RESEARCH
GRANTS
AEROSOL MEASUREMENTS
IN
LOS ANGELES SMOG
VOLUME I
U. S. ENVIRONMENTAL PROTECTION AGENCY
-------
950R71014
AEROSOL MEASUREMENTS
IN LOS ANGELES SMOG
VOLUME I
Particle Technology Laboratory
Mechanical Engineering Department
University of Minnesota
U.S. ENVIRONMENTAL PROTECTION AGENCY
Air Pollution Control Office
Research Triangle Park, North Carolina
February 1971
-------
The APTD (Air Pollution Technical Data) series of reports is issued by
the Air Pollution Control Office to report technical data of interest to a
limited readership. Copies of APTD reports are available free of
charge to APCO staff members, current contractors and grantees, and
nonprofit organizations - as supplies permit - from the Office of Techni-
cal Information and Publications, Air Pollution Control Office, IL S.
Environmental Protection Agency, Research Triangle Park, North
Carolina 27709.
This report was furnished to the Air Pollution Control Office of the U. S.
Environmental Protection Agency by the Particle Technology Laboratory,
Mechanical Engineering Department, University of Minnesota in fulfill-
ment of NAPCA Research Grant No. AP-00839-01. The contents of the
report are reproduced herein as-received from the contractor. The
opinions, findings, and conclusions expressed are those of the authors
and not necessarily those of the Air Pollution Control Office of the U.S.
Environmental Protection Agency.
Air Pollution Control Office Publication No. APTD-0630
-------
Particle Laboratory Publication No. 141
Description of the Experiment
AEEOSOL MEASUREMENTS IN LOS ANGELES SMOG
APCO Research Grant No. AP-00839-01
I. Introduction - K.T. Whitby and B.Y.H. Liu
II. Background Information on the Site and Meteorological Experiments -
G.M. Hidy and S.K. Friedlander, with contributions from W. Green
III. University of Minnesota Size Spectra and Miscellaneous Experiments
R. Husar, N. Barsic, M. Tomaides, B.Y.H. Liu and K.T. Whitby
IV. Determination of Particulate Composition, Mass Concentration and
Size Distribution Using the Lundgren Impactor - D.A. Lundgren
V. University of Washington Aerosol Turbidity Experiments Using
Integrating Nephelometers - R.J. Charlson and N.C. Ahlquist
VI. Measurement of 03, NO, N02, S02 and PAN Using Continuous Gas
Analyzers - P. Mueller. K. Smith and Y. Tokiwa
Report submitted by K.T. Whitby - Principal Investigator
University of Minnesota
Department of Mechanical Engineering
Particle Technology Laboratory
Minneapolis, Minnesota 55455
November, 1970
-------
F-l
FOREWORD
This is the first detailed report describing the experiments conducted during
a collaborative study of smog aerosols in Los Angeles during the summer of 1969-
What was originally conceived as a relatively small experiment to measure aerosol
size distributions using the MAAS and to make some chemical analysis of collected
particulates, grew by the addition of investigators and experimenters to a relatively
large project that taxed the resources of several of the groups involved. However,
the outstanding cooperation from everyone, all done on an informal basis, made
the project a success.
Since this project may well serve as a prototype for further collaborative
research on aerosols, it is worthwhile to make a few candid observations about why
most of the work went well, and only a few things did not.
First, the informal collaborative arrangements used probably succeeded because
all of the key investigators knew each other well and were willing to commit them-
selves and their resources without a lot of time consuming red tape. A less well
acquainted group would probably need more time and perhaps more organizational
formalities to make things go smoothly.
Secondly, this project proved that first string equipment and personnel is
essential if good data is to be obtained and the work is to get done close to
schedule. It also proved the wisdom of thoroughly exercising the equipment in
the home laboratory, using the same people who will operate it during the collab-
orative study. With only a couple of minor .exceptions, no new and untried instru-
ments or procedures yielded much useful data.
n
Thirdly,' several weeks of preparation time is desirable before the main collab-
orative experiments are to be run. In addition to the inevitable instrument recalibra-
tions required for sophisticated instrumentation after shipment, there are always
unexpected "bugs" which occur. For example., a crash program to construct an inlet
bug screen had to be instituted when Lundgren found that small gnats were ruining
his impactor runs.
And then there was the night a whole days worth of data tapes were lost when
a janitor inadvertently emptied a wastebasket that had been pressed into service
as a punched tape receptacle when all the boxes provided for this purpose had
been used up.
Although the detailed data analysis has just begun, it is now safe to conclude
that the project has been successful in obtaining some very valuable coordinated
data. We expect that it will be several years before we have exhausted the possible
applications of the data.
This project, which Friedlander has called the "Great Smog Gaper", has been
most enjoyable. I am most indebted and grateful to the many participants for their
contributions to what has been a great experience.
Ken Whitby
-------
1-1
I - INTRODUCTION
by
K.T. Wiitby and B.Y.H. Liu
History of the Pro.ject
This collaborative research study on the physical and chemical
properties of Los Angeles smog aerosol by the group of investigators listed
in Table 1-1 developed out of discussions between Friedlander, Mueller,
and Whitby several years ago. These investigators came to the conclusion
that further insights into the mechanisms of formation and the behavior
of smog and of smog aerosols could probably best be developed by a collaborative
study carried out by a group of investigators, each well equipped with
apparatus and competent in his own area. Until now, most of our understanding
of behavior of polluted urban atmospheres has been derived from separate
studies by different investigators studying the aerosol in different places
at different times, and often using different techniques. While this disjointed
approach has yielded information about the general behavior of polluted atmospheres,
it has not been too successful in revealing the more complex relationships between
the aerosol and the gas phase reactions. Thus one of the prime goals of this
study was to carry out a comprehensive, collaborative effort, using a sufficient
variety of chemical and aerosol measurement techniques on the same aerosol at
the same place at the same time, so that a significant improvement in the
correlations between various measurements could be made.
Los Angeles was chosen for this study for several reasons. First, Friedlander
and Hidy at the California Institute of Technology and Peter Mueller, Head
of the California Air and Industrial Hygiene Laboratory, at Berkeley, California,
were located in the state of California and could utilize their resources for
a study located there. A second factor was that Los Angeles smog is an unusual
and severe pollution problem. The California Institute of Technology in
Pasadena was chosen as the site because of the availability of excellent
laboratory in the Keck Environmental Sciences Building and the excellent support
that could be provided by Dr. Friedlander's group. Although Pasadena does
not represent the center of the smog area, it is subjected to incursions of
heavy smog, and therefore a variety of smog conditions can be encountered
in a reasonably short time in the summer.
Dale Lundgren was asked to participate because of his experience in
using his impactor for mass distribution measurements and classification by
size for chemical analysis. The simultaneous measurements of the number distri-
butions using the Minnesota Aerosol Analyzing System (MAA.S) and of the mass
distribution using Lundgren's impactor provided a unique opportunity to compare
these two techniques.
-------
1-2
The participation of Charlson and Ahlquist from the University of
Washington made it possible to further investigate the relationship between
light scattering and aerosol mass and also for the first time made it possible
to compare the theoretical scattering calculated from the size distributions
with the measured values.
In addition to the principal investigators, the many other participants
in the project are listed in Table 1-2 along with the experiments in which
they participated.
Although the project was originally conceived by Whitby, Friedlander, and
Mueller, Liu actually provided most of the supervision in Los Angeles during
the project, and Charlson, Hidy, and Lundgren made important contributions
by providing their expertise in their respective areas. Furthermore, the
project could never have succeeded without the contributions of many graduate
students and technicians who worked on the project. Outstanding among these
are the work of Mr. R. Husar, Mr. M. Barsic, and Mrs. R. Husar, from the
University of Minnesota, and Y. Tokiwa and K. Smith from the California
Department of Public Health.
Objectives
Perhaps the most important objective of this study was to provide space,
time and technique coordinated data which could provide new insights into
the mechanisms of formation and the behavior of smog. These data can be
expected to provide new correlations between various instruments for the
continuous measurement of smog.
In addition to the abore general objectives, the various investigators have
stated their interest in using the data to study the applications listed in
Table 1-3.
Experiment Schedule
The major portion of the instrumentation was shipped to Pasadena at the
end of July, 1969- The first two weeks of August were used for setup and
checkout. Actual data collection with the complete system began on August 19
and ended on September 8 with a power supply failure in one of the major
instruments. Because 363 complete size spectra had been measured by then
under a variety of smog conditions, the primary size spectra experiments were
discontinued and secondary experiments such as smog making and classification
were performed until September 19, at which time the project was ended and
most of the equipment returned. A few of the gas analyzers were operated
into October.
The dates and time of the major experiments are shown in Figure 1-1.
-------
1-3
Summary Description of the Project
Although the following sections of this report give detailed descriptions
of the various experiments, the brief description given below is provided to
give an overview of the project.
The experiments may be divided into five main categories as follows:
1. Measurement of the number versus size spectra measurements over
the size range from 0.003 to 6 jm using the automated Minnesota Aerosol
Analyzing System.
2. Sampling of the aerosol by Lundgren impactor for mass and chemical
composition versus size distribution.
3. Continuous gas analyzers
4. Turbidimetric measurements
5. Miscellaneous experiments including:
a) cloud condensation nuclei and ice nuclei
b) samples for electron microscope and single particle chemical analysis
c) humidity effects
d) smog making in a bag
e) smog coagulation in a bag
f) particle beam experiments
The experiments, the techniques used and the investigators involved
are given in Table 1-2.
Sampling Site
Except for the meteorological instruments and the smog making experiments,
which were operated on the building roof, all of the apparatus was located in
a large air conditioned laboratory in the basement of the Keck Environmental
Sciences Building at the California Institute of Technology in Pasadena, California.
Sampling System
The basement location of the laboratory necessitated that the aerosol be
transported from the sampling line inlet 6.7 m above the roof down through
a vertical, 20.5 m long by 7 cm internal diameter PVC pipe to the aerosol
distribution piping in the basement. Important details of the piping system
are shown in Figure 1-2.
Flow velocities and tubing sizes for each instrument were chosen so that
the losses of the aerosol or gas being measured were small. Actual losses of
ozone and condensation nuclei, the two components for which it was suspected
there would be the greatest losses, were actually measured and found to be
on the order of 10$.
The impactors and total mass samplers were located at the bottom of the
vertical line to minimize losses of large particles. The high smog aerosol
-------
1-4
concentrations necessitated diluting the aerosol by a factor of 12 for the
condensation nuclei counter and by a factor of 100 for the optical counter.
All other instruments measured undiluted aerosol.
Minnesota Aerosol Analyzing System (MAAS)
The MAS consists of five parts: 1. The aerosol distribution and dilution
system described above and shown in Figure 1, 2. An optical particle counter,
3. A condensation nuclei counter, 4. The Vlhitby Aerosol Analyzer (WAA),
and 5. A data acquisition system (DAS) which controls the whole system and
records the data. A schematic of the system with the DAS channel assignments
is shown in Figure 1-3. This system is capable of completely automatic "in situ"
measurement of the aerosol number spectra from 0.003 to 6 }jm diameter with
very good accuracy. A detailed discussion of the accuracy and the calibration
procedures used is given in section III.
The optical particle counter (OPC) consists of a Royco model 220 sensor
with a modified,, passive sheath air inlet system and a Royco model 170-1
pulse converter. As operated in Los Angeles, the OPC has a sampling air flow
of 470 cm3/min, a sizing range of 0.4 to 6 jum linearly spread over 57 channels,
a resolution of better than 10$ and a maximum allowable aerosol inlet concentration
of about 50/cm3.
