     d) some weakly-bound agglomerates of smaller particles
        break up while passing through the high shear forces
        within the jet or while striking the surface, and

     e) with rectangular jets, the ends of the jet act like
        a round jet: impactor while identical particles within
        the central areas have significantly different Stokes
        numbers.

     Marple     performed a careful theoretical and experi-
mental study of both round and rectangular jet impactors and
recommends the jet operating conditions shown in Table B.I as
the optimum for obtaining a sharp particle size cutoff.  The
dimensions S, W, and T refer to Figure 4 and Re is the Reynolds
number based on the average jet velocity and the jet  diameter
or width.  Figure 6^-340 shows the impaction efficiency as a
function of (Stk)l/2 for each of the two optimum conditions
outlined in Table B.I.  The rectangular impactor must be much
longer than it is wide.  Re can vary from  500 to perhaps 100,000
without significantly changing the efficiency relationship.  Values
of S/W greater than those shown in Table B.I  (up to S/W = 5) affect
             Table B.I Optimum jet configurations.
                                                  1340
'Jet
Round
Rectangular
S/W
1/2
1
T/W
1
1
Re
3000
3000

-------
                                                   21
                                              o
                                              +-(

                                                CO
                                                3
                                                O
                                                O
                                                a
                                                0)
                                                •H
                                                O
                                               <4-4
                                                0)
                                               M-l
                                                O
                                                o
                                                CO
                                                •H
                                                M
                                                tO
                                                (X

                                                §
                                                u
                                                3
                                                00
                                                •H
                         -  O
3  '  AON3IOUJ3

-------
                                                                22
the efficiency only slightly, but values less than those shown
cause efficiency to change quite drastically.  In several reported
tests,13^0 varying the throat length from T/W = 1 to T/W = 10 did
not influence the efficiency curve significantly.

     A number of other investigators have studied both round and
rectangular impactors.  Tables B.2 and B.3-^    catalog the operating
conditions of each study.

     Particles can be impacted into an air void   '     rather than
onto a flat plate as described earlier.  In this case, the impaction
"surface" consists of static air.  The problem of particle blowoff is
eliminated by the use of a void.  Large amounts of particles can be
impacted and collected for subsequent analysis.  Particle sensors can
sample either the large-size fraction (from the void) or the small-
size fraction (non-impacted particles).  Although this technique appears
to offer several significant advantages over flat-plate impaction, few
models have been built and little experimental data has been found.
Bird & Tole Ltd., Bledlow Ridge, High Wycombe, Bucks, England, manu-
factures a void impactor (called a cascade centripeter) based on the
design reported in Ref. 996.

     Several impactors are often placed in series, each succeeding
stage having a smaller critical Stokes number and thus collecting
smaller particles than the preceding stage. The combination of these
impactors is called a cascade impactor.948,995,996,1002 Qne Q£ ^e
most commonly used methods of measuring the distribution of particle
mass as a function of aerodynamic particle size in the 0.2 - 30.0 ym
range is to classify and collect particles with a cascade impactor, and
then weigh the amount of material collected on each impaction plate.
This technique has been used for measuring the size  distribution of
particles in effluent ducts.  The primary features of this technique
are (1) aerodynamic size classification and (2) classification of the
particles inside the effluent duct at the conditions that exist in
the effluent stream.  Equipment designed for stack sampling is
commercially available from:

         2000 Inc. (modified Andersen impactor)
         5899 South State Street
         Salt Lake City, Utah 84107

         Monsanto Company (modified Brink impactor)
         800 N. Liridberg Boulevard
         St. Louis, Missouri 63166

-------
                                                                                                                                                                          23
o
-*
CO
   CD
   rl
   O
   4-1
   O
   cfl
   9.

  •H
   C
   co
  4->
   o
   01
  CO
  0)
  •rl
  C
  0)
  e
  •H
  S-l
  Q)
w


CM

PQ

 01
rH
f>
 CO
H

























4C
*
rH
CO
. J
M
a
0)






1 — 1
o
CO
o
rl
cu
















CD
cu
,J
•n
rj
ft 4J
CO
*
OI
CO


ft
H


^
3 rH
rH 0
PM PL,









^^
2_
U





rH




O
O
c^
CN
1
O
O
-a-
rH
s
1
^o
*3"
•





s~\
f**x
^o
ON
iH
Xi^

cn
h-I
rJ
W
Is


vO
CO
rH
















4-1
CD
3
TJ

rH
Cfl
O
U




|
Cfl



^-^
LO
u





rH




0
O
r^.
CM
1
O
o

rH





LO
ON
rH
V— '
•
rH
Cfl

4-1
OI

CO
PL]
M
£>
"^
Q

rH
f^.
CO
rH


01
d
0)
rH
rC
4-1
0)
a

^1
oi d
4-1 O
CO -H
13 4-1
1 3
rH rH
O 0
(-( CO
0)
O 0)
>-> 3
rH rH
O ,£!









^-x
2,
U





rH




0
0
0
LO
I
0
o
o
CO










/•— N
^o
x^-
CT^
iH
^^/

£3
M
W
JS
o
CO
LO
vO
CO
rH








CO
»TJ
d
3
O
ft
a
0
o

a
3
•rl
d
Ctf
rl










^-^
2,
U





rH






















/-*\
ON
v^-
CT>
rH
V^

£5
M
^
CO
<1
1— 1

vO
CO
rH


































,^—x
LO
U





rH




O
O
LO
vO
1
O
o
oo
rH










/— N
CO
LO
a\
rH
s_ ^

X
o
CJ
rJ
M
•*

vO
CO
rH

















rH
O
rl
CU
O
r^
rH
C3











CO





rH




O
O
LO
•X
vO
rH

O
LO
CM


CO
1
rH

CO
CN
LO
O"s
rH
\~s
O

o
^

^Q

C^O
^
^
M
00
VO
CO
rH


ro
-H
O

O
a *r~i
3 J-4
•H 3
d H-4
0 rH

1 CO
-( -H
O> »-l
O O
r^l ' — 1
rH JZ
O 0









s~^
2,
O





rH




O
o
LO
•t
vO
rH
1
0
LO
CM


CO
1
rH

n
CM
LO
CT»
rH

O
'Z,
O
*2

c£J

CN]
^
3
M

vO
CO
rH






CO
0)
rl
CU
_r{
ft
CO

CU
d
0)
^
^
.4J
CD
r^»
t-H
O
PM











CO



LO
CM
•
|
CO
o


o
0
LO
CO

o
LO

rH
|
CO
•

CN
VO
O-v
rH
^**s
,
rH
Cfl

4-1
0)

*gr
p2
w
H
CO
o

CO
rH

CO
T)
Cfl
0)
•°
co co
OI CO
rl Cfl
0) rH
f. O
ft
CO <^3

cu 01
C r^i
0) T3
r-4
>S 0)
4^ d
CO -H
>> d
rH Cfl
O t-t
PM CD









s~\
^
0





rH




O
O
O
vO
1
O
o
vO


rH





s^.
r^
vO

rH
x_^

t^-
H
Pi
o

^
CD
^
o\
CM




co
0)
M
01
CX
CO


^
rH
0
PM











CO





rH




O
o
o

1
o
o
CN
rH
LO
1
oo

CO





^
O

u


-------



rH
O
CO
O

CU














*
*
rH CO
Cd CU
•H -H
O T)
CU 3
ft 4-1
CO CO
*
cu
ft
H

CO
0)
0) rl
M 0)
3 43
CO ft
CO CO
cu o
rl 0
PH 4J
vS
0

>C a
p
O.


II

cu
pi



S

CO









0)
o
c
cu

cu
14H
cu
Pi

n 1 •
0) O
















rH
o

(U
o

rH
o









CO








rH





O
0
o

CN
CN
1
o
m
vO



CO
1
rH

,-\
cd
CN
m
ON
rH
^x
o
2^
o
15

t-3

M
K
oo
VO
CO
rH










t3
•H
O
cd

o
•H
rl
3

rH
3










m
u







rH






o
o

**o
1
o
0
-vt"
rH




CO







x-^
oo
in

rH
\^

KX]
H
Pi
pq

m
ON





c£J
co
cu ai
4J M
cd cu
rH 43
cd ft
43 CO
4-1
43 CU
PH 13
CU
rH rl

4-1 CO
fj ?*»
rQ iH
•H O










§
U







rH




0
O
0
*
CN
rH
1
O
o .
CM
rH



00
""^
CO
s-*.
m
in
ON
rH
*~*
•
rH
cd

4-1
CU

P^
w
End
rH
M
PH
CO
p**
CO
rH




c£J
co
cu cu
4J rl
cd oj
rH 43
cd ft
43 CO
4J
43 0)
PH (3
CU
rH ^1
>> 4-1
4J CO
3 >>
43 rH
•rl O
P P^





cd



vO
o







iH




O
O
0
*
CM
rH
1
O
o
CM
rH



oo
^^_
CO
Pi
W
U ON
i-j in
M ON
PH rH

*•#

| "J
j_J
W
re
u
H
M
g


ON
CM






CO
cu
rl
cu
43
ft
CO

cu
C
cu
n
4-1
CO
^
rH
0
P-i









CO






in
CM
•
I
CM
O













CO
1
CO
^^
rH

CN
ON
rH
^-^

•
H
cd

4-1
CU

J2
W
H
CO
o
r»*.
CO
rH







-9

CU
3
•H
43 01
r*-»
Ol T3
C
01 01
rH 13
^ *H
43 C
4J cd
CU M
g P





,Q



1
CO



CM
O
•
1
CO
rH
0
o





o
m
**D
|
OO
rH


rH
1

CO
•

Pi
W x-s
,J CO
. i v^O
W  53




43
1
cd


1
x-s
X— '
U






rH








O
00
CO
1
o
oo



o
rH
1
CN



/— \
oo
in
ON
rH
N— '

£7J
W
CO
Pi
W
p
^3
vO
px.
co
•-I











^
cd
J5

cd
43
3
cd
(3

cd
0









1
/— N
N^X
U






rH








O
00
CO
1
o
00



o
rH
1
CM


^
**D
^G
ON
rH
^^

£^
W
CO
Pi
w



f^x
CO
rH



c3

CU CO
>i 0)
T3 M
CU
CU 43
3 ft
iH CO
43
CU
0) C
13 CU
01 >-i
rH | '
>-, CO
43 >>
4-1 rH
CU O
g PH









g
*>~s
U






rH








O
oo
CO
I
0
oo



o
rH
1
CM


^^
[•^
• *x>
rH ON
Cd rH
^^
4-1
CU

re
u
co
w


m
rH


*>
r-
M
43


«\
cu
T)
•H
M
O
iH
43
O

£j
3
•H
T)
O
CO




4

1
>N cu
-113 C
4-1 -H
CU C
g cd
^
<-3 P

01 CO
13 3
•H rH
C p.
cd
M CU
P 3
rH
- 43
CO
O CU
& C
CO CU
U rH




3

cd


J
*>— '
U





r

L/


1
^~s
O





H

1
O















O
rH
1
CN

N
^
re
re
o
p
pq

*&

P^
w
«
$
fi±
•













^3-
1
in
CO
•


oo
vO
ON
rH
^^s









m
oo
vD










CO
cu

cu
43
ft
CO

cu
pj
cu
^H
4J
CO
^
rH
O
PH





43



CO








iH






O
0
o
f^
1
0
0
VO

o
rH
1
oo
^^.
CO


Pi ON
O *&
fa ON
^ ct

H
CO

t^Q

Pi
w
U
g
OO
r^.
co
-rf










CO
cu
rH
a
•H
4J
£_)
cd
ft

cu
^
p





43



i
^w'
CJ






rH





O
0
ON
iH
1
O
0
m
rH


in
•
rH
1
rH



-------
                                                                 25
Cascade impactors can also classify particles within effluent
streams which have been pre-conditioned, e.g., by dilution and/or
cooling.  Commercial models which might prove useful for conditioned
effluent streams are manufactured or marketed by:

     Aerostatics Instrumentation & Research (modified Andersen impactor)
     1081 East 2200 North
     Logan, Utah 84321

     Environmental Research Corporation (Lundgren impactor)
     3725 North Dunlap Street
     St. Paul, Minnesota 55112

     Monsanto Company (Brink impactor)
     800 N. Lindberg Boulevard
     St. Louis, Missouri 63166

     Scientific Advances, Inc. (Battelle impactor)
     1400 Holly Avenue
     Columbus, Ohio 43212

     2000 Inc. (Andersen impactor)
     5899 South State Street
     Salt Lake City, Utah 84107

     Union Industrial Equipment Corporation (Unico impactor)
     150 Cove Street
     Fall River, Massachusetts 02720

     Willson Products Division, ESB Incorporated (Cassella impactor)
     2nd & Washington Streets
     Reading, Pennsylvania 19603

None of these commercial cascade impactors has an automated deposit
sensing or particle concentration sensing method.  All are intended
to collect a classified sample for subsequent analysis.

     The operable size range for cascade impactors in controlled
laboratory tests is generally, at best, from about 0.2 to 30 um
diameter, assuming particle density of 1.0 to 2.0 grams per cubic
centimeter and nearly spherical shape.  The lower limit is set by
the difficulties imposed by:

     a) very small nozzles which are difficult to drill and tend
        to plug easily,

     b) high jet velocities resulting in some particle blowoff,

-------
                                                                     26
     c) reentrainment of deposited particles limits the
        amount of sample which can be collected, and

     d) the small weight of small particles cannot be
        weighed in the presense of the large tare weight
        of the impaction plate.

The upper size limit can be larger than 30 ym depending on the type
of particle and the stickiness of the impaction surface.  However,
such problems as particle deposition in the entrance regions of the
impactor, blowoff of previously-deposited particles, and bounce of
impacting particles, usually impose a practical upper size limit.

     The lower size limit can be extended by low pressure impaction
for some applications.  A major problem is the vaporization of
particulate materials by the low pressure.  Further testing and
development is necessary to determine whether low pressure impaction
is practical for stack effluents.

     A special case of low pressure impaction, often called an aerosol
beam, uses a sonic nozzle with a very low downstream pressure (<0.01
atmsophere).  After particles pass through the sudden expansion region
just downstream of the nozzle, they are affected very much less by
fluid drag forces than in the atmospheric case.  One study has looked
at the pattern of particulate deposit on a target placed as several
distances downstream of the nozzle.1273,1361  Another study looked at
stop distances of the particles in the aerosol beam.1362  Much work
remains to be done to understand aerosol beams.  It is not clear
how the principle can be utilized for particle size classification.
One possible application may be to use the aerosol beam technique to
accelerate particles to a constant, known velocity and then use a
sensitive momentum transducer (e.g., see Ref. 252) to measure the mass
of individual particles.  Much work remains to determine if this
technique, or any other aerosol beam technique, is practical for sizing
micron-sized particles.

     Impactors, especially cascade void impactors, are definitely prime
candidates for size classifying particles in stack effluents over the
size range from about 0.2 to 30 um.  Features include aerodynamic size
classification and the ability to classify particles at nearly any
environment condition including in-stack conditions.  Cascade impactors
have been investigated extensively both theoretically and especially in
the laboratory.  They have been used to classify stack effluent particles
with mixed results.  Impactors appear to adapt easily to several particle
sensing techniques.

