DEVELOPMENT  OF  A  NUCLEONIC
    PARTIGULATE  EMISSION  GAUGE
           FINAL   REPORT

     ENVIRONMENTAL  PROTECTION  AGENCY

         CONTRACT NO. 68-02-0210
Federal  Systems   Division
EPA LIBRARY SERVICES RTP NC

  APTD-1150

TECHNICAL DOCUMENT COLLECTION  K|RMAN ROAD/ COLUMBUS, OHIO 43202
                         (614) 267-6351
         dustrial
         Vuclconics

-------
            FINAL REPORT


         DEVELOPMENT OF A
      NUCLEONIC PARTICULATE
           EMISSION GAUGE

              Prepared by
            Charles R.  Duke
              Boong Y Cho

     Industrial Nucleonics Corporation
            650 Ackerman Road
          Columbus, Ohio  43202
                   For
      Environmental Protection Agency
Research Triangle Park, North Carolina 27711

          Contract No.  68-02-0210

            25 February 1972

-------
       This report was furnished to the Environmental Protection




Agency by Industrial Nucleonics Corporation in fulfillment of Contract




No. 68-02-0210.   The contents of this report are reproduced herein




as received from the contractor.  The opinions, findings, and con-




clusions  expressed are those of the authors  and not necessarily those




of the Environmental Protection Agency.  Mention of company or




product names does not constitute endorsement by the Environmental




Protection Agency.

-------
                      ACKNOWLEDGEMENT




       The authors wish to acknowledge the IN program staff without




whose total dedication to purpose this program would not have been




possible:  Orval Utt,  Electrical Engineer; Boong Cho, Physicist;




Juan Crawford,  Mechanical Engineer; Dan Landis,  Designer;




Gerald Blessing, Technician.  Special appreciation is offered to




Robert Herling of the Environmental Protection Agency and Project




Officer on this contract, who lended much expertise and guidance in




the field of sampling and filter systems.

-------
                       TABLE OF CONTENTS
Section                                                       Page
Number                                                      Number

  1.0     INTRODUCTION                                     1-1
          1. 1    Principal Components of the Gauge            1-3
          1.2    Specifications                                 1-6
                 1.2.1    Major Specifications                  1-7

  2.0     TECHNICAL DISCUSSION                             2-1
          2. 1    Principles of Operations                      2-1
                 2. 1. 1    Beta Gauging                         2-1
                 2.1.2    Cassette Transfer Mechanism        2-4
          2.2    Design Criteria for Selecting Filter Medium   2-9
                 2.2.1    The Air Gap Effects                  2-9
                 2.2.2    Filter Variation Effects               2-10
                 2. 2. 3    Effect of Filter Thickness on         2-16
                          Counting Time
          2. 3    General  Comments on Filter Selection         2-21
          2.4    Selection of Carbon- 14/PMT Geometry        2-24
                 2. 4. 1    Determination  of Source Strength      2-25
                          Requirement
                 2.4.2    Detector Selection                    2-28

  3. 0     FIELD TESTS

          3. 1    The Cincinnati Series                         3-1
                 3.1.1    Clean Air Tests                      3-3
                 3.1.2    Cassette Effects                      3-4
                 3.1.3    AGC Delay Effects                    3-4
                 3.1.4    Stability Tests                        3-5
                 3.1.5    Stability Tests  -- Using  Filters       3-6
                 3. 1.6    Stability Tests  -- Pallflex Medium    3-6
          3.2    The Dow Series                               3-7
                 3.2.1    Tests on Leaded  Fuel (3. 0 cc/gal     2-10
                          TEL)
                 3.2.2    Tests on Non-Leaded Fuel            3-10
                 3.2.3    Tests on Low Leaded Fuel (0. 5 cc/   3-12
                          gal TEL)
          3.3    Conclusions Regarding the Field  Tests         3-16

  4.0     RECOMMENDATIONS                                4-1

-------
                          LIST OF FIGURES

Figure                                                         Page
Number                                                       Number

  1-1      Vehicle Exhaust Particulate Emission Gauge           1-2

  1-2      Cabinet and Principal Components                     1-5

  2-1      Filter Cassette                                        2-5

  2-2      Schematic Layout of the Three Stations and the         2-6
          Transporting Mechanism

  2-3      The Effect of Moisture  Absorption on Dry Filters       2-11

  2-4      The Effect of Moisture  Desorption on Moisture Laden   2-13
          Filters

  2-5      Variation of Filter Basis V'eight as  a Function of       2-15
          Moisture Absorption and Desorption (Normalized)

  2-6      Required Beta Sampling Time vs. Filter Basis         2-20
          Weight for Four CFM

  2-7      Required Beta Sampling Time vs. Filter Basis         2-22
          Weight for Two  CFM

  2-8      Typical Beta Spectrum                                 2-29
                                                                11

-------
1. 0    INTRODUCTION





       This report is the final report to Environmental Protection Agency





Contract No. 68-02-0210.   Upon acceptance by the EPA, this report





represents completion of contractor-required delivered items.





       In June of 1971, the EPA contracted with Industrial Nucleonics (IN)





to develop a beta gauge and filter sampler for measurement of particulate





emissions from automobile exhaust.  The gauge was to have a minimum





sensitivity of 125 micrograms per cubic meter (|j.g/m ) with no specified





upper limit.  Further, the filter collection system was to have an efficiency





using a dioctyl phthalate aerosol of at least 99. 9%.  A gauge to meet these





requirements has been designed,  built, tested,  and delivered and is as





shown in Figure  1-1,





       A  Hewlett-Packard Computing Counter has been used to perform





the  calculations and switching functions.  In addition,  a Fischer  and Porter





Swirlmeter has been provided for reliable determination of gas flow.   The





inclusion  of these two units allows high accuracy and repeatability with





respect to computation.  This was deemed necessary in order to fulfill





the  wide variety  of applications against which the unit was to be  tested.





The major thrust of the program has been to design a research tool





rather than a prototype so that evaluation of the unit could be made on





the  basis  of the suitability of beta gauging for the various applications,





not merely IN's ability to produce such a gauge.  Certainly, for many





specific applications, a considerable reduction in complexity and cost





is possible.








                                                                      1-1

-------
VEHICLE EXHAUST
PARTICULATE EMISSION GAUGE
                   Figure

-------
       As part of the contract, this unit was tested recently at one of




the EPA's Cincinnati facilities and at the Dow Chemical Company in




Midland, Michigan.   The Cincinnati series was primarily intended for





unit de-bug  and checkout.  The tests at Dow were performed on automobiles




using leaded (3cc/gal tetraethyl lead), low-leaded (0. 5  cc/gal TEL) and




non-leaded (0. 0 cc/gal TEL) fuels.   Good results were obtained using




leaded and low-leaded fuels when compared to gravimetric measure-




ments.  However, less desirable results were obtained with no-leaded




fuels.  No conclusion may be drawn from the non-leaded tests  since




correlation between two different size (4 inch and 2 inch filters)




gravimetric measurements could not  be achieved, thereby leaving no




standard against which to compare the gauge.





       The  details of these tests, and the details surrounding IN's




selection of this  particular design constitute the text  of this report.




While reference  will occasionally be made to operation of the unit,




the reader is referred to the Operation Manual accompanying the  unit




for specific operation,  maintenance and safety procedures to be




followed.






1.  1    Principal Components of the Gauge




       This gauge has been designed primarily as a research tool.




There has been little attention devoted to minimizing the size and weight




and maximizing the  portability as efforts of this type were considered




to be beyond the  scope of this contract.  The device has been fitted
                                                                    1-3

-------
with wheels, however,  to assist in moving the unit from place to place.

Also the Computing Counter and Systems Programmer are quite easily

removed (See Operation Manual) for further ease of transport.   A

complete list of the principal components and their model numbers

is as follows (See Figure 1-2):


       1.    Electronic circuitry  (printed circuit boards).   Access
             to this circuitry is obtained by removing the side  panel

       2.    Welded cabinet brace

       3.    Printer.  Hewlett-Packard Model No.  5055A

       4.    Flowmeter preamplifier

       5.    Vacuum pump.  CAST Model No. 0822

       6.    Flowmeter.  Fischer and Porter "Swirlmeter"

       7.    Flowmeter supports

       8.    Flowmeter electronics

       9.    Swivel caster (four identical double-wheel units)

       10.    Control panel with switches and indicator lights

       11.    Systems programmer,  Hewlett-Packard Model No. 5376A

       12.    Computing counter,  Hewlett-Packard Model No.  5360A

       13.    Flow rate display meter

       14.    Particulate Collection station.  Station at which gas is
             drawn through the filter and particles are  collected

       15.    Nucleonic source  enclosure and filter locating mechanism

       16.    Magazine to hold 20  filters
                                                                   1-4

-------
                  Z5-I/Z
37-1/4
L

"







•




-'


--


-











-


1
I
1
1
                                         2,
                                        CD
                                                                                 I a)
      0 I !
oo o ob opj
                   SIDE
      FRONT
                         CABINET AND PRINCIPAL COMPONENTS
                          (SIDE  PANEL AND  FRONT DOOR  REMOVED)

                                    Figure.  I ~ Z

-------
      17.    Filter transporting mechanism actuator, Cliftronics
             R28S1 32 rotary solenoids

      18.    Radiation detector bousing.  Station at which filter
             and collected particulate masses are detected.

      19.    Flow regulating valve.


1. 2    Specifications

       This gauge has been overdesigned for many specific applications.