From Figure 1-3, it will be noted that the OPC is connected so that upon
command from the DAS, it will make a measurement cycle, record the data and
then wait for the next command. This permits time correlation of its data
with the other instruments being read and controlled by the DAS. Since all
data from the DAS and from the OPC is available on punched tape, computer
processing of the data is easy. The minimum time for a measurement cycle is
4 minutes.
The condensation nuclei counter (CNC) is a standard automatic General
Electric counter operated at an under-pressure expansion of 8 inches of Hg.
vacuum. It was equipped with a fixed ratio 12 to 1 capillary diluter to keep
the indication within the linear range of the counter. The counter output
was recorded both with the DAS and a strip chart recorder.
The Miitby Aerosol Analyzer (¥AA) is an automated commercial version,
manufactured by Thermo Systems, Inc., of St. Paul, Minnesota, of the instrument
described by Miitby and Clark (1966). This instrument has feedback controlled
high voltage power supplies which permit automatic operation in conjunction
with a Dymec DAS.
Some characteristics of the WAA used in Los Angeles are as follows:
minimum scan time = 4 min., sizing range 0.0075 - 0.6 pm, aerosol flow rate
= 11.5 1/min., charger volume = 1.7 1, and mobility resolution =
-------
1-5
Sampling for Mass and Chemical Distribution
During all of the major intensive sampling periods the aerosol was sampled
by Lundgren impactors and the filter sampler to obtain the aerosol mass dis-
tribution and to obtain size classified samples for chemical analysis for the
elements and ions shown in Table 1-2, line 3-
The aerosol was sampled for mass and chemical size distribution with
two Lundgren impactors (Lundgren, 1969). Most of the samples were of 12 hour
duration, with some of 4 hours, at a sampling rate between 60 and 85 1/min.
At 70 1/min. the cut sizes are 18, 5.6, 1.9> and 0.6 jjm. Samples were collected
on a 1 mil Teflon film for weighing and subsequent chemical analysis.
Total mass samples were collected simultaneously, at the same flow rate
as the impactor, onto 90 mm diameter Teflon filters.
These experiments are described in detail in Section IV.
After the conclusion of the main experiments, the WAA was also operated
in the classification mode to classify the aerosol in the 0.01 - 0.5 Jim range
into 4 fractions for mass and chemical analysis. Because this was the first
attempt at using the WAA for mass distribution analysis using a Teflon coated
collector electrode, only a few good data have been obtained. The usefulness
of these samples for chemical analysis must await processing of the samples.
Continuous Gas Analysis
An Atlas Electric DmoesED,, NC>2, S02, Oo, a Beckman 03, and a Mast 0^ continuous
recording gas analyzer provided by the California AIHL, Berkeley, were operated
for the entire project period. In addition, an experimental Chemiluminescence Oo
analyzer provided by J. Hodgeson, E.K. Stevens and Andy O'Keeffe of the Division
of Chemical and Physical Research and Development of NAPCA, Durham, North
Carolina, and an experimental peroxyacetyl nitrate (PAN) analyzer provided by the
Statewide Air Pollution Center, Riverside, California, were operated part of
the time.
Data from these analyzers was recorded both on strip charts and on punched
tape by the DAS, to permit obtaining both instantaneous readings and hourly averages.
These experiments are described in detail in Section VI.
Atmospheric Turbidity Measurements
1. Description
The University of Washington experiment, conducted byProfessor Robert J.
Charlson and Mr. Norman C. Ahlquist, consisted mainly of light scattering
-------
1-6
measurements made with three integrating nephelometers. (Ahlquist and
Charlson, 1968-1969.) Dew point and instrument temperature were also recorded.
The instruments were:
1) A four channel instrument operating in four very narrow wavelength
bands located at 360., 436, 546, and 6?5 nm-
2) A broad-band device covering the wavelength range from 420 to 550 nm.
3) A device with a medium wavelength band located at 550 nm approximating
the response of the human eye.
2. Purpose
The initial purpose of this portion of the experiment was to gather data
on the light scattering properties of real Los Angeles smog. These data
necessarily must be compared and correlated with the measurements by the other
groups in order to put the results into proper perspective. The goals of
the experiments fall into three classes:
l) experiments with three nephelometers and one hygrometer alone:
a) correlation of broad-band and narrow-band light scattering coefficient
with wavelength dependence as a parameter. Los Angeles provides a sufficiently
variable aerosol for such a study.
b) relation of light scattering coefficient to humidity
c) relation of light scattering to visibility, utilizing weather
bureau data for visibility
2) experiments relating nephelometer and hygrometer to that from other
experimenters
a) relationship of wavelength dependence of light scattering (angstrom
exponent, °<- ) to size distribution (Junge exponent, ft )
b) mass concentration-light scattering correlation with oC as parameter
c) extinction due to light scattering compared to that by nitrogen
dioxide
d) correlation of light scattering with gaseous pollutants
3) experiments or interpretations which will arise as a result of the observations
These experiments are described in detail in Section V.
Miscellaneous Experiments
The considerable number of miscellaneous experiments conducted are listed
in Table 1-2.
Cloud and Ice Nuclei measurements were made since this project offered an
excellent opportunity to correlate them with the great variety of chemical
and physical measurements being made on the aerosol.
Electron microscope samples were taken both to see what the particles looked
like and for possible single particle electron microprobe analysis by the AIHL.
Solar radiation measurements by pyroheliometer; wind velocity, direction
and turbulence; and temperatures at the sampling line inlet and the instruments
inlet were measured to provide necessary supportive data.
-------
1-7
A few rough experiments were made in which the aerosol was humidified
by passage through a chamber containing water,, to find out how hydroscopic
the aerosol was.
Some smog coagulation experiments were performed by rapidly filling
a 56m3 polyethylene balloon with a blower and then measuring the size
spectra for times up to 5 hours as the smog in the closed bag coagulated.
The measured size spectra are being compared with those predicted by various
theories and with the observed day and night aerosols.
A number of smog making experiments were performed in which about 15 m
of ultra filtered smog was pumped into the balloon and the condensation
nuclei count measured with the G.E. counter while the bag sat in bright
sunlight on the roof. After the initial experiments in the large bag showed
a large and variable smog making potential at different times of the day, a
smaller automatic apparatus was constructed and operated continuously to see
how the "smog making potential" of the air fluctuated during the day.
Douglas Advanced Research Laboratory also made some atmospheric probes
in the vicinity with a LIDAR apparatus to see how these measurements would
correlate with the turbidimetric and aerosol spectra data.
Data Analysis and Publication Plans
Since the primary purpose of this report is to provide a detailed
description of the experiment as a background for interpreting the data,
it includes only such data as is necessary to illustrate the procedures.
As of the date of this report, all but a small part of the data on the chemical
analysis of the particulate samples has been reduced, and distributed to the
investigators. Analysis of the data is well along and papers are being written
for presentation as a group at the American Chemical Society's Kendall
award symposium in April, 1971.
One project review meeting of all of the principal Investigators was
held on February 28, 1970 at the University of Minnesota, and another is
scheduled for November 30, 1970 at Berkeley.
Final plans for distribution of the data to other than the participants
have not been finalized. Currently, the University of Minnesota's size
spectra data, the meteorological data and most of the gas analyzer data is
available as a computer printout and on a magnetic tape. Charlson nephelometer
data and hourly averages of the gas analysis data are available on punched
card decks. Lundgren's data is in table form. The chemical analyses of the
particulate samples are still in process but should be finished by the end of 1970.
Some thought has been given to providing all of this data on microfiche
for those who would like to have the complete data for further analysis. The
large amount of good data of a type that has never been obtained before that
has resulted from this project will undoubtedly mean that it will require
several years for complete analysis and publication of the results.
Sample Computer Printout
Figure 1-4-is a sample of the 363 pages of computer printout of the data
obtained by and results computed from the data recorded by the data acquisition system.
-------
1-8
Table I - 1
Investigators, Participants, and Their Addresses
University of Minnesota. Minneapolis. Minnesota
Mr. Nick Barsic
125 Mechanical Engng. Bldg.
University of Minnesota
Minneapolis, Minnesota 55455
Mr. Rudolf Husar
125 Mechanical Engng. Bldg.
University of Minnesota
Minneapolis, Minnesota 55455
Mrs. Rudolf Husar
139 Chemistry Building
University of Minnesota
Minneapolis, Minnesota 55455
Dr. Benjamin Liu
125 Mechanical Engng. Bldg.
University of Minnesota
Minneapolis, Minnesota 55455
Dr. Ken Whitby
125 Mechanical Engng. Bldg.
University of Minnesota
Minneapolis, Minnesota 55455
California Institute of Technology. Pasadena, California
Dr. Barton Dahneke
Institut fur Aerobiologie
der Fraunhofer-Gesellschaft
5949 Grafschaft
uber Schmallenberg (Sauerland)
West Germany
Dr. S.K. Friedlander
W.M. Keck Lab of Envirn. Hlth. Engng.
California Institute of Technology
Pasadena, California 91109
Dr. George Hidy
Science Center
Aerospace & Systems Group
North American Rockwell Corp.
1049 Camino Dos Rios
Thousand Oaks, California 91360
California Air and Industrial Hygiene Lab. Berkeley, California
Mr. W.A. Moser
Life Sciences Department
Autometics Division
North American Rockwell Corporation
Anaheim, California
Dr. Josef Pich
W.M. Keck Lab of Envirn. Hlth. Engng.
California Institute of Technology
Pasadena, California 91109
J.H. Seinfeld
W.M. Keck Lab of Envirn. Hlth. Engng.
California Institute of Technology
Pasadena, California 91109
Dr. Peter Mueller
Air and Industrial Hygiene Lab
California State Department of Public Hlth.
2151 Berkeley Way
Berkeley, California 94704
Y. Tokiwa
Air and Industrial Hygiene Lab
California State Dept. of Public Health
2151 Berkeley Way
Berkeley, California 94704
K. Smith
Air and Industrial Hygiene Lab
California State Department of Public
2151 Berkeley Way
Berkeley, California 94704
Health
-------
1-9
University of Washington, Seattle, Washington
Mr. N.C. Ahlquist Dr. Robert Charlson
University of Washington University of Washington
Room 305,, More Hall Room 305, More Hall
Civil Engineering Dept. Civil Engineering Dept.
Seattle, Washington 98105 Seattle, Washington 98105
Environmental Research Corporation, St. Paul, Minnesota
Mr. Dale Lundgren
Environmental Research Corp.
3725 North Dunlap Street
St. Paul, Minnesota 55112
Meteorology Research, Inc., Altadena, California
Mr. W. Green, Paul MacCready, et_.al. Dr. Underwood
Meteorology Research, Inc. Meteorology Research, Inc.
