State
Environments •
Agency
Indus'-

Resea
                         -
Evaluation
of the PILLS  IV
Interagency
Energy/Environment
R&D Program Report

-------
                RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into nine series. These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology.  Elimination of traditional grouping  was consciously
planned to foster technology transfer and a maximum interface in related fields.
The nine series are:

      1.  Environmental Health Effects Research
      2.  Environmental Protection Technology
      3.  Ecological Research
      4.  Environmental Monitoring
      5.  Socioeconomic Environmental  Studies
      6.  Scientific and Technical Assessment Reports (STAR)
      7.  Interagency Energy-Environment Research and Development
      8.  "Special" Reports
      9.  Miscellaneous Reports

This report has been assigned to the INTERAGENCY ENERGY-ENVIRONMENT
RESEARCH AND DEVELOPMENT series. Reports in this series result from the
effort funded under the 17-agency Federal Energy/Environment Research and
Development Program. These studies relate to EPA's mission to protect the public
health and welfare from adverse effects of pollutants associated with energy sys-
tems. The goal of the Program is to assure the rapid development of domestic
energy supplies in an environmentally-compatible manner by providing the nec-
essary environmental data and control technology. Investigations include analy-
ses of the transport of energy-related pollutants and their health and ecological
effects;  assessments of,  and development of, control technologies for energy
systems; and integrated assessments of a wide range of energy-related environ-
mental issues.
                           REVIEW NOTICE

 This report has been reviewed by the participating Federal Agencies, and approved
 for publication. Approval does not signify that the contents necessarily reflect the
 views and policies of the Government, nor does mention of trade names or commercial
 products constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia  22161.

-------
                                 EPA-600/7-78-130
                                             July 1978
        Evaluation
   of the PILLS  IV
                by

  William E. Farthing and Wallace B. Smith

       Southern Research Institute
       2000 Ninth Avenue, South
      Birmingham, Alabama 35205
        Contract No. 68-02-2131
             T.D. 11007
      Program Element No. EHE624
  EPA Project Officer: William B. Kuykendal

 Industrial Environmental Research Laboratory
   Office of Energy, Minerals, and Industry
     Research Triangle Park, NC 27711
            Prepared for

U.S. ENVIRONMENTAL PROTECTION AGENCY
   Office of Research and Development
        Washington, DC 20460

-------
                        TABLE OF CONTENTS


                                                               Page
List of Figures	iv

Acknowledgments 	  v

Summary 	  1

SECTIONS

     I - Introduction 	  2

    II - Background	4

   III - Description of the PILLS IV	10

    IV - Experimental Investigation of the PILLS IV
         Response Function  	 14
                         t
     V - Theoretical Investigation  	 23

    VI - Theoretical Versus Observed Response 	 40

   VII - Conclusions and Recommendations	43

References	45
                               111

-------
                             FIGURES

Number                                                         Page
   1   The Behavior of the PILLS IV Compared to an
       Inertial Cascade Impactor When Sampling Fly Ash	    3

   2   Mean Values of the Scattering Functions of Single
       Particles vs.  Particle Size.   The Scattering Angles
       are 75° to 105° and 20° to 40°	    5

   3   Scattered Intensity as Function of Scattering Angle
       in the Plane of Polarization,  m = index of refrac-
       tion 	    6

   4   Ratio of Scattering Cross Sections for the Refrac-
       tive Indices of Figure 3	    8

   5   The PILLS IV Optical Particle Counter	    9

   6   A Cross Section of the Idealized PILLS IV View
       Volume	   11

   7   Counting Logic of PILLS IV	   12

   8   Aerosol Generation and Sampling System for PILLS  IV
       Evaluation.   Note:  The Diameter of the Sampling
       Tube at Point E is Varied According to Desired
       Velocity	   15

   9   Comparison of  PILLS IV and Climet Responses to
       0.357 ym Latex Particles	   16

  10   Comparison of  PILLS IV and Climet Responses to
       0.760 ym Latex Particles	   17
   ^

  11   Comparison of  PILLS IV and Climet Responses to
       1.099 ym Latex Particles	   18

  12   Comparison of  PILLS IV and Climet Responses to
       2.02 ym Latex Particles at Low Concentration	   19

  13   Comparison of  PILLS IV and Climet Responses to
       2.02 ym Latex Particles	   20

  14   Comparison of  PILLS IV and Climet Responses to
       2.77 ym Latex Particles	   21
                               IV

-------
                             FIGURES (Continued)

     Number                                                       Pae
        15   The  Reference  Function of the PILLS IV Versus the
            Ratio  R.   It determines the counting volume for
            particles  of any diameter D.  The top scale gives
            the  equivalent diameter for any R-value programmed
            into the instrument ..............................   27

        16   cJii»/acii» and R Versus D for a Refractive Index of
            1.33-O.Oi.  The point symbols give Oii»/aci<» ......   29
        17   CTn/acii*  and R Versus D for a Refractive Index of
            1.50-O.Oi.   The point symbols give an,/acii» ......   30

        18   CTii»/acii»  and R Versus D for a Refractive Index of
            1.59-O.Oi.   The point symbols give am/aci* ......   31
  8      19   an»/acii»  and R Versus D for a Refractive Index of
            1.50-O.li.   The point symbols give an»/acii» ......   32
  9
        20   Om/acii,  and R Versus D for a Refractive Index of
            1.50-0.5i.   The point symbols give a^/acii* ......   33

        21   Cnj/acii,  and R Versus D for a Refractive Index of
 12          1.96-0.66i.   The ptoint symbols give am/adi, .....   34

        22   The  Counting Volume of the PILLS IV Expressed in
            Terms  of  am/acii*/  D, Dc, and ADC for R < 1.1 and
            in Terms  of  Oii,/acii» for R > 1.1 .................   38

        23   A (Jn/aciif Versus D Relationship that would Produce
            the  PILLS IV Data from the Impactor Data in
            Figure 1 .........................................   41
17


18


19


20


21

-------
                         ACKNOWLEDGMENTS

     This work was carried out with the assistance in the labora-
tory of Douglas R. Sittason.  Gratitude is also expressed for the
helpful criticisms of Joseph D. McCain, Larry G. Felix, and
Kenneth M. Gushing at Southern Research Institute.  The theoreti-
cal evaluation of the PILLS IV could not have been carried out in
its present form without the information supplied by Dr. G.
Kreikebaum, formerly at Environmental Systems Corporation  (ESC)
in Knoxville, Tennessee, and Dr. H. W. Schmitt at ESC.

     We also appreciate the guidance of our Project Officer,
Mr. D. Bruce Harris.
                               VI

-------
                             SUMMARY

     The operating characteristics of the PILLS IV in situ par-
ticle sizing instrument have been investigated theoretically and
experimentally.  The results of both types of work show large
errors in this instrument's ability to size particles.  Attempts
to correlate the experimental findings with qualitative theoreti-
cal explanations have been successful.  This investigation estab-
lished a sensitivity to particle refractive index and detector
response that seems to account for the observed characteristics
of the instrument.  Further measurements would be required to
test this explanation quantitatively.

     The prototype device, an extension of the PILLS  (Particulate
instrumentation by Laser Light Scattering) technology to fine
particles, was designed to measure particle size using the ratio
of intensities of light scattered from a particle at two small
angles (14° and 7°) with respect to an incident laser beam.  The
intensity ratio was chosen as the sizing parameter because of its
relative independence of particle refractive index.  However, the
magnitude of the scattered intensity at 14° is also used for several
important decisions in the electronic processing logic, which, for
this particular optical system, render it especially sensitive
to refractive index and detector variations for determinations of
particle size distribution.  Possible solutions to these problems
with only minor hardware changes are offered.