This is a direct result of emphasis being placed on accuracy over a wide

range and flexibility.  The beta counting times and pumping times may

seem astronomical and unnecessary.   The magnitude of these numbers

is,  however, an added bonus  received by inclusion of the HP Computing

Counter  and Systems Programmer.  It is also correct to say that the

computational errors are infinitesimal by comparison to the errors in

the sensing heads (flowmeter and beta counting).   This is as it should

be for an instrument of this nature.   The limitation of the  gauge is

contained solely in the limitations of the sensors used, namely the

state-of-the-art in stabilizing photomultiplier tubes, the state-of-the-

art in measuring absolute flow, and process repeatibility.   This  necessarily

implies that improvement in any one of these three areas will result in

improved measurement accuracy automatically.

       The major specifications follow,  but some prefacing comments

are necessary.  First, it is difficult to state a range of the gauge in terms

of stream concentration unless some  limit is placed on botv> flow rate and

time.  This is  a result of there being both a minimum and  a maximum
                                                                     1-6

-------
        total mass which can be detected with the current design.  The minimum

        and maximum mass determinations are stated with the assumptions on

        flow rate and time clarifying the stream concentration.

               Second, probably the greatest limiting factor on accuracy at

        high concentrations is the ability to measure the total  sample volume

        correctly.   The Swirlmeter measures volumetric flow and therefore

        to achieve  a measurement in terms of an "Actual Cubic Foot", use must

        be made of the suggested correction factor in the Operations Manual.

        For  most applications where this factor is not included, the flow accuracy

        is ±2%, excluding corrections  to "Standard  Cubic Foot."


        1. 2. 1  Major  Specifications (assumes zero process error)
    Specification

1.  Effective Filter Area

2.  Flow Rate

3.  Minimum Concentration

4.  Maximum Concentration
5.   Beta Counting Range



6.   Beta Counting Range

7.   Pumping Time


8.   Pumping Time


9.   Minimum Mass Resolution
Value
 3 cm
20-120 1/min
80 ± 16
                         Remarks
              (2o-)
3 x 106 ±2% >ig/m3
(assumes no flow-
meter  correction)

10"  sec to 2 x 10 1
sec.
30 sec. to 300  sec.

  -6             11
10   sec to 2 x 10
sec.
30 sec to  3 x 10
sec.

 ±60 ng (2
-------
10.

11.
Maximum Mass Determination  30,000 ±600 (o.g (2cr)
Flow Absolute Accuracy
12.    Flow Absolute Accuracy


13.    Weight

14.    Power

15.    Source

16.    Source Activity
±0. 75% ACFM



±2. 0% ACFM


350 Ibs.

1500 watts

Carbon-14

100 microcuries
Manufacturer's specifi-
cation (using correction
factors)

no correction factors
flow >40  1/min
                                                     while pumping
                                                                            1-8

-------
2.0     TECHNICAL DISCUSSION



2. 1     Principles of Operation



2.1.1   Beta Gauging



        The Vehicle Exhaust Particulate Emission Gauge utilizes a



filter "cassette" on which to collect the particles.  Detection of



particles  on the filter is accomplished by an isotopic radioactive



source (Carbon- 14) and a scintillation detector (Photomultiplier  Tube).



        The attenuation of beta particles through a medium follows an



exponential law:
                  N  =  Noe                                           (])



where            N  =  detected counts



                  N  =  detected counts in absence of medium
                   o


                  [i  =  attenuation coefficient of medium



                 px  =  basis weight of medium.



In this application, the "medium" discussed consists  of several separate



media: 1)  the air gap between source and  detector; 2) the clean filter;



3)  the detector window,  and 4)  the particles collected.  Therefore



equation (1) can be rewritten as:



                            -(upx)a   -(npx)f  -(M-PX)W -(HP*)0

                  N  =  N e        e         e         e      H         (2)




•where            N  =  detected  count



                  N  =  counts emitted from surface of source in
                   o


                       direction of detector



,              (HPX) a  =  attenuation exponent for the air
                                                                      2-1

-------
             (}j.px)    =   attenuation exponent for clean filter



             ((j.px)    =   attenuation exponent for detector window



             (fj.px)    =   attenuation exponent for particles.



       Obviously,  the only factor of interest in Equation (2) is the  last


one referring to the particles.  Every effort has been made to either



control or minimize the effect of these  other variables.  The  effects



of variations in the density of the air has been minimized by reducing


the air gap between the source  and  detector to about one centimeter.



In order to account for variations between filters and to allow deter-


mination of the  reduction in total collected counts due to detector



attenuation, a standardization measurement is made and retained for


later comparison.  Equation (2) may then be written as:



             N     =    N  e
               Lt         1


where       N     =    detected counts after collection of particles

                       (Measurement)



             N.    =    detected counts before collection of particles

                       (Standardization)



Taking the natural logarithm of both sides:
             Ln
or
then         px   =   —   Ln   —-                                    (3)
                                                                      2-2

-------
       Equation (3) is the expression to determine the mass per unit




area collected.  To determine total mass collected, multiply by total




filter sample collections area A.






                                        /NA
                Mass =  Apx =  —  Ln  I ——  1.                      (4)







       Notice that by making an initial standardization measurement,




all reference to the filter itself has been removed.  This is indeed one




of the advantages  of the  beta  gauge.  Almost any filter medium may be




used, the only requirement being that the medium is not modified during




the  sampling process.  The "cassette" approach accentuates this advantage,




allowing use of filters not suitable for use as a tape.




       The total gas flow through the unit is determined by use of a




Fischer and Porter Swirlmeter (1" size).  This unit creates a precessing




vortex in a  region containing  a thermistor.  The thermistor responds to




the  different heating afforded by the center of the vortex and, through




pulse shaping,  produces a pulse output.  One pulse is representative




of a specific volume of gas.   To determine the  total flow,  the gauge  simply




collects the pulses from the flowmeter and multiplies by a constant.  A.




more complete description is presented in the Operation Manual




accompanying the gauge.  The volume then becomes:






                Volume   =  K N                                      (5)






To determine concentration,  Equation (4) is divided by Equation (5):
                                                                      2-3

-------
       Concentration  =  ——	 = K    —~^—  Ln   I —— I           (6)
                         Volume      2    ™        i ™  i           <• >
where K  is a constant relating the various units involved.   Equations
        C*



(4), (5),  and (6) represent the outputs required by contract.  These




along with the counts N  , N  ,  and N  are printed on paper to provide




a permanent record of the test.  By virtue of the Hewlett-Packard




computer, all calculations, including the natural logarithm, are done




digitally, affording extremely high accuracy of computation.  The only




analog circuitry in the gauge  is the pulse gathering and shaping circuitry




and the automatic gain control.






2.1.2  Cassette Transfer Mechanism




       It has been previously mentioned that this  gauge uses a filter




cassette on which to collect the particles (See  Figure 2-1).  The  cassette




is constructed of 1/16" aluminum  and is irregularly  shaped to avoid




improper loading into the magazine.  Installation  of the filter on the




cassette is accomplished by cutting a square piece of filter paper




(nominally 1" square) and taping the filter  to the cassette.  Notice also




that the cassette has two holes on  opposite corners to precisely locate




the cassette at the beta measurement station.  This is to insure that




the source is irradiating the  same ares after collection on which standardi-




zation measurement was made before collection.