Altadena, California Altadena, California
Douglas Advanced Research Laboratory, Huntingdon Beach, California
Dr. Freeman Hall, Research Scientist
Douglas Advanced Research Labs
5251 Bolsa Avenue
Huntington Beach, California 92647
-------
Table 1-2
Summary of Experiments, Instruments, and Investigators
Experiment^
Instruments or Technique
Investigators
1. (a) Size spectra 0.003-6/im
by number
(b) Mass distribution below 0.5/A
(a) General Electric Nuclei Counter
(b) Whitby Electric Aerosol Analyzer
(c) Modified Royco 220 sensor + Hewlett-Packard
multi channel analyzer
(d) Dymec 25 channel data acquisition system
Whitby
Liu
Husar
Barsic
2. Sampling of particulates for mass
distribution and chemical analysis
(a) Lundgren impactors
(b) 75 mm Teflon filters
Lundgren
Mrs. Husar
3. Chemical analysis of particulates
Analysis for Pb, Fe, V, Zn, Si, Na, Mg, C(Non
, S0,~, Cl , and Bromine
Mueller
Tokiwa
K. Smith
4. Continuous gas analysis for NO,
N02, S02, 03, PAN, and H.C.
(acetylene index?)
Devices, NO sensor
Devices, N02 sensor
Devices, S02 sensor
Devices, 03 sensor
NO: Atlas Electric
N02: Atlas Electric
S02: Atlas Electric
03: Atlas Electric
03: Mast Development Co., #724-11
03: Chemiluminescence - AIHL
PAN: Statewide Air Pollution Center, Riverside, Cal.
H.C.: Calif. Air & Ind. Hygiene Lab., Berkeley, Cal.
Mueller
Tokiwa
K. Smith
Mrs. Husar
5. Turbidity measurements
(a) Charlson-Ahlquist integrating nephelometer
(b) Charlson-Ahlquist four wavelength integrating
nephelometer
Charlson
Ahlquist
Barsic
6. Cloud Condensation Nuclei
Meteorology Research Incorporated (MRl)
Diffusion cloud chamber
MRI-Underwood
Hidy
H
-------
Table 1-2, continued
Experiments
Instruments or Technique
Investigators
7. Ice nuclei
MRI Filter method
MRI-Underwood
Electron microscopy
(a) Electrical precipitation
(b) Particle beam
(a) Thermo Systems electrical sampler
evaluation AIHL & U. of M.
(b) Gal. Tech. apparatus
Mueller
Whitby
Dahneke
Friedlander
9. Meteorological measurements
(a) solar radiation
(b) wind velocity, direction
and turbulence
(c) temperature & humidity
(d) smog forecasts
(a) pyroheliometer
(b) MRI wind instrument
(c) Thermocouples Hygrometer & dew point
(d) from Los Angeles
Hidy
Barsic
Husar
10. Effect of humidity on aerosol
size distribution
Aerosol spectrometers (1) with humidification
and dehumidification
Husar
Liu
11. Coagulation of smog in a 56 nr
polyethylene bag
Aerosol spectrometers (1) - measurement as
function of time
Husar
Liu
Friedlander
12. Smog making experiments
G.E. Nuclei Counter, $6 nr polyethylene bag,
pyrex chamber and sunlight
Husar
Liu
Friedlander
13. LIDAR probing of atmospheric
aerosols
Douglas Advanced Research Lab., LIDAR
Hall
-------
Table 1-3
Proposed Applications of the Experimental Results
1-12
5.
6.
7.
Comparison of experimental measurements
of size spectrum with theory
Relationship between aerosol and gas phase
a. What fraction of aerosol originates in gas?
b. Is condensation process homogeneous or
heterogeneous?
Development of models for photochemical smog
formation in urban basins for both particle
and gas phases, eventual application to
standards setting
Relationship between turbidity measurements
and mass concentration models
Inadvertent weather modification
Comparative performance of oxidant analyzers
Effect of humidity on size distribution
Correlation of size spectrum with
- gas concentrations
- aerosol chemistry
- mass concentration and distribution
- electrom micrographs
(Pich
(Husar
(Liu
(Friedlander
(Whitby
(Friedlander
(Mueller
(Whitby
(Friedlander
(Seinfeld
(CharIson
(Hidy
(Mueller
(Husar
(Liu-Whitby
Univ. of Minn.
with other
investigators
-------
SCHEDULE OF
THE AEROSOL MEASURE-
MENTS AT CALTECH.
ATM. SIZE SPECTRA
BAG EXPERIMENT
CLASSIFICATION
SMOG MAKING
MON.
AUG. 4
II
18
25
SEPT. 1
8
15
TUE.
5
liMo
12
AMR
MINU
19
££
26
99 *;;;;*;:;•::*
2
9
16
;:•:::::
WED.
6
rRUMEr
13
PRE
20
27
: * '•.':': •'.•."•"••.*' v.v**{*."«
• %•"•*•/"• ''•'• "v/V*".*'*:
3
249
10
17
THUR.
7
sIT SE'
14
PARATIC
21
T 39
28
4
349. 350-;
:: ^/
II
18
;::;;:
FRI.
8
n 10
1 Ur
15
iM
'IN
22
98
29
•//;;;;•;.'•*
5
163 ..,..,,;
!".;**:'•'.
12
'. *; •
'•' '•.
19
SAT.
9
16
23
30
6
13
••' 'f'l 'I;
'•'•*:'::'.'.
20
SUN.
10
17
24
31
7
/'"\ '".'i •
14
21
Figure
1-1 Los Angeles Smog Aerosol Experiment Schedule
-------
SMOG MAKING
HI-VOLUME
SAMPLER
OjCHEMILUMIN.
CONDENSATION
NUCLEI COUNTER
CHARLSON AHLOUISTr^-
NEPHELOMETER 7.34!
13.8
-
r.
0.3
7.3
1.5
400
<*=
LJ
^
>l
L
N02 ATLAS
NO ATLAS
03 BECKMAN
-1
]
— sj
**H
:ALTECH
PARTICLE BEAM jj
—,'*\
=$'\\
1
1 '
STATIC SAMPLER |U_____^£
1 rU^ipi
r , ••
NUCLEI COUNTER I
- \=^^\
>l
SAMPLER ji
l'1
MRI
ICE NUCLEI
AEROSOL INSTRUMENTS
AEROSOL FLOW
CHEMICAL INSTRUMENTS
-FLOW RATE, L/SEC.
-RESIDENCE TIME OF THE
AEROSOL IN THE PIPE
SYSTEM , SEC.
-PIPE DIAMETER, CM.
GAS METERS
Figure
1-2 Schematic of the piping system used to transport
aerosol to the various experiments
-------
1 pvROHFI IOMFTFR
2WtNn *sPFFn
3. WIND TURBULENCE , CT
. WIND DIRECTION
5 AMBIENT TEMP
6 . SAMPLING LINE TEMP
7 AEROSOL ANAI Y7FR TFMP
8RFI ATIVF HUMIDITY
9ADCM
. UPEN
IO . DATE CODE
II 19 i^ nPFM
II , l£ , IO . VJrC.li
14 ATI A4? n .. .. ,
1*5 MAtT O
Ifi ATI L<\. lun -
1 7 ATI A 1 MH
1 f . Ml I_HO IMVJo
10 /^nffci
IB . OPEN
IO II AF UU MPDUI C"7R MM ... .
i y • \J. vjr v» . i»CrH. or o IMIVI.
or\ 1 1 /->F uu MFPH Sdf; MM
21 . U. OF W. NEPH. 436 NM.
OO I 1 r\F UU MC"DU ^CO MM
C.C. . U. Ur VY. ntrrl. ODU IMIVI.
j1"*! MRI IMFPHFI OMFTFR
24 G E NUCLEI COUNTER
_^_oiv U/UITDV Aconcni AMAIVTCD
^^ tO . Wnl 1 BY AtnUoUL AIMALT/.CK
MPUT SCANNER
I
WHITBY
AEROSOL
ANALYZER
OPTICAL
SENSOR
DIGITAL
VOLTMETER
— PRINTER — j,
— •• COUPLER
PUNCH ,5(5
! !
1 1
DIGITAL
CLOCK
i
1
SIGNAL
COMMAND
PULSE
Figure
1-3 Block diagram of the data acquisition system
showing inputs, interconnections and the data
outputs
"* 1
t
—
MULTI -
/•oLJAMMPI
VxrlAiMlNtL
ANALYZER
TELETYPE — ^ PRINTED PAPER
__^_ DDIMTETD O
^^ rnllMTtR a
PUNCH — ^PUNCHED TAPE
-------
RUN NO.291
SEPT , 3 196« 15/20
***DATF OF PROCESSING MAV , 197 -RBH
PYHOHELIOMF.TER
MIND VECTOR
MIND DEVIATION
MIND SPEED
l .oi G.CAL/CH**?»MIN
2QC -I4 CEGR, CW FROM N
vf.66 KILOMETERS/HOUR
ENVIRONMFNT*«***«•»**»**
ROOF TFMP 87.4 F.30.8 C
LINE TFMP 83,6 F,?8.7 C
rtAA TPMP 85.1 F.29.5 C
HEL, HUM. 32.27 PERCENT
*»*******NEPHEbCMPlER DATA************
.OOOl77 1/MFTEHS
.000^56 i/HFTERS
,000385 I/METERS
.000531 I/METERS
UOFH 675 NM. BiSCH
UOFW 546 NM. Btecn
UOFW 436 NM. B-1N
NCDMN
CUMV
NCUMV
o.oooE-oi o.onoF-pi o
0.688F+07 o,338F»n4 o
|7 0.*79F+p4 o
|7 0.2B3F+04 o
,oooE-oi o.oooF-oi o.onoF-oi
ll o.344F»o5 n.J'lE+no
,2 0.593F + o5 |
l2 o,737P+Q5 |
0.972E+05 0
0,8i3F. + f|5 Q
0.469E+05 0
0.343R+05 0
0.129F+05 u
0.272E+0* 0
E R DAT
0.766F. + Q3 0
0.125E+Q3 o
0.5l6F*f|2 o
0.221F. + 02 o
0
0
0.326E+01 0
|4 o.37lE*t>2
.247E+o4 Q.371E+02
,323F+n4 o.606E*02
,278E+p4 0.638F+02
0.912E + fl5 O.S78F. + 00
0.953E»p5 ft,9lHF+no
o,972F+n5 ni'37E+no
0.993F»n5 0.95AE+00
o.nnof-oi o.oooE«o
0.352E-01 0.**4F-n
0.105F+00 0.139F'd
0.223F+00 0.294F.Q
0.387F*no •=--'• -
0.2.-'.7F-0
.254E+04
.ll9E+n3
,7n6F*o2
,4l2F+o2
.279E+n2
,2ftOF+02
0.610F+02
0.341E+02
0 . '• • - u »
0.529F+01
0.102F*o6 |
0«103F*n6 o«996F+oo
0.10*E+o6 o.9g9F+no
O.IOOE+OI
fl ,100E*01
n.innE*oi
0 .lOOE + 01
II o«535e'0
0.565F»ni 0«746F»o
0.ln5F+o2 0.138F*o
0.21oF*02 0.277F. + Q
0.27lE + fl2 o.351F.»p
0.3o9F+n2
ll 0.1l8F + p2
0.346E*00 0,535e»nl
n.H9E*oO 0.282F + 01
0.158E+00 0.54lF+ol
ll 0.527F*oi
0.35ZP+01
0.359F+Q1
0.104F»o6
0.10*P+n6
0.104F»p6
0.10
-------
SECTION II
BACKGROUND INFORMATION ON SITE
AND METEOROLOGICAL EXPERIMENTS
Pasadena Smog Experiment
August - October, 1969
by
G.M. Hidy and S.K. Friedlander
with contributions from
W. Green
1. Introduction
This portion of the description deals with three aspects of the experi-
mental program. First, the general character of the observational site is
outlined, covering an inventory of sources, and the meso-scale meteorology
of the Los Angeles basin. In the second part, more details are given about
the physical site in Pasadena. Finally, a brief summary of the meteorological
instrumentation and support for the program is presented.