-------
                                                                 27
iv.  Centrifugal Spectrometry

         A fourth method of using aerodynamic drag to  classify aerosol
    particles by size is centrifugal spectrometery.1333  (PP-  123-126),
    123,  290, 354, 369,  546, 505, 649,  818,  925,  1119, 1120,  1341,  1342,
    1 O / Q
         This technique  is equivalent to gravitational sedimentation
    except that centrifugal acceleration speeds  the process allowing its
    application to small particles in the 0.03 -  3.6 ym  equivalent  aero-
    dynamic diameter range.

                 649
         Figure 7    shows the principle of  operation  of  2 centrifugal
    spectrometer designs.   In the first design,  aerosol  passes through
    a spiral-shaped channel which rotates about  its axis.  Centrifugal
    acceleration pushes  particles radially toward the  outer wall  of the
    channel while aerodynamic drag force (see Equation B.I) resists the
    radial motion of the particles.   The effective radius of  the  spiral
    channel increases as the aerosol passes  through the  spectrometer.
    Large, heavy particles deposit near the  inlet where  the centrifugal
    acceleration is low, while small particles deposit near the exit
    under the influence  of high centrifugal  acceleration.  The spectro-
    meter is normally precalibrated  using aerosol particles of known
    size  and noting the  position of  their deposit on the  outer wall of
    the  channel.   The second design utilizes a  sheath of clean air
    between the aerosol  stream and the  deposit surface.   The  clean  air
    sheath improves the  resolution of the spectrometer by forcing all
    particles to travel  nearly the same radial distance  before being
    deposited on the outer wall of the  channel.
                                                  AEROSOL INLET

                                          FLOW NOZZLE
     (1)  Spiral  design
(2)  Sheath air  design
        Figure  7. Two  centrifugal  spectrometer designs.649'1119

-------
                                                                    28
   This results in considerably sharper size classification than the
   first design, but requires lower aerosol sampling rates.649,1119
   The models which have been built used various types of driving
   motors and bearing systems to allow high centrifugal speeds without
   vibration.  A commercial version of Stober design is marketed by:

        Ivan Sorvall Inc.
        (street address unknown)
        Norwalk, Conn. 06856

        It is difficult to imagine an automatic sensing technique for
   use with centrifugal spectrometry.  Additional problems for application
   to stack effluents is the limited size range (essentially submicron)
   and the small amount of material which can be collected.  This technique
   has seen considerable application in research studies of laboratory-
   generated and atmospheric aerosols, and may be useful for some research
   on stack effluents.  However, its application to routine, automatic
   stack effluent monitoring is doubtful.
v. Cyclone Separation
                                                                   aerosol
     A fifth method of using aerodynamic drag force to classify ae
particles by size is the cyclone separator.1333(pp.126-135), 1334,
1335,1216,675.  „.      0   , „ im  ,     i     •   •  -,   *      ^-    c
                Figures 8 and 9 ±J->-> show the principle of operation of
a cyclone.  Aerosol enters the rectangular inlet tube A and passes into
the cylindrical part of the cyclone where it acquires a spiral motion.
Air spirals downward along the outer wall of the cylinder, throwing
large particles against the wall where they can slide to the bottom of
the cyclone.  When the air reaches the bottom of the cyclone, it ascends
an inner  spiral to exhaust through the upper axial tube C.  Small
particles are carried out by the exhaust air stream.   Large particles
can be collected for analysis from the bottom of the cyclone.
 Figure  8.  Cyclone.
                  1333
                                      Figure 9. Motion of gas in a
                                                        -I OTQ
                                                cyclone.

-------
                                                               29
     Cyclones are used  industrially as dust collectors on large
effluent streams.  They are used in aerosol sampling as "respirable"
size range separators.   In this application, a small cyclone  separates
the large particles from the air stream, presumably simulating the
upper passages of the human respiratory system.  The aerosol  which gets
through the cyclone,  containing only small particles, then simulates
the aerosol which penetrates deeply into the respiratory system.  These
small particles are generally collected for analysis on a second stage
consisting of a high-efficiency filter.  Figure 10 shows the  "respirable
dust" curves which the  cyclone attempts to simulate.1052  Qne Of the
curves was first developed at a conference at Los Alamos Scientific
Laboratory of the U.S.  Atomic Energy Commissioners  and was  modified
and adopted as a standard by the American Conference of Governmental
Industrial Hygienists in 1968.1334  ^he other curve on Figure 10 was
defined earlier by the  British Medical Research Council.1052  The
aerodynamic diameter  in Figure 10 is based on a particle density of 1.0
gram per cubic centimeter.   The vertical axis is percent of  respirable
dust by weight of the particles.
                                                      I
                             • LASL  CONFERENCE CURVE
                             o ACGIH MODIFICATION
                             A MRC  CURVE
                  345678
                   AERODYNAMIC  DIAMETER  (/im)
10
          Figure 10.   "Respirable dust" curves.
                                              1052

-------
                                                           30
     The relationship between the size of collected particles and
 the dimensions and operating parameters of a cyclone can be approxi-
 mated by:1333 (p.  129)
     DP,min= 3  Ws->                '                   (B'7)
      v           op

where:

     D    ,   =  diameter of the smallest collected particle,
      p,min                                       r

          n  =  viscosity of air,

         R-  =  radius of the outer cyclinder  (see Figure 8),

         v   =  average inlet duct velocity,

         Y   =  particle density, and

          S  =  number of turns of the outer spiral in the
                cylindrical part of the cyclone.

The difficulty  in using Equation B.7 is in evaluating S.  In most
practical cyclones, 1 < S  < 3.  Equation B.7 shows that for any
given cyclone,  an increase in v  results in the collection of
smaller particles.  Practical cyclones can be designed such that
D      is from  about 0.5 to 20 urn.
 P ,min
     It is not  easy to analyze cyclones mathematically.  In fact,
most cyclones have been designed by use of experimental data.
Factors which influence the cyclone performance are the level of
turbulence in the cylinder, the relative size of the inlet and
exhaust (and thus the inner and outer spirals), redispersion of
particles which have struck the wall, the shape of the particles,
pulsations in air velocity, the dust concentration level, and the
roughness of the cylinder walls.  The flow resistance through the
cyclone increases roughly as v  .  Although some cyclones have been
operated successfully in the inverted position, it is important
to define the orientation of the cylones.  Cyclones, should always
be operated with the large particles moving downward unless its
operation in an alternative position is carefully investigated.
Reference 675 reports the results of an experimental evaluation
of cyclone operating conditions.

     A set of six identical cylones, each operated at a different
flow rate, has  been used recently to measure  atmospheric particle    1354
size distribution in 6 size categories ranging from about 1-12 ym.
Walter C.  McCrone Associates,  Chicago,  is currently investigating the
applicability of cascade cyclones to size distribution measurements in
stacks under contract to the Environmental Protection Agency.

-------
                                                                  31


         Cyclones are a definite candidate for effluent particle size
    classification because of the operable size range, the ease of
    continuous operation, and because the size classification is aero-
    dynamic.  The large particles can either be collected continuously
    for analysis or can be eliminated continuously.  As will be seen
    later, cyclones adapt easily to the inlet of particle sensing instru-
    ments, making automatic sensing of "respirable" range particles a
    relatively simple task.  A single cyclone has a limited operable size
    range.  Cyclones have already found application to separating the non-
    respirable particle fraction from an air stream so that the respirable
    fraction can be analyzed.

vi. Centrifugal Elutriation

         Another particle size classifying method usind aerodynamic drag
    force is centrifugal elutriation.  This method, usually known as the
    Bahco method, is often used to classify fly ash effluent particles by
    size after they have been collected from the effluent stream. ^»^'
                  1332
         Figure 11     shows the principle of operation of this method as
    sold commercially for powder sizing.  The entire assembly except compo-
    nents 1-6 which make up the powder feeder assembly, rotates about a
    vertical axis.  Clean air enters past the throttle nut 13, past the
    symmetrical disks 11, through the sifting chamber 10, radially toward
    the axis and then outward through the fan vanes 8.  Particles enter the
    device through the feed nozzle 6 and enter the air stream through the
    rotary duct opening 9.  Large particles are carried outward by centri-
    fugal force and are collected in the catch basin 12.  Small particles are
    carried with the air stream and emerge through the fan vanes 8 where
    they impact on the inner wall of the rotor casting 7.  Thus, the classi-
    fication results in 2 size fractions, one with particles larger than the
    cutoff size and one with smaller particles.

         The size cutoff can be varied over a range of sizes by changing
    the throttle nut 13 position by means of throttle spacer 14.  This
    changes the amount of air passing through the device and, in effect,
    changes the aerodynamic drag force acting on the particles.  To measure
    the size distribution, the operator begins with the smallest size cutoff,
    runs an analysis, weighs the large fraction which collects in the catch
    basis 12, and then reruns the large size fraction with a somewhat larger
    size cutoff setting.  This is repeated as many times as necessary.

         The ranges of sizes on the commercial unit is from about 50 ym down
    to 1 or 2 ym.  The lower limit is set by the impaction efficiency of
    particles onto the rotor casting 8.  Particles below a certain size will
    pass out with the air stream and not be collected with the fine fraction.
    With appropriate design, the fraction which is not collected on the rotor
    casting 8 could be collected on a filter, resulting in a measurable
    fraction of small particles.  However, the difficulty in collecting small
    particles is a serious limitation for particles below perhaps 5 ]jm or for
    distributions which contain significant mass fractions below this size.

-------
                                                                   32
Figure  11. Cross-sectional diagram - 1. hopper, 2. spring
           plate, 3.brush,  4. orifice tube, 5. vibrator,
           6. feed nozzle, 7. rotor casting, 8. fan wheel
           vanes, 9. rotary duct opening, 10. sifting chamber,
           11. symmetrical discs, 12. catch basin, 13. throttle
           nut, 14. throttle spacer, 15. motor.1332
     Another factor affecting the classification of any particles which
must be resuspended from a powder state is the agglomeration and/or
fractionation of particles.  This is especially important for smaller
particles in the micron size range or smaller.  Small particles often
adhere to larger particles and are classified with the bigger  particles.
Thus, it may be desirable to investigate the operation of the centrifugal
elutriator on the suspended effluent particles as they emerge from the
stack.  This may  be possible with a  redesigned  system.

     The quality of the classification also depends on the type of air
flow through the classifying regions and how the particles are introduced
into this region.   The turbulence level, boundary layer formation, and
the geometry of the particle entrance region are all important.  Complete
analysis of the existing units is beyond the scope of this report.  Such
analysis could certainly result in improved performance,but considerable
performance testing would be required to evaluate the operation of any
such device.

     This technique has been accepted by the ASME for measuring the size
of collected fly ash powder samples.239  j^g gahco apparatus was used to
obtain much of the size information on fly ash which is reported in the
literature, including much of the data reported in Volumes I and II of
this report.

-------
                                                                      33
           The commercial version of the centrifugal elutriator for
      powders is manufactured by:

           Harry W. Dietert Company
           9330 Roselawn Avenue
           Detroit, Michigan 48204

      This company holds the American license from the Swedish Company:
      AB Bahco.

           The centrifugal elutriation technique is a candidate for effluent
      particle classification.  Although the technique is commonly used for
      sizing fly ash particles, the present technique requires the collection
      of a fly ash sample and redispersing the particles for the measurement.
      This process is subject to large errors when submicron or micron-sized
      particles are present in significant amounts.  The lower size limit of
      the present apparatus is in the 2 - 5 ym range.  It would seem that
      this technique could operate with the feeding of airborne particles
      directly into the apparatus.  Such a technique would probably require
     . considerable redesign, but could result in a reasonable classifier for
      the 1 - 60 Mm (approximate) range.  This technique results in aerodynamic
      equivalent diameter classification with its many advantages.  This
      technique appears to offer no significant advantage over cyclones and
      impactors and is generally more expensive.

b. Electrostatic Classification

        An aerosol particle which carries an electrical charge can be acted
   upon by an electric field.  The force acting on such a particle is expressed;
             F  = n  e E                                         (B.8)
              e    p
   where:
             F  = force acting on the particle because of its
                  electrostatic charge,

             n  = number of elementary charges attached to the
                  particle,

                                                     -19
              e = elementary unit of charge, 1.6 x 10    coul/elementary
                  charge,  and

              E = electric field intensity.

   This  force causes  the particle to move through the gas in a direction
   determined by the  polarity of the field and the polarity of the charge
   on the particle.   The force resisting this motion is aerodynamic drag
   force,  expressed in Equation (B.I).   Equating these two forces, one can
   define the final speed  attained by the particle:

                  n   e E

                        '                                         (B'9>

-------
                                                                    34
Another term often  used  in electrostatic theory is the electrical
mobility of a particle Z  defined:
                  n  e
                  _P	
    V
p ~ E    3irnD C
                                                               (B.10)
Figure 12 shows  the mobility of singly-charged particles in standard
air as a function of particle diameter D .   The mobility of multiply-
charged particles is simply the value shBwn in Figure 12 multiplied by
the number of elementary  charges n .   The electrical mobility of a par-
ticle is the velocity  of  that particle when acted upon by an electric
field of unit intensity.
              i-o
               0-001
                        0-01        0-1
                             Particle diameter,
                                            1-0
           Figure 12.  Electric mobility of singly-charged
                      particles at NTP.1211

-------
                                                                  35
      If  aerosol particles  can be electrostatically charged such
 that  n  is a reproducible  function of particle size D ,  then a
 mobility classification is also a particle size classification.
 Several  mobility classifiers  are shown in  Figure 13.1211  xhe
 mobility classifier in Figure 13.c is particularly suited  to this
 application.   Unipolar-charged particles enter the-classifier as
 an  annular ring surrounding a core of clean air.  A voltage with
 polarity opposite the  polarity of the particles is applied to the
 central  rod.   The outer tube  is electrically grounded.   The charged
 particles are attracted toward the central rod with a radial velocity
 expressed in Equation  (B.9).   Particles with high mobility deposit
 near  the top of the rod while lower mobility particles  travel further
 down  the rod before being  deposited.   If the geometry and  electric field
 intensity are chosen properly, particles with mobility  less than a given
 value (mobility cutoff) will  not be deposited on the central rod,  but
 will  pass out the bottom of the classifier where aerosol sensors can
 detect the concentration.   The mobility cutoff of a given  geometrical
 design can be varied by changing the electric field intensity or the
 aerosol  and clean air  flow rates.   Thus, in operation,  one would measure
 the. concentration at one mobility cutoff (corresponding  to a pre-
 determined particle size cutoff)  and then  at a different mobility  cutoff
 (corresponding to a different particle size cutoff).  The  concentration
 of  particles  between the two  sizes is the  difference between the two con-
 centration measurements.

      Thus,  an instrument based on this principle measures  the number con-
 centration within several  size ranges.  The size ranges  are determined
 by  the relationship shown  in  Equation (B.9).   The velocity v from  Equation
 (B.9)  determines  where the particle will land on the central rod.
                                       Aerosol
                         (o)
                                            Clean air
(b)
                                                        -Aerosol
        Figure  13. Aerosol mobility classifiers.
                                                  1211
     Two factors limit the particle size resolution of  this  type  of
classifier.  The first is the difficulty in placing an  equal number  of
charges on all particles of a given size.  The second is  the problem of
introducing particles at the top of the mobility analyzer in a very  thin
annular ring so that all particles travel an equal radial distance through
identical flow conditions to reach the central rod.