-------
                            SECTION I

                          INTRODUCTION
     The PILLS* IV is a prototype instrument designed to provide
on-line, in situ determinations of particulate size distribu-
tions in process gas streams.  This study of the PILLS IV was
undertaken to investigate discrepancies in a set of previous
measurements1 in which this device was simultaneously tested with
a cascade impactor.  In that work the size distribution of resus-
pended fly ash in a wind tunnel was obtained with both devices.
The results of those tests are illustrated in Figure 1 where the
number frequency per cm3 is plotted versus particle diameter (D)
on semilog scales.  The number frequency, AN(D)/AD, is the number
per cm3, per diameter increment.  The study reported here was
performed to determine if modifications to the PILLS IV to elim-
inate such discrepancies were feasible.

     An evaluation program was carried out in which the PILLS IV
response was investigated experimentally and theoretically.  In
the experimental work, monodisperse aerosols of PSL spheres were
sampled and checks of the instrument hardware were performed,
thus eliminating several equipment related malfunctions as the
source of the observed discrepancies.  In the theoretical work
mathematical expressions were developed for the counting rate of
each channel as a function of the aerosol light scattering prop-
erties.  Upon analysis of these expressions, a basic design
problem involving a counting criterion was found to be capable
of producing the large errors indicated in Figure 1.  In fact,
the PILLS IV data in that figure can be explained quantitatively.
Recommendations that could provide acceptable measurements of
size distributions of polydisperse aerosols are given at the
end of this report.

     In Section II, background information related to our theo-
retical study is presented.  Then in Section III the relevant
components and operation of the instrument are described.  Sec-
tion IV summarizes the experimental work performed in this study.
Section V gives a theoretical description of the instrument's
response; Section VI gives a comparison of this theoretical re-
sponse to the observed response; and Section VII gives our recom-
mendations to anyone performing measurements with the PILLS IV.

*Particulate Instrumentation by Laser Light Scattering developed
by Environmental Systems Corporation, 1212 Pierce Parkway,
Knoxville, TN 37921.

-------
a

<
                                                      Impactor

                                                   O  PILLS IV
          Figure  1.
          D(ym)

The Behavior of the PILLS IV Compared to an
Inertial Cascade Impactor When Sampling Fly
Ash.  (After Gooding.1)

-------
                           SECTION II

                           BACKGROUND
     Optical techniques provide the potential for particle sizing
systems where representative data can be obtained in real time
with no perturbation of the aerosol.  The PILLS IV is one such
instrument.  However, interpretation of optical data to obtain
size is not only difficult but sometimes restricted by ambigu-
ities, as illustrated in this section.

     When a beam of light is incident upon a particle, part of
the light is scattered and part is absorbed.  The nature of the
scattered light is dependent upon the particle size and other
parameters, including chemical composition, which is characterized
in optical work through the refractive index m (a complex number
expressed by m = n-in1).  Figure 2 illustrates this relationship.
The average, scattered intensity at angles of 20-40° and 75-100°
from the beam direction is plotted as a function of particle size
for various refractive indices.  The imaginary part of the re-
fractive index, n1, determines the amount of absorption by the
particle, which reduces the overall level of scattering.  Of
course such sensitivity to particle composition is undesirable
for particle size measurements, so methods have been sought to
minimize that effect.

     At angles close to the incident beam the scattered wave is
determined mainly by the diffracted wave, which passes outside
the particle, and the wave emanating from the particle itself.
For a large, homogeneous particle this latter wave is influenced
by diffraction also, to some extent, but unless |m-l| « 1,
refraction at the particle-medium boundary is the dominant factor.
If absorption occurs, the refracted wave passing through the par-
ticle is attenuated.  The combined wave observed in the far field,
the result of adding the wave passing around the particle to that
passing through it, is thus determined mainly by diffraction at
small angles, and refraction at larger angles.  The above dis-
cussion leads to the well known characteristic that scattering at
small angles (within the forward lobe) is less sensitive to re-
fractive index that at larger angles, as illustrated in Figure 3,
where scattered intensity for spheres is plotted versus the scat-
tering angle 0 and two very different refractive indices.

-------
    10*
CO
^  103
UJ
Ul
a:
2  io*
O

u.
CO
   100
                   / 20° TO 40°
                     7S°TO10S°
                               /   /A
                            1. REFRACT INDEX 1.33 - 0-i
                            2.             1.50 - 0-i
                                          1.58 - 0-i
                            4.             1.50 - 0.1-i
                            5.             1.33 - 0.1-i
                            &             1.95 - 0.66-i
                                    I  I  I I
                                             10*
                                             103
       102
                                             10°
       0.1
     0.2    0.3  0.4    0.6  0.8 1.0

            PARTICLE RADIUS infim
2.0    3.0

 3630-218
Figure 2.
Mean  Values of the Scattering  Functions of Single
Particles vs.  Particle Size.   The Scattering Angles
are  20° to 40° and 75° to 105°  (after  Quenzel2).

-------
      103
      102
   I
      0.1
         F	1	1	1	1	1    I     I     I   =
                   DIAMETER - 1.0 jim, X - S14.5 nm
                               n- 1.96-0.661
                       I  *
             20   40
          60   80   100  120   140   160   180
                  SCATTERING ANGLE, degraes
                                                3630-266
Figure  3
Scattered Intensity as a Function of Scattering
Angle  in the Plane  of Polarization,  m =  index
of refraction.  X = wavelength of incident radia-
tion.   (After Gravatt.3)

-------
     Hodkinson ** was the first to point out that although refrac-
tive index does have some effect on the magnitude of scattering
in the forward lobe, the shape of the lobe does not change.  Thus
the ratio of scattered intensities at two angles within the for-
ward lobe, while sensitive to particle size, is practically in-
sensitive to refractive index.  Figure 4 illustrates this point
in a plot of the ratio versus the dimensionless size parameter
ot(=TTD/A) for two refractive indices.

     Gravatt,3 Shofner et al,5 and Chan6 have developed prototype
systems for particle sizing that are based on the intensity ratio
concept of Hodkinson.  Figure 5 is a schematic of Shofner's sys-
tem, the PILLS IV.  The intensities of the scattered light pulses
at the angles 61 and 62 are normalized to the reference pulse at
6=0° for synchronization and to account for fluctuations
in intensity of the laser source.  The optics and sensors are
kept clean and cool by the use of a purge air system.

     The laser used in the PILLS IV is a semiconductor junction
diode  (X = 0.9 urn).  The useful size range for particle sizing is
from 0.2 to 3.0 ym.  Shofner states that the view volume of his
system is approximately 2 x 10"7 cm3.  The upper concentration
limit for single particle counters is determined by the require-
ment that the probability of more than one particle appearing in
the view volume at a given time be much less than unity.  For
Shofner's system this requirement and the dimensions of the view
volume would set the concentration limit at approximately 106 par-
ticles/cm3, a value much higher than for conventional single par-
ticle counters.

-------
00
                                        m =  1.55  -  O.Oli
                                            m =  1.96  -  0.66i
                                                        D = 5.0 ym
                                                        for X = 0.6328 urn
                   D » 0.5 urn
                   for X * 0.6328 ym
               Figure  4.  Ratio of  Scattering  Cross  Sections  for the Refractive
                          Indices of  Figure  3.

-------
<£>
      CONTROL    LASER AND
      ELECTRONICS OPTICS
                                                OPTICS*
DETECTOR,
AMPLIFIER
                              SAMPLING
                              VOLUME
                                                                                    SIZE
                                                                                    ANALYSIS
                                                                                    CIRCUIT*
                                 READ-OUT
                      DESIGN-ESC PROPRIETARY

                     *• DIGITAL PRINT-OUT
                      PAPER OR MAGNETIC TAPE
                      CRT HISTOGRAM
                      LED DISPLAY OF COUNTS
                      MINI-COMPUTER INPUT
                      RATE METER
                                               TYPICAL SIZE RANGE-0.2-3.0Mm
                   Figure  5.  The PILLS  IV Optical  Particle Counter.    (After  Shofner
                                et al.5)

-------
                           SECTION III

                   DESCRIPTION OF THE PILLS IV


A. Basic Optical Components

     The light source of the PILLS IV is a GaAs laser diode with
a lens system that focuses the beam to a very small cross-sec-
tion on the order of tens of microns in diameter.  Such a beam
is normally described as a Gaussian beam because the intensity
profile perpendicular to the beam direction is given by

          I = I0e-2(r/B«>2     ,                           (1)

where r is the distance of the point of interest from the beam
axis and BO is a measure of the beam width at its focus.