2. Inventory of Sources Around Pasadena
The general statistics concerning the nature of the pollution sources
in the Los Angeles basin are given in Table II-l. The magnitude of the
emissions from major sources are indicated in Table II-2. From these data
one can readily see that motor vehicles make up the bulk of the sources
of gaseous and particulate pollutants except for sulfur dioxide. However;
stationary sources of various kinds also contribute a substantial fraction
of material to the Los Angeles atmosphere. Pollutant sources have been broken
down in further categories in Ref. (1).
Like most of the urban area in the Los Angeles basin, Pasadena is
interlaced and surrounded by high speed highways, so that this community
receives a dose of automobile exhaust from local sources each day. Unfortunately,
it also is susceptible to pollutants sweeping in via the winds from densely
populated areas to the south-and west, as will be shown later.
The geography of the Los Angeles basin with major highways is sketched
in Fig. II-l. Fuller et al.(1) break down the typical stationary sources in
this region into the following categories: (a) chemical processing equipment,
(b) boilers and heaters, (c) paint bake ovens, (d) incinerating equipment,
(e) melting equipment, (f) concrete plants, (g) petroleum processing plants,
(h) rendering equipment, and (i) power plants. Except for the last source,
all are concentrated mainly in the zone south of Pasadena from Inglewood to
Whittier to the coast at Long Beach. There are major sources of electrical
power just to the south and to the west in Pasadena and Glendale, accounting
for about 10$ of the total rated output in the basin. At the present time,
the various sources listed above produce a variety of pollutants. The power
plants, however, release mainly nitrogen oxides to the atmosphere.
-------
II-2-
Table II-l
GENERAL STATISTICS
for 1968-1969 (Ref. 1)
ITEM
Population
Los Angeles County Land Area
Los Angeles Basin Land Area
Solvents Used (Emitted)
Refinery Crude Throughput
Fuels Burned During Rule 62 Period
Oil Containing More Than 0. 5% Sulfur
Oil Containing 0. 5% or Less Sulfur
Natural Gas
Refinery Make Gas
Fuels Burned During Rule 62. 1 Period
Oil Containing More Than 0. 5% Sulfur
Oil Containing 0. 5% or Less Sulfur
Natural Gas
Refinery Make Gas
Gasoline-Powered Vehicle Registration
Gasoline Consumed
Diesel-Powered Vehicle Registration
Diesel Fuel Consumed by Motor Vehicles
Jet-Powered Aircrafe Flights/Day
Jet Fuel Consumed (Within L. A. County)
Piston-Driven Aircraft Flights/Day
Helicopter Flights/Day
Aviation Gasoline Consumed (Within
L. A. County)
QUANTITY
7,300,000
4, 083 Sq. Miles
1, 250 Sq. Miles
1, 000, 000 Lbs. /Day
731, 000 Bbls. /Day
193, 200 Gals. /Day
59, 430 Gals. /Day
1, 894, 400 MCF/Day
272, 400 MCF/Day
487, 400 Gals. /Day
2, 828, 100 Gals. /Day
2, 128, 200 MCF/Day
277, 200 MCF/Day
3, ^50,000
8, 000, 000 Gals. /Day
14, 500
150, 000 Gals. /Day
1, 505
330, 000 Gals. /Day
9, 650
590
1 10, 000 Gals. /Day
(c)
(c)
(a) Rule 62 period is from April 15 through November 15
(b) Rule 62. 1 period is from November 16 through April 14.
(c) Burned under variance granted by Hearing Board.
-------
II 3-
Table_ ll-2
EMISSIONS
-------
II -4
Typical contaminant concentration levels in the area around Pasadena
(West San Gabriel Valley) for the summer months are listed in Table II-3-
In terms of other regions of the Los Angeles Basin, these values suggest
that ozone concentrations are relatively high while the other pollutants
listed are moderate to low compared with the more urban areas to the west
and south.
3. The Meteorology of the Los Angeles Basin
Several general studies have been made of the meteorology of the Southern
California region. Excellent summaries of measurements prior to the late
1950's are given by the Air Pollution Foundation.(2,3) More recent data are
available from the Weather Bureau. (4) In general, the meteorology near the
ground in the Los Angeles basin is dominated by the influence of the neigh-
boring ocean and the peculiarities of the local topography.
The broad scale features that characterize Los Angeles weather are (a)
the Pacific high pressure zone which dominates the synoptic scale atmospheric
motion from early spring until early fall, (b) the continental high pressure
region over the deserts and high plains to the east and north which is present
much of the period from fall through winter, and (c) the winter passage of
cyclonic storms originating to the north, south, and west over the Pacific.
The conditions which seriously restrict the dispersion of material in
the planetary boundary layer develop most frequently in the summer months
and in early fall when the Pacific high induces a subsidence inversion over
the Basin. The resulting elevated inversion together with the blocking of the
surrounding mountains forms a box-like trap for pollutants discharged into the
atmosphere. Periods where a subsidence inversion exists are accompanied by a
strong on-shore sea breeze during the day and diminished winds at night.
Sometimes, the wind reverses direction at night to blow off-shore as a result
of nocturnal radiation cooling and down slope drainage.
Outbreaks of air from intense high pressure areas over the deserts to
the east create the Santa Ana condition, often observed during the fall.
This condition is marked by strong dry winds sweeping across the Basin from
the northeast, pushing the contaminants out to sea. As the Santa Ana winds
diminish, however, shallow inversions sometimes form by radiative cooling
during the night which can hinder dispersion of pollutants emitted at the
earth's surface. This class of inversion tends to dissipate quickly during the
day in contrast to the elevated subsidence inversions associated with the
Pacific high.
In late fall and winter the passage of cyclones creates conditions of
deep mixing near the ground giving rise to favorable dispersion conditions.
Generally then, the diffusion meteorology of the Los Angeles basin is
identified with meso-scale (ten to a hundred of km.) horizontal ebbing and
flowing of air limited to a rather shallow surface layer. The depth of this
-------
Burbank
Ventur
Glendale
San'Gabriel
Mtns
Pasadena
\
Santa Monica
Mtns
Santa
Monica
PACIFIC
OCEAN
LOS
Santa
— — - ' -
Monica Fwy
ANGE!LES
San B. jFway
Inglewood
Downey
\
\
Torrance
Compto
Azusa
Long
Beach
Pomona
Anaheim
Santa
Ana
Newport
Beach
Figure II-l
A map of the Los Angeles area showing major highways
and communities around Pasadena.
-------
II-6-
Table II-3
Typical
in the
Air Pollution Concentrations
West San Gabriel Valley
for 1968-69 (in
°3
NO
X
CO
so2
a
0. 18
0. 76
37
0. 12
(From Ref.
b
0.09
0. 50
19
0. 07
ppm)
1)
c
0 04
0. 29
11
0. 06
d
0.02
0. 17
6
0. 02
a_ Average of the 4 highest daily maxima.
_b Average of the 4 highest consecutive 8-hour periods
£ Mean hourly average of 4 highest days based on 24-hour
average.
cl Mean hourly average.
-------
II-7
mixing region under the inversion base at any particular time defines the volume
of air available for handling the pollutant emissions. The altitude of the
inversion base varies, but averages about 600 meters over Los Angeles.
For this experiment in air pollution, the period of late summer and
early fall was chosen because this is the period of most frequent days for
serious pollution. During this period the weather is associated largely with
the Pacific high, and typical streamlines for surface air flow during the day
look like those in Fig. II-2. Here the sea breeze penetrates the basin from
the west pushing the pollutants emitted in the industrial zones northeastward
over Pasadena. At night, average wind patterns, as drawn in Fig. H-3, show a
reversal in flow out to sea bringing polluted air from the east back over
Pasadena before being swept out toward sea.
Since there is often limited flushing of air in the marine layer as
it ebbs out to sea, pollution may accumulate from day to day in the "tidal"
flowing air mass. Therefore, air passing over Los Angeles may experience a
buildup in pollutants over several days because of the persistence of the
diurnal ebb and flow wind patterns. Some evidence for this sort of buildup
is given by Pack and Angell.(5) Such conditions make the normal trapping of
pollutants in the marine layer worse and may be involved in inordinately severe
pollution episodes. In any case, we can expect that the pollution over Pasadena
will be a complicated mixture from many sources, probably dominated by motor
vehicle emissions.
4. Representativeness of Sample
Because of limited resources, it was possible to conduct the sampling
program only at one location. Pasadena is not centrally located in the
Los Angeles Basin. Based on the prevailing winds expected during the sampling
period, pollution should come into the sampling point from the west-south
sector. Because there are few major stationary sources within a 5 mile radius
of Pasadena, it is expected that the samples taken will represent a very
complicated mixture of material coming from concentrated stationary and
mobile sources over about a 20 mile or more radius to the Pacific Ocean.
The measured material will have aged varying degrees in the atmosphere before
reaching Pasadena, and will be mixed to a varying extent with material emitted
locally, primarily from power plants to the west, and motor vehicle traffic
near the site. Samples also may contain a spurious component resulting
from dust emissions associated with building construction adjacent to the
Keck laboratories.
Aside from the possible contamination of building dusts, it is expected
that the samples taken in Pasadena should satisfactorily represent at least
the northeast sector of the Los Angeles Basin. Whether or not the results
will be characteristics of the polluted atmosphere to the south and west in
the Basin must remain an open question until new sampling is conducted over
a broader area.
-------
Figure II-2 streamline chart for July, 1200-1800 PST
From USWB Tech. Paper 54, (1965)
-------
Figure II-3 streamline chart for July, 0000-0500 PST
From USWB Tech. Paper 54, (1965)
-------
11-10
In addition to the possibility of inhomogeneity of samples horizontally,
it is anticipated that the material collected will not be entirely representative
of the vertical distribution of aerosols in the Basin. It is known from other
work that the number, size and mass distribution will change with height
in urban and non-urban areas. However, it is not known to what extent the
chemical composition changes with height.
The samples taken at Pasadena are not likely to be representative of
the annual average character of the Los Angeles aerosol. Routine sampling
at other periods of the year, particularly during winter, indicate marked
seasonal differences in aerosol concentration in Pasadena. It is expected
that similar differences should exist in the chemical composition of the
aerosol.
At this time, we can only consider the samples taken in this study
reasonably characteristic of a mixture of partially aged and newly formed
aerosol in a northeast, quasi-suburban regime of the Los Angeles Basin during
the months of more intense smog formation.
5. Characteristics of the Observational Site
The various instruments were placed either on the roof or the basement
of the Keck Laboratories of California Institute of Technology. The building
is three stories, 36 feet 6 inches high with a parapet 41 inches high on the
roof. The anemometers were located -on the roof to the west 16 feet above the
roof. The PVC sampling tube, 2-7/8 inches O.D. was raised 22 feet above the
roof on the west side. The total length of the tube to the diluters in the
basement laboratory was 6? feet.
The long sampling tube may provide a deposition surface for particles and
gases. For fully developed turbulent pipe flow, and a perfectly absorbing
pipe wall, the fractional removal can be calculated from the expression
where c^_ and c2 are concentrations of particles or gases entering and leaving
the pipe, respectively, k the mass transfer coefficient, V the gas velocity,
L the length of pipe, and d the pipe diameter. For gas diffusion, the
Reynolds analogy can be used to estimate k:
k = f
V 2
where f is the coefficient of friction for the pipe flow. If, for example,
the Reynolds number is 10,000, 99$ of the pollutant gas will be removed over
a 60-foot length of 2-1/2" pipe. Thus gas concentrations at the top and bottom
of the sampling tube must be carefully checked because of the potentially high
-------
11-11
removal efficiency if the pipe wall is perfectly absorbing. The results of
these tests are described in the section on gas sampling (Mueller).