-------
                                                                 36
     There are many ways to place an electrostatic charge on aerosol
particles.  Most of these are reviewed in Reference 1211 and will not
be discussed here.  The most successful for classifying 0.005 - 0.6 ym
particles is shown in Figure 14 and is called a sonic jet unipolar charger.
The details are discussed in Reference 68.  The method of charging is
called diffusion charging and is discussed in Reference 1211 and 56.  The
mobility of particles charged by this method allows good size classification
from 0.01 - 0.2 urn.  The classification below 0.01 ym deteriorates because
only a very small fraction of such particles can be made to accept a single
charge.  Above 0.2 ym, the mobility versus particle size relationship slowly
flattens until 2.0 ym particles have about the same mobility as 10.0 ym
particles.  The electrical charge n  placed on equally-sized particles is
quite narrow, resulting in relatively sharp size classification in the
0.01 - 0.2 ym range.  For example, completely monodisperse particles
(geometric standard deviation of 1.0) have a  measured geometric standard
deviation of 1.15 with an existing commercial instrument.1344
              Figure  14.  Sonic  jet  diffusion  charger.
                                                     68
     Another charging technique, field charging, could probably be used
to classify particles from 1.0 ym to 10 or 20 ym.  However, this technique
has not been developed as thoroughly and other classifying techniques
appear more promising for the larger-sized particles.

-------
                                                                     37
        The second factor limiting particle size resolution is the introd-
   uction of aerosol in as thin an annular ring as practically possible.
   This assures that all charged particles of equal mobility travel radially
   through identical conditions and their deposit is not spread out on the
   collection surface.  The need for a reasonable through-put requires an
   annular ring of practical width.  In practice, the ring width can be kept
   down to about 10% of the radial distance which the particle must travel.

        This technique has been used to measure particle size in a number of
   studies of atmospheric and artifical aerosols.27,40,42,61,68,1276,1285,
   1344,1390.
        An aerosol charge-mobility classifier instrument is manufactured by:

             Thermo-Systerns Inc.
             2500 Cleveland Ave. N.
             St. Paul, Minnesota 55113.

   The instrument, the Whitby Aerosol Analyzer, can be used in several modes:
   (a) as an automatic concentration versus size detector, (b) as a size
   classifier, and (c) as a mobility classifier for an externally-charged
   aerosol.  The present commercial model is large and would not lend itself
   readily to stack monitoring in its present form.  The maximum aerosol temp-
   erature must be kept below about 100 F on the commercial unit because of
   contruction materials.

        Electrostatic classification is a definite candidate for classifying
   0.005 - 0.6 ym effluent particles.  Recent  improvements in the charger
   design and the mobility operating conditions make the size classification
   very good, especially in the 0.01 - 0.2 ym range.1344  with proper design,
   the technique could probably be made to operate at either in-stack or out-
   of-stack conditions.  Although this technique does not lend itself to stack
   applications in its present state of development and requires some develop-
   ment for such application, it remains the most practical classifying tech-
   nique in its size range.   Although considerable development could probably
   result in classification  of particles above 1.0 ym, the availability of
   other techniques makes such development unimportant.

c. Sieve Classification

        One of the best known methods of classifying powder particles into
   size fractions is sieving.  This is usually done with a set of screens
   mounted in matched holders and stacked such that particles encounter
   progressively smaller openings as they fall through the screens.  The
   particles which remain on each screen are then weighed to obtain the
   size distribution.

        One of the greatest  problems with sieves is getting all particles
   which should pass a given sieve to actually fall through and keeping
   those which should not from passing through.   The method most often
   used to make particles fall through is to agitate the particles

-------
                                                                   38
violently enough to bounce  each particle into many orientations
and to break up loose agglomerates.  However, the agitation must
not be violent enough to break up the primary particles.  These
techniques have extended the lower size limit of sieving techniques
down to about 10 ym diameter.

     One such technique is acoustical agitation.  An acoustical speaker,
mounted either below or above the stack of screens, causes the particles
to bounce violently.  Such a unit is marketed commercially by:

            Allen-Bradley Company
            (street address unknown)
            Milwaukee, Wisconsin

The lower size limit of the screens on this model is about 10 ym.  The
largest screen size available has about 600 ym holes.

     A second method of agitating the particles is to periodically blast
air upward through the screen.  Such a unit is manufactured by:

            Alpine American Corporation
            3 Michigan Drive
            Natick, Massachusetts 01762

The smallest advertised hole size of these screens is 15 ym.  The largest
is greater than 100 ym.

     Another method of agitating particles on the screens is to wash the
stack of screens in a liquid such as water during the sieving process.
This helps achieve better classification, but care must be exercised to
keep from losing or dissolving particles in the process.

     One manufacturer of ordinary matched sieve sets is:

            W. S. Tyler Company
            (street address unknown)
            Mentor, Ohio 44060

The holes in the sieves range from large sizes down to 44 ym.

     An electromagnetic sieve shaker is manufactured by:

            Geoscience Instruments Corporation
            435 East 3rd Street
            Mount Vernon, New York  10553.

-------
                                                                        39
        Although sieving could be adapted to airborne particles,  little
   application is expected for classifying stack effluents because it does
   not operate on small enough particles and the technique does not lend
   itself to automatic, continuous sensing of particle size.   Some research
   application may be found in the manual sizing of large particles collected
   by effluent control devices.  The agglomeration, deagglomeration, and
   fractionation of powder particles is a major problem as is screen binding.

d. Filtration

        The efficiency of a filter as a function of particle  size generally
   has a shape similar to Figure 15.  If a particle distribution is located
   primarily above or below the minimum point, this feature can be theoreti-
   cally used to size classify the aerosol.   Filtration theory has been
   reviewed by many investigators, including References 1211  and 1333.
                           0.1
          1.0
Particle Diameter, ym
10.0
100
          Figure 15.  Typical filter efficiency curve.   The exact  location
                     of the minimum shifts both horizontally and  vertically
                     with changes  in filter media,  face velocity, filter
                     loading,  and  other factors.

        If several  filters are used, each with an efficiency curve falling
   in  a  different size  range,  the  filters can  be  placed in series similar
   to  a  cascade impactor.   The first filter will  ideally collect  everything
   above a given cutoff size (50%  point on the efficiency curve).  Each
   succeeding  filter  will collect  a smaller size  fraction.

        An automated  system may consist of cascaded filters with  sensors
   removing a  small part of the material passing  through each filter.   The
   sensor then measures the concentration of material  below each  filter's
   cutoff size.

-------
                                                                       40
        This technique has several important problems.  Filters become
   plugged, changing the efficiency curve and thus changing the size
   cutoff.  The efficiency curve for most filters is not very steep,
   making the size cutoff rather poor.  The practical range of this
   technique is from about 1.0 to 5 urn, a very narrow range.  Size
   classification in the range below 0.1 microns has not proven practical.
   Since several other techniques appear to be considerably more promising
   within the same size ranges, filtration classification will probably
   not be used for stack effluents except for specific research applications.

e. Brownian Diffusion

        For particles smaller than 1.0 pm, Brownian diffusion becomes an
   important factor governing the motion of an individual particle.  Brownian
   motion of a particle is caused by the random bombardment of the particle
   by gas molecules.  A single molecular impact does not change the direction
   and speed of an aerosol particle appreciably, but random collisions with
   a large number of molecules causes motion such as that shown in Figure 16.
   Fuchsl333 and Daviesl^H discuss Brownian diffusion in considerable detail
   and much of the following discussion has been derived from them.
            1333
         Figure 16.   Trajectories of gas molecules (a)  and particles
                     undergoing Brownian motion (b).1333
        The diffusion coefficient of a particle, a quantity characteriz-
   ing the intensity of Brownian motion, can be expressed:
             D = kTB
(B.ll)

-------
                                                                 41
where:

       D = diffusion coefficient,

       k = Boltzmann's constant,

       T = absolute temperature,

              £
       B = -T—-— = particle mobility,
               P
       C = Cunningham's correction factor,

       n = viscosity of the gas, and

      D  = particle diameter.
       P

         1211
Table B.4     lists diffusion coefficients at normal atmospheric
conditions for particles from molecular size to 100 \im radius.  Notice
that the diffusion coefficient is much larger  for small particles than
for large ones.

     As an aerosol flows through a tube, the Brownian motion of some
particles will cause them to cross fluid streamlines and hit the tube
wall.  For particles in the size range where Brownian motion is important
(below 0.1 pro), they will adhere strongly to the tube and not be reentrained.
Thus, they are lost from the aerosol stream.

     The rate of deposition of particles on the tube wall is a function of
the particle size; the concentration of small particles decreases more
rapidly than that of large particles.  Considering only the deposition of
particles caused by Brownian motion, the concentration of aerosol leaving
a long, circular tube is    :


 ~= 0.819exp(-14.63A)+0.0976 exp(-89.22A)+0.019 exp(-212A)      (
 0*
  o

where:

       C = average exit number concentration,

      C  = average inlet number concentration,
       x = length of tube,

       d = tube diameter, and

       V = average gas velocity.

-------
                                                                     42
                      Table B.4

        The approximate molecular diffusion coefficients
        of small particles in air at 760 mm and 20 C.1211
Particle radius
(ym)
-4
10 (hydrogen molecule)
5 x 10"4
io"3
5 x 10~3
io-2
2 x 10"2
5 x 10"2
io-1
2 x 10"1
5 x 10"1
1
2
5
10
20
50
100
Coefficient of diffusion, D
(cm /sec)
7 x 10"1
5-2 x 10"2
1-3 x 10~2
5-3 x 10~4
1-4 x 10"
3-6 x 10~
6-8 x 10"6
2-2 x 10~6
8-4 x 10"7
2-76 x 10"7
1-3 x 10~7
6-16 x 10"8
2-4 x 10~8
1-2 x 10"
5-9 x 10"9
2-4 x 10~9
1-2 x 10"9
	                         1211
C/C  is shown in Figure 17     as a function of A.   Equation (B.12)  assumed
(a)°Re <2000 (laminar flow), (b) the tube is long enough so that end effects
are negligible, (c) the velocity profile is parabolic,  (d)  the aerosol con-
centration is small enough so that particle interactions are negligible,  and
(e) other effects such as electrostatic forces do not affect Brownian motion.

      Equation (B.12) shows that
           £-= F (A)
           LI
            o
where, using the relationship shown in Equation (B.ll):
                                                              (B.13)
           A =
               dv
                           (ir} Hh-
                            DP   d2v
The first term in Equation (B.14) is a function of the gas, the second term
is a function of particle size, and the third term is related to the system
geometry and flow rate.  Thus, if  a monodisperse aerosol passes through
a given tube at known conditions, a measurement of C/C  determines particle
size.

-------
                                                                     43
         0-5 —
           0-001
                    0-005  0-01
                                   0-05  0-1
                                                 0-5    I
   Figure 17. Diffusion to the absorbing wall of a long tube through
              which the flow is viscous and follows Poiseuille's
              Law.1211
     For a polydisperse aerosol, several tubes can be used, each with
different geometry and/or flow rajtes.  Each tube has associated with it
a certain particle size of which C/C  = 0.50.  This particle size is
called the cutoff size with most particles larger than this size passing
through and most particles smaller being deposited.  In this case, the
fraction of the size distribution located between two_given particle
sizes is simply the difference between the fractions C/C  measured at
each condition.
     The size range where this technique is useful is from 0.001 to 0.05 ym
diameter.  Above 0.05 ym, Brownian motion becomes insignificant compared
to other motions.  Below 0.00_1 ym is the molecular regime.  The detector
normally used for measuring C/C  is a nuclei counter  (see Volume I of
this report).  Since the size cutoff is not very sharp  (see Figure 17),
size measurements made in this way are not very accurate.  However, this
is the best known technique of size classifying 0.001 - 0.01 ym particles.

     Husar     suggests a somewhat different approach for using diffusion
for sizing particles.  It also describes a very recent design of a sizing
instrument and measurements made by the instrument.  Tables of design
criteria will prove useful for the design of other such instruments.

     Several problems limit the use of diffusion techniques for classifying
effluent particles.  Large particles above about 1.0 ym should be eliminated
before the smaller particles pass through the diffusion tubes.  If the temp-
erature of the aerosol stream and the tubes are not in equilibrium, thermal
forces and condensation may confuse the deposition.  Other deposition
mechanisms may also confuse the deposit, including gravity and even low levels
of turbulence.

-------
                                                                            44

             The diffusion technique may see application to stacks only for
        classifying 0.001 - 0.01 ym particles.  This will probably be only in
        research application, although this technique is the only known size
        classification method in the 0.001 - 0.005 pm range.

     f. Other Forces

             Many other forces can act on small particles.  Some of these include
        thermophoresis, diffusiophoresis, photophoresis (a form of radiometric
        force), and magnetic forces (see, for example, Reference 1333 and 1211).
        Although any or all of these forces may have an important effect on the
        motion of an aerosol particle in a given system, no way is known to classify
        particles by size using these phenomena in practical situations.

2. Particle Sensing Techniques
     Volume I and II of this report discuss all known techniques for the sensing of
aerosol concentrations in practical situations.  Although Volumes I and II are
oriented toward measurement of the mass of aerosol particles, all other comments are
appropriate for the present application.  The reader is referred to Volume I for a
brief, but comprehensive, survey of all sensing techniques; and to Volume II for
detailed discussions of the viable techniques for effluent particle sensing.

3. Combinations of Classifiers and Sensors
     Many factors must enter into the choice of a size classifier-concentration
sensor combination for any particle sizing application.  A number of these factors
have been discussed in the introduction of this volume.  After studying his appli-
cation carefully, the investigator must decide what particle size range is of
primary interest, what type of size classification he prefers, and which parameter
of the particles he wishes to sense.  The remainder of this section presents those
classifier-sensor systems which appear most practical for automatic or semi-auto-
matic measurement of the concentration of particles within specified particle size
ranges in effluent streams.

     Nearly any size classifier can be used with nearly any concentration sensor
to measure the size distribution of aerosols.  It can be seen from the list of
particle size classifiers (at least 10 separate methods) and the longer list of
particle concentration sensors (at least 31 separate methods) that several hundred
possible combinations exist.  Not all of these combinations are technically
feasible and most of them are not practical for effluent particle sizing.  Table B.5
lists 27 combinations which appear to be technically feasible and which also appear
to be the most practical for application to effluents.  Not all of the combinations
in Table B.5 are equally applicable, however, and most combinations have not yet
been developed.  The remainder of this discussion will point out strengths and
weaknesses of the most promising of these combinations.

     None of the combinations in Table B.5 covers the entire range of particle
sizes found in effluent ducts (from about 0.001 um up to at least 100 um diameter).
Thus, one must choose a smaller particle size range of primary interest, choose a
system which covers that range, and determine the limitations of the chosen system.