     There are three photo detectors at angles of 14°, 7°, and
0° from the incident beam.  The detector at 0° is used to moni-
tor the undeviated beam to establish a reference for the 14° and
7° detector signals.  The collection optics of the 14° and 7°
detectors are of the annular design so that the detector fields
of view illustrated in Figure 6 form hollow cones of thickness
2b with the beam passing through the common apices of the cones
and along their common axis.  It is important to note that al-
though the beam is drawn with a finite width of 2B0, significant
levels of radiation (determined by Equation 1) propagate outside
of this column.  The width of the beam as determined from measure-
ments by the manufacturer is 46 ym for 2B0 (I/Io = 0.5 for r =
13.5 ym) and the width of the detector fields of view are approx-
imately 200 ym (for 2b).

B. Counting Scheme

     As the laser is pulsed  (103 times/sec), the 14° and 7°
detectors respond with signals il(, and i7 to the scattered light.
A baseline for iii»  (or i?) is established by background sub-
traction in which the average of iii»  (or i?) for the two pulses,
before and after the pulse of interest/ is subtracted from in,
(or i?).  The background subtraction technique is based on the
assumption that if a particle of interest was in the sizing vol-
ume daring a pulse of the beam, then no particle(s) of significant
                               10

-------
                 -Bo
Bo
                                                              to reference
                                                                  detector
Figure 6.  A Cross Section of the Idealized PILLS  IV  View Volume
           The beam has a Gaussian intensity profile

           (i.e., 1= Ioe-22,.

-------
                      111.      1?
                       1
            Lref
            1
                 Referencing i? & iiu to
                   Subtraction of background
                           R =
     Large Particle Branch

                         T
                       Small Particle Branch
      Calculate i   (R)
(0.2  i   (R)  > ii,  > i   (R)L_E.
x"    GII,              Ci * ^S

^
T
f
Locate j where
R. - R - R. + AR.

              I
        increment j
        channel by 1
rejected
                          Locate k where
                                      )k
                                  c 1 1| K
                                          (i   ).  > i   ^  (i   )  + A(i
                                            c 1 4 K    1 1,
                                                           I
                          increment k
                            channel by  1
             Figure  7.   Counting Logic of PILLS IV,
                                     12

-------
 size were  in  the  sizing  volume on either the preceding or  follow-
 ing pulses.   That assumption is valid when the aerosol concentra-
 tion is  below the limit  discussed at the end of Section II.   The
 counting logic after  background subtraction and referencing  of  in,
 and i? to  the beam monitoring detector is depicted in Figure 7.
 If the ratio  i7/illt,  denoted by R,  is greater than 1.1, then a  par-
 ticle size is inferred from its value.  If the ratio is less than
 1.1, the parameter ilt| is  used to determine particle size.   Note
 that the PILLS IV uses the inverse of the parameter plotted  in
 Figure 4.

     A further criterion is invoked for the instrument to  count
 a particle whose  detector  scattering ratio is above 1.1.   The
 measured signal in, must be in the range

           0.2iCii»(R)   1  iii 1 ici-. (R)                        (2)

 where iCm(R)  is  a predetermined function of R.   The purpose of
 this criterion is two- fold.  It attempts to provide a definition
 of the instrument sizing volume by eliminating counts when par-
 ticles are near the edge of or outside of the 14°  detector view
 volume but well within the 7°  detector view.   It also prevents
 the sizing of small particles  as larger ones due to large  errors
 in R produced by  system  noise.   If in, produced by a particle
 with ratio iy/iiu  satisfies the criterion then the instrument
 adds one increment in the  size channel corresponding to the  ob-
 served intensity  ratio.  This  criterion is not applied to  signals
 for which  R is less than 1.1.

     The output of the PILLS IV is the number of particles per
 channel  in two series of channels,  one series corresponding  to
 values of  R and one series to  values of in,.   The  upper and  lower
 limits of  each channel can be  varied by the operator.   When  R > 1.1
 the channel number to be incremented is determined by the  value of
 R; that  is,
          R> 1 R 1 R- +

where j denotes the channel to be  incremented.  When  R  <  1.1  the
channel number to be incremented is determined by  the value of
iii,; that is,

          (-iciOk <_ in* <_  (icn»)k  + A(icn»)k                 (4)

where k denotes the channel to be  incremented.  The operator  then
relates the channels, j or k, to calibration tables for particle
sizes.
                                13

-------
                           SECTION IV

EXPERIMENTAL INVESTIGATION OF THE PILLS IV RESPONSE FUNCTION


     An aerosol generation system was developed to evaluate the
PILLS IV empirically.  The system, depicted in Figure 8, consists
of a nebulizer* containing a suspension of uniform latex spheres,"*"
a drying chamber, rectangular ducts to and from the PILLS IV probe,
and an optical particle counter**  with diluter.  The rectangular
ducts were constructed with the same shape and size as the opening
in the probe which contained the sensing region.  The entire aer-
osol stream flowed downward to eliminate settling losses.  The
aerosol stream exiting the PILLS IV probe was sampled isokinet-
ically by a particle counter or filter for independent determina-
tion of particle concentration.

     Sampling experiments were performed with the PILLS IV purge
air lines stopped and an aerosol stream velocity of 25 cm/sec.
Velocity measurements made in the duct with a thermal-anemometertt
demonstrated a flat profile across the duct within ±10%.  The
nebulizer flowrate of 10 LPM introduced 0.61 cm  of the latex
sphere water suspension per minute into a total air flow of 50
1/min.  This gave a residence time of 8 seconds before particles
passed through the PILLS IV.  For each size of PSL spheres, aer-
osols of two concentrations were produced.  One was low in par-
ticle concentration for calibrating the dilutor, and one high in
concentration to evaluate the PILLS IV.

     Figures 9-14 give aerosol size distributions in terms of
AN/AD as derived from the PILLS IV and the Climet Instrument in
 * Retec X70/N; Civitron Burton Division, Retec Development
     Laboratory, Van Nuys, CA 91406.

 t Polystyrene (0.357 ym, 760 ym, or 1.099 ym dia.), Poly-
     vinyltoluene (2.02 ym dia.), or Styrene/Vinyltoluene (2.77 ym
     dia.); Dow Diagnostics, P.O. Box 68511, Indianapolis, IN 46268.

** Climet Instruments 0208A Particle Analyzer; Climet Instruments
     Co., Redlands, CA 92373

tt Sierra Instruments Model 440/441; Sierra Instruments, Inc.,
     P. O. Box 909, Carmel, CA 93924.
                               14

-------
          10:1 Scale
                Charge
                Neutralizes
10 LPM Clean, Dry Air  -*.
to Nebulizer
   PSL Spheres in Wa
   Suspension from a
   Syringe Pump
   Pressure Measurements
   at Points A,  B, and C
   Velocity Traverses
   at Points D and E
                                      Additional Air When Desired
  Aerosol Dumped
            ^Hollow Cylinder
            P (2.5 cm dia.)
              inside silica gel
              drying agent
                                                       40 LPM Clean, Dry Air
                                                 PILLS IV Probe
                                   Diluter (absolute filter with nicropipets)
              7 LPM to Optical "Particle Counter
      Figure  8.
Aerosol Generation and  Sampling  System for PILLS IV
Evaluation.   Note:   The Diameter of the Sampling Tube
at  Point E  is Varied According to Desired Velocity to
Preserve Isokinetic Sampling.
                                    15