In the case of aerosol deposition, the Reynolds analogy is no longer
applicable because of the very small values of the particle diffusion coefficient.
Instead it is necessary to use an expression for the mass transfer rate which
includes a term for the Schmidt number, NSC, the ratio of the kinematic
viscosity of the air to the particle diffusion coefficient, D. If, for example,
the equation of Metzner and Friend(6) is used
k /f\5
V = 0.049(jJ NSc
for particles 0.01,4 in diameter (D=5.24 x 10~5 cm /sec), less than 2% are
removed over a 60-foot length. A smaller percentage of the larger particles
will be removed because the diffusion coefficient decreases with increasing
particle diameter. Thus particle loss by diffusion would not be expected
to be a serious problem.
Surrounding the Keck building are other multi-story buildings of the
Institute and a parking lot. Farther away is a residential area ringing the
entire Institute for a radius of approximately one mile. During the observational
period, a new building was being constructed next to Keck. The layout of the
Institute is sketched in Fig. II-4-
In a neighboring park, about one mile from Keck, the Los Angeles APCD
has a monitoring station measuring continuously winds, temperature, oxidants,
and particulates.
6. Summary of Instrumentation for Meteorological Observations
During the observational period of the experiment, limited meteorological
data were obtained either by direct measurement, or from local sources such
as the LA APCD or the weather bureau.
Direct Observations. Most of the direct observations were made from the
roof of Keck Laboratories. They include the parameters listed in Table II-4.
The output from instruments could be considered a bare minimum for monitoring
local changes in atmospheric dynamics during the experiment. The mean wind
and direction gives a good estimate of the steadiness and direction of air
flow near the ground in the region near the sampling site. It may be extrapolated
very roughly to suggest the direction and trajectory of air parcels entering
the area from distances farther than a few hundred meters. Any such extrapolation
should be made with great caution, however, because of the highly broken
urban "surface" surrounding Keck Laboratories. Any such rough surface can
contribute to spurious, non-representative wind patterns at heights less than
the characteristic height of the roughness elements. In a similar way, the
-------
H
Figure II-4 California Institute
of Technology campus, showing
location of the sampling mast
on the Keck Laboratory
CALIFORNIA BOULEVARD
-------
Table II-4
Meteorological Instrumentation at Keck Laboratories
Parameter
1) Mean Wind
(Magnitude &
direction)
2) Standard Deviation
of Wind Direction
3) Temperature
4) Humidity
5) Net Radiation
6) Aitken Nuclei
7) Cloud Condensation
Nuclei
8) Ice Nuclei
Instrument
or Method
Meteorology
Research, Inc.
Meteorology
Research, Inc
G. E. Continuous
Counter
MRI diffusion
counte r
MRI Ice Nuclei
Sampler
Sampling Rate
1 avg/hr
(strip chart)
1 inst. /ZO min
tape punch
same as above
1/20 min
punch tape
1/20 mm
punch tape
1/20 mm
punch tape
1/20 min
punch tape or
strip chart
2 lit/min for
5 min
Period
continuous
continuous
continuous to
Sept. 20
continuous to
Sept. 20
continuous to
Sept. 20
intermittent
operation with
MAAS, and
in October
intermittent
operation with
MAAS, and
in October
intermittent
operation with
MAAS, and
in October
Instrument
Reference
(7)
(7)
(8), (12)
(9)
(10)
(14)
H
H
-------
11-14
standard deviation of the wind direction will give a crude measure only of the
dispersion capability of the atmosphere. Fluctuations, for example., can arise
from a number of processes like eddy shedding from the parapet of the building
which will be non-representative of the air boundary layer away from the
building.
Both temperature and humidity were measured to characterize the thermo-
dynamic state of the air mass in the Basin that was associated with local
wind patterns. Care was taken to shield the thermometer mounted on the roof
so that temperature observations should be reliable to within 1°C or less.
Humidity was measured as relative humidity at the lower end of the sampling
tube just upstream of the Minnesota Aerosol Analyzing System (MAAS). Therefore,
care should be taken to correct the measured value to the outdoor temperature
for any interpretation of the data requiring identification of atmospheric
conditions.
The net solar radiation reaching the roof top was measured with an Epply
pyroheliometer. This device is a standard one and should give reasonably
accurate data for the total radiation received. For further discussion see,
for example, Ref. (11). However, it gives no data for radiation received as
a function of wavelength, which may be of use in detailed interpretation of
the photochemistry involved in smog formation.
Aitken nuclei counts have only indirect bearing on meteorological
phenomena. Nevertheless, these counts provide a measurement which appears to
have some correlation with visibility and total aerosol loading. The monitoring
with the G.E. Continuous Counter provided an independent estimate of the smog
present and its variation, as well as being an integral part of the MAAS.
The device's limitations are discussed elsewhere and will not be outlined
further here.
Perhaps more relevant to the meteorology than Aitken nuclei are the
condensation nuclei activated at low supersaturations of less than ~L%. These
"cloud" condensation nuclei associated with air pollution are of considerable
interest now with respect to inadvertent weather modification and particularly,
warm rain modification.
The production of ice nuclei from air pollution and particularly from
automobile exhaust has been realized (Schaefer, 1968).(12) Although these
nuclei may not appear to alter the formation of clouds and warm rain over the
Los Angeles Basin, they may be very important to the physics of supercooled
clouds downwind of Los Angeles. Rain and snow in the mountains ringing the
basin and the activity of rain-producing clouds as far east as the Great Plains
may be affected by these pollution particles (Hidy et al).(l3)
To measure the "cloud" condensation nuclei a manually operated
MPJ Twomey-type diffusion chamber was used. Although the instrument is
analogous to a Rich counter in operation, the MEI-Twomey device relies on
-------
11-15
supersaturation of diffusing water vapor between two moist plates at two
different temperatures rather than expansion. Because of the curvature of the
vapor pressure vs. temperature curve for water vapor, a small supersaturation,
generally less than ~L%, is achieved as a result of the differential transport
of heat and water vapor between the two plates.
The MRI cloud condensation nuclei counter used in this study was of
standard design, using visual detection of 30° forward scattered light from
droplets formed in the supersaturation zone. After setting the temperature
of th? two moist plates to maintain a supersaturation at 0.3% to Q.1%, the observer
uses a hand pump to flush the entry lines and chamber with fresh air. The
chamber is closed, and after the sensitive region quiets down, the operator
then counts visually the number of droplets visible in the zone of supersatura-
tion. Several samples of air are counted for nuclei content in sequence.
This constitutes one sample. A run required several to give a reasonable
average value of nuclei counts.
The MRI-Twomey counter yields counts which agree qualitatively with the
range observed in several different areas with characteristic air masses.
However, the device is an experimental unit. As a result of manual operation
of the device and an element of observer error, any correlation between cloud
condensation nuclei and other aerosol measurements in this study will be
qualitative only. More quantitative comparisons should be made, of course,
in future experiments with newer improved instruments as they become available.
Because of difficulties with the cloud condensation nuclei counter, data
over only two extended periods, August 21 and August 26, were obtained.
Since the ice nuclei samples were of lesser interest, these were taken only
on the same dates as the cloud condensation nuclei.
Ice nuclei were sampled by collecting the aerosol on a millipore filter
and placing the filters in a specially designed diffusion chamber. After a
period of time at a fixed temperature below 0°C, water saturation, crystals which
have been grown in the chamber are counted on the filter surface. Sampling
for ice nuclei was carried out using a millipore filter in a vacuum line for 5
minutes at a flow rate of 2 lit/min. This method has the advantage over all
other ice nuclei counters in that the activation tempaature and supersaturation
can be controlled precisely.
Meteorological Data from Local Sources. In addition to the meteorological
observation attempted at the Keck Laboratories, several parameters were
recorded from local sources during the experiment. These are summarized
in Table II-5.
The first two items were monitored largely to check the consistency with
actual occurrences of these "interpretations" of available data by the local
agencies. The items, 3-5, on which the LA APCD smog forecast is partly
-------
II-le-
Table II-5
Meteorological Parameters Available
from Local Sources
Parameter
1 ) Forecast
2) Smog Forecast
3) Oxidant Level
4) Inversion Base
5) Visibility
6) Radiosonde (Temp. , dew
pt. , surface winds up to
2000 meters altitude)
7) Lidar sounding of
inversion base
Period
once per day
once per day
once per day
once per day
once per day
twice per day
August 26, 28
Source
USWB
LA-APCD
LA-APCD
LA-APCD
LA-APCD
USWB
Freeman Hall
(McDonnell -Doug las)
-------
II-l?
based were of interest for comparison with daily observations,, as well as giving
an index of what sort of episodes to expect in scheduling longer duration
operations of the aerosol counters.
In the Los Angeles area radiosonde soundings are taken twice daily
from Los Angeles International Airport. These soundings give temperature
and dewpoint as a function of altitude,, with only surface winds. For the purposes
of this study, only such data for the surface and up to altitudes just above
the inversion base have been included since these are most relevant to
our smog measurements.
Although the radiosonde observations provide a useful guide to the
atmospheric structure on a meso-scale, they will not be entirely representative
of local activity in Pasadena. Therefore, interpretation of the observations
at Keck in terms of the nations of air over the Los Angeles Basin necessarily
will be somewhat limited.
In parallel with our experiments, Dr. Hall and Dr. Ediger operated lidar
sounding device at the lower reaches of San Gabriel Canyon. Dr. Freeman has
found that the lidar can be used to trace the air motion near the San Gabriel
Mountains by watching changes in back scattering from suspended particles.
In addition he can deduce quantitative values of visibility from the lidar
return which will be compared with nephelometer measurements and with direct
measurene nts of size spectra using MAAS. Even though the lidar measurements
were made only in late August, they nevertheless will be useful as supplemental,
supporting data for main thrust of the experiment.
Acknowledgement s
The following people were involved in this portion of the project and
their efforts are acknowledged.
1. Paul MacCready (MRI)
2. William Green (MRI)
3. Walter Underwood (MEI)
4. T. Lockhardt (MRI)
5. Abdul Alkezweeny (MRI)
6. Len Doberne (CIT)
7. Freeman Hall (McDonnell Douglas)
8. Andy Moser (Autonetics-North American Rockwell)
Part of this portion of the project was supported by PHS Grant AP-00680-02.
-------
11-18
REFERENCES
1 Fuller, L.J., R.L. Chass, and E.G. Lunche, Profile of Air Pollution
Control in Los Angeles, Air Pollution Control District, Los Angeles
County, (1969).
2. Neiburger, M. and J. Edinger, "Summary Report on the Meteorology of the
Los Angeles Basin with Particular Respect to the Smog Problem , T.R. #1,
Air Pollution Foundation, Los Angeles (1954).
3. Neiburger, M., N.A. Renzetti, and R. Tice, "Wind Trajectory Studies of
the Movement of Polluted Air in the Los Angeles Basin", T.R. #7;
Air Pollution Foundation, Los Angeles (1956).