-------
                                                                                                                                                                         45
  CD   O
  !-i
  O   >,
  cn  4-1
  C  -H
  0)   CD
  w   C
      0)
 T3  13
  C
  tO   tx)

  CO  ,C
  (-1  4-J
  0)  -H
 •H   IS
 4-1
 •H   QJ
  CO  rH
  CO   O
  CO  -H
 iH  4J
  O   V4
      CO
  CU   CX
  N
 •H  M
  CO   CO
      U
 4-1  -H
  O   !-i
      01
  co  ,c
  C   ex
  o   co
 •H
 4-1   co
  03
  C  4-i
 -H   O
 J3

  O   CO
  a   a
 cfl   O
 a  -H
 •H   B
 4-J
 a   c
 cfl  -H
 S-i
 CX   -
      J-l
 4-1   0)
 O  4J
      0)
 co   E
 0)   cfl
 60 •
 a  -a   B
 tfl      -H
 !-l   QJ  4-J
     XI   C
 a)  4-1   cu
 N       o
 •H   01
 cn  H   a
     cO  -H
 a>      xi
 rH  CO   3
 X>  >-i   U
 CO  01 -~
 !-i  X)  0
 C1J  0  «9
 CX  3  J-J
 O  C  00

 0)  H  O
X:  rH   •
 H  <  rH
0)
pq

 0)
 CO
H










UJ
Z>
g
2
I
O
UJ
h-
CD
Z
CO
Z
UJ
CO

Z
o
1-
<
tr
h-
UJ
Z
8












GO'O - lOO'O
a I "^ 1 ft. 1 /^ /^/N 1 ""1 ^ ^"\ rt fc 1
d31NOOD l3TOnN



OOI-SOO'O
NOIiVnN311V
QNV 3dnidVO
NO! OllVlSOdJ.0313





<
o

K
a.








cn
8
2



OS - E'O
l\/llN310d
9NI1IOS

OS- Z'O

Adl3W010Hd

001 - Z'O
NOISSIWSNVdl
1H9I1

02 - lO'O
nviSAdo zidvno
OldlD3"l30Z3ld



OOI-IO'O
NOIlvnN311V
NOIlVIQVd V138













1




0
10
1
1
CM
d

o
ro

CM
d
O
ro
I
1
i
CM
d
o
ro
1
CM
d
0
CM
i
*
CM
d


o
ro
i
*
CM
d




£°
Oro
5;'
|S




I




O
ro
|
1
tn
d

o
ro
1
•n
o
0
ro
1
in
d
0
ro
1
in
d
o
CM
1
*
«Q
O


O
ro
*
in
d




£°
o1?
u«n.
>-o
o




i




o
o
1
q

o
in
1
o
—
o
in

q
—
o
o
1
q
—
0
CM
1
*
O
^_


O
g
i
*
q
—
_i
<
Z Q;
o§
H«-Q
<<§
GRAVIT
ELUTR
1.0- li




I




O
m
1
o
•

0
m
j
q

o
to

q
—
o
m
1
q

0
CM
i
JK
^
q



8
1
*
q
—
-JZ
-A
2
UjO
_id
UJ

in
o
d
I
1
_
o
o
d


i


I







i



I





i





m
22Q
-j

01
N
•H
cn

01
C
o

0
4-1
C
•H

B
0)
x:
4-1

60
C
•H
CX
B
3
rH

#l
'i— i
4-1
O
4-1
3
O

CU
N
•H
CO



0)
rH
•H
11
cfl
rH
O
CD
I,
M
0)
X!
£

en
C
O
•H
4—1
CO
O
•H
rH
CX
CX
cfl

01
E
o
en

j-i
o
4-t
CO
C
o
(-1
o
•H
E

u-i
o
o
o
4-J

pj
3
0
13

rH
3
4-1
OI
CD
3

0)
>
O
>-i
CX

?*.
cfl


i-i
0
4J
a
cfl
cx
B
•H

cu
S-i
3
en
CO
o>
S-4
CX

£
0
i-j
•K
*











































































•
4-1
C
0)
to
0)
M
CX

4J
o
C



-------
                                                                       46
     The only  combination which can measure size distributions down to  0.001 ym
is the Brownian diffusion classifier and nuclei counter sensor as shown in Figure
18.  Such a system has been used on artifical laboratory aerosols with  some degree
of success in  several research laboratories.  The system consists of a  set of
diffusion tubes of different lengths (called a diffusion battery) and a condensation
nuclei counter to sense the aerosol concentration entering and leaving  each tube.
Each tube removes particles smaller than a given size, called the cutoff size.  Each
tube has a different cutoff size.  Thus, the concentration of particles within the
size range between two specific tube cutoff sizes is the difference between the two
exit nuclei counter concentrations.  The sharpness of the size classification is
not very good, but this is the only potentially-automatic measurement method which
operates in this size range.
         CONDENSATION
        NUCLEI COUNTER
                I
                            J   CONDENSATION    [__.
                               NUCLEI COUNTER  Jlr
                                      2           F
                                J    CONDENSATION
                                -,  NUCLEI COUNTER   ^
                                i          3           r
                                     J   CONDENSATION   |__
                                        NUCLEI COUNTER  ±2-
                                               4          F
                      DIFFUSION TUBES
                                          J   CONDENSATION
                                          "1  NUCLEI  COUNTER
                                          1         5
           AEROSOL  IN
         Figure 18. Diffusion battery  classifier with nuclei  counter concentration
                   sensors.  This technique could probably operate in the 0.001  -
                   0.01 pm size range.  Condensation nuclei  counter 1 measures
                   the concentration  before aerosol passes through diffusion tubes
                   to the other condensation nuclei counters.

-------
                                                                           47
     The only alternative  method  (not automatic) is electron microscopy which
can be performed only on non-volatile solid particles (not liquid).   Electron
microscopy in the size range  below 0.01 ym is very difficult, if  possible at all.

     In using the diffusion battery, all particles greater than about  0.1 or 0.2 ym
should first be eliminated, e.g., by impaction.  This reduces any interference
caused by the large particles.  Although this technique has been  used  in laboratory
research studies, much development remains to make it useable for effluent particles.

     At present, the optimum  choice for classification in the size range between
diffusion batteries and aerodynamic methods is electrostatic classification.  This
system consists of an electrostatic charger, an electrical mobility  classifier,
and either a charge collector (e.g., a particle filter) or a condensation nuclei
counter to measure aerosol concentration downstream of the mobility  classifier  (see
Figure 19).  As the mobility  classifier is adjusted to collect particles with various
mobilities, the aerosol concentration is noted at each mobility.   The  concentration
within a given size range  is  equal to the difference in concentrations measured with
corresponding mobility cutoff settings.  In the case of the charge collector sensor,
the current draining off the  collected particles is measured by an electrometer.  By
knowing how many charges have bled off each particle, one can calculate the aerosol
concentration which was responsible for the measured current flow.   A commercial
system designed for operation in atmospheric aerosol has a high enough resolution in
the 0.01 - 0.3 ym range to measure a geometric standard deviation of about 1.15 when
sampling a completely monodisperse aerosol with a geometric standard deviation of 1.0.
The instrument has somewhat lower resolution over the complete 0.005 - 0.6 ym range.
The electrostatic sensor has  proven superior to the condensation  nuclei counter in
atmsopheric aerosols.   Although considerable development and testing is necessary to
apply this technique to effluent streams, no basic problems are foreseen.
AEROSOL
t
IN
ELECTROSTATIC
DA DTI ^*l C
rAK riQ/Lc.
CHARGER
k.

ELECTRICAL
MOBILITY
CLASSIFIER
                                                            CONDENSATION
                                                                NUCLEI
                                                               COUNTER
                 AIR
                 OUT
                                                        ALTERNATE SENSORS
                                                                   i
   CURRENT
 COLLECTOR
(PARTICLE FILTER)
                                                                              AIR
                                                                             OUT
                                                            ELECTROMETER
        Figure 19.  Electrostatic classifier with two alternate particle sensors:
                   (a) electrostatic sensor, and (b) condensation nuclei counter
                   sensor.  This technique operates  in  the  0.005 - 0.6 ym size
                   range.

-------
                                                                         48
     There are several choices of classifier-sensor combinations for the 0.2 -
 100 pm range.  All of the aerodynamic classifier systems can be arranged with
 the classifier upstream and in series with the sensor, as shown in Figure 20.
AEROSOL
IN '
AERODYNAMIC
CLASSIFIER


PARTICLE
CONCENTRATION
SENSOR
AIR
OUT
          Figure 20.  One possible arrangement of classifier and sensor for
                      all aerodynamic classification systems listed in
                      Table B.5.

In this case, the sensor detects everything passing through the classifier, i.e.,
the particles smaller than the designed cutoff.  To measure several size fractions,
one must use several classifier operating conditions, or, more commonly, several
classifiers, each designed for a different size cutoff, with corresponding particle
sensors.  In the case of the mass sensors, the concentration of particles passing
the last classifier stage can be lumped into one size fraction.

     A look at several specific design concepts will make clear how other combina-
tions can be assembled.  First, we will look at the impactor-sensor combination.
One system could be arranged as shown in Figure 21.  In this design a continuous
flow of aerosol is drawn into the cascade impactor.  Sensor #1 measures the  total
concentration of the aerosol stream by sampling a small fraction of the impactor
flow.  The aerosol entering the impactor passes through impactor Jet //I which
collects particles greater than size D   .  Sensor #2 measures the remaining con-
centration (particles smaller than D  *j.  The aerosol then passes through impactor
Jet #2 where particles larger than D  _  (and 
-------
                                                                            49
                 IMPACTOR
                   STAGE
                      I
IMPACTOR
  STAGE
     2
IMPACTOR
  STAGE
     3
IMPACTOR
  STAGE
     4
AEROSOL,
    IN
                                                                           r
                                                    AIR
                                                   OUT
ft


SENSOR SENSOR


2
i! I

AIR OUT AIR OUT



SENSOR
3

4

AIROUT


SENSOR
4

4
AIR OUT


SENSOR
5
I


AIROUT
Figure 21. An impactor-sensor particle size measuring system using
                         aerosol concentration sensor behind every impaction stage.
                         Each sensor measures the concentration of particles smaller
                         than the preceding  impactor cutoff size.
        A second impactor design, using the  impactor-piezoelectric sensor as  an
    example, is shown in Figure 22.  In this  design, all of the aerosol passes through
    every impactor jet.  The impaction deposition plates are piezoelectric quartz
    crystals driven by external oscillators.   In this case, each crystal senses the
    size fraction larger than the size cutoff of its corresponding jet and smaller
    than the size cutoff of the preceding impactor stage.  A piezoelectric con-
    centration sensor using electrostatic precipitation can sample the aerosol
    passing the last impactor stage.   The same arrangement could be used with  beta
    radiation attenuation and soiling potential.  However, neither is as sensitive
    as the piezoelectric technique.

        A third practical impactor design uses the void-space impactor concept as
    shown in Figure 23.  As particles impact  into the void (air) space, a sample
    of the void air is drawn through  a sensor.  The sample flow of void air must be
    much smaller than the air flow through the impactor jet.  This technique avoids
    the problem of particle blowoff which occurs as the impacted sample builds up on
    conventional impaction plates. Nearly any sensor can be used with the void impactor.
    A unique feature of this technique is the concentrating effect of the system. After

-------
                                                                                 50
IMPACTOR
STAGE
1
IMPACTOR
STAGE
2
IMPACTOR
STAGE
3
IMPACTOR
STAGE
4
AEROSOL
    IN
                                 1




SENSOR
ELECTRONICS
1


SENSOR
ELECTRONICS
2


SENSOR
ELECTRONICS
3


SENSOR
ELECTRONICS
4



CONCE
TRATK
SENSC
                                                                                 nr
                                                                                  AIR  0
                Figure 22.  Another impactor-sensor particle size measuring system using
                            an impaction plate which actively senses deposited particles.
                            This example shows piezoelectric crystal sensors.   Beta
                            radiation attenuation  or soiling potential could also be
                            applied with this design.

       the system reaches steady-state operation, the aerosol in the void sample flow
       becomes concentrated:
                 vs
o Q.
                                                                           (B.15)
                          vs
      where:
                C   = concentration of void  sample  flow,

                C   = initial concentration  of  the  identical aerosol
                      size fraction,

                Q.  = flow rate through the  impactor jet, and
                 J
                Q   = flow rate of the void  sample  to the sensor.

      The concentrating effect can make some of the less sensitive sensors applicable.
      The void-space impaction system responds  more slowly to fluctuations in aerosol
      concentration than the design shown in Figure 21 and 22 because of the damping
      effect of the void space.  Hardware of this type employing beta radiation
      attenuation sensing is being fabricated by Environmental Research Corp., St.  Paul,
      under contract to Environmental Protection Agency/NERC.

-------
                                                                               51
                   IMPACTOR  IMPACTOR  IMPACTOR IMPACTOR
                    STAGE      STAGE     STAGE     STAGE
                       1234
AEROSOL
   IN
J
                                   SENSOR
                                      2
                                   T
                                   SENSOR
                                       3
                                   T
                                                                   SENSOR I— AIR
                                                                    OUT
SENSOR
   4
                       AIR  OUT   AIR OUT    AIR OUT    AIR OUT
              Figure 23.  An impactor-sensor particle size measuring system
                          using an aerosol concentration sensor to measure the
                         .concentration of particles impacted into a void space.
                          The air flow rate passing out  of each void space must
                          be very small compared with the air flow through each
                          impactor stage.
         The cyclone classifier  is  particularly  useful as  a  single-stage size
     classifier in the 0.5 - 20 ym size range.  It has been  used in industrial hygiene
     sampling applications as a simulator of particle retention in the upper respiratory
     system in humans.  Thus, anything passing through the appropriate cyclone (small
     size fraction) is defined as respirable dust.  The cyclone classifier with
     appropriate particle concentration sensors will probably  find a similar use in
     stack effluent monitoring.   The size classification with  a- cyclone is not as
     sharp as with an impactor, so detailed size analysis with several  .size stages
     would preferably be done with impaction.  However, the  cyclone has several
     advantages for single-stage  classification of effluent  particles:

         1. Can operate continuously for an indefinite time without any
            deterioration in classification ability and without phenomena such
            as particle bounce and particle blowoff which hinder continuous
            operation of all impactors except void impactors,
         2. Recognized as  the  standard simulator of the  respiratory system for
            industrial hygiene applications, and

-------
                                                                        52
     3. Simulates the operation of cyclone effluent control devices, making
        cyclone classifiers especially useful in evaluating such control
        devices.

     A cyclone classifier with appropriate sensors is  shown schematically in
Figure 24.   The sensor which measures the concentration of  small particles passing
through the cyclone can be nearly any sensor listed on Table B.5.  Probably photo-
metry and piezoelectric quartz crystal sensors would be most useful because of
their high sensitivity.  The elimination of the large  particles would probably
eliminate any particle adhesion problems normally encountered with the piezo-
electric sensor.   However, the sensor which measures the large particles coming
out from the cyclone must sense large, powder-like particles. The beta radiation
attenuation and soiling potential sensors, both with filter-tape particle
collectors, would probably be superior to other sensing techniques for this
measurement.
                                        AIR  OUT
                                            I
                                   SMALL  PARTICLE
                                        SENSOR
   AEROSOL  IN
                                    LARGE  PARTICLE
                                        SENSOR
         Figure 24.  Single-stage cyclone classifier with  two concentration
                     sensors: one for the small particle fraction and one
                     for the large particle fraction.

-------
                                                                          53
     Gravitational sedimentation has been used, particularly in England, to
simulate the deposition in the human respiratory system.  This principle
could be used to separate large particles from an effluent stream as does
the cyclone.  However, there does not seem to be a simple way to sense the
coarse particle fraction separately.  One could sense the total concentration
and the fine fraction, and then subtract to obtain the coarse-particle fraction,
but this is not as accurate as sensing the coarse and fine fractions separately
as with the cyclone classifier.

     Gravitational elutriation could be used in much the same way as cyclone
classification.  Classification into two size fractions is relatively simple,
but the size cutoff is probably not sharp.  Both the fine and coarse fractions
could be sensed directly as with the cyclone.  Although this technique offers
promise as a single-stage classifier, it has not been developed for aerosols
such as stack effluents.  Thus, development and testing is needed to more
fully evaluate its characteristics for this application.