-------
3.
^>

Q
                                         Total PILLS IV Counts

                                             O  181

                                             O  610

                                                Climet Instrument
        10Z r
        10
       Figure 9.
Comparison of PILLS IV and Climet Responses to
0.357 urn Latex Particles  (back to back runs
with 2.0 x 10' particles/cm8 in nebulizer solution)
                               16

-------
        105 C
        10"  -
Q
<

2
                                  O  Total PILLS IV Counts —21

                                     Climet Instrument
Figure 10. Comparison of PILLS IV and Climet Responses to
           0.760 ym Latex Particl

           in nebulizer solution) .
                   0.760 ym Latex Particles. (2.7 x 109 particles/ cm
                             17

-------
       10

        10'
        10
        10 =
        10
                                   O  Total PILLS IV Counts —536

                                        *—« Singlets on  filter

                                      i    « Doublets on  filter

                                      	 Climet  Instrument
Figure 11. Comparison of PILLS IV and Climet Responses to 1.099 ym
           1.099 Mm Latex Particles.
           (1.12 x 109 particles/cm3 in nebulizer solution.)
                               18

-------
           io3cr
m

'B
Q
<

25
           io2  -
                                      O  Total PILLS IV Counts—14

                                               Climet Instrument
        Figure 12,
Comparison of PILLS IV and Climet Responses to
2.02 ym Latex Particles at Low Concentration.

(1.43 x 107  particles/cm3  in nebulizer solution.)
                                 19

-------
B
Q
<
\
Z
                     Climet  Instrument
                                          Total PILLS IV COUNTS

                                                   O  17
                                                   D  30
                                         —«  Singlets on filter
                                         —•  Doviblets on filter
        Figure 13.
Comparison  of PILLS  IV and Climet Responses to
2.02  ym Latex Particles, (back to back runs

with 1.43 x 109 particles/cm*  in nebulizer solution.
                              20

-------
        105C
m
I
                                      Total PILLS IV Counts —456
                                      Singlets on filter
                                      Doublets on filter
                                      Climet Instrument
         Figure  14.
Comparison of PILLS IV and Climet Responses to
2.77 ym Latex Particles.
(1.11 x 109 particles/cm9 in nebulizer solution.)
                               21

-------
conjunction with a multi-channel analyzer.  In some cases, the
concentration of PSL spheres was also determined from filter
catches.  The PILLS IV sizing volume was assumed to be that given
by the manufacturer, 2 x 10   cm3.

     The general features of the particle size distributions ob-
tained with the Climet Instrument are as one might expect from
the aerosol generator.  There is a high concentration of parti-
cles counted around the diameter of the nebulized latex particles.
The presence of doublets is usually observable as distinct peaks
of considerably lower concentration.  There are also substantial
concentrations of particles apparently produced by dissolved wet-
ting agents in the stock solution in which the latex particles
must be stored.7  When a droplet from the nebulizer is dried the
wetting agent forms a particle of unknown structure with light
scattering properties which makes most of them appear to be be-
low 0.3 Mm as sized by the Climet instrument.  This explanation
also agrees with measurements in our laboratory where a stock
solution from which the latex particles had been filtered was
nebulized and dried.  The number of "small" particles seen by
the Climet was unaffected.  Further, it was found that dilution
of the stock solution with water reduced the number of small par-
ticles detected by the Climet instrument.

     The data in Figures 9-14 show two major characteristics of
the PILLS IV relative to the Climet Instrument.  Its responses
to PSL spheres with diameters of 2.77 ym (Figure 13)  and 1.099
ym (Figure n) are comparable to that of the Climet Instrument,
while its responses to the other PSL spheres, 2.02 ym (Figures
12 and 13), 0.760 ym (Figure 10) , and 0.357 ym (Figure 9) are one
to two orders of magnitude low.  Such responses to latex parti-
cles are qualitatively explained by the theoretical response
function given in the next section. 'Sizing of the three largest
size particles is performed by the ratio  i?/iii» with the count-
ing criterion invoked on in,.  The PILLS IV response to the two
smaller sizes of latex particles depends on iii, only.  The other
major characteristic of the data given in Figures 9-14 is that
relative to the Climet Instrument, the PILLS IV frequently indi-
cates a large concentration of particles of sizes other than
that of the latex particles under study.  The origin of these
counts is not known, although a possible source is discussed in
Section VI.
                               22

-------
                             SECTION V

                     THEORETICAL INVESTIGATION
      The  results  of  the experimental investigations revealed
 large errors  in sizing particles with the PILLS IV.  However,  it
 was  not determined whether the discrepancy is due to a basic de-
 sign problem,  an  error in calibration,  or a subtle malfunction-
 ing  of a  component.   In order to determine if the design or logic
 of the PILLS  IV prototype was faulty,  a theoretical analysis of
 the  system  has been  performed.

      Equations (5) represent the total signal from the three in-
 strument  detectors on the n^*1 pulse of the beam:


            =  i   (n)  + i   (n)  + Ci^ci„£m(r,*)e~2(r/B°}I0e"T
                ei if        mi ^

       i?(n) =  i  (n)  + i'  (n)  + C7a7f 7 (r,JUe'2 (r/Bo)  I e-T    (5)
                e?        my                             o

       io(n) =  i   +  C0Ioe~T
                eo

 Of all the  parameters given,  only (Jii«  and a7,  the differential
 scattering  cross  sections at 14°  and  7°,  are directly related  to
 the  particle of interest.  The  other  parameters are defined in
 Table  1.  Through background subtraction and referencing of i 11»
 and  i? to i0,  the processed  signals,  ipm and i  ,  can be ob-
 tained with errors denoted by Ai     and Ai  :   p?

                f*
     where i    = p^i %f m (r, l)e~2 (r/Bo) *± Ai                 (6)
            Pii,  Co                           PIP

                 ^
       and i    = 7^0 7 f 7 (r,Jl)e"2 (r/Bo)   ±  Ai
           p?    Co                         P?

     Optimum performance  is  obtained  from the instrument when
Ai    = Ai   =  0,  and accurate values of  f7(r,£)/fi*(r,t),  C0,
  Pl«»     P7
C?,  and Cii» have been determined  by calibration measurements.
As described in Section III,  the  ratio  of signals i   /ipllf
and  i    are used to deduce  a particle  size and to  decide  if
     Pi"»
the particle should be counted.   In this  analysis,  the counting


                               23

-------
                               TABLE I-

                             DEFINITIONS

 i    ,  i   ,  and  i          - The portions of the  signals  due  to
       e?       e°          electronic background noise.

 i    and  i                - The portions due  to  aerosol  particles
 milj      m?                other than the one of interest on the
                            nth pulse.

 i    and i                 - The signals from  the 7°  and  14°  detect-
 P7       Pl<*                ors after background subtraction and
                            referencing to the 0° detector signal.

 Cii», C?,  and C0           - System conversion factors  that relate
                            light intensity at the detectors to
                            the signals which are processed  for
                            size evaluation.
     T
 Ioe~                      - Intensity of the  beam after  trans-
                            mission  through the  aerosol.
 -2 (r/B )2
e~     o                  - Variation of beam intensity  with posi-
                            tion in  the aerosol.  (r is  depicted
                            in Figure 6).

 f7(r,£,) and fii»(r,Z)      - Optical  transfer  functions which re-
                            late actual scattered intensity  to
                            that reaching the detector (r and  £
                            are depicted in Figure 6).  The  func-
                            tions are expected to decrease mono-
                            tonically as r and £ increase.

 a7  and am               - The differential  scattering  cross sec-
                            tion of  a particle at 7° and 14° with
                            respect  to the incident  beam.  The
                            units are those of radiant steranee,
                            radiant  flux per  unit area per unit
                            solid angle.

 o                        - A reference function programmed  into
   ciu                       the PILLS IV, aside  from an  adjustment
                            for the  detector  response, to estab-
                            lish a counting criterion which  the
                            14° detector signal  must satisfy.