       During operation each cassette is transported automatically




through the gauge.   The cassette transfer mechanism has three  stations




(Figure 2-2).   The first is the loading station,  where the lowest cassette








                                                                       2-4

-------
     UPPER SIDE
                          Adhesive tope
                          Rlter medium
     UNDER SIDE
FILTER  CASSETTE
      Figure 2-1
                                      2- 5

-------
SEALS
 GAS OUT

   TO

  SUCTION

  PUMP
                          LOADING

                          STATION
c
(

^ 	 j j'Hri-,
(n Uki ! ^
'
^4>
^
                                                 TOP  VIEW OF TRANSPORTING SLIDE
                                                                               180

                                                                               ROTARY

                                                                               SOLENOIDS
             DETECTOR




            COLLECTION


            STATION
                   SOURCE-DETECTOR   LOADING
                   STATION
STATION
  ro
  i
SCHEMATIC LAYOUT OF THE THREE  STATIONS d THE TRANSPORTING MECHANISM
                                           Figure 2-2

-------
in the magazine drops into a pocket in the transporting slide.  At the




second station the cassette is positioned between the  source and the




detector,  so that the beam of radiation passes through the filter.  At




the third station the cassette is positioned so that the gas to be monitored




may be drawn through the filter, and the particles collected.




       Each cassette is transported between the  stations by a slide of




the same thickness as the cassette.  The  slide is  moved between stations




by two 180-degree rotary solenoids.  In Figure 2-2 the slide is shown in its




rightmost position,  ready to pick up a cassette from  the magazine.  The




next operation is for the  right-hand solenoid to rotate 180 degrees.  The




cam roller on the end of  the  solenoid arm will follow the  path indicated,




causing the  slide to  travel to the left and transport the cassette to the




source-detector station.  At this station two tapered  pins are pushed




down into the two small holes in the  cassette.   At this time the standardizing




measurement is made to determine a background  count in the absence of




particulate matter.  The locating  pins are then withdrawn and the left-hand




solenoid is rotated 180 degrees.  The cam roller  on the end of the solenoid




arm will follow  the path indicated, causing the slide to travel further  to the




left and transport the cassette to the collection station.  At this  station the




cassette is clamped between two elastomeric seals which prevent leakage




of gas into or out of the stream passing through the filter.  The cassette




is then clamped, the vacuum pump is turned on,  and  the particles
                                                                       2-7

-------
are collected on the filter.  At the completion of the collection cycle,


the sealing clamp is released and the left-hand solenoid is rotated


180 degrees in the opposite direction to transport the cassette back


to the source-detector station.  Here the locating pins are reintroduced


into the small holes in the cassette.


       After the  measurement of the material deposited on the filter


is made, the right-hand solenoid is rotated 180 degrees in the opposite


direction, and the slide is returned to its initial position.  As it does


so the cassette in its rightward travel encounters the stripper springs,


two inclined flat springs which guide the cassette downwards through


an opening in the base plate and out of the transporting slide pocket.


The used cassette drops into a receiving container and the slide pocket


is left vacant, ready to pick up another cassette from the magazine.


       Originally, it was proposed that the transfer mechanism be a


wheel in which holes had been cut to contain the cassette.  After the


program started,  a more complete examination of this approach revealed


certain problems.  The problems, though not insurmountable, presented


complex switching and sensing conditiongs which were considered to be


simplified by using a linearly translating slide.   If the filter manufacturing


quality control were such that the basis weight from filter to filter did not

                           2
vary by more than 10 jig/cm  , a considerable reduction in complexity


could have been accomplished.  Current state-of-the-art in  filter manu-


facturing does not permit this, however.
                                                                     2-8

-------
2. 2    Design Criteria for Selecting the Filter Medium


       Several filter media were evaluated during the course of the


program.  Before discussing the various filters, however, it is worth-


while to discuss the criteria for evaluation.


       The contract required determination of the particulate concen-


tration in a stream having at least 125 micrograms per cubic meter.

                                                  3
This determination was to an accuracy of ±25 |j,g/m  at a 95% confidence


level and an efficiency of collection 99. 9%.  These requirements place


a rather severe constraint on the sampling head repeatibility and the


electronic system stability. In determining the amount of material


which must be collected to allow this determination,  consideration must


be given to the magnitude  of such error sources as the change  in air


gap conditions and filter deformation.



2.2.1  The Air  Gap Effects


       The air gap between source and detector represents a "background"


attenuation for the beta particles.  It is an attenuation which cannot be


avoided and any variation  in the  air gap to  cause a change  in air density


between  standardization and measurement  represents an error  in measure-


ment.  To minimize the effect of air density variations,  the air gap itself


must be reduced as far as it practically feasible.  In this gauge the air gap


is about one centimeter.   This represents  an effective background basis


weight of about ] . 3 mg/cm .  If the variation in air density can be held


to within 1% (about 5  F) between standardization and measurement, then
                                                                  2-9

-------
one may consider ± 0. 5% or ± 7 fig/cm  fluctuation (]
-------
Rgure 2-3
                                       2-11

-------
for extremely high structural strength.  It has a basis weight of approxi-

                  2
mately 3.2 mg/cm .   The sample chosen was a strip 31.8 cm x 110 cm.


(This is equivalent in area to a circular filter approximately 210 mm in


diameter.  While admittedly large, an approximate number for any other


size filter may be obtained by scaling the areas, as the variation on a


basis weight consideration should be identical, regardless of filter size. )


       The test procedure was to weigh the filter as delivered from the


manufacturer,  dry the filter at 110 C for 10 minutes (to obtain the actual


"bone dry" weight would require much longer times such as 24 or 48


hours at 110 C), and then weigh the filter at periodic intervals after


exposure  to 72 F and 50% relative humidity.  As can be seen, a weight


loss of 49 milligrams (140 |j.g/cm ) occurred upon drying.  Much of this
loss is recovered rapidly,  so that after 10 minutes only 14 mg (40


remains unrecovered.  At this  point, however,  the recovery becomes quite


slow, so that after 30 minutes, 11 mg (31. 5 |o.g/cm  )  still remain lost, and


after 4 hours,  40  minutes,  9 mg (25. 7 [jig/cm ) of moisture still had not


been regained.  For a typical filter  size used (47 mm) for comparison,


this would represent an error  of 445 micrograms due solely to moisture


absorption effects.


       The effects of moisture desorption after having been exposed to


100% relative humidity at 72 F are even more dramatic (See Figure 2-4).


After exposure of the filter to  this environment for 25 minutes,  a weight


gain of 76 mg (217 (j.g/cm )  as  observed.   After  10 minutes of
                                                                  2-12

-------
Figure  2-4
                                      2-13

-------
stabilization, 22 mg (63 fig/cm  ) were  still retained, after 30 minutes,

                     2
12.8 mg (36.6 |j.g/cm ), and after four hours, 40 minutes,  there was

                      2
still 9 mg (26. 7 p.g/cm ) of excess moisture.


       If the data is normalized to initial filter weight to present it


on a percentage variation basis, Figure 2-5 is the result and illustrates


clearly the "hysteresis" effect of moisture on fibrous materials.  This


hysteresis would represent an uncertainty in the amount of particulate mass


collected of as much as ±  1 % of  the initial filter basis weight unless the


complete moisture history is known.  For the Pallflex medium,  this is


an error of approximately ± 32 jjig/cm  due solely to moisture effects.


In addition, if the measurement is made immediately after sampling (as


it is in the case of this gauge),  even higher errors  can  result.


       From the above data,  it is recommended that the  Pallflex medium


not be used in this gauge.   There are,  of course, other filter media which


may be used which exhibit less  sensitivity to either moisture absorption


or desorption.


       The GE Nuclepore, for example,  showed a weight loss of only


6. 4 |j.g/cm  after drying and recovered to its initial weight (no apparent


hysteresis) after only three minutes of stabilization.  When exposed to

                                       2
high humidity, it gained only 1.  3 |j.g/cm  and has regained its initial


weight within one minute.   Unfortunately, this material does  exhibit


the highest pressure drop for a  given flow rate of almost any other type


of medium, and has  shown a tendency to  clog up.
                                                                   2-14

-------

I
             2-15

-------
       The Gelman Type A filter avoids the problem of high pressure



drop for a given flow rate and does not clog easily.  Upon drying as


                                                                  2
previously described, this medium showed a loss of only 9. 7 (j.g/cm



and has recovered half this amount in 1 5 minutes.   When exposed to



high humidity, however,  this medium showed a weight gain of 221 |j.g/


   2
cm , but had regained its start weight within 10 minutes.   It is also



quite thick (approximately 7 mg/cm ) and  will require a longer counting



time (see the next section).



       A further consideration is that these errors were produced under



static conditions (i. e. , no forced flow through the filter) and may not



be reproducible under dynamic conditions  with which the EPA is  involved.




2.2.3  Effect of Filter Thickness on  Counting  Time



       The following discussion concerns  only stream concentrations



of 125 fjLg/m  as far as the flow rates are concerned.  It is equally



applicable when the required particulate basis weight  resolution is  of



the same order.