-------
                                                                         54
C. TECHNIQUES WITH INSEPARABLE CLASSIFICATION AND SENSING


     On several aerosol size measuring techniques, the apparatus used to classify
the particles into various size ranges cannot be physically separated from the
apparatus used to sense the particle concentration within each size range.   In
these cases the processes of classification and sensing occur simultaneously.
Most of these techniques involve the interaction of the particles with electro-
magnetic radiation and are discussed in considerable detail in Volume II of this
report.  The discussions below summarize each technique and describe briefly how
each could be adapted to measurements of particle size  in effluent streams.  The
techniques include:

     1. Optical Techniques
        a. Optical Particle Counters
        b. Angular Light Scattering
        c. Multi-Wavelength Light Transmission
        d. Light Scattering: Polarization Ratio Method
        e. Holography
        f. Automated Microscopic Method

     2. Impact and Momentum Sensors

     3. Piezoelectric Single Particle Counter
                                    1211
1. Optical Techniques (See Hodkinson    , pp. 316-317, Table III, on various ways
   to use light scattering to measure particles)

     a. Optical Particle Counters

             The scattering of light by individual particles as they pass,
        single-file, through a beam of light has been used extensively to
        measure the size distribution of airborne particles.  The principle
        of operation can be seen in Figure 24.  Aerosol is drawn through the
        sampling tube by suction.  As a particle passes through the sensing
        volume, it scatters a pulse of light which is detected by the photo-
        multiplier tube.  The output of the photomultiplier tube is a series
        of voltage pulses, one for each particle passing through the sensing
        volume.  The amplitude of the voltage pulse is proportional to the
        size of the particle.  By classifying the amplitudes of the voltage
        pulses by means of a pulse-height analyzer, the size distribution of
        the aerosol can be measured.

-------
                                                                        55
PHOTOMULTIPLIER
                     CONDENSATION LENS

                PROJECTION  LENS—i
                     AEROSOL
                      INTAKE
PUPIL  LENS
            SLIT

    COLLECTION LENS


         CHOPPER DISC



  CHOPPER  MOTOR

         LIGHT PIPE
                               ENLARGED PETAIL
                                OF SENSITIVE
                                  VOLUME
                                     AREA
                         AEROSOL
                            EXHAUST
SUCTION (INLET) TUBE
           ."'^•SENSITIVE
               VOLUME
               .98 CUBIC MM
                                               ILLUMINATED
                                                 VOLUME
                                               4 CUBIC MM
                                                      EXHAUST TUBE
      Figure 25. Principle of operation of an optical particle  counter
                 which senses light scattered at 90  from  the incident
                 beam. (From a Royco Instruments,  Inc.  instruction
                 manual)
         Many particle characteristics, in addition to size,  affect  the
    amplitude of the scattered light, including  the refractive index,  the
    shape, and the surface optical properties of the particle, the angle
    between the incident light and the scattered light,  the orientation
    of a non-ideal particle in the light beam, and the wavelength character-
    istics of the light source-photomultipler system.  With any given  instru-
    ment design applied to a specific aerosol, most of these  extraneous
    effects can be minimized with appropriate calibration.  An aerosol with
    different optical properties requires a new  calibration.  If all of
    these secondary effects are ignored, an optical counter measures the
    number concentration within size ranges classified by particle surface
    area.

-------
                                                                  56
     The design of  the optics of an optical particle counter, including
 the  choice of scattering angle and the use of lasers instead of a multi-
 wavelength light source, is quite controversial.  Several commercially-
 available designs detect forward-scattered light, i.e., light scattered
 in the same direction as the light beam travels.  This design usually
 minimizes the effect of particle refractive index.143  with proper
 design and calibration of a multi-wavelength 'system, the use of laser
 light sources does  not appear to offer significant advantages for most
 applications.

     The concentration of particles sensed by an optical particle counter
 must be low enough  so that the chance of more than one particle occupying
 the  sensing volume  at one time (physical coincidence) is small.  If more
 than one particle is present in the sensing volume the scattered light
 detected by the photomultiplier will be the sum of the light scattered
 by both particles.  Thus, the instrument will detect only one particle,
 measuring it somewhat larger than either of the actual particles.  For
 application in effluent streams, dilution of the sample with clean air
 is necessary to reduce coincidence errors to an acceptable level.  Dilution
 by 1,000 to 10,000  times would be required with present commercially-
 available counters.  If the size of the viewing volume can be significantly
 reduced, less dilution will be required.

     A problem analogous to physical coincidence is electronic coincidence.
 Electronic coincidence occurs when the electrical signal from a second
 particle arrives before the system has had sufficient time to process the
 signal from the first particle.  Possible solutions to this problem are
 1) improvement in the speed of the electronics, 2) dilution, and 3)
 reduction in the size of the viewing volume.

     The theoretical lower limit of particle size which can be detected
 is determined by the scattering of light by the gas in the sensing volume.
 For practically-sized sensing volumes, this limit is approximately 0.1 -
 0.2 ym diameter for carefully-controlled laboratory aerosols.1388  Most
 commercially available optical counters operate down to 0.3 ym which is
nearly the optimum.  If larger numbers of 0.1 ym particles are present in
 the aerosol, the counter may count fluctuations in the concentration of
 these particles as  if they were particles in the range above 0.3 ym.""^
This could also limit the lower detectable size of an optical counter.

     The upper size limit of the optical counter is determined by the
fluid mechanics problem of getting the airborne particles into the sensing
volume.   Low flow rates and concentration are needed to reduce the size of
 the viewing volume which in turn reduces coincidence losses and increases
 the sensitivity of  the sensor to small particles.  With these low flow
rates, large particles (above 5 or 10 ym) tend to settle out of the
 sampling tube before reaching the sensing volume.  A carefully designed

-------
                                                                     57
   inlet flow system can greatly improve the chance of sensing large par-
   ticles, but about 30 ym appears to be the practical limit with present
   commercial instruments with carefully modified inlets.  There is no
   theoretical upper size limitation of the light scattering phenomena
   itself.

        Despite these important limitations, optical particle counters
   will undoubtedly find use in research measurements of particle size
   distributions in effluent streams.  The optical counter will not cover
   the entire size range of effluent particles, as is true of any sizing
   technique, but it does cover an important size range.  It appears that
   the technique will not be used for routine monitoring of effluents be-
   cause of the difficulties in bringing a representative sample to the
   sensing volume.   When used with other instruments so that a wide size
   distribution is measured, the optical particle counter can be a powerful
   tool for understanding the dynamics of any aerosol system.

        The optical particle counter is covered in greater detail in Volumes
   I and II of this report.  A complete list of references is included in
   Volumes I and II.  Many good reviews of light scattering phenomena are
   available, including those by Hodkinson,12H Hodkinson & Greenfield,1^3
   Kerker,1215 and Van de Hulst.-^Ol  Technical descriptions of commercially
   available instruments are authored by Zinky,370 Qgle,580 Randall &
   Keller,1210 Martens & Keller,756 Martens & Fuss,670 Martens,578 and
   Sinclair.615  Other special purpose models are described by Thomas,
   et al,120 Moroz, et al,U28 Whitfield and Mashburn,949 Neitzle,824
   Nelson,276 Mumma,598 Gebhart, et al,1388 and Kiktenko, et al.1066
   Most of these references discuss instruments designed for clean room
   monitoring.  Several discuss atmospheric air pollution measurements.

b. Angular Light Scattering

        Angular light scattering refers to the scattering of light by an
   aerosol particle at various angles with respect to the incident light
   beam.   To measure the angular distribution of scattered light, the
   observer or light detector must measure the relative intensity of light
   in each direction around a particle.  The shape of the angular scattering
   distribution changes as particle size varies.  Thus, the shape of the
   angular distribution of scattered light is a measure of particle size.

        This technique is discussed in detail in Volumes I and II of this
   report, by Kerker, et al,70  an(j by Kratohvil & Smart.842  Discussions
   of the general principles of light scattering also cover the angular
   distribution of scattered light.

-------
                                                                    58
        An instrument utilizing angular scattering for analysis of single
   aerosol particles has recently been developed by:

          Science Spectrum, Inc.
          1216 State Street
          P.O. Box 3003
          Santa Barbara, California 93105

        This technique has been used only in research laboratory appli-
   cations .  Considerably more theoretical and experimental work remains
   to demonstrate the utility of this technique.  The procedure is now
   primarily manual with no simple method of automation available.  The
   data reduction technique is also manual.  For these reasons, this
   technique currently is probably unsuited for particle size distribution
   measurements of effluent streams.

c. Multi-Wavelength Light Transmission

        The extinction of a light beam as it passes through a cloud of
   aerosol particles will change if the wavelength of the light beam
   changes.  The type and amount of change depends on the size of the
   aerosol particles.  This dependence, a rather complicated one, is
   discussed in detail in Volume II of this report and by Kerkerl215 and
   will not be repeated here.  Measurement of the transmission of light at
   several wavelengths through an aerosol cloud can result in an estimate
   of the mean volume-surface particle diameter in the range of size from
   about 0.2 - 2 ym diameter.  The estimate is not very accurate unless
   the size distribution is narrow.  Effluent particle size distributions
   are nearly always quite broad.  However, the measurement is very simple
   and inexpensive to make, requiring only a simple transmissometer with
   variable wavelength.  The wavelength variation can be performed by a
   disk made up of several different filters.  Several methods of data
   reduction are described in Volume II.

        Although this technique cannot measure the complete size distri-
   bution, it can measure the volume-surface diameter with reasonable
   accuracy in the 0.2 - 2 ym range.  Its simplicity as applied to effluent
   streams make further investigation desirable.  If some way can be found
   to automate the data reduction, probably by a computer technique, the
   technique could result in useful particle size information at low cost
   and complexity.

d. Light Scattering; Polarization Ratio Method

        When unpolarized incident light is scattered by a particle, the
   scattered light intensity can be described by two plane polarized
   components: i, perpendicular, and i  parallel to the plane of observa-

-------
                                                                     59
   tion.  There is a relationship between the intensity of scattered
   light and the wavelength of incident light, the particle size,  shape,
   size distribution, index of refraction,  and observation angle.   In
   the polarization ratio method, the ratio i-,/io is a measure of  the
   particle size.

        This technique, described in detail in Volume II,  is practical
   only for measuring size distributions with a standard deviation of
   less than 0.3.  Effluent particles have  a much wider spread of  particle
   sizes, making this technique impractical.

e. Holography

        Holography is a technique by which  three-dimensional information
   is recorded on a two-dimensional photograph.  The technique consists
   of photographing the interference'pattern that exists when a diffracted
   or object field (Fresnel or Fraunhofer diffraction pattern of the object)
   is allowed to  interfer with a reference field or background wave.  The
   image can be later reconstructed and any part of the three-dimensional
   reconstructed image can be focused or analyzed by simply placing the
   focusing plane or analyzer in the reconstructed image.   The photographed
   object, e.g., a cloud of aerosol, is not disturbed in any way.   The
   hologram is a permanent, three-dimensional, photographic record of the
   cloud.

        Holograph is presently being used successfully by TRW Systems
   Group under an EPA contract to study the spatial distribution of
   clouds of particles within a coal-fired, steam-generating combustion
   chamber.1257,1279  The limit of resolution of this system (a lensless
   system) is about 25 pm and is determined by the photographic film
   quality.   The depth of field of the system is over 30 feet.  A high-
   quality telescope could increase the resolution to about 1 urn,  but
   the depth of field would then be only millimeters.  For measuring the
   size distribution of particles in effluent streams, a resolution of at
   least 1 pm would be necessary.  The lack of depth of field would be
   severely restrictive.

        The analysis of a reconstructed hologram of aerosol particles can
   be automated by scanning the entire reconstructed image in three-dimensions,
   The scanner could be a light-intensity sensor (photomultiplier) which
   could have a data analysis system nearly identical to that used in optical
   particle counters.  The result would then be the number concentration  of
   particles in given size ranges.  This size distribution could be measured
   within various portions of the reconstructed image yielding the spatial
   distribution of the particle size distribution.   Although such  a system
   does not exist, it appears that nearly all components of such a system
   have been developed for other uses.  Although present holographic tech-
   nology is limited for continuous monitoring applications by its resolution
   and depth of field, such a system would  be a very useful too]  in many
   research studies.

-------
                                                                     60
        Holography is a complex science in an infant state of development.
   A chapter in Volume II discusses holography in more detail.   Develop-
   ments currently being made will undoubtedly make large improvements  in
   the applicability of holography to the measurement of individual effluent
   particles.

f. Automated Microscopic Method

        Microscopic methods of particle sizing have undergone considerable
   automation in recent years.  It is now possible to automatically perform
   nearly all of the tedious counting and sizing process.  One present
   method requires the human operator to obtain a representative sample
   which is compatible with the microscope.   The operator places the sample
   in the microscope and focuses on the desired portion of the sample.  The
   image is then relayed to a TV screen where a computerized electron beam
   scans the image.  The scanner detects the size of each object on the
   screen by sensing dark and light spots.  The computer then categorizes
   the detected spots by size and prints out a particle size distribution.
   The operator must choose other fields of view of the sample and other
   magnifications.  Electron microscope analysis can be performed in a
   similar way.

        It appears possible to automate the entire sizing system for a
   given application.  Nearly all parts of such a system have been developed
   before for other applications.  The prime obstacle appears to be the
   rather extreme mechanical and electronic complexity of such a system.
   The system would consist of an automated sampling system which would
   obtain representative deposits of effluent particles on a suitable
   surface and pass these sample deposits on to the microscopic system.
   The automated microscopic system would then choose a suitable field  of
   view, focus on the field of view, perform its counting and sizing
   procedure, and print out the results.  The system would have to perform
   analysis at several different magnifications to assure accurate statistical
   analysis of all reasonable size ranges.  The design of such a system would
   appear to be a matter of connecting appropriate components and programming
   a computer as a process controller and data analyzer.  However, the  cost
   would be very high and the complexity would undoubtedly lead to poor
   reliability.

        The problem areas of such a system include the resolution limits
   and automatic focusing complexities, the statistical limitations of
   any microscopic study which requires examination of adequate numbers
   of individual particles and several magnifications to cover the wide
   particle size range of effluent particles, the complexity of the entire
   system, and the prohibitive cost of the system.  These restrictions
   probably limit such a system to research programs and prohibit application
   for routine continuous measurements.

-------
                                                                          61
2. Impact and Momentum Sensors

     When a moving particle strikes a surface, its momentum is transferred to
the surface.  If the surface is suspended in such a way that it can vibrate,
the amplitude of vibration is proportional to the transferred momentum.  This
principle has been used to detect the momentum of micrometeoroids striking
spacecraft.252  in most applications, a piezoelectric transducer is used to
convert the mechanical vibration into an electrical signal.

     One of the most sensitive momentum transducers is shown in Figure 26.
When a moving particle strikes the target, the piezoeletric beams deflect
resulting in a damped oscillating electrical signal.  The amplitude of the
oscillation is calibrated in terms of particle momentum.  Thus, if the velocity
and density of the particle is known, the particle size can be calculated.  The
piezoelectric beams used in this design^-1  are polycrystalline-modified lead
zirconate titanate ceramic.  This material is described as being a nearly ideal
sensing element.252  fhe threshold sensitivity of the momentum transducer is
about 10   dyne-sec.
     In an analytical evaluation of this instrument for use as a particle size
transducer for aerosol particles, it was found that particles below 30 microns
could not be accurately sized assuming the ultimate momentum resolution (10~^
dyne-sec.), reasonable particle velocities (10^ cm/sec.), and unit density
particles.2  It appears to be difficult to accelerate the particles to higher
velocities without introducing aerodynamic instability into the sensor.  Perhaps
the aerosol beam technique-*^/3 >1361,1362  could be used for this purpose.  However,
the inability of this sensor to detect particles below 30 microns makes it
relatively useless for effluent aerosols.
                                                         Target
     Piezoelectric beams


            Support -
          Figure 26. Schematic of a particle momentum sensor which uses
                     piezoelectric beams as the transducer.252  Particles
                     strike  the target, causing it to vibrate which, in
                     turn, causes an electrical signal in the piezoelectric
                     electronic circuit.