                                24

-------
 scheme  is  first analyzed assuming the optimum conditions, and
 then  the changes in behavior,  which would result if those assump
 tions are  violated are discussed.  Also, it is assumed that the
 programmed function i   (R)  was  established for particles at
                      C 1 it
 the center of the view volume  where i    = (Cn,/C0)On» and
 ip? = (C7/C0)a7.                     Pl*

      Ideally,  the ratio R (= i  /i   )  is determined by 07/014.
                               p?  Pi k
 When  this  ratio is above 1.1 then the particle is counted if
 i     is  in the range 0.2i  (R)   < i   < i   (R) .   However,
 Pit                      CIH    —  Pin—  Cii»v
 since i     is a function of  r  and Si, the number  of particles hav
       PI <«
 ing a given value of R which are counted after many pulses
 depends  upon the position of each in the beam as well as the
 value of a i 4 .   In this manner  a sizing volume of the instrument
 is established.   On the other  hand, if am(R)  varies due to dif-
 ferent composition or shape  from that used in calibration to
 establish  i   (R) ,  then the  sizing volume varies, decreasing
            c 1 1»
 with  On,(R)  for  a given R.  More specifically,  the volume in
 which particles are counted  is determined by values of r and i
 for which
°'2 <   °l
-------
     We  see  from Equations  7 and  8 that the two  functions  fn,(r,£)
and On»/ocn, must be known  or otherwise accounted for to reduce
PILLS  IV data in terms of the aerosol  size distribution and con-
centration.  The most important question is whether for a  given
particle size a\*/Qci*i which is  determined by the composition of
the particles themselves, varies  enough to cause significant
changes  in the response of  the PILLS IV.

A. Variation of ai«,/acn» with Refractive Index

     The exact function oc i «» programmed into the PILLS IV  was not
available.  However, Dr. G. Kreikebaum, one of the developers,
kindly provided a description of  the procedure by which it was
obtained.  That procedure was reproduced in this study as  follows:

   * for Particles with a 7 /am >  1.1
          The differential scattering cross sections a in and a?
     proportional to the respective scattered intensities, were
     calculated using Mie theory for a broad range of refractive
     indices* which one might expect the PILLS IV to encounter
     and for the diameters 0.3 ym - 4.0 ym.t  From these, the
     function  aiw   versus R was obtained for each refractive
     index.  Then for each R-valuef aci«» (R) was set equal to
     the maximum value of a 1 1» versus refractive index for that
     value of R.  Values of R > 10 were excluded from this
     analysis since a unique size cannot be associated with
     those R values . **
PCI** for Particles with ay/a in < 1.1

          The calibration curve acii»(R) for R < 1.1 was set equal
     to GII» for a particle refractive index of 1.59 - O.Oi.

     The resulting function ocii»(R) is given in Figure 15.  The
top scale shows acll| in relation to particle diameter, D, assum-
ing the relationship between D and R predicted by diffraction
theory or by Mie theory with high absorption.  Knowledge of
     *The refractive index m is a complex number, n-in1.  The
values of n1 employed were 0.0, 0.01, 0.05, 0.1, 0.5, and 1.0.  For
n' = 0.0, n was varied  from  1.33  to  1.96  in  increments  of  0.01.
For the other values of n1,  n  values of 1.33,  1.40,  1.45,  1.50 to
1.65 in increments of 0.01 were employed.  In  addition,  calculations
for the refractive index of  carbon,  1.96-0.66i,  were included.  These
may not be the exact refractive indices used for programming the
PILLS IV;  however, the differences are believed to be insignificant.

      The calculations were carried out by varying the Mie size
parameter a(irD/X)  in increments of 0.2 from 0.9 to 13.9 where  D
is the particle diameter and X is the wavelength, 0.904 ym.

                               26

-------
o
          0.81  1.5    2.0
   Figure 15-
The Reference Function of the PILLS IV Versus the
Ratio R.  It determines the counting volume for
particles of any diameter D.  The top scale gives
the equivalent diameter for any R-value programmed
into the instrument.
                              27

-------
acn, (R) enables one  to determine  the relative response of  the
PILLS  IV to particles of differing refractive index.

     To predict au»/acii» the results of the Mie  theory were  used
again.  In addition, some calculations using diffraction theory
were performed to check the appropriateness of Mie  theory.   The
Mie theory of scattering assumes  that the incident  radiation is
an infinite plane wave, and some  differences in  the actual par-
ticle  scattering could be expected because the intensity of  the
PILLS  TV beam is Gaussian in form and the intensity of light is
not uniform across the scattering region associated with the par-
ticle.  The results of the diffraction calculations show no  sig-
nificant change in a?/cJi«» at different positions in the Gaussian
beam (B0 = 23 ym) until the particle diameter reaches 6 ym or
higher.  Thus Mie theory using the intensity of  the wave at  the
center of the particle is suitable for the particles of interest
here.

     Figures 16-21 give theoretical response functions versus
particle diameter for particles of five representative refrac-
tive indices, as expressed by the ratio ai «»/crci i» (R) .   The value
Oci«»CR)  is  given in Figure 15 using the appropriate value of R
for each D.  The ratio R versus D is also given  for these refrac-
tive indices.  When Om/ocm has  a value of one and R > 1.1, par-
ticles of that size and refractive index are counted with maximum
sensitivity; that is, the sizing  volume for these is maximum.
The maximum sizihg volume Vmax is determined by the optical-field
transfer function of the collection optics and the  beam profile
as expressed in Equations (7) -(8).  As ai i»/aci «» (R)  decreases,
the effective sizing volume decreases, since for a  smaller value
of a ii» the particle must be closer to the more intense central
portion of the beam.  When OII»/OGI<» is below 0.2, no particles
with R > 1.1 would be counted.
     The significance of o\n/ac\n for particles in which R < 1.1
is most directly related to the deduced particle size.  The
higher values of OII»/OCK» for the same position in the view vol-
ume are counted in channels of the instrument corresponding to
larger particles and two particles with the same value of
^i«i/ocm could be counted in different channels if one were posi-
tioned closer to the center of the view volume than the other.

B. The Optical Transfer Function fii»(r,&)

     The fluctuations in 
-------
     _c—i—i—i—i—i—i—i—i—i—i—i—i—i—i
             O R < 1.1
             D R > 1.1
   i-h
lo-'-f-
                                 D
                D
               D
100
                                                         10
            I  I  I  I  I   I  I  I  I  I  I   I  I   I  I   I  I
                            D(ym)
Figure 16.  om/oc   and R Versus D for a Refractive Index of
            1.33-O.Oi.  The point symbols give oJ
                         29

-------
      1  - -
O 1
              I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I



                     O R < 1.1


                     n R > 1.1
100
                   I  I   I  I   I  f   I  I   »  I   I  I  I  I  I  I  I
                                                              10
                                D(ym)
    Figure 17.  ai%/o    and R  Versus D for a Refractive Index of
                     c 11»


                1.50-O.Oi.  The point symbols give 01
                             30

-------
   1  +
 lo'1-!-
        I—I—I  I  I—I—I—I—I—I—I—I—I—I—I—HH—I—I
             O R < 1.1




             D R > 1.1
             I  I   I  I  I  I  I  I  I   I  I   I  t   I  I   I  »
100
                                                               R
                                                       +  10
                             D(ym)
Figure 18.  On,/ac   and R Versus  D  for a Refractive Index of



            1.59-O.Oi.  The point  symbols give  Oi
                         31

-------
O 1 it
-I—I—I—I—I—I—I—I—I—I—I-H—I—I—I—I—f—I





      O R < 1.1




     D R > l.l
                  I  I  I  I  I  I  I   I  I   I  I   I  I  I  I  I  I
                                                               100
                                                            -•  10
     Figure 19.  Cit/cr    and R Versus D  for a Refractive Index of
                 1.50-O.li.  The point symbols  give