       In the process of a quantitive  determination  of the collected mass,



the filter interferes with the measurements.  For example, if the filter



weighs 1,000 to 10,000 times the weight of the minimum desired resolution



of the collected particulate,  the total weight of the filter/particle system



has to be determined to 0. 1 to 0. 01% accuracy.  Even though it is feasible



to determine the weight of certain objects  to these accuracies,  extreme



precaution is required.  The  weight of an object like a paper or  glass



fibre filter  is "'unstable"  in the sense that  its weight will change with the
                                                                     2-16

-------
temperature and relative humidity of the environment and furthermore




a hysteresis effect is observed in the adsorption and desorption of water




vapor (see the previous section for a more complete discussion).   Since




the gain and loss of the weight with the environment and its history will




be proportional to the weight of the filter, it is highly desirable to use




as thin  a filter as possible for a sample collection medium.




        In a nucleonic method of weight determination,  a further considera-




tion is required in the sense that the filter will attenuate the radiation




without contributing to the  signal.   This attenuation is  not detrimental




except for the possible physical  deformation of the filter due to pumping,




if no restriction is imposed on the required source strength.  However,




if the source strength is limited to a given value, the loss  of radiation




intensity has to be made up by a long counting time to obtain a desired




statistical accuracy.




        Consider a case of  "zero" filter thickness.  Then,  with a 100 |j.Ci




C-14  source and the geometry of IN design,  a count rate of 50, 000 cps is




obtained.   If a two sigma resolution of ±20 |j.g/cm  (corresponding to a




stream resolution of ±25 ^.g/m  with a 4 CFM sampling flow rate, sampling




for 20 minutes from an aerosol stream containing 125  |J.g/m  of particulate)




or ± 10  jig/cm  one  sigma resolution is required, the required counting




time is determined as follows: The attenuation of radiation is given by
                            N   =  N  e  rr                             (7)
                                                                      2-17

-------
where N and N  are the detected total counts with and without the sample;




and ^JL, p , and x are the mass attenuation coefficient,  density and thick-




ness of the sample, respectively.   By differentiation, one obtains
                      dN

or
                                                                       (8)







       For C-14 beta rays, (j, is approximately 250 cm /gm and the




minimum required resolution of (px) is ±10 |j.g/cm^ (one sigma).  There-




fore, the minimum required resolution for (|j.px) is 250 x 10 x 10   =0. 0025.




This means, from equation (8), that the nuclear signal, N, has to be




resolved to an accuracy of ± 0. 25% (one  sigma).  It is well known that the




nuclear signal fluctuates with one  sigma  value of ± ^N when the total




count is N.  Therefore,
                                           = — ">-2
or
                      N =  1. 6 x 105.                                (10)
                                                                   2-18

-------
                                                4
Since the count rate with no sample, n , is 5 x 10  cps, the counting



time to accumulate 1. 6 x 10  counts is




                            1.6 x 105       1.6 x 105    ,  ,       ,     Ml.
                      T  =  	  =	—  =3.2  seconds.    (11)

                              n°           5x10



       Now, if the filter is  assumed not to contribute further error due



to its unstable weight (absorption and desorption of moisture) and physical



deformation,  its effect on the nucleonic measurement of collected particles



is to attenuate the radiation. Therefore, the count rate, n ,  will be
                                                          o


reduced according to
                      n  = n  e
                            o



where the subscript, F,  represent the value for the filter.  Substituting



n for n  in equation (11), one obtains the required counting time for



various filter thicknesses from
                      T =    '          =  3.2e       seconds.         (13)
                               n



This equation is plotted in Figure  2-6 labeled with "no electronic error."



       An additional error is caused in the measurement of N(for the



purposes  of discussion, this is called electronic error, but also includes



all other errors besides statistics) in equation (7).  Thus, 0.25% total



allowed error in equation (9)  should be allotted between the electronic



error and the statistical fluctuation.  Assuming an RMS averaging,



one gets




                      Ee2   +  -^j-  = (2.5xlO~3)                        (14)




                                                                       2-19

-------
  4        5
Figure 2-6
    8
2-20

-------
or
                      N  =  -    -

                             (2. 5 x 10   )   - E
        Following the same procedure as in the case of "no electronic



error," one obtains the counting time from
                            N


                                   5 x 104  [(2.5xlO~3)   -E2J seconds.





        This equation is plotted in Figure 2-6 with E  =  0. 05%, 0. 075%,
                                                   G


0.1%, 0.15%, 0.2%, and 0.23%.



        If,  however, the same stream resolution (±25 ng/m ) is required,



but only 2 cfm (56 liters/min) flow rate is used, the basis weight  resolu-



tion is ± 5 fj.g/cm  and  Figure 2-7 is the result with E  plotted for 0. 025%,



0. 05 %,  0. 1%, and 0. 12%.  As can be seen, the  effect is rather dramatic



on the counting times,  with resolution of this magnitude impossible  at



4  cfm if all other errors amount to 0.25%,  and at 2 cfm if all other errors



amount to  0. 125%.




2. 3     General Comments on Filter Selection
        While the advance copy of the Operation Manual delivered with the



instrument recommended that the  Pallflex E70/2075W medium be used,



later test results  show this medium to be undesirable from a moisture



absorption and desorption standpoint.  It is IN's opinion that the



"optimum"  filter medium has  not yet been found.  Much work  remains



to be done in this area.  In selecting a medium to use, however, several



conclusions can be drawn from the preceding sections of this report.
                                                                     2-21

-------
                                                                           •4-
                       1
                                                                    1
                                                              oth
                         Q
ifci
                                           1ft
IQJ
                                  4
                                Figure
             2-7
7          e
    2-22

-------
       First, the filter should have as low a basis weight as possible



for two reasons.  One, to reduce the amount of time required for beta



measurement, and two, to reduce the effect of any slight distortion



caused in the filter by the pumping process.  From the previous



section,  the beta collection time may vary as much as four to one,



depending on the basis  weight of the filter.  Also a 1% basis weight


                      2                   22
variation for 1 mg/cm   filters is 10 [o,g/cm ,  but for  8 mg/cm  , a



1% basis weight variation is 80 p.g/cm  .



       Second,  the property  of paper-based filters to  absorb and lose



moisture during the process  casts doubt on results obtained using



these materials.  In fact, even the Gelman Type A filters yield



dubious results after exposure to high humidity, which unfortunately



happens  to be the situation with automotive exhaust.   This  does, of



course,  mean that there is currently no standard against which to



reliably  measure results,  either with the beta gauge or with gravi-»



metric comparisons.



       Third,  the problem of moisture is  not confined to beta gauging.



In fact, it isn't  even confined to moisture.  Any volatile material would



be subject to identical fluctuations.  Again, automotive exhaust contains



a high percentage of volatile  material,  at least in the cases of  low-leaded



and non-leaded fuels.



       Finally, it is conceivable that a different type of filter material



may be required from filter manufacturers.  Little consideration has
                                                                   2-23

-------
been given by this industry to such properties as low moisture sensi-




tivity.   This is presumably due to primary application areas being




chemical analysis of collected particulate or simply to clean the air





passing through the  filter.






2. 4    Selection of Carbon-14/PMT Geometry




       Several sources were evaluated prior to contract:  Kr-85, C-14,




Ni-63, and Pm-147.  A complete discussion of the tradeoffs involved in




selection is beyond the scope of this report  and  only the major reasons




will be presented here.  Krypton-85 was eliminated on the basis of poor




sensitivity.   Nickel-63 was eliminated on the basis of low modal energy.




Promethium-147 was eliminated on the basis of short half-life.  This




left Carbon-14 as the optimum  selection.




       IN had proposed that the source to use was in the form of a




carbide, where there is a chemical bond between the carbon and




surrounding  atoms.   In so doing,  an extremely high integrity source




could be achieved,  capable of surviving almost any environmental




condition, up to and including fire.   IN still maintains that this is the




most desirable source to have.   Unfortunately a complete  survey of




the source manufacturing field  disclosed no vendor willing to undertake




the development of a carbide source.   While the technology appears to




be available  to produce such a source, neither time nor money would




permit the development under this contract.
                                                                  2-24

-------
       The vendor eventually selected was New England Nuclear,


Billerica,  Massachusetts.  Originally a source was received from


them which was a C-14/Kapton mixture deposited on a small alumi-


num disc.   This proved to be unsatisfactory from a uniformity


standpoint.  Several iterations were required before achieving the


physical integrity, uniformity,  and emission efficiency we desired.