-------
                                                                         62
3. Piezoelectric Single Particle Counter^

     A piezoelectric microbalance can sense the addition of a single particle
to its surface.1223  Rather than measuring frequency shifts caused by a number
of particles added to the crystal surface as done the standard piezoelectric
microbalance (see Volume II of this report), the single particle counter
differentiates the frequency signal resulting in a pulse whenever a single par-
ticle becomes attached to the crystal surface.  The magnitude of the pulse is
related to the mass of the particle.  Thus, if particle, density is known, the
size can be calculated.  A pulse-height analyzer can be used to classify the
pulses into particle size range.

     The ultimate size resolution claimed for this technique is about one micron
                         179*}
for unit density spheres. ^    The particle must adhere solidly to the crystal
surface immediately upon contact, or the pulse magnitude will not be proportional
to particle mass.  The concentration and sampling rate must be low enough to
prevent more than one particle from striking the crystal at one time.  It appears
that no more than about 1()3 particles per second can be sampled.  The crystal
sensor is somewhat sensitive to temperature, relative humidity, condensible vapors,
and any other foreign material which can adhere to the crystal surface.

     One can also measure the mass concentration with the piezoelectric single
particle counter by monitoring the rate of change of frequency during the time
period of interest.  Thus, it can also be used as a mass concentration monitor
as described in Volumes I and II of this report.

     This technique has just recently been introduced to aerosol science and is
not yet very developed.  Thus, it is not clear which or how many of the potential
problems will limit its use in effluent streams.

-------
                                                                        63
D. LABORATORY POWDER SIZING TECHNIQUES

     A number of semi-automatic techniques exist for the sizing of powder
particles in the laboratory.  Most of these techniques require the dispersion
of the particles in a suitable liquid.  The size analysis is then performed
on the liquid-suspended particles.

     Several of these techniques could conceivably be automated for sizing par-
ticles in effluent streams.  A representative sample would first have to be
collected from the effluent stream, dispersed accurately into the suitable liquid,
and then sized automatically by the sizing apparatus.  At least two major problems
present themselves: (1) the mechanical complexity of a completely automated par-
ticle size monitoring system, and  (2) the difficulty in defining the relationship
between the particles as they exist in the stack and as they exist in the dispersion
liquid.  The first is primarily a mechanical design problem, but the second is a
more basic problem.

     Several configurations are suggested here as possible solutions to the
problem of automating such a system.  The system would consist of two major
components: (1) the particle size analyzer, and (2) the equipment needed to collect
a representative sample and bring it to the particle size analyzer suspended in
a suitable liquid.

     Several semi-automatic particle size analyzers are commercially available for
sizing 0.5 - 100 \im particles suspended in a liquid.  Two such instruments are
the Micromeritics particle size analyzer* and the Coulter counter**.  These are
two of the most automated laboratory powder sizing instruments available.

     The Micromeritics instrument classifies particles by size using the principle
of particle sedimentation through the liquid.  The particle concentration is sensed
by an x-ray beam.  A computerized readout system allows the adjustment of particle
density, liquid density, and viscosity.  The result is plotted by an X-Y plotter
in terms of the "% by mass less than size" versus "equivalent particle diameter".
The particles must be introduced into the instrument in the form of a liquid
suspension.

     The Coulter counter feeds the particles single-file through a small orifice
and measures a change in the electrical conductance across the orifice as each
particle passes through.  The amplitudes of the resulting electrical pulses, one
for each particle, are proportional to the size (volume) of the particles.  The
results, after appropriate classification of the electrical pulses, can be present-
ed in several forms, such as "% by number less than size" or "number counted per
size range" versus "equivalent particle diameter".  As in the Micromeritics instru-
ment,  the particles must be introduced into the Coulter counter in the form of a
liquid suspension. The Coulter counter liquid must be an electrolyte.
 *Manufactured by: Micromeritics Instrument Corporation, 800 Goshen Springs Road,
  Norcross, Ga. 30071.

**Manufactured by: Coulter Electronics Industrial Division, 590 West 20th St.,
  Hialeah, Florida 33010.

-------
                                                                        64
     Each of the size analyzers has its own set of problems.  These have been
quite well described in the literature and will not be discussed here.  Some
of these problems may be limiting factors in the design of a fully-automated
system.

     However, a major problem faced by all sizing systems of this type is the
design of the second component mentioned above:  the equipment needed to collect
a representative sample of particles from the effluent stream and bring the sample
to the size analyzer suspended in a suitable liquid.  No automated system of
this type is known and all known design concepts appear to lack the capability
of reliable, long-term, unattended operation.  The two most feasible air-to-
liquid particle samplers are the LEAP and MSI samplers*.  The LEAP sampler uses
electrostatic precipitation to collect particles from 300 to 15,000 liters per
minute of air onto a film of liquid.  The liquid flows continuously over the
collection surface and then into a collection bottle.  The MSI sampler is nearly
identical except that it uses impaction rather than electrostatic precipitation
and it operates at only one flow rate, 1000 liters per minute, as commercially
designed.  Conceptually, one of these samplers could continuously collect particles
from the effluent stream and feed the resulting liquid suspension to the particle
size analyzer.

     Even if the design problems could be satisfactorily solved, the problem of
relating the measured particle size in the liquid suspension to the particle size
in the effluent stream remains.  Generally, one wants to measure the size of an
agglomerate just as it exists in the effluent stream.  However, the agglomerate
may break up or grow in the liquid suspension.  The control of the agglomeration
or deagglomeration process would be difficult.  Also, any soluble particles would
be dissolved by the liquid, and liquid or partially-liquid particles would be
grossly changed in the liquid suspension.  In instruments which use an electrolytic
suspension, the acidity of the stack gas could change the electrolytic properties.
Thus, the interpretation of data from  -this type of sizing system would be difficult.

     For these reasons, this technique of sizing effluent particles should be
approached with extreme caution.
*Manufactured by Environmental Research Corporation, 3725 North Dunlap Street,
 St. Paul, Minnesota 55112.

-------
                                                                          65
E. A DIFFERENT SIZING CONCEPT: PARAMETRIC MEASUREMENT

     The measurement of a single particle concentration parameter has rather
severe limitations for characterizing an aerosol such as combustion effluent.
For example, the number concentration is usually dominated by the great numbers
of small particles below 0.1 ym diameter.  The addition of enough 25 ym particles
to double the mass concentration will cause an undetectibly small change in the
number concentration.  On the other hand, the mass concentration is usually
dominated by the great mass of large particles primarily above 1 ym diameter.
Doubling the number of 0.01 - 0.1 ym particles causes an undetectibly small
change in the mass concentration in such cases.  Thus, mass concentration
measurements are sensitive to large particles while number concentration
measurements are sensitive to small particles.

     The parametric particle size measurement method would use the inherent
size limitations of particle concentration sensing techniques to measure a
form of particle size distribution.  Several concentration sensors would be
used, each sensing a different particulate parameter.  For example, number,
surface, and mass concentration are three potential parameters.  Number con-
centration would be highly sensitive to small particles, surface concentration
to medium-sized particles, and mass concentration to large particles.  Although
this method would not offer enough detailed information for many scientific
research studies, it would appear to offer very useful measurements for source
monitoring applications.

     Particle number concentration can be measured by a condensation nuclei
counter or by an electrostatic technique.  Particle surface area, or a parameter
similar to surface area, is measured by well-designed optical transmissometers
and photometers.  Particle mass is measured by beta radiation attenuation instru-
ments.  All of these techniques are discussed in detail in Volumes I and II of
this report.  Nearly all have been used in other applications, and adaptation of
at least one sensor in each group to effluent measurements appears easily possible.
Electrostatic sensors, optical transmissometers, and beta radiation attenuation
have all been used on large combustion effluent stacks in the past.  Thus, the
design of equipment to make all three measurements simultaneously, on a continuous,
monitoring basis, does not appear to be an obstacle.  Further development of
these techniques would not be prohibitively expensive and could probably result
in reliable instruments in about 2 years.

     Even though this approach would result in the measurement of only three
particle concentration parameters, not in the measurement of size distributions,
much information about the particle size distribution would be available from
the three measurements.  The number concentration measurement would be sensitive
to fluctuations in the concentration of small particles.  The mass concentration
measurement would be sensitive to fluctuations in the concentration of large
particles.   The surface area measurement would be sensitive to particles in the
middle size range.   Thus, an increase in the emissions of large particles would
be detected by a corresponding increase in measured mass concentration and

-------
                                                                            66
little change in the measured number concentration.  The surface-area measure-
ment would resolve whether the particles were very large (perhaps 20 ym) or
close to 1 ym.  An increase in the emissions of small particles would be detected
by a corresponding increase in the measured number concentration, but with little
change in measured mass concentration.  Again, the surface area measurement would
resolve whether the additional particles were very small (approximately 0.01 ym)
or in the 0.1 ym range.

     There are a number of other factors making this approach to particle size
monitoring of source effluents attractive.  Both the large and the small particles
are important to air pollution.  The particles above 10 ym settle onto nearby
surfaces, contaminating buildings, automobiles, streets, agricultural crops, trees,
rivers, lakes, etc., with possible damage due to corrosive effects.  Particles
below 0.1 ym remain suspended for long periods of time, contributing to global
pollution, to rain nuclei, to condensation processes, to particulate-gaseous
chemical interaction including photochemical smog formation, etc. These small
particles grow, by condensation of vapors and by collision and agglomeration with
other particles, into the intermediate size range between 0.1 and 10 ym.  Particles
of the intermediate size range penetrate most deeply into the human respiratory
system, penetrate filters more easily than other sizes, remain airborne for long
periods of time, are most visible, cause degradation of visibility, etc.

     Thus, each size range is important for different reasons, and the measurement
of a single particle parameter does not characterize the entire range of particles
found in combustion sources sufficiently enough to relate source emissions to air
pollution effects.

     Theoretically, if any form of the particle size distribution can be measured
with enough accuracy, any other form of the size distribution can be calculated
from it.  However,  the resolution limitations of practical instrumentation do not
allow accurate mathematical transformations to be made on aerosols such as
combustion effluents.  For example, any measurement of particle mass concentration
within various size ranges will ignore the insignificant mass of the large numbers
of 0.01 - 0.1 ym particles.  Any measurement of particle number concentration
within various size range will ignore the insignificant numbers of larger par-
ticles above 10 ym.  Instrument accuracies to 8 or 10 significant digits of con-
centration and over 8 or 10 orders of magnitude of concentration would be necessary
to avoid this problem.  Thus, measurement of the entire range with a single instru-
ment is not possible at this time and does not appear probable.  Measurement of
a single parameter over the entire range with several instruments is also not
possible.

     The mass concentration and the mass size distribution are nearly useless to
the meteorologist studying global pollution patterns and rain formation, to the
city air pollution official concerned with predicting the rate of photochemical
smog formation, and to the researcher studying the dynamics of pollution in the
submicron size range.  These applications require a measurement which is sensitive
to the particles of primary interest, i.e., the submicron particles.  The number
concentration and number size distribution are two such measurements.

-------
                                                                          67
     The number concentration and number size distribution are nearly as use-
less to the pollution control official concerned with dust fall on various
surfaces and the stack owner concerned with complying with pollution control
regulations written in terms of particulate mass.  These persons require the
mass concentration and mass size distribution; parameters which are of primary
interest to them.

     Likewise, the pollution control official concerned with reducing the visible
smoke from stacks and the visible haze over a city, and the people concerned with
reducing the harmful health effects of particulate pollution will be most con-
cerned with the particles in the middle size range: from 0.1 to 10 ym.  Optical
measurements and particle surface area sensors are usually most sensitive to
particles in this size range.

     This does not mean, however, that three separate and complete size distri-
butions would have to be measured to satisfy all applications.  Rather, it appears
that sufficient data may be provided by three single concentration measurements
of these particle properties: (1) number concentration, (2) surface or cross-
sectional area concentration, and (3) volume or mass concentration.  This would
satisfy most of the requirements for monitoring applications.

     Measurements of other effluent particulate parameters may also prove useful.
For example, the mass of particles within the respirable range as defined by the
American Conference of Governmental Industrial Hygienists-^jS  may be a useful
measure of the harmfulness of effluent particles from a health effects viewpoint.
The respirable mass concentration is essentially the portion of the total mass
concentration which remains airborne for long periods of time.  The difference
between total mass concentration and respirable mass concentration is essentially
the fraction of particulate emissions which settles to the ground near the effluent
source.   Thus, measurement of respirable mass concentration and total mass con-
centration provides another interesting combination for continuous monitoring
applications.

     Some of the parameters which may be of interest for particle size monitoring
of source effluents are:

     Number concentration
     Surface concentration
     Volume or mass concentration
     Visible or visibility measurement
     Respirable
     Suspended
     Settleable
     Rain and  snow nuclei
     Potential photochemical smog aerosol formation.

-------
                                                                          68
     Theoretically, the result of this parametric approach appears to be
practical methods to measure not only the concentration of particulate
emissions in terms useful to a wide range of pollution control personnel, but
also much of the necessary particle size information.  The necessary hardware
is either already developed or can be developed within about two years.  The
equipment will not be prohibitively expensive, even for use as a continuous
monitor on every large combustion emissions source.

     The authors recommend an intensive investigation into the potential use-
fulness of this approach.  The response to each measurement in the system to
reasonable, expected effluent changes should first be studied.  This would
isolate conceptual problems and define additional benefits of such measurements.
The conceptual evaluation should be followed or accompanied by development of
the appropriate concentration monitoring instruments capable of operating in a
total system.  The development of concentration sensing instruments for this
application requires no development in addition to the concentration sensors
needed for the monitoring of single concentration parameters.  Thus, particle
size information sufficient for most monitoring applications can be obtained
with little additional development of equipment.

-------
                                                                          69
F. SUMMARY AND CONCLUSIONS

     1. Particle size distribution is one of the three most important
        parameters in defining the relative potential harmfulness  of
        particulate emissions.  A single particle concentration measure-
        ment cannot define the relative potential harmfulness  of an effluent
        stream without information about the relative concentration within
        several size ranges.   Particle size measurements are urgently needed
        for research applications, for control equipment evaluation, and for
        air pollution control monitoring.

     2. Most size distribution measurements of combustion effluent particles
        made in the past used available equipment which severely limited the
        size range effectively sampled.   The technique used for most such
        measurements could not measure particles below 2 ym or over 100  ym
        leading to the conclusion that nearly all effluent particles are
        within the 2 - 100 ym size range.

     3. Demonstrated size distribution measurements  of combustion  effluent
        particles in the micron and submicron range  are almost nonexistent.
        This size range appears to be the most important for most  air
        pollution considerations.