4. U.S. Weather Bureau Tech. Paper #54, (1965).
5. Pack, D.H. and J.K. Angell, Monthly Weather Rev., 91, (1963).
6. Metzer, A.B. and W.L. Friend, Can. J. Ghem. Eng. , 3_6, 235 (1958).
7. Meteorology Research Inc., Manual: Wind Diffusion Recording System
Model 2040, IM-112.
8. The Epply Laboratory, Bulletin No. 2.
9. General Electric Co., Instruction Manual - Condensation Nuclei Counter,
GEI # 45069, Feb. 1967.
10. Fletcher. N.H., Physics of Rainclouds, Cambridge Univ. Press, 1962.
11. Kondratyev, K., Radiation in the Atmosphere, Academic Press, N.Y.,
Chap. 2 (1969).
12. Schaefer. V.T., "New evidence of inadvertant modification of the atmosphere",
in Proc., of the 1st Natl. Conf. on Weather Modification, Albany, N.Y.,
April-May 1968, p. 163-172.
13. Hidy, G., R. Bleck, I. Blifford, P- Brown, G. Langer, JP. Lodge, Jr.,
J. Rosinski, and J. Shedlovsky, "Observations of Some Dynamical Properties
of Aerosols over N.E. Colorado", MCAR TN.49.National Center for
Atmospheric Research, Boulder, Colorado (1969).
ll+. Alkezweeney, A., "Ice nuclei measurement by millipore filter technique",
Research Rept. No. Ill, MRI Annual Report FT 1969, Ariz. Wx Mod Res.
AOG to U.S. Bur. Rec., Denver, Colo., Contract # 14-06-D-6581.
-------
SECTION III
MINNESOTA AEROSOL ANALYZING SYSTEM
1) Introduction - K.T. Whitby
2) Pump system, diluters, plumbing - M. Tomaides & R. Husar
3) Data Acquisition System - N. Barsic
4) Condensation Nculei Counter - N. Barsic & K.T. Whitby
5) Whitby Aerosol Analyzer - R. Husar & K.T. Whitby
6) Optical Particle Counter - B.Y.H. Liu
1. Introduction
The Minnesota Aerosol Analyzing System (MAAS) is probably the most
complex set of apparatus that has ever been assembled to date for the in situ
size distribution analysis of aerosols. Although our laboratory had been
developing this system for a number of years, the system that was used in
Los Angeles was new enough in many respects so that it took the best efforts
of everyone in our laboratory to get it all working and to get it calibrated
in time for the project. Because this system is unique and because it may
well serve as the prototype for similar systems in the future, we have
described it and its characteristics in considerable detail in this section.
¥e have discussed the calibration of all of the instruments in
considerable detail and the calibration of the Whitby Aerosol Analyzer in
particular detail because a knowledge of its performance and limitations
is essential to several conclusions which will come out of this study.
Although a fairly detailed calibration of the WAA had been made before
the project, the calibration presented here was performed by R. Husar as
part of his Ph.D. thesis after the project.
Although the accuracy of each instrument is discussed in detail in each
part of this section, a summary and a few comments are given here.
a) Condensation Nuclei Counter
As used in L.A., the CMC breads 1.75 times the number read by the
WAA. The latter is considered to be closer to the correct absolute number
concentration. Because of losses in the connecting tubing and diluter, the
smallest size sensed by the CNC is estimated to be 0.0035 pm.
b) Whitby Aerosol Analyzer
The maximum usable sizing range of the WAA is from 0.005 ;um to 1.0 um.
However, in the L.A. study the instrument was used only over the range from
0.0075 to 0.4 jum. Estimated accuracy for calculating the number concentrations
are as follows: 0.0075 um + 200$, 0.01 pn + 100$, 0.04 um + 20$ and for
D > .1 um + 10$.
c) Optical Particle Counter
As operated in L.A., the OPC was used to count particles in the
size range from 0.33 to over 6.8 um. However, for reasons which are explained
in part 6, only the count data from 0.4 to 6.8 um was used. The last channel
-------
III-2
on the multichannel analyzer counted all particles larger than 6.8 urn. The
instrument was calibrated against polystyrene latex so that the particle sizes
will be correct only for aerosol particles having a refractive index of 1.6.
The data indicates that the smog aerosols have considerable water associated
with them so it is probable that the true refractive index of the smog aerosols
was less than 1.6, perhaps as small as 1.4. If the true refractive index
were 1.4, then the true particle size would be as much as 3Q% larger than that
indicated. This would correspond to an underestimation by a factor of 2 of
the number of particles in the optical range. It has therefore been estimated
that the maximum error in the measured number concentration is about a factor
of 2 and that the maximum error in size measurement is about
It should be noted that in the size range where the WM and OPC overlap,
the agreement between the dv/dDp vs Dp curves calculated from the number
data are ordinarily within a factor of two. This excellent agreement suggests
that the absolute error in the particle concentration by number at a particle
size of about 0.4 wm must be less than about + 50$.
/
In view of the fact that the number concentration of the aerosol was
found to vary from time to time by one to two orders of magnitude over most
of the size range, it is considered that the MAAS accuracy and resolution was
quite adequate to detect all significant changes in the aerosol over the size
range from 0.0035 to 6.8 }m.
2. Pump system, diluters, plumbing
A schematic diagram of the piping system for the aerosol supply is shown
in Figure 1-2. Further information, besides the piping and instrumentation
arrangement is contained in the rectangular boxes placed in the vicinity of
distinct points of the system. The four numbers in each box correspond to the
following specific information of the particular piping section: the flow
rate in liters/second; the- residence time of the aerosol from the time it
entered the top of the main supply pipe to the point under consideration;
the inside diameter of the pipe in cm; and the distance from the last marked
distinct point of the piping system in cm.
The main purpose for giving the detailed specification of the flow rates,
pipe diameters and distances, is to permit an estimation of the aerosol losses
in the piping system. Although the system was designed to minimize aerosol and
gas losses of Oo and condensation nuclei, the loss was also checked experimen-
tally as described later.
The smoggy air entered the piping system through a sampling cap placed on
the top of the main sampling pipe, 6.7 meters above the roof of the three
story building.
The cap itself included a screen (40 mesh) which had the purpose of
preventing the entrance of small insects into the piping system because soon
after the first experiments started, it was found that insects were transported
along with the gaseous and particulate matter, deep into the system. Each insect, of
course, was a potential cause of plugging and erratic readings. As a matter of fact,
-------
Ill-3
several of the initial impactor classification runs were ruined because of
plugging of the impactor jets.
At the end of the main supply pipe, the aerosol was distributed through
a manifold with five isokinetic probes to the various instruments.
The instruments located on the left of the main sampling pipe in Fig. 1-2
are those for the monitoring of the chemical contaminants and several
infrequently used aerosol instruments. The instruments for the aerosol
size distribution measurements are on the right of the main supply pipe,
consisting of the Miitby Aerosol Analyzer, the sensor unit of the Royco optical
counter, two Lundgren impactors, GE nuclei counter, the Charlson-Ahlquist
nephelometer; and the Meteorological Research Institute nephelometer.
A flow rate of 7 I/sec through the main sampling stack was maintained by the
high volume sampler blower. This air was discharged into the room after
filtration. The high velocity through the sampling stack minimized the loss
of 03 and condensation nuclei. Tests made by comparing the 03 concentration
at the stack inlet with those measured by the instruments in place in the
laboratory gave readings that were within about 10% of each other. This
was within the estimated experimental error.
A similar comparison of the condensation nuclei readings at the stack
inlet with those of the counter in its normal location in the laboratory gave
an average penetration of 85%. This was considered to be near the experimental
error. Since most of the losses would be of the smallest particles (e.g. those
below 0.01 um) the losses of particles in the ¥AA range are negligible.
No direct measurements of the losses of large particles in the system
were accomplished. It had been hoped that comparisons of the results from
the Noll impactor might provide an estimate of the accuracy of the MAAS on
large particles. However, no useful data was obtained with the Noll impactor.
Preliminary examinations of the data suggest that most of the time there were
few particles above about 5 u in the atmosphere at the sampling stack inlet
location. For this and other reasons which will be discussed in more detail when
the data is described in subsequent reports and papers, it is believed that the
MAAS has provided an accurate measure of the particle concentration up to
its upper limit of measurement of 6.8 ^u.
All instruments for the gas analysis (see left hand side of the
diagram, Fig. 1-2) were supplied by the smoggy air through glass piping with
rubber couplings. Glass was chosen for this purpose because of low adsorption
activity of its surface for the gaseous contaminants.
The four instruments, the California Institute of Technology Particle
Beam, the Liu Electrostatic Sampler., the Meteorological Research Institute
(MRI) cloud nuclei counter, and the MRI ice nuclei sampler, were connected
-------
in-4
to the distribution manifold only temporarily with flexible plastic tubes
of 1 cm in diameter.
In order to minimize the loss of large particles, the two Lundgren
impactors were placed underneath the manifold as shown in Fig. 1-2. The
connection between the manifold and the ibnpactors was made out of PVC pipe,
5 cm in diameter. Two impactors were used so that one could be prepared while
the other was in use.
The wide range humidity sensor (Hygrodynamics, Inc..) and one of the
temperature sensors (copper-constantan thermocouple) were installed at the
end of the large T-branch of the manifold, as shown in Fig. 1-2. This location
was chosen because of the comparatively high flow rate at that point, and
because of convenience of installation. The humidity sensor itself was placed
in a bypass, the flow being generated by the stagnation pressure of a Pitot-
tube arrangement, as shown in Fig. 1-2. The time constant of this humidity
sensing arrangement was checked out by rapidly switching the flow from outside
air to room air and found to be on the order of one minute.
The end of the manifold where the humidity sensor and thermocouple were
installed had a 2.0 meter long flexible connection of 3-8 cm I.D. to the
high volume sampler.
Just before the high volume sampler, a branch of the flexible 3.8 cm
diameter tube was conducted to the MRI nephelometer and then to the Charlson-
Ahlquist nephelometer. The suction for both of the instruments was provided
by a fan in the Charlson-Ahlquist instruments, so the MRI nephelometer was in
line rather than on a separate branch, with an independent suction source.
This arrangement was found to be necessary because the fan of the MRI
nephelometer was not strong enough to pump air against the #» 2 cm of w.g.
negative pressure that existed in the piping system. The negative pressure
was mainly due to the pressure drop in the main supply line from the roof
to the basement.
The aerosol supply for the WAA, Royco optical counter and for the GE
nuclei counter was provided through a common flexible pipe directly from the
main manifold. All aerosol connections of these instruments were kept short
to minimize diffusion losses of small particles. The detailed dimensions and
distances are indicated in the rectangular boxes in Fig. 1-2.
All three instruments had separate in-line aerosol diluters. The highest
required dilution was for the Royco sensor, with a total dilution of 1:100,
accomplished by using two diluters in series, while for the nuclei counter
a dilution of 1:12 was found to be sufficient. The main diluter, before the
inlet of the ¥AA, was not used as a diluter because dilution was found to
be not necessary and also because some difficulties arose with its operation.
-------
III-5
The design of the passive in-line diluter used to dilute the aerosol
for the optical counter and the condensation nuclei counter is shown in
Fig. III-l. It is constructed from a Geliuan No. 12100 membrane filter
cartridge having a pore size of 3 pm. The only modifications to the standard
filter cartridge are the installation of a short length of glass capillary-
tubing of about 0.5 mm I.D. in the center of the cap that closes the top of
the filter and the insertion of a U-shaped piece of cardboard or metal in the
outlet hose connector to mix the aerosol stream as it leaves. In this type of
capillary leak diluter a small stream of aerosol passes through the capillary
and is then remixed with the clean air that has passed through the filter.