IN believes that the source which is now in the gauge is one  of the


finest non-carbide sources made.  It  is constructed by depositing the


C-14 on  an  aluminum planchet, laying a thin  plastic film over this


and spraying the active area with Kapton.   This source has the highest


percentage  yield of any other in IN's experience and made it possible


to reduce the activity to such a level as to prove the feasibility of


making this gauge eventually license free.



2.4.1  Determination of Source Strength Requirement


       The license free aspect of the gauge was paramount in IN's


philosophy of design.  Carbon-14 occurs naturally and as a result has


one of  the highest allowed quantities to qualify for National AEC  license


exempt status.  This quantity is 100 microcuries.  It has already been


discussed (Section 2.2. 3) how much collection time  is involved,  pro-

                           4
viding  a  count rate of 5 x 10  pulses per second can be achieved  in the


absence  of a filter.  It will now be shown that this count rate is in fact


achieved in this gauge.
                                                                     2-25

-------
       The effective yield of the 100 microcurie source delivered with


the gauge is estimated by New England Nuclear at 15%.  The number of


counts actually emitted from the source in all directions is
      , „   ,~10    counts	   ir,~4^  •       ir   cc   m5 counts
      3. 7 x 10	;	 x 10    Curies x . 15= 5. 5 x 10
                  sec. •  Curie                          '         second



If the source is place 1 cm from a disc- shaped detector of 1 inch


diameter, the geometrical factor is 1/2 (1  - cos 0) ~ 18% where 0 is


the half -angle subtended by the detector.  Thus, in the absence of air


and sample  5. 55 x 10  x . 18  = 10  counts per second reach the


detector.


       The  effective attenuation  coefficient (,) has been estimated at
       2
250 cm /gm.   The desired count rate is in the absence of the filter


but the attenuation due to air and the detector window must still be


considered.  It was previously discussed that beta radiation  attenuation


obeys a negative exponential law:
                      N  = N e
                             o
 This can be expressed differently as
 where
                      N  = N (
                             o
                      N  =  10  counts per second (previously derived)


                          = attenuation exponent due to air


                   ((j.px)w = attenuation exponent due to the detector window.
                                                                     2-26

-------
For air, the density    is equal to approximately 1.2 mg/cm  and




the distance is 1  cm.  The attenuation factor (e ^   -A) due to air then



becomes 0. 74.  The detector window has a basis weight (px) of approxi-




mately 0.9 mg/cm  (manufacturer's  specification).   Therefore the




detector window attenuation factor is 0.8.  The total detected counts



                5                       4
then becomes 10  x.74x.8  =  5.92x10  counts per second.   Allowing




for detector efficiency and inclusion of a threshold detector to exclude




PMT  characteristic noise, it can be seen that the detected counts will


                                    4
not vary significantly  from the 5x10  counts per second required.  In




fact,  the count rate achieved  in the absence of a filter is approximately



44, 000 cps so that  good agreement with theory is achieved.



       This count rate allows the use of the gauge over a wide range of



filter area densities without inordinately increasing the beta measure-




ment  time  (See Section 2. 2. 3).  This was deemed desirable by IN in




order to increase  its flexibility as a research tool.  In Figure 2-6,  for




example, it can be seen that using a filter of basis  weight 7 mg/cm




(such as the Gelman Type A)  and  a process repeatibility of 0. 1% (estimated



reliability  for the IN  gauge) a counting time of 220 seconds is required to




achieve ±10 ^.g/cm  resolution.  If the count rate were one-tenth




our estimated rate (this would yield an estimated detected count rate




of 4. 4 x  10 counts per second, typical of maximum rates available  from




Geiger-Mueller detectors), this time would be 2,200 seconds to achieve




the same accuracy.
                                                                  2-27

-------
2.3. 2  Detector Selection




       The deficiencies of a Geiger-Mueller tube in the rapid determina-




tion of filter basis weight has been discussed.  There are three other




types of detectors which may be used for low energy beta radiation




detection:  solid state detectors,  proportional counters,  and a PMT/




scintillation crystal combination.  The current state-of-the-art in solid




state detectors are such that they must be used in a nearly constant




temperature environment.  This was felt to be an unnecessary require-




ment to place on the detector station.  Proportional counters and ionization




chambers were considered carefully, but the window thickness is such




as to prevent the use of license exempt quantities for  C-14 sources.




That left only the  PMT/scintillation  crystal combination.




       There are  two basic  scintillation crystals which could be used:




CaF_(Eu) or Nal(Tl).   Since the modal energy of C-14 is so low, an




extremely thin window is required (on the order of 1. 0 mg/cm ). For




this reason and the fact that Nal is highly hygroscopic and would tend




to deteriorate with time,  CaF  was  selected as the scintillator.
                             £t



Harshaw Chemical Company was  then contracted to fabricate the




detector  unit, using an Amperex XP-1011 tube,  a CaF_ crystal, and
                                                     £j



aluminized Mylar  as the detector window.




       This detector was quite stable in gain in  contrast to other




very similar detectors.  As is the case with all PMT's,  however, the




gain drift was significant enough to require the addition of an automatic
                                                                    2-28

-------
gain control (AGC) circuit.  A typical beta spectrum is as shown in



Figure 2-8.  If two thresholds are established, L and  H for low and



high, respectively, it is possible to select these  thresholds such that



the difference in the two numbers remains relatively constant.  For



any major  changes which may occur,  however, the ratio of the two



counts provides a control  signal by which the gain can be adjusted  to



keep the above condition true.  Through this stabilization technique,



PMT stabilities on the order of 0. 1% are achieved and it is this



capability that allows the measurement to be made at the low end



of sensitivity.
                 g
                 o
                           H
Pulse Height ( Millivolts)
                    Figure 2-8
                                                                  2-29

-------
3. 0    FIELD TESTS
       Two series of tests were performed on the gauge.  The first


series was performed during the week of 4 October  1971, at the  EPA's


Cincinnati facility and was for the  primary purpose  of debugging the


unit.  The second series was performed during the week of 25 October 1971,


at the Dow Chemical Company in Midland,  Michigan, and was for the


primary purpose of acceptance tests on the unit.   The results obtained


from both these series are the  subject of this section of the report.


Prior to the field tests, the  attenuation coefficient JJL was experimentally

                        2
determined to be 252 cm /gm by noting the effect on the count rate of


differ ring thickness of Mylar.



3. 1    The Cincinnati Series


       Little data was  obtained from automotive  tests in this series due


to several problems  encountered with the unit.  As the purpose of the


tests was to reveal these problems, however,  considerable progress


may be considered to have been made.  Minor problems were encountered


with the  locating solenoid binding up.   Due  to self-check microswitches


installed on the unit, no errors in data due to this malfunction are believed


to have occurred.  A problem was encountered but not recognized until


after the Dow series with interference between the pressure fluctuations


in the air line due to the pump and the Swirlmeter.   Apparently,  these


fluctuations were viewed by  the Swirlmeter as vortex precessions and


counted as a volume  of flow.  It is believed that this was only a problem
                                                                    3-1

-------
at the low flow rates.   To correct the problem, the flowmeter was




changed from its "downstream of the pump" location to its present




"upstream of the filter" location and the interference was no longer




observed.  During the brief redesign period between the  Cincinnati




and Dow series of tests,  all encountered deficiencies were corrected.




       One of the major thrusts of this series was to evaluate different




filter materials.  There were four different media which were hopefully




to be tested:  1)  the Gelman Type A; 2) GE Nuclepore (0. 4 \± and 1.0 u),




3) Pallflex No. E70/2075W; and 4) a paper filter  of high efficiency and




low basis weight developed by Dr. Wendell Anderson, Naval Research




Lab. ,  Washington, D.  C.    Details of the tests follow,  but a brief




discussion of these media is appropriate.  Some deformation (tearing)




of Dr.  Anderson's material was detected.  This is believed to be a




result of the short time delay  between the  command to  release the clamp




on the  cassette at the  sampling station and the command  to transport the




cassette to the beta measurement station.  After  this,  it was decided




to switch to the Pallflex medium,  and as this  filter seemed to perform




adequately at the time, no tests were performed on either the Gelman




or GE  filters.  Tests  were later performed which showed the Pallflex




medium to be inadequate  due to moisture interferences (See Section 2. 2. 2)




and it is now recommended  that tests be performed by  the EPA on the




Gelman and GE filters to determine  their adequacy.
                                                                    3-2

-------
        This test series can be divided into six separate segments.





The included segments, reasons for their inclusion, and results




obtained are contained in the following sections.   In this entire series,




volume collected is unimportant and will  not be included in the data.