     4. There is no single preferred way of presenting particle size distri-
        bution data which is  useful for  every application.   The preferred
        size measurement technique is usually the method which most directly
        obtains the desired information  in its final  form.   An important
        consideration is whether particle number,  surface area, volume,  mass,
        or some other parameter is needed for any given application.

     5. No sizing instrument  can classify particles  over the entire particle
        size range of interest (from 0.001 to over 100 ym).  The alternative
        is to combine several techniques to cover the entire range or to choose
        the range of primary  interest and find an appropriate  technique  for
        that range.   Both of  these approaches require very careful interpretation
        of the data.

     6.  Aerodynamic particle  size  is the most useful  size parameter in most
        applications.

     7.  Table B.5 lists  the operable size ranges of various  combinations of
        size classifiers and  concentration sensors.   The recommendations for
        particle size measurements of most combustion effluents for most air
        pollution applications  are outlined below:

        a.  An impaction  size  classifier  (an aerodynamic  classsifier)  coupled
           with a beta radiation attenuation,  piezoelectric  quartz crystal,
           or photometric  concentration  sensor appears to be most  applicable

-------
                                                                    70
      to detailed effluent stream particle size measurements at this
      time.   The impactor cannot classify particles less than about
      0.2 or greater than about 30 ym.   However, this appears to be
      the range of primary interest for many, perhaps most, applications.
      The particles below the last impactor cutoff can be lumped into
      one size range and, with appropriate attention given to correct
      design of the sampling system particles larger than the first
      cutoff can be lumped into another size range.  Thus, although all
      size range cutoffs are between about 0.2 and 30 ym,  the entire
      range of size passing through the device is accounted for.  At
      the present, this approach appears to offer the most promose of
      success as a single automated particle sizing tool in effluent
      streams.

   b. If size classification below 0.2  ym is necessary, electrostatic
      techniques (0.005 - 0.6 ym) or Brownian diffusion techniques
      (0.001 - 0.05 ym) must be used.  Although not yet applied to
      stack effluents, an electrostatic technique has demonstrated
      high resolution in the 0.01 - 0.2 ym range on laboratory,
      atmospheric, and small, flame-generated aerosols.  The Brownian
      diffusion technique, used only for specialized laboratory measure-
      ments to date, is the only known  method for sizing 0.001 - 0.005 ym
      particles.  Further research and  development of both of these
      techniques is necessary before automated instruments for stacks
      could result.  No other method of  classifying particles by size
      significantly below 0.2 ym is known.

   c. The cyclone classifier can aerodynamically separate an effluent
      aerosol sample into two size fractions.  The size split must be
      between about 0.5 ym and 20 ym and is not very sharp.  The cyclone
      classifier has been used extensively for separating the "respirable
      fraction" from industrial hygiene aerosols and may prove useful for
      similar application in effluent streams.

8.  Although optical particle counters are severely limited for effluent
   stream measurements with their present design, major modification may
   significantly reduce the problems.  The lower size limit of optical
   particle counters is about 0.2 ym.  The upper limit, determined by the
   sampling system, is about 30 ym; and may be improved by development of
   an instrument which can operate directly within the effluent stream.
   Another problem at present is the high dilution with clean air required
   to prevent coincidence.

9.  Holography offers some promise for research sizing applications in
   effluent streams.  However, routine use appears limited by the cost
   and complexity of the equipment.  In effluent streams, its ability to
   obtain a 3-dimensional record (photograph) of the effluent particles

-------
                                                                      71
    without disturbing the flow is highly attractive.  However, present
    holography is limited by the smallest particle which can be individually
    resolved (a few microns) while still maintaining a practical depth-of-
    field (perhaps greater than a centimeter) and distance-from-apparatus-
    to-particle (greater than 3 meters for use in effluent streams).
    Holography is developing rapidly and may prove more useful in the future
    as the equipment cost and complexity decreases, and as resolution in-
    creases .

10. Powder and slurry particle sizing techniques do not appear applicable
    to effluent streams because of the difficulty in relating the particle
    size measurement (made in a liquid) to the actual airborne particle.

11. A different technique, which appears to offer a practical method for
    long-term,  continuous monitoring in effluent streams uses the size
    limitations of several different concentration sensors to effectively
    measure particle concentration within several size ranges.  The
    technique would use (1) a beta radiation attenuation sensor to measure
    the total particulate mass concentration (sensitive to D^, or large
    particles in the 1 - 100 ym range), (2) a transmissometer to measure
    opacity (roughly sensitive to D^, or particles primarily from 0.1 - 10 ym)
    and (3) a condensation nuclei counter or electrostatic counter to measure
    the particle number concentration (sensitive to the number of particles,
    or to particles from 0.001 - 1.0 ym).  Analysis of the 3 simultaneous
    measurements would appear to offer sufficient particle size information
    for most  continuous air pollution monitoring applications.  Measurement
    of total  mass concentration and respirable mass concentration offers
    another interesting combination of particulate parameters.  Nearly all
    hardware  needed for these measurements already exists.  Further investi-
    gation and testing of this approach is recommended.

12. Although  this report addresses itself to the heart of particle sizing
    apparatus (the size classifier and concentration sensor), the equally
    important problem of particle sampling must also be considered.  The
    problem of delivery of truly representative samples of effluent to the
    measuring instrument has not yet been solved.  Although one can tolerate
    some agglomeration of fragmentation of particles in a sampling system
    for total concentration measurement, such changes in particle size cannot
    be permitted in a particle sizing system.  Questions related to the
    conditioning of the effluent (dilution, heating, cooling, etc.) prior
    to measurement by most sizing instruments must also be investigated
    thoroughly.  Emphasis must be placed on a thorough investigation
    of the particle size changes which take place in effluent sampling
    systems.   This area merits at least as much attention as the particle
    sizing instrument itself.  The advantages offered by any potential
    sizing instrument which does not require removal of a sample from the
    stack must be emphasized.  Unfortunately, all of the practical, available
    techniques require sample extraction.

-------
                                                                         72
G. REFERENCES

  27      Peterson, C.  M.,  and Paulus ,  H.  J.,  "Continuous  Monitoring of  Aerosols
          Over the 0.001-  to 10-micron  Spectrum",  American Industrial Hygiene
          Association Journal, p.  111-122  (Mar - Apr 1968).

  29      Lundgren, D.  A.,  "An Aerosol  Sampler for Determination  of  Particle
          Concentration as  a Function of Size  and  Time", APCA Journal, V.  17,
          no. 4, p. 225-229 (Apr 1967).

  40      Liu, B.Y.H.,  Marple, V.  A., and  Yazdani, H.,  "Comparative  Size Measure-
          ments of Monodisperse Liquid  Aerosols by Electrical and Optical  Methods",
          Environmental Science and Technology, V. 3, no.  4,  p. 381-386  (Apr 1969).

  42      Clark, W. E., and Whitby, K.  T., "Concentration  and Size Distribution
          Measurements  of  Atmospheric Aerosols and a Test  of  the  Theory  of Self-
          Preserving Size  Distributions",  Journal  of Atmospheric  Sciences, V.  24,
          no. 6, p. 677-687 (Nov 1967).

  56.      Liu, B.Y.H.,  Whitby, K.  T., and  Yu,  H.H.S.,  "On  the Theory of  Charging
          of Aerosol Particles by Unipolar Ions in the  Absence of an Applied
          Electric Field",  Journal of Colloid  & Interface  Science, V. 23,
          p. 367-387 (1967).

  61      Whitby, K. T., and Peterson,  C.  M.,  "Electrical  Neutralization and
          Particle Size Measurement of  Dye Aerosols",  Industrial  & Engineering
          Chemistry Fundamentals,  V. 4, no.  1, p.  66-72 (Feb. 1965).

  68      Whitby, K. T., and Clark, W.  E., "Electric Aerosol  Particle Counting
          and Size Distribution Measuring  System for the 0.015 to 1  u Size Range",
          Tellus, V. 18, p. 573-586 (1966).

  70      Kerker, M., Farone,  W. A., Smith,  L. B., and  Matijevic, E., "Determination
          of Particle Size  by the  Minima and Maxima in  the Angular Dependence  of
          the Scattered Light, Range of Validity of the Method",  Journal of Colloid
          Science, V. 19, p. 193-200 (Mar  1964).
          	  " —-1	                                                                   I

 120      Thomas, A. L., Jr.,  Bird, A.  N., Jr., Collins, R. H.,III,  and  Rice,  P.C.,
          "A Portable Photometer and Particle  Size Analyzer", ISA Journal, V.  8,
          p. 52-56 (Jul 1961).

 123      Goetz, A., and Preining, 0.,  "The  Aerosol Spectrometer  and Its
          Application to Nuclear Condensation  Studies", Monograph No. 5.,
          Geophysical Union. (1960).

 124      Thompson, J.  K.,  "Determination  of Aerosol Size  Distribution by  Jet
          Impactor-Light Scattering Technique", Analytical Chemistry, V. 29,
          p. 1847-50 (1957).

-------
                                                                         73
143      Hodkinson, J. R., and Greenfield, J. R.,  "Response Calculations for
         Light-Scattering Aerosol Counters and Photometers", Applied Optics,
         V. 4, p. 1463-1474 (1965).

154      Flesch, J. P., Norris, C. H.,  and Nugent, A. E.,  Jr., "Calibrating
         Particulate Air Samplers with  Monodisperse Aerosols:  Application to
         the Andersen Cascade Impactor", American  Industrial Hygiene Assoc.
         Journal, V. 28, no. 6, p. 507-516 (Nov -  Dec 1967).

208      Crandall, W. A., "Development  of Standards for Determining Properties
         of Fine Particulate Matter", ASME - Paper 64-WA/PTC-3, for meeting
         Nov. 29 - Dec. 4, 9 p. (1964).

239      Anon., Determining the Properties of Fine Particulate Matter, American
         Society of Mechanical Engineers, PTC-28-1965 ,  40  p. (1965).

252      Rogallo, V. L., and Neuman, F. , "A Wide-Range  Piezoelectric Momentum
         Transducer for Measuring Micrometeoroid Impacts", Ames Reseach Center,
         Moffett Field, Calif., Clearinghouse No.  NASA  TN  D-2938 (Jul 1965).

274      Anon., "Dust Emission Control  in Calcium  Carbide  Production", Trans-
         lated from German, Clearinghouse No. TT 68 50469/2 (Dec 1965).

276      Nelson, M. B., "Fabrication of Particle Counters  for Clean Rooms",
         IIT Research Institute, Chicago, 111., NASA Tech. Brief 67-10076 (Jul  1966)

290      Goetz, A., and Kallai, T., "Determination of Size and Mass Distribution
         of Aerosols", Symposium on Air-Pollution  Measurement Methods, Special
         Tech. Pub. No. 325, ASTM, p. 40-55 (1964).

294      Mitchell, R. I., and Pilcher,  J. M., "Cascade  Impactor for Measuring
         Aerosol Particle Size", Industrial and Engineering Chemistry, V. 51,
         p. 1039-1042 (1959).

354      Goetz, A., and Kallai, T., "Instrumentation for Determining Size- and
         Mass-Distribution of Submicron Aerosols", APCA Journal, V. 12, no. 10,
         p. 479-486 (Oct 1962).

369      Goetz, A., "Methods for Measuring Particle Composition in Photo-
         activated Aerosols" APCA Journal, V. 14,  no. 6, p. 213-219 (Jun 1964).

370      Zinky, W. R., "A New Tool for  Air Pollution Control, the Aerosol Particle
         Counter", APCA Journal, V. 12, no. 12, p. 578-583 (Dec 1962).

505      Goetz, A., "An Instrument for  the Quantitative Separation and Size-
         Classification of Air-Borne Particulate Matter Down to 0.2 Micron",
         Geopfis, Pura. Appl.,  V. 36,  p. 49-69 (1957).

-------
                                                                            74
546       Goetz,  A. » "The Aerosol  Spec fro. .^.:i t  -  A h     i.,i i ,>i..riiit for the
          Analysis of Air-Borne Particles in th,-. '-,.:',  ,  - ,..  Range", Public
          Works,  V. 90, no. 2> p.  91-93 (f-Vb ''/>•?.

578       Martens, A. E. , "Errors  in Measaiea^n-, an.i  Count J ng  of Particles
          Using Light Scattering,  APCA Journal, V.. I*.,  bf, 1-663 (Oct 1968).
580       Ogle,  H.  M. , "Particle  Count J rig T«."h:, :•,,"'  .  <.L''< •' > .

638       Rumpf, H., "The Particle  Size Analysis of  inh-7 trial  Dusts", Staub-
          Reinhalt  der Luf^t^jEngl .  Traits. _) ;  V. J5 ,  i,,.   1,  p,  17-26 (Jan 1965).

649       Stober, W. , "Measuring  Methods  ;..> La;,.!,'.^  ,;.   V-sieal  Properties
          of  Aerosol Systems", §j^iuj^~J^ein.2>r.'^ i  ju r   ;i.'t  i< ^.i ,  f rans . ) , V. 25,
          no.  9, p. 34-41 (Sep 1965)*.  """    ..... '  " .....      "

670       Martens,  A. E. , and Fuss,  K.  H. ,  "An OpticaJ  Counter  for Dust Particles",
          Staub-Reinhalt __der ;_Luf t ;  (_En^l . i'r.;ub . ") -  v, ?<-,  so.  6, p, 14-18 (Jun  1968).
675      Simecek,  J., and Kuhalek,  J. ,                      , ;  . ...  Preseparation  in
         Dust  Sampling", S t :mb-Rei nh 3 '• t  ':    ,  <~   .  ^'    -•;-.;,.}, V. 28, no. 10,
         p.  24-31  (Oct. 1968),       ~       .------ ........ -

680      Whitby, K.  T. , and Liu,  B.Y.H,S  ' '.oau! -ir f .,n   <  • ~ ountable Pulses by High
         Concentrations of Subcountable •  /< '- <:  '.   ' ;- ;  .  '•,  the  Sensing Volume  of
         Optical Counters", 'L9lIi.rJlSJL_£'J"  '••   *  •"'     ' !'f  '  '  '  '^(i?Ii£,?.' ^- 25, no.  4,
         p'.  537-546  (1967). ""             " ............       .....

685      Parker, G.  W. , and Buchholz,  H.   "  > •' i:e i' ' uf..-, i. i"; ;,ation  of Submicron
         Particles by a Low-Pressure Ca:  adr fnpa.-to1 ',  Oak Ridge, Tenn. ,
         Clearinghouse No. ORNL-4226  (Ju i  i-K>8),

756      Martens,  A. E., and Keller, .K  ') , .  "Ar 7n;,f ••umout  for  Sizing and Count-
         ing Airborne Particles", Ameij. ...in i ut-"  ._ija;  hygiene Assoc. Journal,
         V.  29,  p. 257-267 (May  - Jun  1'VoBf"   ...... ~

818      Brown,  P. M. , and Hochrainer ,  D . ,  'T.Oiii. al A: roLioi Spectrometer,
         Cylindrical Aerosol Spectrofu-.-t ej"   I-'-iulona!  C^nrer for Atmospheric
         Research, Boulder Colorado, ^Lu-n  ; >-.'Ou. ,,  '; .   IMi "• BO 880.

-------
                                                                          75
 824      Neitzel, W. E., "A High-Volume, Real-Time Aerosol Monitor",  Sandia Lab.,
          Albuquerque, N. M., Clearinghouse  No.  SC-DR-69-56 (Jun 1969).

 833      Heywood, H., "Fundamental Principles of Sub-Sieve Particle-Size Measure-
          ment", paper presented at Inst. of Mining & Metallurgy Symp.  on Mineral
          Dressing, London, Eng., Paper No. 6 (Sep 23-25, 1952).