The dilution ratio can be changed by varying the capillary length and diameter.
However, capillary diameters less than 0.3 mm I.D. should not be used. Also,
it has been found that dilution ratios of more than 1/25 result in unacceptable
variations of dilution ratio with particle size. Higher dilution ratios are
obtained by using several diluters in series.
The dilution ratios of this type of diluter must be obtained by experiment.
The ratios of the diluters used in this project were measured experimentally
using polystyrene latex aerosols ranging in size from 0.55 to 2.8;um using the
Royco 220 optical counter as the particle sensor. Over the given size range,
the dilution ratios were found to be independent of particle size. "When these
diluters are used in series at high dilution ratios the overall ratio of the
combination must be checked to make sure that there is adequate mixing between
stages. It is best to operate the diluters in a vertical position to minimize
the losses of large particles.
Most of the instruments of the MAAS are equipped with their own internally
located aerosol pumping system and were thus capable of independent operation
when connected to the aerosol distribution piping. Only the operation of the
¥AA necessitated the use of externally located vacuum pump which would have
been very noisy had it not been enclosed in a sound absorbing box. The box
is about 22" x 23" x 33" in size, contains the oil free, rotary vane vacuum
pump, Model 3, 1200 rpm, 13.9 scfm (Conde Milking Machinery Co., Inc.), WAA sonic
jet charger, Uriico high volume aerosol sampler with speed control, small
compressor axial cooling blower, and the system of valves and flow rate
gages for the flow rate adjustment, as seen in Fig. Ill-2. The box can easily
be transported with the aid of four swivel casters on the bottom, and all
instruments within the box can be reached after removing the box lid. All
control valves, switches, and gages can be operated from outside of the box.
3. Data Acquisition System
The large number of electrical signals from the many instruments used
in this project were recorded with an automatic data aquisition system. This
section describes the system, gives its specifications and presents its
performance characteristics that are essential to interpreting the data recorded.
Data collected for the 1969 Los Angeles air quality project consisted of
strip chart records, digital print-out, and punched tape. An important objective
of data collection was to record data from many different experiments and
-------
III-6
correlate these data records on a common time base. In addition, it was
desirable to have one complete set of data containing results from all experiments
to facilitate correlation of data taken by different investigators.
Description of Data Acquisition System
The major data recording item was a Hewlett-Packard 2010-B data acquisition
system. Included in this package was a 2901-A input scanner,, a 2401-C digital
voltmeter, a 2509-A digital clock, a 2545-A tape punch coupler, a 2545-B
power supply, a 2545-C paper tape handler, and a 562-A digital printer. The
major components of the data system are illustrated schematically in Figure 1-3.
In addition, the basic 2010-B system was purchased with a monitor option which
permits stopping the scan on a given channel in order to take more than one
reading.
A. Input Scanner
The major component of the data system is the input scanner.
This device transfers analog voltage, current, or frequency signals to one
set of measuring and recording equipment. Data points may be scanned contin-
uously or upon local command. The remote command capability permits automatic
data logging at intervals specified by the Hewlett-Packard 2509-A digital
clock. Although the system is capable of measuring frequency, resistance
and ac voltages, only the dc voltage measuring mode was used.
a. Monitor Mode
The input scanner is capable of recording 25 channels of data,
however, a "monitor mode" option provides additional capacity. Inserting a
diode pin in the "monitor mode" row of the patch board for any of the 25 channels
permits more than one data signal per scan to be accepted by that channel.
Upon reaching a monitor mode channel, the scanner records one reading on that
channel and stops. A ground closure output (from -26v to ground lasting
approximately 3 milliseconds) is presented at connector J 108. This signal is,
in turn, presented to the sensing device (in this case the Whitby Aerosol
Analyzer). Upon receiving this signal, the Wiitby Aerosol Analyzer issues
another ground signal lasting approximately 30 milliseconds to connector J 106
of the input scanner to initiate a successive reading on the selected monitor
mode channel. Another ground closure at J 108 follows. This process continues
as long as the Whitby Aerosol Analyzer issues ground closures to the scanner.
Stepping to the next channel is accomplished when a 3 millisecond duration
ground closure from the 2509-A digital clock is issued to J 10? of the scanner.
b. Measurement Delay
Each measurement requires a finite time delay to provide
stabilization time for the dc amplifiers and ac measuring devices. The system
provides 4 possible delay times ranging from 30 to 910 milliseconds. For
all tests conducted at Pasadena, the 30 millisecond delay was used.
c. Specifications
Number of channels: 1 to 25 floating signal pairs with shields
Maximum voltage: 750 v peak or 500 v rms
-------
III-7
Maximum, frequency: Maximum recommended frequency through
scanning switch is 300 Kc
Amplitude (volts, rms) times frequency
should not exceed 10
Shunt capacitance: (Hi to lo) 100 pf
Crosstalk capacitance: (Between high or low of adjacent channels)
4pf
Leakage capacitance: (Signal circuit to chassis ground) 90 pf
Signal path resistance: (input to output) less than 0.5 -"-
Leakage resistance: (Hi to lo or to chassis ground) 10 -a-
minimum at 70% HH; 10-n. minimum at 95$ RH
(up to 40° C)
Thermal offset: Less than 5 Mv
Common mode rejection: 85 db at 60 cps, 100 db at dc (with up
to 1000-^ between source ground and
voltmeter L0)
Connectors: Cannon 3-pin, gold-plated
Source impedance: 68K ohms
Scan start: Circuit closure or -26 v pulse, 30 ms wide
(from 2509-A digital clock)
Scanning speed: recording: 85 ms per channel, minimum
skipping: 30 ms per channel
d. Channel Assignments
All channels listed below recorded one data signal except number 25
Sixteen signals were recorded in monitor mode on this channel.
Channel Number Signal Signal Range
1 Pyroheliometer 0-10 millivolts
2 Wind direction 0-5 volts
3 Wind deviation (direction) 0-5 volts
4 Wind speed 0-5 volts
5 Roof temperature 0-10 millivolts
6 Sampling line temperature 0-10 millivolts
7 Aerosol analyzer temperature 0-10 millivolts
8 Relative humidity 0-10 millivolts
9 OPEN
10 Date code 0-100 millivolts
11, 12, 13 OPEN
14 Ozone-Atlas 0-100 millivolts
15 Ozone-Mast 0-100 millivolts
16 NO - Atlas 0-100 millivolts
17 N02 - Atlas 0-100 millivolts
18 OPEN
19 U of W, 675 nm 0-10 volts
20 U of W, 546 nm 0-10 volts
21 U of W, 436 nm 0-10 volts
22 U of W, 360 nm 0-10 volts
23 MRI Nephelometer 0-1000 millivolts
24 Condensation nuclei counter 0-10 millivolts
25 Whitby aerosol analyzer 0-10 volts
-------
III-8
B. Digital Voltmeter
a. General Description
After data signals are received by the input scanner, they are
diverted to the digital voltmeter. This is an all solid state electronic
instrument capable of measuring dc potentials up to + 1000 volts. There are
5 voltage ranges, the lowest full scale range, + 0.1 volts, permits high-
resolution millivolt measurements. In addition, the voltmeter is capable
of frequency measurements from 5 cps to 300 Kc. A nibcie tube display indicates
the measurement in six digits in addition to the polarity and units being
measured.
Digital coding available at the DVM is 4-2'-2-1 binary coded
decimal (bed) for recording on the digital printer and for further digital
data processing. For this particular system, the data coding was further altered
to teletype ASKI 33 for recording on punched paper tape.
b. Specifications
i. Circuit type: floated and guarded signal pair
ii. Ranges: 5 ranges from 0.1 v to 1000 v full scale
iii. Overranging: Overranging to 300$ of full scale is permissible
on all but the 1000 v scale
iv. Input impedance: 10 M on 10, 100, 1000 v ranges
1 M on 1 v range
100 K on 0.1 v range
Impedance is within + 0.02$ of nominal value
in all ranges
v. Internal calibration source: + 1 volt internal standard
provided for self-calibration.
Voltage reference is derived
from specially aged, temperature
stabilized zener diode, with
guaranteed drift of less than
± 0.006$ in 6 months.
C. Digital Clock
a. Purpose
The 2509-A digital clock was necessary to provide a time base for
correlating data and to provide remote control signals required by the data
system. For example, data recording by the digital printer and paper tape
punch required control through the digital clock. Also, a scan cycle through
the 25 channels could be initiated at desired time intervals specified by the
digital clock. Finally, the aerosol analyzer required the use of a digital
clock since data recorded in "monitor mode" required a timed pulse to initiate
stepping to the next channel.
b. Description
The 2509JA digital clock is an all electronic precision instrument
-------
III-9
that accumulates time in seconds and presents it in hours, minutes, and
seconds in 24 hour cycles. The time output is visually displayed in 2 digits
of hours, 2 digits of minutes, and 2 digits of seconds on nixie tubes and is
presented digitally in bed 4-2'-2-1 at an output terminal at the rear of the
instrument.
The digital clock also generates a train of pulses for system timing
(timing monitor mode duration) and control purposes. A group of block-out
latching push buttons on the front panel controls the pulse repetition rate.
Intervals are available every one second, ten seconds, one minute, ten minutes,
and one hour. A system option provides a thumbwheel switch with 9 multipli-
cation positions (1-9) which, when combined with the 5 latching push buttons,
provides additional timing flexibility.
Since the clock must keep an accurate time record and provide time signals
for recording by the data system, the clock must operate independently of
the recording system, without the need for synchronization. This is possible
since the 2509-A allows time recording at any instant and allows any time
reading to be held for as long as 9>998 milliseconds for recording purposes
without any loss of time.
c. Specifications
Circuit reference: All voltage levels are referred to chassis ground.
Time reference: (internal) Derived from power line frequency
( 60 cps standard)
Accuracy: Equal to time reference, +0, -1 second (non-accumulative)
D. Output Coupler
The bed, 4-2'-2-1, information from the dvm and digital clock must be
transferred to both the paper tape punch unit and the digital printer. In
its standard form the 2545-A output coupler accepts 10 characters of externally
supplied parallel bed information and translates this to IBM 8-level code.
A system modification translated this bed information to Teletype ASKI-33
code so that the paper tapes could be processed on a standard Teletype
shared time terminal.
E. Data Recording
This device, made by the Teletype Corporation (model BPKE High Speed
Tape Punch Set) is designed to rapidly perforate paper tape for data storage
and subsequent processing. It must be used with thesbove mentioned 2545-A
output coupler and a power supply (2545-B). In the current configuration
ASKI-33 code is punched for direct processing on a Teletype shared time terminal.
The data is also printed simultaneously on paper tape.
F. Physical Electronic Connections
Figure 2 is a schematic of the laboratory in which the 1969 Los Angeles
air quality measurement study was made. Electronic cable routing and instru-
ment locations are emphasized here. Equipment locations were the result of
optimizing the allowable space and component functions.
-------
111-10
a. Gas Analysis Equipment
All gas analysis instrument s were located along the south wall
of the laboratory (see Fig. III-3). The gas analysis devices that were
read by the data system were: two ozone monitors, one NO sampler, and one
NOo sampler. These instruments were owned by the State of California, Air
and Industrial Hygiene Laboratory. The N02, NO, and one ozone monitor were
made by the Atlas Electric Devices Company, 4114 North Ravenswood, Chicago
60613. The other ozone sampler, Mast, model 724-H, was from the Mast
Development Company, Davenport, Iowa. All four devices are standard, commer-
cially available shelf items.