3.1.1   Clean Air  Tests




        These tests were performed on Dr. Anderson's medium and




were for the purpose of determining the clean air characteristics of




the unit.  Clean air was obtained by inserting an in-line 47 mm Gelman




Type A filter in the air line at the entrance port  to the unit.  Only two




tests were performed.  In both, the beta  collection time for standardi-




zation  and measurement was 90 seconds.   In test 1 , the pumping time




was 23 minutes; in test 2, the pumping time was  5 minutes.




        Test 1 indicated a total collected mass of 432. 8^.g while test 2




indicated a total collected mass of 169. 5p.g.  Both of these  values were




unacceptable since the gauge specification was ± 30p.g  for 2 cfm flow




rate and ± 60)j.g for 4 cfm flow rate  and this latter flow rate was the




maximum attainable with the unit.  It was suggested that the cassettes




may introduce errors and the next group  of tests  were devoted to seeing




if that  error was constant.
                                                                    3-3

-------
3.1.2  Cassette Effects

       Seven cassettes were labeled A through G and installed in the

unit with no filter attached.  Here are the results:


                                      Total Indicated Mass
                Cassette No.               (Micrograms)

                    A                       +3.9
                    B                       +393.1
                    C                       + 358. 3
                    D                       -  42.9
                    E                       +243.6
                    F                       +249.6
                    G                       +82.7

These  tests showed that the error, if cassette introduced, was at least

not constant for all cassettes.  It was then suggested that possibly the

temporary interruption of the beam by the cassette as it moved into

position may cause a disturbance of the AGC.  To determine this, a

delay was introduced between the time the cassette was moved into the

source-detector station and the time the computer began accepting data.


3.1.3  AGC Delay Effects

       Four of the lettered cassettes used in the previous test were

again tested with these  various delays.  Here are the results:
Cassette No.
B
C
D
G
Delay
(seconds)
60
15
15
60
Total Indicated Mass
(micrograms)
- 191. 3
- 44.4
- 88. 9
+ 125. 5
                                                                    3-4

-------
While the grouping is tighter and generally of opposite sign from the

previous tests, these data were judged inconclusive and unacceptable.

It was  then decided to investigate the performance of the gauge with-

out either filter or cassette in the transfer mechanism.


3.1.4  Stability Tests

       These tests were performed by placing the unit in "automatic"

and allowing it to run with no cassettes in the loading magazine. During

test 9, it was noted that the count rate being  observed on the Nixie dis-

play of the Computing  Counter exhibited a sensitivity to whether or not

sunlight was shining on the floor of the test area.  On tests 10, 11,  and

12, precautions were taken to  ensure that the external light reaching the

gauge was  minimized.  On tests  13,  14, and  15, the front door of the unit

was opened to allow light to shine on the cassette transfer mechanism.

On test 16,  the light was again shut off.  Here are the result:

                                    Total Indicated Mass
              Run No.               (micrograms)	

                   1                       -1343.6
                   2                       -   50.6
                   3                       +  143.4
                   4                      -   33. 7
                   5                       +  100.8
                   6                       -   26.6
                   7                       -  183. 1
                   8                       +  270. 7
                   9                       +  131.4
                  10                       -   59.6
                  11                       +   68.1
                  12                       +   49.6
                  13                       +  131. 6
                  14                      +  470. 0
                  15                       -  301.4
                  16                       -   21.1
                                                                      3-5

-------
These tests confirmed conclusively that a light leak did exist in the

detector.   In tests 10,  11,  12 and 16 the values obtained,  while larger

than desirable,  were considered adequate for  this series provided flow

rates of approximately 4 cfm are used.  It was then decided to use filters

in the transfer  mechanism, taking care to see that external light

reaching the interior of the unit was minimized.


3.1.5  Stability Tests  -- Using  Filters

       Tests were performed again using Dr.  Anderson's medium on

the clean air previously discussed.  Up until this point little attention

had been paid to deformation of  the filter.  On tests 5 and 6, filter

deformation in the form of tearing of the filter away from the cassette

was  definitely observed.

                                  Total Indicated Mass
                Run No.           	(micrograms)

                   1                        - 86.4
                   2                        + 48. 1
                   3                        -   3. 0
                   4                        + 29. 0
                   5                        +292.6
                   6                        + 95. 9

On the basis of  these data, the decision was made to change media and

use the tougher Pallflex media for the remainder of the tests.  It was

decided to repeat the previous test using the Pallflex medium.


3.1.6  Stability Tests -- Pallflex Medium

        Tests were again performed using a 47mm in-line filter to

obtain clean air.  It had also been theorized at this point  that some

error may be occurring due to either moisture loss in the medium
                                                                   3-6

-------
or actual material loss from the filter itself.   This was as a result

of data from tests at IN prior to the Cincinnati test series that showed

the no-cassette stability to be  well within the ± 30|j.g total indicated

mass specification.   Thus six  tests were run using the same filter

cassette on all tests.

                                      Total Indicated Mass
                    Run No.              (micrograms)	

                       1                     -126.4
                       2                loss of record
                       3                     - 83.2
                       4                     - 58.6
                       5                     - 31.4
                       6                     - 27.4


On the basis of these data, it became fairly clear that some kind of

process  error was occurring.   It was not determined at this time

whether  tbe error was due to moisture or material loss.  Later studies

(Section  2.2.2 of this report) point tTie finger rather  strongly towards

moisture,  but no conclusion can be drawn from these studies regarding

the data  from either  the Cincinnati or Dow series of tests.

       This concluded the Cincinnati series except for some tests on

a laboratory aerosol which were unsuccessful  due to the shortness of

time available.  Even though limited data had been obtained,  the  series

was considered successful in that the obvious deficiencies in design  and

alignment had been pinpointed.


3. 2    The Dow Series

       The tests at the Dow Chemical Company in Midland, Michigan,

during the  week of 25-29 October, 1971, may be divided into three
                                                                 3-7

-------
 general areas: tests on 1) leaded fuels (3. Occ/gal TEL),  2) unleaded




 fuels (0. 0 cc/gal TEL),  and 3) low-leaded fuels (0. 5 cc/gal TEL).




 Three different cars were used for the three different areas and will




 be discussed later.  Tests were performed with the automobiles operating




 under steady state conditions (constant speed), the Federal LA-4 cycle,




 and the older California Seven-Mode cycle.




        The Dow facilities were quite well equipped and similar in




basic construction to the EPA's Cincinnati facilities.  An isokinetic




mixing tube was used,  in which the overall flow was 500 cubic feet




per minute.  Two samples were taken by Dow for gravimetric com-




parison to the beta gauge.  Initially these samples were two 143 mm




filters but later consisted of one 143 mm filter and one  2  inch filter.





        Prior to the actual automotive tests, a test was run using a




 cassette prepared with a filter as the medium (all tests but one used




 Pallflex E70/2075W) was delivered from the manufacturer.  This filter




 showed a weight loss  of  70. 6 micrograms. In order to test the stability




 of the unit (± 30 micrograms was the specification) two  tests were run




 with no cassette at all in the instrument showing +10. 9fJ.g and +4. l|j.g,




 respectively.  Convinced that electronically the unit was  well within




 specification,  several cassettes were prepared by baking at various




 temperatures for various times and allowed to stabilize for various short




 times in the ambient conditions.  This baking discolored the filter




 somewhat, possibly introducing contaminants (no discoloration had
                                                                  3-8

-------
       ever been noticed before at idential temperatures), and as a result

       this conditioning was no longer employed after this series.  The pump

       was then allowed to draw clean air through the filter for a short time.

       The results are as follows:
            Bake Time    Bake Temp   Equalize Time    Pump Time    Weight Change
Test No.       (min)          ( ° F)           (min)	      (min)        (micrograms)
1
2
3
4
5
6
10
10
10
10
20
—
210
210
210
210
200
	
10
20
30
45
30
—
5
5
5
5
20
20
-54. 0
-93. 2
-83.2
-55. 4
-48.2
-77. 8
       Test No. 6 is a repeat of the cassette used in test No. 5 with no inter-

       mediate baking.  That is to say, the cassette was  immediately reprocessed

       through the beta gauge to determine whether the same phenomena which

       existed during the Cincinnati series (Section 3. 1. 6 of this report) could

       be duplicated here.  Obviously, it was not.