 834      Higgins, R. I., and Dewell, P., "The Measurement of Airborne Dust Con-
          centration in Iron-Foundries Using the Hexlet Dust Sampler,  B.C.I.R.A.
          Journal, V. 8, no. 3,  p. 425-436 (May 1960).

 842      Kratohvil, J. P., and Smart, C., "Calibration of Light-Scattering Instru-
          ments, III.  Absolute Angular Intensity Measurements on Mie  Scatterers",
          Journal of Colloid Science, V. 20, no.  8, p. 875-892 (1965).

 912      Walton, W. H., "Theory of Size Classification of Airborne Dust  Clouds
          by Elutriation", British Journal of Applied Physics, Supp. no.  3,
          p. S29-S39  (1954).

 925     • Stober, W., and Flachsbart, H., "Size-Separating Precipitation  of
          Aerosols in a Spinning Spiral Duct", Environmental Science & Technology,
          V. 3, no. 12, p. 1280-1296 (Dec 1969).

 948      Mercer, T. T., "The Stage Constants of Cascade Impactors", Lovelace
          Foundation, Albuquerque, N.M., Clearinghouse No. LF-12 (Oct  1963).

 949      Whitfield, W. J., and Mashburn, J, C.,  "Development of an Increased
          Sampling Rate Monitoring System", Sandia Lab., Albuquerque,  N.  M.,
          Clearinghouse No. SC-RR-66-585 (Oct 1966).

 957      Brink, J. A., Jr., "Cascade Impactor for Adiabatic Measurement",
          Industrial & Eng. Chemistry, V. 50, no. 4, p. 645-648 (Apr 1958).

 995      Mercer, T. T., "The Interpretation of Cascade Impactor Data", American
          Industrial Hygiene Assoc. Journal, V. 26, p. 236-241 ( May - Jun 1965).

 996      Hounam, R. F., and Sherwood, R. J., "The Cascade Centripeter: A Device
          for Determining the Concentration and Size Distribution of Aerosols",
          American Industrial Hygiene Assoc. Journal, V. 26, p. 122-131 (Mar -
          Apr 1965).

1002      Lippmann, M. , "Review of Cascade Impactors for Particle Size Analysis
          and a New Calibration for the Casella Cascade Impactor", American
          Industrial Hygiene Assoc. Journal, V. 20, p. 406-416 (Oct 1959).

1052      Ettinger, H. J., "Survey of Technique Employed to Define Aerosol
          Respirable Dust Concentration and Particle Size Characteristics",
          University of California, Los Alamos, N. M., Clearinghouse No.  LA-4249.

-------
                                                                          76
1066      Kiktenko, V. S. , Safronov, Y. P., Kuclryaut.sev,  S.  1., Fedorov,  B.  F.,
          Pushcin, N. I., and Fedorovich, A. A,, "Photoelectric Count of  the
          Number of Aerosol Particles of Organic and  Inorganic Origin", Engl.
          Trans, from: Gifiyenai Sanitariary_„ Moscow» 7.  2.6, no. 2, p. 47-53,
          Feb. 1961, Clearinghouse No. JPRS-8334 (May 1961).

1069      Ramanujam, M., and Venkateswarlus , D., "Particle-Size Analysis"  Chemical
          Age of India, V. 19, no. 11, p. 975-983  i'ir'(",8).

1119      Stober, W. , "Design and Performance cf a Si,'-.»-• -Separating Aerosol
          Centrifuge Facilitating Particle-Size Sp<">ct r^rr.otry in the Submicron
          Range", Assessment Airborne RadioactIvitv, T'roc.  Symp., Vienna,
          p. 393-404 (1967).

1120      Raabe, 0. G., "Calibration and Use of the ooetz Aerosol Spectrometer",
          Assessment of Airborne Radioactivity, Proc. Gymp., Vienna, p. 417-431 (1967]

1128      Moroz, W. J., Withstandley, V. D., and Anderson,  G. W., "A Portable
          Counter and Size Analyzer for Airborne Dust'!, \_eview of Scientific
          Instruments, V. 41, no. 7, p. 978-983 (Jul  L97(J).~

1201      Van de Hulst, H. C., Light Scattering by Small  Particles, Wiley (1957).

1210      Randall, L. M., and Keller, J. D., "Electro-Optical Aerosol Counter
          Inst., AIHA Conf., Houston, Texas (1955),
 /il      Davies, C. N. , ed. AerosoJ Sclent . /•... ', ..    •. <; ••  .'19^6).
1215      Kerker, M. , The Scattering of Ligi'T and Ctht-r  F i t;c, I romagnetic  Radiation,
          Academic Press (1969),

1216      Green, H. L., and Lane, W. R. , Particulate CJ.'Hids:  Dusts,  Smokes  and
          Mists. , E. and F. N.  Spon Ltd., London  (1964),

1223      Chuan, R. L., "An Instrument for  the DirtcL Meai.-irement  of  Particulate
          Mass", Aerosol Science, V. 1, p.  111-114  0 ^Q) .

1257      Matthews, B. J., and  Kemp, R. F.,  "Investigation  of  Scattered  Light
          Holography of Aerosols and Data Reduction Techniques", TRW  Systems
          Group, Redondo Beach, Calif., TRW Report No.  14103-6002-RO-OO  under
          NAPCA Contract CPA 70-4 (Nov 3970).

1273      Israel, G. W. , "Investigations of  Aerosol Beams", Staub-Reinhalt  der
          Luft,  (Engl. Trans.), V. 29, no.  5, p. 5-8 (May 1969).

-------
                                                                           77
 1276      Whitby, K. T., and Liu, B.Y.H., "Atmospheric Particulate Data -
          What Does It Tell Us About Air Pollution?", Paper Presented at llth
          Conference Methods in Air Pollution and Industrial Hygiene Studies,
          Berkeley, Calif., Mar. 30, 31, and Apr 1, 1970.

 1279      Matthews, B. J., and Kemp, R. F., "Holographic Determination of
          Injected Limestone Distribution in Unit 10 of the Shawnee Power
          Plant", TRW Systems Group, Redondo Beach Calif., TRW Report No.
          14103-6001-RO-OO under NAPCA Contract CPA 70-4 (Jun 1970).

 1285      Paulus, H. J., and Peterson, C. M. , "Urban Aerosols: Count Size
          Related to Meteorological Data", School of Public Health, University
          of Minnesota, Minneapolis, Minnesota, Final Report (Nov 1969).

 1299      Knight, G., "A Simple Method for Determining Size Distribution of
          Airborne Dust by Its Settling Velocity", paper presented at the
          American Industrial Hygiene Association Annual General Meeting,
          Detroit, Mich., May 19, 1970.

 1331      Todd, W. F., Hagan, J. E., and Spaite, P. W., "Test Dust Preparation
          and Evaluation", Public Health Service, U. S. Department of Health,
          Education,and Welfare, Cincinnati, Ohio.

 1332      Graham, A. L., and Hanna, T. II., "The Micro-Particle Classifier",
          Ceramic Age, (Sep 1962) .

 1333      Fuchs, N. A., The Mechanics of Aerosols, Published by the MacMillian
          Company, New York, N.Y. (1964).

 1334      Anon., "Threshold Limit Values of Airborne Contaminants for 1968",
          American Conference of Governmental Industrial Hygienists (1968).

 1335      Lippmann, M., and Harris, W. B., "Size-Selective Samplers for Estimat-
          ing Respirable Dust Concentrations", Health Phys., 8:155-163 (1962).

 1340      Marple, V.  A., "A Fundamental Study of Inertial Impactors", Ph. D.,
          Thesis, Dept. of Mechanical Engineering, University of Minnesota
          (Sep 1970).

1341      Dyment, J.,  "Use of a Goetz Aerosol Spectrometer for Measuring the
          Penetration of Aerosols Through Filters as a Function of Particle
          Size",  Aerosol Science, V. 1, p. 53-67 (1970).

1342      Khmelevtsov, S. S., "A Size-Separation Collector for Sampling Aerosols
          From Curvilinear Flow", Dept. of the Army, Fort Detrick, Frederick,
          Maryland, Clearinghouse No. AD 678 123 (Nov 1967).

-------
                                                                         78
1343      Redkin, J. N., "Properties of Atmospheric Aerosol Measured with  a
          Centrifugal Spectrometer", Journal of Geophysical Research,  V. 75,
          no. 18, (Jun 1970).

1344      Whitby, K. T., Husar, R.,  McFarland,  A.  R.,  and Tomaides, M.,
          "Generation and Decay of  Small Ions", University of  Minnesota,
          Minneapolis, Minnesota.,  Prepared for National Air Pollution Control
          Admin., under USPHS  Research Grant No. AP 00136-08 (Jul  1969).

1354      Lippmann,  M., and Kydonieus, A.,  "A Multi-Stage Aerosol  Sampler  for
          Extended Sampling Intervals", American Industrial Hygiene Association
          Journal, p. 730-737  (Nov  - Dec 1970).

1357      Herdan, G., Small Particle Statistics, 2nd ed., Academic Press Inc.,
          New York,  N.Y. (1960).

1361      Israel, G. W., and Whang,  J. W.,  "Characteristics of Aerosol Beams",
          paper presented at Second International  Clean  Air Congress,  Washington,
          D. C., Dec 6 - 11, 1970.

1362      Dahneke, B. E., and  Friedlander,  S. K. ,  "Velocity Characteristics  of
          Beams of Spherical Polystyrene Particles",  Aerosol Science,  V. 1,
          p. 325-339 (1970).

1363      May, K. R., "The Cascade  Impactor: An Instrument for Sampling  Coarse
          Aerosols", Journal of Scientific  Instrumentation, V. 22, p.  187  (1945).

1364      Wells, B.  J., "An Evaluation of the May Type of Cascade  Impactor",
          Health Physics, V. 13, p.  1001 (19fa7).

1365      Sonkin, L. S., "A Modified Cascade Impactor",  Journal of Industrial
          Hygiene and Tox.„ V. 28,  p. 269 (1946).

1366      Laskin, S., In C. Voegtilin and H. C. Hodge, eds. Pharmacology and
          Toxicology of Uranium Compounds,  V. 1, p. 463-505, McGraw-Hill,
          New York (1949).

1367      Wilcox, J. D., "Design of a New Five-Stage Cascade Impactor",  A.M.A.
          Arch. Ind. Hyg. & Occup.  Med., V. 7,  p.  376 (1953).

1368      Ranz, W. E., and Wong, J.  B., "Impaction of Dust and Smoke  Particles",
          I & E Chem., V. 44,  p. 1371  (1952).

1369      Ranz, W. E., and Wong, J.  B., "Jet Impactors for Determining the
          Particle-Size Distribution of Aerosols", A.M.A., Arch. Ind.  Hyg. Occup.
          Med., V. 5, p. 464 (1952).

-------
                                                                         79
 1370      Stern, S. C., Zeller, H. W., and Schekman, A. I., "Collection
          Efficiency of Jet Impactors at Reduced Pressures", I & EC
          Fundamentals, V. 1, p. 273 (1962).

 1371      Davies, C. N., Aylward, M., and Leacey, D., "Impingement of Dust
          from Air Jets", A.M.A. Arch. Ind. Hyg. Occup. Med., V. 4, p. 354
          (1951).

 1372      Mercer, T. T., and Chow, H.Y., "Impaction from Rectangular Jets",
          Journal of Coll. and Interface Science, V. 27, p. 75 (1968).

 1373      Pilcher, J. M., Mitchell, R. I., and Thomas, R.  E., "The Cascade
          Impactor and Particle-Size Analysis of Aerosols", Proc. of 42nd
          Annual Meeting of Chemical Specialties Manufacturers Association,
          Inc., Dec. 6-7, 1955.

 1374      McFarland, A. R., and Zeller, H. W., "Study of a Large-Volume Impactor
          for High-Altitude Aerosol Collection", General Mills, Inc., Electronics
          Division, Report No. 2391, Contract AT(11-1)-401 (1963).

 1375      Zeller, H., "Large Volume Impactor Collector", Applied Science Division,
          Litton Systems, Inc., St. Paul, Minn., Report No. 2893, Project No.
          89125 (1965).

 1376      Andersen, A. A., "New Sampler for the Collection, Sizing, and
          Enumeration of Viable Airborne Particles", Journal of Bact., V. 76,
          p. 471 (1958).

 1377      Andersen, A. A., "A Sampler for Respiratory Health Hazard Assessment",
          American Industrial Hygiene Association Journal, V. 27, p. 160 (1966).

 1378      Mercer, T. T., and Stafford, R. G., "Impaction from Round Jets", Ann.
          Occup. Hyg., V. 12, p. 41 (1969).

1379      McFarland, A. R., and Husar, R. B., "Development of a Multistage In-
          ertial Impactor", University of Minnesota Particle Technology Laboratory
          Publication 120 (1967).

1380      Mercer, T. T., Tillery, M. I., and Newton, G. J., "A Multistage, Low
          Flow Rate Cascade Impactor", Aerosol Science, V. 1, p. 9 (1970).

1383      Hounam, R. F., "The Cascade Centripeter", United Kingdom Atomic
          Energy Authority Research Group, Harwell, Berkshire, England (1964).

1390      Husar, R. B., "Coagulation of Knudsen Aerosols", Ph.D. Thesis, University
          of Minnesota, March 1971.

-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO. 2.
APTD-1524
4 TITLE AND SUBTITLE „, , c ,-, , , i nTi
State of the Art: 1971
Instrumentation for Measurement of Particulate
Emxssions from combustion sources, Volume III:
Particle Size
,,UTHOR( Qilmore J. Sem; ; John A. Borgos;
Kenneth T. Whitby; & Benjamin Y.H. Liu
9. PERFORMING ORG-XNIZATION NAME AND ADDRESS
Thermo-Systems Inc.
| 2500 North Cleveland Avenue
St. Paul, Minnesota 55113
1.2. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
National Environmental Research Center
Research Triangle Park, North Carolina 27711
3. RECIPIENT'S ACCESSI ON- NO.
5. REPORT DATE
July 1972
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
1AA010
11. CONTRACT/GRANT NO.
CPA 70-23
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES .
Volume I was issued as APTD-0733
Volume II was issued as APTD-0734
       Volume III (this volume) discusses candidate techniques for automatic
  or semi-automatic measurement of particle size distribution in combustion source
  effluents.  Automatic or semi-automatic particle size measuring instruments do
  not yet exist for this application.  This report considers the application to
  effluent streams of particle size measuring instruments used in other fields.
  The discussion emphasize the particulate concentration parameter (mass, number,
  surface area, etc.)  which each technique senses as well as the method of classify
  ing particles into size ranges (aerodynamically, electrostatically, optically,
  '•'• „s.  Included are descriptions of the basic operation of each technique,
  Discussions of limitations of each technique, suggestions of possible major
  problems in applying each technique to effluent streams and an overall evaluation
  of each technique relative to others.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
Particulate Emissions from Combustion Sour
Size Distribution Instruments
Mass Measurement Instruments
,
13 DISTRIBUTION STATEMENT
Release Unlimited
b. IDENTIFIERS/OPEN ENDED TERMS
ces
19. SECURITY CLASS (This Report/
Unclassified
20. SECURITY CLASS (This page)
Unclassified
c. COS AT I Field/Group

21. NO. OF PAGES
85
22. PRICE
EPA Form 2220-1 (9-73)

-------