                              32

-------
     f—I—I—I—I  i   I—I—I—I—I—I—I—I—I—I—I—I—I-H—|- 100




             O  R < 1.1



             a  R > 1.1
 i
     ••  o
     ..   o
10->-r
                                                        • -  10
             I  I  I   I  I   I  I   I  I  I  I  i  I  I   I  I   I
                             D(ym)
Figure 20«  0i
-------
        I—I-H—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I—I
                  O R  <  1.1
                  D R  >  1.1
i  i  t   i  i   i  i   i  i
                                      t  t  i   i  i   i  I
                                             100
                            D(ym)
Figure 21.  cr»*/0c   and R Versus D  for  a  Refractive Index of
           1.96-0.66 i.  The point symbols  give  a in/a
                         34

-------
volume while monitoring the detected intensity of scattered
light.  It was found that values of r and i for which

          0.2 < fm(r,i) (am/acm)e~2(r' B°)2 < 1
                                                              (9)
formed a cylindrical  surface  20  pm  in diameter  and  600  ym  in
length.  This  suggests  that fn,(r,£) is uniform in  the  central
diamond-shaped portion  of the view volume  and  decreases very
rapidly near the edges  of this region .  Attempts to measure the
view volume in this study were not  conclusive but were  consist-
ent with the manufacturer's results, i.e.,  fllt(r,fc) is  given by

          fii(r,£) =  1  for r  <  (b/sin 14°-£)tan 14°         (10)
                   =  0  for 0  > r  >  (b/sin 14°-£)tan 14°,

where b = 82 pm.

C. The Response Function of the PILLS IV

     With fm(r,i)t R(D) , and om (D)/oCi >» (D) known, theoretical
behavior of the PILLS IV for a given aerosol can be predicted
from Equations (7) and  (8) and the  relationship of  R as a  func-
tion of diameter D.  Equations  (7)  and  (8)  determine the count-
ing volume for each particle size with given light  scattering
properties described by a'7 (D) and Om(D).   For  channels in
which R > 1.1  the count rate is the product of  the  particle con-
centration in  the channel-size interval and the average counting
volume over the size interval.  For channels in which R <  1.1
the count rate is more  complicated, because the channel in which
a particle is  counted depends upon  its position in  the beam.
Thus particles having the same o?(D) and aii»(D)  can be counted
in different channels.  The counting volume for each particle
size depends upon the lower and upper limits of the channel of
interest.  The total count rate of  a channel is then the summa-
tion over all  size intervals  (for which R < 1.1)  of the products
of particle concentration in each size interval times the  average
counting volume over the size interval.  This may be considered
unusual behavior, but the count rate of each channel is actually
similar to a measurement of the cumulative  concentration of par-
ticles with sizes between that for  which R  = 1.1  and the size
associated with each channel.  A method for obtaining the  size
distribution from the count rate is discussed further in Section
VII.  The remainder of  this section is concerned  with character-
izing the counting volumes as a function of particle light scat-
tering properties and the channel limits.   The  most desirable
behavior is obtained when V is independent of these parameters.

     The counting volume is cylindrically symmetric about  the
optical axis.   The intersection of  its bounding  surface with any
plane parallel to the optical axis  forms a  trapezoid (see Fig-
ure 6)  whose base and top are determined by minimum and maximum

                              35

-------
values of r which  satisfy  Equations  (7)  or (8).   The size of this
volume is given by
      V =
            2TT
          tan 14°
              b(rmax'rmin)  ~
                              (rmax~rmin)
                                        (11)
where r  .  £ 0.  By  inserting  fm(r,&)  of Equation (10)  into
Equations"11 (7) and  (8)  the  values  of  rmax and rmin can be obtained
from
     B
  Info.   (R)/am(D)l  <  r2  <  -BQ2  ln|"o.2a   (R)/am (D)l(12)
  '   L ci*          J          -5—   L    ci»          J
for channels in which R  >  1.1  and  from
      B
         In
            (a   )
              Cl<»
                A (a    )
                   ci •»
                          < r2 < -
                                                       (13)
             an, (D)      am (D)

for channels in which R <  1.1.  Characterization of the PILLS IV
behavior is straightforward  for the  upper  channels governed by
Equation (12), because the channel limits,  set by the operator,
are not involved.  However,  for the  lower  channels the behavior
depends upon the channel limits.  Characterization of the instru-
ment's full potential then requires  including  the channel limits
as a parameter.

     Equation  (13) can be  put into the  same form as (12)  by not-
ing in Figure  15 that ocm for R  < 1.1  is  proportional to D5 and
by identifying particle diameters Dc -»•  Dc  + ADC from calibration
tables associated with  (ocn»)v •*•  (acii,)^ + A(aci«»)_ for channel
k.  Then we find that        K                     k
              = °c.>(Dc' = °c..(D)
          (D)
and
A(a   .),  _ a
   c i » * -
(D7

(D)
                      an, (D)
                              (H
    (D)
                 aii» (D)
Thus, Equations  (12) and  (13) actually have  the same form expres-
sed by
"Bo  In  [L(R)a   /a  1  < r2  <  "B0   In
                                            [L (R) H
                                                         (14)
where
L(R) = (DC/D)
                   AD
                           C/DC)
     H(R) =  (1 + ADC/DC)~5
                                       for  R <  1.1
                              36

-------
and L(R)  =1
                          for  R >  1.1
    H(R)  =  0.2

     Equation  (14)  inserted into  (11)  gives V as a function of
am(D)  for  the  upper  channels and as  a function of Oi4(D),  D,
and the channel limits  for the lower  channels.   V is  given  in
Figure  22 in the form of  V/Vmax where Vmax is the maximum count-
ing volume  occurring  independent  of particle light scattering
properties.  Vm=i .  given  by
              In 3.X
                              lnH'R>  -l/3--lnH(R)        , (15)
depends upon H(R),  constant  for  the  upper  channels  and variable
for the lower channels.   Note  that V approaches  zero when the
abscissa approaches H(R).  That  is when  rmax  approaches zero.

     To describe  the response  to particles with  R > 1.1 from
Figure  22, the curve for  H(R)  =  0.2  is employed.  Sizing of par-
ticles is performed from  R,  but  the  number counted, proportional
to V, depends upon  aii»/ocii».   If aii»/ocii,  is  below  one (see Fig-
ures 16-21) then  V  changes rapidly for small  changes in aiit/oci<,
Indeed, for the calibration  function acii»  presently programmed
into the PILLS IV om/ocii» is  always less  than or equal to one.
It is desirable,  however,' that if acin(R > 1.1)  were reduced so
that a ik /ac 1 1» is  always greater  than one,  V/Vmax is essentially
constant for all  refractive  indices.

     All of the curves in Figure 22  depict the variation of the
counting volume of  lower  channels.   Each channel has a curve
determined by the channel limits through H(R) .   Note that ADc/
the width of each channel, must be decreased to increase H(R).
The counting volume depends  upon the independent variable
(01 i»/oci <») (D/Dc) 5H(R) .  To facilitate understanding  of the depend-
ance of V on am/ocm and D  the  upper scales  in  Figure 22 show
how V varies with (ai i»/acn») (D/Dc)   for  specific values of H(R).
Each scale refers to  a different curve.