              The average of the above six tests was  -68. 6 micrograms, and

       represented  a "process" error (since all readings are negative and in

       the 50 to 90 microgram category).  Since no solution to this  problem was

       immediately apparent, it was decided to delay a more detailed study in

       this area since the first  automotive series would be run with leaded fuels

       and the mass of collected particles was expected to be much higher than

       this figure.
                                                                              3-9

-------
3.2.1  Tests on Leaded Fuel (3. 0 cc/gal TEL)




       On tests 1  through 8, the flowmeter was placed downstream of




the pump.  After these 8 tests,  it was determined that the pressure




fluctuations introduced into the  air flow by the  pump was  interferring




with the vortex precession action of the Swirlmeter.  For all subsequent




tests at Dow, the flowmeter was placed upstream  of the filter.




       A  synopsis of this series is presented in Table 3-1.  Little




conclusion can be drawn from tests 1 through 8 due to pump interference.




In tests 5, 9, and 10,  it is suspected that there was electrical noise




introduced into the power line,  causing  the computer to malfunction.




This problem did not occur for  the remainder of the Dow  series.




Tests 11 through 15,  however,  showed excellent agreement between




the beta gauge and a gravimetric comparison.






3. 2. 2  Tests on Non-Leaded Fuel
       The next series of tests were performed using an automobile




which had been fueled solely with non-leaded fuel since its manufacture.




Considerable discrepancy existed in cases between the beta gauge and




the gravimetric comparison.   This  phenomenon is not clearly under-




stood as  of the date of this report.  To provide additional data,  and since




good agreement was obtained between the two 143 mm gravimetric com-




parisons, it was decided to substitute a 2 inch diameter  filter for one




of the 143 mm diameter filters for test 20 and all subsequent tests.
                                                                   3-10

-------

Test
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Test
Mode
Moist. Chk.
30 mph
30 mph
30 mph
--
60 mph
LA-4(C.S. )
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
60 mph
60 mph
60 mph
30 mph
1 5 mph
Pump Time
(minutes)
5
20
20
20
--
20
23
10
23
23
10
10
10
15
15
IN Mass
Gain ((jig)
-14
2980.
2491
2485
--
2366
3230
2151
--
--
3627
3870
1929
1502
1437
TABLE 3-1
LEADED FUEL TESTS

IN Total
Flow (liters)
595
2315
1935
1840
--
874
1022
787
--
--
453
795
417
1184
1382
Dow 143 mm
Gain (^g)
--
3650
3850
3600
--
13200
14000
4050
--
--
9150
5800
5750
2400
1750

IN Rate
(mg/min)
--
18.2
18.2
19.1
--
38. 3
45.8
39. 1
--
--
113. 3
68. 9
65.5
18. 0
14. 7
Dow Rate
(mg/min)

22.8
24. 1
22. 5
--
82. 5
76. 1
50.6
--
--
114.4
72. 5
71.9
20.0
14.8
% Dev.
IN - Dow
Dow

-20. 1
-24.2
-15. 1
--
-53.6
-39. 8
-22. 7
--
--
- 1.0
-5.0
- 8.9
-10. 0
- 0.6
Remarks

A
B
B
C

B

C
C
D
A



NOTES:    1.   Dow Flow Rate for all tests -  4 CFM
           2.   Mass Flow Rates are from automobile
           3.   Fuel is 3. 0 cc/gal TEL
           4.   Vehicle is 1971  Chevrolet - 350 CID Engine
REMARKS:   A.    Flow not constant throughout test
              B.    Difficulty adjusting Flow -  Pump
                    Interference Suspected
              C.    Run Aborted -  Computer did not cycle
              D.    Flow meter upstream of filter for all
                    subsequent Dow tests.

-------
       This approach would provide data as to whether the phenomenon




was area sensitive.   This was indeed the case and there are some later




indications that the indicated mass gain of the filter may be a function




of different filter manufacturing lots.  The data regarding this, however,




is inconclusive at this point.  Tables 3-2 and 3-3 synopsize the data




from the non-leaded tests.  A significant observation is that the per-




centage deviation between the 2 inch and 143 mm sample is of the same




order as the beta gauge (20 mm diameter  sample) and the  143 mm sample.




This phenomenon is currently attributed to a higher moisture absorption




and absorption of high volatility organics by the larger filter.  Inasmuch




as excellent data had been obtained with leaded fuels,  it was decided




to discontinue the non-leaded fuels and  switch to low-leaded fuels.






3.2.3  Tests on Low-Leaded (0. 5  cc/gal TEL) Fuels




       This series was also run using a 1971  Chevrolet with 350  CID




engine.   There was some attempt during this series to vary certain




parameters surrounding the sampling process and determine their effect




but little conclusion can be  drawn from these attempts.  Tables 3-4 and




3-5 present a synopsis of the low-leaded tests.




       Here again the beta  gauge agreed quite closely in several  cases




with the  2 inch diameter filter.  It is interesting to note that as the test




mode switched from the LA-4 cycle  to steady-state operation (tests 26




and 27),  the polarity  of the  percentage deviation between the two inch




filter and the 143 mm filter reversed, causing the beta gauge to give
                                                                    3-12

-------

Test
No.
16
17
18
19
20
21
Test
Mode
LA-4(C. S.)
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S.)
Calif. Cyc.
Pump Time
(minutes)
23
10
10
23
23
60
IN Mass
Gain (^g)
1049
156
33
230
197
359
TABLE 3-2

IN Total
Flow (liters)
2454
1208
736
2075
2378
4750
Dow 143 mm
Gain (jig)
4050
1350
1400
1400
1400
2200

Dow 2 in
(Gain (p.g)
--
--
--
--
200
600
IN Rate
(mg/min)
6. 05
1.83
.64
1. 57
1. 17
1.08
Dow 143 mm
Rate (mg/min)
22.01
16.87
7.61
7.61
7.61
5. 79
Dow 2 in
Rate (mg/min)
--
--
--
--
1. 32
1. 92
TABLE 3-3
Test
No.
16
17
18
19
20
21
% Dev.
IN -143 min
143 min
-72. 5
-89. 1
-91.6
-79. 4
-84.6
-81.0
% Dev.
IN -2 inch
2 inch
--
--
--
--
-11. 3
-43.8
% Dev.
2 inch- 143 min
143 min
--
--
--
--
-82. 9
-66.8
Remarks





Test terminated at 47. 5 min. (Automotive Failure)
NOTES:    1.     Dow 143 mm flor - 4CFM
            2.     Dow 2 inch flow - 3. 3 CFM
            3.     Fuel is non-leaded
            4.     Vehicle is 1971 Chevrolet - 350 CID engine
            5.     2 inch filter had 6. 5 in Hg differential pressure
            6.     143 mm filter had negligible differential pressure

-------
TABLE 3-4
Test
No.
22
23
24
25
26
27
28
29
30
31
32
33
Test
Mode
LA-4(C.S.)
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
LA-4(H. S. )
60 mph
30 mph
1 5 mph
Idle
30 mph
60 mph
60 mph
Pump Time
(minutes)
23
10
23
23
23
15
15
15
15
15
5
5
IN Mass
Gain ((ig)
2796
960
251
298
959
4119
350
332
248
510
2308
1605
IN Total
Flow (liters)
1784
1222
506
506
1840
993
996
1469
1488
1472
496
495
Dow 143 mm
Gain ((ig)
1700
1600
2300
2400
2500
7000
1200
700
500
1500
3000
2200
Dow 143 mm
Flow (liters)
2600
1130
2600
2600
2600
1400
1400
1400
1400
1700
566
566
Dow 2 in
Gain ((jig)
8200
900
1300
1400
1400
600
400
300
200
500
1600
1200
Dow 2 in
Flow (liters)
2140
931
2140
2140
2140
1400
1400
1400
1400
1400
465
465
IN Rate
(mg/min)
22. 2
11. 1
7. 0
8. 3
7.4
58. 7
5. 1
3.2
2.4
4.9
65. 5
45. 7

-------
TABLE 3-5
Test
No.
22

23
24


25

26
27

28

29

30

31

32
33

Rate (mg/min)
38.6

20. 0
12.5


13.0

13.6
70.9

12.2

7. 1

5. 1

12.5

75.0
55.0

Rate (mg/min)
54.2

13.7
8.6


9.3

. 9. 3
52.5

4. 1

3. 0

2.0

5. 1

48.6
36.5

% Dev.
IN - 143mm
143 mm
-42. 4

-44. 4
-43. 9


-36.3

-45.7
-17.2

-58.0

-55.0

-53.4

-60. 8

-12. 7
-16.8

% Dev.
IN - 2 in
2 in
-59.0

-18. 7
-18.4


-10. 3

-20. 3
+12. 0

+26.0

+ 5.0

+16. 4

- 3.2

+34.6
+25.4

% Dev.
2 in - 143 mm
143 mm
+54. 1

-31. 7
-31. 3


-29.1

-31. 9
-26.0

-66. 7

-57.1

-60. 0

-59. 5

-35. 1
-33. 7

Remarks
IN Flow/2in Flow
not constant
Flows steady
Face velocities
matched on IN and
and 2 in
Gelman Type A
filter used

143mm - 2 in flow
rates matched
143mm - 2 in flow
rates matched
143mm - 2 in flow
rates matched
143mm - 2 in flow
rates matched
143mm flow returned
to 4 cfm
Doors to facility open
Doors to facility
closed

-------
an indication which is between the two gravimetric measurements.