     As an example,  consider a monodispersed  aerosol of latex
particles where am/ocii»is 0. f or all  diameters and  Di  j_,  the
diameter at which R = 1.1, is  0.75 ym.   Then  the number counted
in the channel with associated diameters Dc -» Dc  +  ADC,  assuming
ADC/DC = 0.27, is given by the curve  labeled  H(R) = 0.3 in Fig-
ure 22.   A channel  for which Dc  = D would  receive no counts
since V approaches  0 when the  independent  variable  approaches
H(R)  or less.  The  number x of counts would  increase  for decreas-
ing Dc until Dc = D[H(R)]  /5  = 0.87D  where it would become
essentially constant for lower channels.   For a  polydisperse
aerosol of latex  spheres the counts  in any channel  would be due
to particles for which Dc 5 D  £  DI.I.  The number of counts in a
given channel increases as Dc  decreases  for two  reasons:

                               37

-------
          Abscissae for Specific Values of  H(R)
   H(R)  = 0.75


   H(R)  = 0.3


   H(R)  = 0.2



   H(R)  = o.oo:


      1   +
lO"1 1


ID"1 1
1 1

ID"1 1

1 r
1 10
n i i 	
10


10
i

10
i

102
	 1 	
102 103


102 103
1 ... . ..J

io2 10 3
1 -I
1 1
10 3
1
io"1--
 max
      10
                                          0.0031    Q.2  0.3  0.75
                                                               H(R)
                                                   3 x IO"6
                                                   9 x 10~7
                                                   7 x 10
                                                   2 x 10
                                                   -7
                                                   -7
                  0.0031
                  0.2
                  0.3
                  0.75
                             0.3
         10
           -2
               10
                       -1
               L(R)aci
                                  10          10:
                             )(D/DC)SH(R) for R
                          _  )            for R
                                                       <  1.11
                                                       >  1.1
                        H
                         103
Figure 22.  The Counting Volume of the PILLS IV Expressed in Terms
            of a in/a   , D, D , and AD  for R < 1.1 and in Terms
                    C i it      C        C
            of 0i%/cr    for R > 1.1.
                                V/V    is given  for  4  values
                                   met A,
            of H.  The top  scales  show V/Vmax versus  (0i */<*Cl 1|)
            (D/DC)
If
        for specific H-values.   — -*-,,-c
then V goes from 0 -»• Vmax as  D  goes from DC

                  38
                                                   were  one,

-------
because V/Vmax is larger for lower Dc - values, and also the range
of D-values which contribute to the channel is larger for lower
DC-values.  Inspection of the curves  shows that the one for
which H(R) = 0.75 or higher is the most desirable for each channel
since V is nearly a step function.  Then for a polydisperse aero-
sol of latex spheres the count rate of each channel would give
the cumulative concentration of particles with diameters in the
range Dc ^ D 5 DI.I.  The differences  in count rate between suc-
cessive channels would give the concentration in the size range
Dc to Dc + ADc. This approach is discussed further in Section VII.

     For particles other than latex spheres am(D)/acm(b) is not
one and it varies with D as seen in Figures 16-21.  For such par-
ticles the lower channels count particles for which
(ai i»/acii») ~/5Dc ^ D 1 D, , .  This means variation in particle type
produces some variation IA the lower sizes of particles which a
channel counts.
                               39

-------
                           SECTION VI

              THEORETICAL VERSUS OBSERVED RESPONSE


     The theoretical description of the preceeding section ex-
plains the instrument ' s observed response to fly ash in Figure 1
if the function for ai 14 (D)/aci i» (D) given in Figure 23 is chosen.
Since very little is known about the optical properties of flyash
the accuracy of this deduced function cannot be evaluated with
certainty.  In comparison to homogeneous spheres it agrees within
20% with the scattering of spheres having refractive index m =
1.5 - 0.5i.  With the exception of a very small region around
0.8 ym, it could also be constructed of a combination of refrac-
tive indices with low imaginary parts and real parts varying from
1.5 to 1.96.  Other factors effecting aiii/oci«» are particle shape
and internal structure.  It must also be recalled that the func-
tion given in Figure 23 is deduced with the assumption that the
detector constants Cii*  and Co of Equation (6) remain constant
after calibration as well as the ratio f ? (r , l)/f i •» (r , I) related
to optical alignment of the lenses.  These factors that can effect
R are especially critical for sizes around 0.75 ym.  For example,
in this region a 15% lower value of am(D)  causes R to be 18%
high.  If R > 1.1 this particular error would cause 0Cll|(R)  (see
Figure 15) to be more than a factor of 5 higher, and aiii/Dci"*
would be less than 0.2 of its correct value.  Thus, such an error
in the deduced oii*(D) would eliminate essentially all counts.  It
appears then that such an error would also cause am/acn, for m =
1.5 - O.li (Figure 19)  to appear similar to that in Figure 23.

     The response function [Equation (11)]  also qualitatively ex-
plains the response to latex spheres, determined in this study.
This is seen by comparing the values of ai«,/aci<» at the appropri-
ate diameter (Figure 18) to the response of the PILLS IV at the
five diameters studied, given in Figures 9-14.  in the case of
0.760 ym particles, it appears again as described above for fly
ash that R was deduced to be slightly too high due to a shift in
the optical constants, causing the chosen GCI<* to be so high that
      i"* was too low for counting to occur.
     The high count rate, which occurred in some instances, at
sizes other than that of the nebulized latex, can only be explained
by the presence of particles other than latex.  The light scatter-
ing by these other particles must be such that the deduced size
with the Clime t is much smaller than with the ratio 07 /am used

                               40

-------
      1 +
   io- * +
i   i  i   i  I  i  i   i

    •  R < 1.1

    •  R > 1.1
•   »  «   t  *
                                                   '  «

               i  i  I  I  i  I   I  i   I  I   I  I   I  I   I  t   i
                    123

                               D(yra)
Figure 23. A am/a,.   Versus D Relationship  that would  Produce
                  c 1 1»
           the PILLS IV Data from the Impact or Data  in  Figure 1.
           For D > 0.8, aii»/a_   is unique while many possibilities
                             Ci i»
           for D < 0.8 could explain Figure  1.
                             41

-------
by the PILLS IV-  This suggests that the high PILLS IV counts
compared to the Climet counts were due to particles formed from
wetting agents in the stock solutions in which the latex particles
are stored.  Fuch's7 discussion of these particles (called "emp-
ties"), and observations of them, indicates that very little is
known about their formation.  Apparently the size and structure
are very sensitive to humidity and concentration of the nebulized
solution.  If the effective refractive index of these particles
were close to one, the differences in the PILLS IV and Climet
data in Figures 9, 11, and 14, would be explained.  The light scat-
tering at large angles (used by the Climet) decreases relative to
small angle scattering (used by the PILLS IV) as the refractive
index decreases.  Thus, such particles would appear to be smaller
when sized with the Climet instrument.  Another explanation not
related directly to light scattering is turbulence encountered by
particles in the inlet line to the Climet.  If the "empties" are
fragile, then breakage could result to form small fragments.
                               42

-------
                            SECTION VII

                 CONCLUSIONS AND  RECOMMENDATIONS


      It  appears  that  the  major  problem  that cause the  PILLS  IV to
disagree with other sizing  devices is the  dependance of the  instru-
ment  counting volume  upon Om/acif at low  values  of  am/aci*  as
seen  in  Figure 22.  It  is interesting that as  aii»/ocii»  increases
the sizing volume reaches a maximum volume and remains  essentially
constant.  This  suggests  that the major discrepancies  could  be
eliminated if aCilt(D  or R)  were reduced so as  to  increase  oii»/aci<».

      The calculations performed in this study  for homoneneous
spheres  showed that if  OG m for R > 1 . 1 were reduced by a  factor
of 4  then am/acii* would  always be close to 1  or  greater.  Then
for R >  1.1 the  counting  volume would be constant; thus eliminat-
ing the  sensitivity to  par.ticle composition.   This improvement
would also be accompanied by a  slight increase in the  counting
volume to include more  of the fringes of the 14°  detector  view.
Particles in this portion of the  volume would  be  sized  incorrectly;
however, their contribution to  the total count would be small.

      For R < 1.1, decreasing aCait(D) would also improve the  in-
strument response; however, the present scheme of reducing the
data must be modified.  If aci*  for  R  <  1.1 were  adjusted  so  that
(oii»/acii»)   is close to one,  then from  Figure  22 it  can be seen
that the counting volume V goes  from 0 to Vmax  as D goes from
Dc to Dc + ADC and is essentially constant at Vmax for  larger dia-
meters.  In the range of D .from  Dc to  Dc + ADC/ the average count-
ing volume for  (am/Ocm)   =  1  is given by
                      5/lnH +    ADc
                ~ -i-for small AD/D   (<0.3).

Thus, for  (a i i»/oci O l'5 close to one, half the particles  in
with Dc 5 D  <  Dc +  ADC would be counted in the channel  associ-
ated with Dc.  In addition all of the particles in V^x  for
DC + ADC - D 1 DI.I would be  counted.   This  means  that in the high.-
est channel  (k = K), where Dc +  ADc  =  Dl.lr  only particles in the
size range Dc  < D ^ Dc + ADC would  be counted.  The  actual concentra-
tion in that size range is thus  given  by

           (AN/AD)R(ADC )R = 2  ANR/


                               43

-------
assuming that AN/AD does not vary much over  the  size  increment.
ANK is the number of counts in  the highest channel  and n is the
number of laser pulses.  The number of counts  in the  (K-l)tn chan-
nel would be given by


      AN    = n(V    )   , [(AN/AD) V(AD  )„  + i(AN/AD)__ ,(AD )„ .]
       A—j.      max j\—-L         A  c  K.    2        J\—J.   c IN.—i.


The same logic would hold for lower channels,  so we find that the
actual size distribution could  be determined;  i.e.,

                                  K
      (AN/AD). = 2[ANv/n(V    ),  - I(AN/AD).(AD).]/(AD K .    (16)
            K       K    max K  •  i.. •«      i    o A     ^j^
     The accuracy.of the scheme for R < 1.1 depends upon  keeping
 [ai i» (D)/aci i» (D) ]™ close to one.  The effect of deviation of  its
value from 1 would be to shift the particle diameter  range counted
in each channel by a factor [1- (01 <*/aCi •») ~x/* ] .  The errors intro-
duced by this factor could be minimized by a simple on  site ad-
justment using the PILLS IV itself.  The instrument should be mod-
ified so that ocm could be varied by a known adjustable  factor.
The procedure for that adjustment would use the highest channel
for which R < 1.1 (i.e., channel K where  (DC)K +  (ADC)K =
Dl.l = 0-75).  This factor would be varied until
0ii-tDi.l)/0ci* (Dl.l) =  [Di.i/(Dc)K]-5.  The value of  ac(Dlel)
that produces that condition could be identified when the count
rate in channel K goes to zero while increasing aci^fDi.i).   This
procedure would produce a measurement of am(DI.I) that would
then be used to set acn» so that am(Di i)/aci •» (DI.I) = 1.
According to the calculation of 0ii»(D) for homogeneous  spheres
illustrated in Figures  16-21 this could leave ai i» (D)/aci"» (D)  ?  1
for lower D-values of particles for some refractive indices.   How-
ever, the maximum expected error in particle size would be less
than 30% and more typically less than 10%.  Of course the errors
involved in this procedure could be reduced with some representa-
tive measurements of OII»(D) for aerosol particles encountered in
the different types of process streams.  Such measurements could
also provide the possibility of changing the form of  ticii»(D)  as
well as its magnitude to something more appropriate for the aero-
sol under study.

     Another recommendation to improve measurements with  the
PILLS IV is the development of a method to easily calibrate the
detectors.  This would eliminate the potential problem  discussed
in Section VI concerning small shifts in Cii», C?, and Co,  produc-
ing large errors in counting particles with diameters close to
DI.I.  One possible method would be the temporary insertion of a
known diffuse light source into the instrument's view volume.
                                44

-------
                           REFERENCES

1. Gooding, Charles H.  Wind Tunnel Evaluation of Particle Siz-
   ing Instruments.  EPA Contract No. 68-02-1398, U.S. Environ-
   mental Protection Agency, Research Triangle Park, N.C., 1976.
   64 pp.

2. Quenzel, H.  Influence of Refractive Index on the Accuracy
   of Size Determination of Aerosol Particles with Light-Scat-
   tering Aerosol Counters.  Appl. Opt. 8(1): 165-169, 1969.

3. Gravatt, C. C.  Real Time Measurement of Size Distribution of
   Particulate Matter by a Light Scattering Method.  J. Air Pollu-
   tion Control Assoc., 23(12): 1035, 1973.

4. Hodkinson, J. R.  Particle Sizing by Means of the Forward
   Scattering Lobe.  Appl. Opt., 5(5): 839-844, 1966.

5. Shofner, F. M.,  G. Kreikebaum, H. W. Schmitt, and B. E.
   Barnhart.  In Situ, Continuous Measurement of Particle Size
   Distribution and Mass Concentration Using Electro-Optical
   Instrumentation, presented at the Fifth Annual Industrial
   Air Pollution Control Conference, Knoxville, April, 1975.

6. Chan,  P. W.  Optical Measurements of Smoke Particle Size Gen-
   erated by Electric Arcs.  Environmental Protection Technical
   Series, 1974.  EPA-650-2-74-034.

7. Fuchs, N. A.  Latex Aerosols - Caution.  Aerosol Sci., 4:
   405-410, 1973.
                                45

-------
                                 TECHNICAL REPORT DATA
                          (Please read Instructions on the reverse before completing)
 1. REPORT NO.
  EPA-600/7-78-130
                                                       3. RECIPIENT'S ACCESSION-NO.
 J. TITLE AMD SUBTITLE
 Evaluation of the PILLS IV
            5. REPORT DATE
               July 1978
                                                       6. PERFORMING ORGANIZATION CODE
 7 AUTHORtS)
 William E. Farthing and Wallace B.  Smith
            8. PERFORMING ORGANIZATION REPORT NO

               SORI-EAS-78-611
9 PERFORMING ORGANIZATION NAME AND ADDRESS
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35205
            10. PROGRAM ELEMENT NO.
            E HE 624
            11. CONTRACT/GRANT NO.

            68-02-2131, T.D. 11007
 12 SPONSORING AGENCY NAME AND ADDRESS
 EPA, Office of Research and Development
 Industrial Environmental Research Laboratory
 Research Triangle Park, NC 27711
            13. TYPE OF REPORT AND PERIOD COVERED
            Task Final; 5/77-3/78
            14. SPONSORING AGENCY CODE
              EPA/600/13
 is SUPPLEMENTARY NOTES jERL-RTP project officer is William B. Kuykendal, Mail Drop 62,
 919/541-2557.
 16 ABSTRACT
               report gives results of theoretical and experimental investigations of
 the operating characteristics of the PILLS IV (Particulate Instrumentation by Laser
 Light Scattering) in situ particle sizing instrument. Results of both investigations
 show large errors in sizing particles with the instrument. Attempts to correlate
 the experimental findings with qualitative theoretical explanations established a
 sensitivity to particle refractive index and detector response that seems to account
 for the observed characteristics. Further measurements would be required to test
 this explanation quantitatively. The prototype was  designed to measure particle size
 using the ratio of intensities of light scattered from a particle  at two small angles
 (14 and 7 degrees) with respect to an incident laser beam. The intensity ratio was
 chosen because of its relative independence of particle refractive index. However,
 the magnitude of the scattered intensity at 14 degrees is also used for several impor-
 tant decisions in the electronic processing logic which, for this particular optical
 system, render it especially sensitive to refractive index and detector variations
 for determinations of particle size distribution. Possible solutions to these  problems
 were offered with only minor hardware changes .
 7.
                             KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                          b.lDENTIFIERS/OPEN ENDED TERMS
                         c.  COSATI Field/Group
 Air Pollution
 Measurement
 Size Determination
 Dust
 Lasers
 Light Scattering
Air Pollution Control
Stationary Sources
Particulate
PILLS IV
13B
14B

11G
20E
20N,20F
 9. DISTRIBUTION STATEMENT
 Unlimited
                                           19. SECURITY CLASS (This Report)
                                           Unclassified
                         21. NO. OF PAGES

                              51
20. SECURITY CLASS (This page)
Unclassified
                         22. PRICE
EPA Form 2220-1 (9-73)
                                        46

-------