There was a very evident change in moisture content between the




two modes with some condensate appearing on the clear plastic tubing




which was used during the La-4 cycle and no visible condensate during




the steady state runs.  The temperature of the  sample flow did not




appear to change appreciably on any of the runs besides the 60 mph




steady-state runs.   There is also a higher concentration of volatile




organics during the  La-4 cycle and this may also be a major contributing




factor.




       An experiment was also run by varying  the temperature of the




room in which the automobile was  located.  This was accomplished




by running a test with the front and rear doors  of the dynamometer




facility open,  then duplicating the test with the  door  nearest the front




of the automobile closed.  This produced  a temperature variation of




30°F in the room and tests 32 and  33 are the results.  Notice that there




is little difference between tests 27 and 32, both being run with the doors




open.  Test 33, however,  showed a significant  reduction in particulate




emission  resulting from raising the intake air  temperature to the engine.






3. 3    Conclusions  Regarding the  Field Tests




       The field tests at Dow and Cincinnati were sufficient to establish




the capability of beta gauging as a process monitor.   The detector




stability achieved and the use of digital processing throughout  assures




high computational accuracy and repeatability.   Much,  however,  remains
                                                                     3-16

-------
to be learned about the sampling process.   The fact that poor correla-



tion is achieved between filters of different diameters must surely be



more fully explored and understood.



       The data on moisture absorption by paper filters also indicates



this medium to be undesirable for most applications requiring resolu-



tion of 1  - 2% of filter basis weight.  Such materials as the  GE Nuclepore



or Gelman Type A do not exhibit as serious a problem in this area as the



Pallflex  E70/2075W and it is understood that these materials are under



investigation by the  EPA at the time of this writing.  Other  materials



must be  investigated and the .filter industry must be called upon to



address  this problem.



       Variations of a percentage of initial filter basis  weight would



affect both gravimetric and beta gauging identically.   The larger filter



face velocity of the beta gauge, however,  would tend to  concentrate the



material and result  in a larger basis  weight change for  the particles.



For example, the effective sample area for the beta gauge is approxi-



mately 3 cm  while  for the 2 inch  gravimetric comparison,  the  effective


                            2
area is approximately 10 cm .  At 4. 2 cfm for the beta gauge, this


                                           3           2
represented a normalized flow rate of 1. 4 ft /(min • cm ).   At  3. 3 cfm


                                                           3           2
for the gravimetric  unit, the normalized flow rate is 0. 33 ft /(min • cm ).



Obviously, then,  for the same stream concentration,  more  material will



be collected on the beta  gauge's filter  on a per-square-centimeter basis



and the moisture variance will  not be as significant.
                                                                     3-17

-------
       It is not the intent of this document to attribute the inconsistent




data solely to moisture.  Rather,  this was an area in which IN has




considerable expertise and facilities to investigate (at least under static




conditions) from our previous experience in the paper industry.  It is




the intent, however,  to point out to the reader that moisture can (and




in many  cases likely will) be a major cause  of inconsistent data.  The




effects of filter composition on "errors" in data are relatively unknown




to IN,  and as  a result, little positive or conclusive  data can be  presented




in this area.
                                                                   3-18

-------
4.0    RECOMMENDATIONS





       There are several areas of study and clarification revealed as





a result of this program.  First, it is believed that some definition is





required by the Environmental Protection Agency as to what constitutes




"particles".   This may eventually require an exact preparation of the





mixing tube air with respect to temperature and humidity.  It  may also





require withdrawing a sample from the tube at a reasonably fixed flow





rate (implying a constant pressure drop across the filter, barring any





filter clogging).





       In order to do the above, however,  the  problem of moisture





interference must be solved.   In IN's opinion,  this is one of the major





problems facing particle sample today.  It  should be of concern





not only to those involved with mobile  sources  but  also those concerned





with stationary source or ambient monitoring.  It is  not sufficient simply





to allow the gravimetric samples to stabilize in a controlled environment





room as Figure 2-5 of this report will verify.   In effect, in order to





place any validity on data, the complete previous history of the filter





must be either known or standardized  and the nature of each sample must





be known  and taken into consideration.  Solution may require either the





discovery of a medium already produced but not widely touted or  the





manufacture of a new medium,  using materials specifically designated





to ward off water.
                                                                  4-1

-------
       There is also limited data which shows that the composition




of the filter medium may introduce wide fluctuations in the data.   This




is attributed to an affinity (or lack of same) for highly volatile constituents




in the sample which are condensed from the  stream and which would




ordinarily remain in the gaseous state.  A study should be initiated




to determine the total nature of vehicular exhaust under both steady state




and Federal Cycle  operating  conditions.  This information could then be




used to specify (or eliminate) certain materials used in the manufacture




of the medium.




       It is believed that extensive experimentation will be required by




the EPA to establish correct and reliable sampling procedures.  The




proof of principle of beta gauging has been successful: a nucleonic gauge




can be built with the required sensitivity.  Further development in this




area should be directed towards specific application areas rather  than




general purpose application.  A  filter medium must be found, not  only




for the beta gauge but for  gravimetric comparison as well.  The IN




design is excellently suited to this type of experimentation by virtue




of the cassette approach and  should be able to yield valuable data in




this area.




       Finally, the Federal Cycle Test as currently understood by




IN is at best cumbersome. It is believed that solution of the above two




problems will yield far more consistent results than available in the




past.   It is also our belief that reliable data  may also show that a
                                                                  4-2

-------
definite correlation exists between a typical steady-state sample




(60 mph,  for example) and the LA-4 Federal Cycle Test.  If this




be the case,  a considerable  reduction in the complexity (and there-




fore cost) of an automobile particulate emission gauge is possible.
                                                                   4-3

-------
BIBLIOGRAPHIC DATA
SHEET
1. Report No.
3. Recipient's Accession No.
4. Title and Subtitle

     Development of a Nucleonic Particulate Emission Gauge
                                                5. Report Date
                                                  March, 1972
                                                6.
7. Author(s)
     Charles R.  Duke  and Boong Y.  Cho
                                                8. Performing Organization Rept.
                                                  No.
9. Performing Organization Name and Address
     Industrial Nucleonics Corporation
     Federal Systems  Division
     650 Ackerman Road
     Columbus, Ohio  43202
                                                10. Project/Task/Work Unit No.
                                                11. Contract/Grant No.

                                                 68-02-0210
12. Sponsoring Organization Name and Address
     Environmental Protection Agency
     Research Triangle Park,  North Carolina  27711
                                                13. Type of Report & Period
                                                   Covered
                                                 Final Report
                                                                     14.
15. Supplementary Notes
16. Abstracts
             A project was conducted to design,  fabricate, and test a particulate
     emission monitor utilizing a beta radiation attenuation and filter  sampler
     technique.   The  instrument was to have a minimum  sensitivity of 125 ±25
     micrograms per  cubic  meter and be capable of operation in either a manual
     or automatic mode  of operation.  Carbon-14 was selected as  the optimum
     isotope and a filter "cassette" was used to collect the particles.  The effect
     of moisture absorption by various types  of filter media were investigated
     and reported.  A series of tests were performed on automobiles using leaded,
     low-leaded and non-leaded fuels.   The results of these tests are reported as
     well as recommendations for future  effort.
17. Key Words and Document Analysis.  17a. Descriptors

     Automotive Pollution
     Beta Gauges
     Filter Samplers
     Filters
     Air  Pollution
     Moisture Interference
     Automotive Organics
     Automatic  Particulate Monitors
17b. Identifiers/Open-Ended Terms
17c. COSATI Field/Group
18. Availability Statement
        Unlimited
                                    19. Security Class (This
                                       Report)
                                         UNCLASSIFIED
                                                              ~*.^~^..~^>. .^^,
                                                         20. Security Class (This
                                                            Page
                                                         	UNCLASSIFIED
          21. No. of Pages
          sixty-two (62)
                                                          22. Price
FORM NTIS-35 (10-70)
                                                                              USCOMM-DC 40329-P71

-------