Data from the Mast-ozone analyzer were recorded on a 0-10 mv
strip chart recorder. A parallel data signal was recorded on the Hewlett-
Packard data system whenever particle size distributions were being recorded.
This instrument was connected to channel 15 of the data system through a
2 - conductor cable.
Channel 14 accepted the Atlas-ozone data. A two conductor plus shield
cable was used for this application because signals from this instrument
were easily affected by noise. Connections were as follows:
shield: Atlas case to #1 terminal of data system input
low: Atlas - to #2 terminal of data system input
high: Atlas + to #3 terminal of data system input
Channel 16 accepted the Atlas - NO signal through a 2 conductor plus
shield cable as discussed above. Connections were the same as above except
for the addition of a 25 u fd capacitor attached across the + and - connections
of the dymec. The capacitor was necessary in order to obtain a anooth data
signal.
Channel 1? carried the Atlas - N02 signal. Connections were the same as
for channel 16, including the 25 )i fd capacitor.
b. Particle Size Distribution Measurements
All equipment for measuring particle size was located in the
northwest corner of the laboratory, as illustrated in Fig. III-3.
Particle size data from the Wiitby Aerosol Analyzer were recorded on
channel 25 of the data system. Signals were transmitted through a EG/58
cable and one scan was recorded once each 10, 20, or 30 minutes, depending on
the experiment being conducted. All cables between the aerosol analyzer and
the data system were RG/58 type with the conductor mated with the "high"
or #3 pin of the data system input, and the shield tied to both the "low" and
"ground" pins of the data system input.
Three electrical connections are necessary between the Whitby Aerosol
Analyzer and the data system. The first connection is necessary to notify
the aerosol analyzer that channel 25 has been reached by the data system, and
-------
III-ll
a particle size analysis cycle should begin. A second signal then notified
the data system to accept data from the aerosol analyzer. This command signal
is carried via a similar cable connecting the "DVM read command" terminal at
the rear of the aerosol analyzer to terminal J 106 of the data system.
Finally, particle size data are transmitted from the "output" terminal on
front of the aerosol analyzer to channel 25 of the data system.
G. Optical Particle Counter
Particle size data determined by the Royco PC 220 optical counter were
recorded on punched paper tape and printed on paper by a teletype typewriter
and tape punch. Signals were carried via a special multi-conductor cable
connecting the teletype to a Hewlett-Packard 5415-A analog to digital converter
and related electronics (Hewlett-Packard 5431-A display, 5421-A digital processor,
and Royco 1?0-1 pulse converter). This group of equipment received data
signals from the Royco PC-220 optical sensor and command signals from J 83
of the data acquisition system.
H. Total Nuclei Counts
Total particle concentration data were determined by a General Electric
Condensation Nuclei Counter. These data were recorded continuously on a
Hewlett-Packard Mosely 680 strip chart recorder, and periodically on channel 24
of the data system.
I. Nephelometer Data
Light scattering data were recorded by two nephelometers located in the
northeast corner of the laboratory. Channels 19 through 22 of the data system
recorded light scattering information from the University of Washington
nephelometer. The following table indicates the signal carried on each channel.
Data from the MET (Meteorological Research Institute) nephelometer were
recorded on channel 23.
Channel Wavelength
19 675 nanometers
20 546 nanometers
21 436 nanometers
22 360 nanometers
J. Recording Time Sequence
Data signals for the first 24 channels were all recorded within 2 minutes
(approximately 90 ms per channel and 20 channels). At the end of this scan
the electrical particle counter began operating for a duration of approximately
4 minutes. The optical particle counter began counting when the record
command was received by channel 1 of the data system, and continued counting
for 10 minutes. After this 10 minute period, the data stored by the optical
-------
111-12
system were recorded on the teletype unit. A summarycf the recording sequence
appears in the sketch below.
Optical Counter
o
10
Fig. II1-4 Data acquisition system recording sequence
K. Meteorological Data
Meteorological data were recorded on channels 1 to 5 of the data system
via 5 pairs of 200 foot long shielded cable (provided by the Gal Tech Physical
Plant). These cables carried one temperature, one pyroheliometer, and three
wind signals.
L. Miscellaneous Signals
Temperature for the Whitby Aerosol Analyzer flow system was determined
by a thermocouple located in the aerosol line between the three-foot cylindrical
flow chamber and the collection filter. This temperature signal was carried
by a 2 conductor plus shield cable and originated at a copper constantan
junction, as were all thermocouples used for the project.
Sampling line temperature was recorded at the humidity sensor located
at the laboratory end of the atmospheric aerosol stack. Relative humidity
was also sampled here, however, after 10 September, 1969, the thermocouple
was eliminated and the relative humidity measurements were made at the aerosol
analyzer inlet.
4- Condensation Nuclei Counter
A standard General Electric, Cat. Wo. 112 L428 G 1 Condensation Nuclei
Counter, Skala 1963,was used to measure the total particle concentration.
Catalogue specifications for this instrument are: Theoretical lower limit
of detection 0.002 jim diameter, count ranges 0-300, 0-1K, 0-3K, 0-10K,
0-30K and 0-100K linear ranges and M and 10M non-linear ranges, response
time 2 sec., and sampling rate 6 1/min. The fluctuations of the CNC output
were damped somewhat by the installation of a 1000 ^F 6 volt condenser
connected across the recorder output.
Four years of experience with this counter plus numerous comparisons with
-------
Ill-13
other particle counting instruments have shown the following characteristics
and problems.
a) Factory calibrations are good only within about a factor of three.
Absolute calibration of these counters is difficult and must be made by
comparison with some other counter- The counter used in this study was
overhauled by General Electric just before the study and was used with their
calibration. The G.E. calibration is traceable to the calibration of Pollack.
The only method which we have available to calibrate the CNC is to compare it
with the WAA on aerosols for which the entire size range of the aerosol is
within the counting range of the WAA. Such comparisons were made from data
on smog aerosols during the project under conditions where coagulation had
removed many of the particles below about 0.01 pm. Such conditions occurred
several times during the night. These data for the CMC and the WAA are compared
in Fig. Ill-5. It is seen that the CNC reads about 2g times the WAA.
After the MAAS was returned to the University of Minnesota after the
project, further comparisons of the CNC and the WAA on laboratory aerosols
and atmospheric aerosols having no particles below 0.01 }jm yielded a ratio
of 1.75- From these comparisons we have concluded that the CNC reads about
1.75 times the WAA. The night time aerosols probably contain some particles
below the range of the WAA, thereby giving a higher ratio than with the
laboratory aerosols. This is the figure that is being used in comparing the
data from the two instruments. Because as is explained in part 5 below, we
have reason to believe that the number concentrations calculated from the
WAA have greater absolute accuracy, the CNC concentrations reported are the
measured values divided by 1.75.
b) The minimum particle diameter sensed by CNC's is usually greater than
that calculated from the Thompson-Gibbs equations for the expansion ratios
used because of diffusion losses of particles in the plumbing conveying tine
aerosols to the instrument. At the 8 inches of Hg underpressure used, the CNC
should count all particles larger than about 0.002 jum diameter. However,
it is estimated that with the inlet tubing and the diluter capillary will
remove most particles smaller than about 0.0035 urn. Thus where a lower limit
for the CNC must be assumed as for the calculation of dN/dDp for the size
interval below the lower limit of the WAA, the lower size of detection for
the CNC is assumed to be 0.0035 ^m.
c) Comparisons of particle counts using adiluter and the linear scales
in comparisons with counts on the non-linear scale without the diluter have
shown that the non-linear scales are very inaccurate. For this reason, almost
all of the data was taken using the linear scales and the 1/12 diluter. On
only a few occasions did the CNC concentration exceed the 1.2M limit of this
combination. Most of the time the CNC was operated on the 0-100 K scale
with the 1/12 diluter.
-------
111-14
5. Vlhitby Aerosol Analyzer (WAA)_
The aerosol size distribution in the diameter range between 0.0075 pi
and 0.6 ^im was measured with a modified commercial model 3000 Whitby Aerosol
Analyzer (WAA)*. In the present section, the principle of its operation's
reviewed and the main components of the instrument are described, including
the modifications made on the standard commercial instrument. Particular
emphasis is placed on the discussion of the calibration procedure and results
since the present calibration was more elaborate and detailed than the
earlier calibrations of the instrument. Finally, estimates of the accuracy of
the WAA are given as a function of particle size.
Principle of Operation
The electrical mobility of a charged particle i.e. the ratio of the
particle velocity to the applied electric field acting on it, is a function
of the particle size. In a proper electrical and flow field, therefore,
particles of different sizes may be classified according to their electrical
mobility. For diffusion charging and for particles less than 1.0 yum in size
(diameter) the electrical mibility is a monotomically decreasing function of
the particle size and it varies only slightly for sizes larger than 1.0 pan.
Accordingly, for particles less than 1.0 urn in diameter, the electrical mobility
is uniquely related to the particle size. If diffusion charged particles are
then introduced into an electrostatic precipitator through a point or line
source, the precipitation distance increases with increasing particle size.
The Whitby Aerosol Analyzer (WAA) as described by Whitby and Clark (1966)
was developed starting with the ideas and findings as stated above. The
objective of their work was to extend these findings and to combine them in
such a way that the number of particles in different size ranges could be
measured by properly varying some of the parameters of the system. The five
years of work on the development of the instrument resulted in the following
arrangement and measuring procedure.
The aerosol first enters a jet-charger, (fig. III-6) especially developed
for this purpose, where the charging results from the fast mixing of uncharged
particles with high concentration of negative ions, generated on a corona needle.
The physical mechanism of charging is governed by molecular diffusion and in
a small part by the existing electrical field in the charger vessel.
As a result of the prevailing diffusion charging, the particles leaving
the charger vessel are charged such that the number of elementary charges
on a particle is approximately proportional to its diameter, Dp.
The aerosol leaving the charger is then introduced into the precipitating
tube, as an annular cylinder surrounding a core of clean air. The aerosol
collector, a metal rod, passes axially through the center of the tube.
Depending on the voltage on the collecting rod, all particles with an
electrical mobility exceeding a certain value will be collected on the
precipitating rod, while those with smaller mobility (larger size) pass through
the precipitation section and are then collected on an absolute filter.
Manufactured by Thermo-Systems, Inc., St. Paul, Minnesota 55113
-------
111-15
The absolute filter itself is electrically connected to a sensitive
electrometer, which measures the total current due to the charges given up
by the particles collected on the filter. A step increase of the rod voltage
will cause particles of a larger size range to be precipitated with the
resulting decrease of the filter current.
The number of particles AN, in a given size range AD , corresponding
to the current change A I may be calculated from the known values for the
mean number of elementary charges per particle, the flow rate and the
fraction of particles lost by space charge and diffusion.
The aerosol size distribution in the size range between 0.0?$ fw and
0.6 pm is obtained by scanning the precipitation voltage from 225 volts
(for 0.075 £un) to 14,000 V (for 0.6 jum) in 15 steps, measuring the corresponding
currents on the electrometer and then, using the calibration curve, converting
the voltage vs. current curve to a discrete (14 point) size distribution curve.
To allow the new particle trajectories in the precipitation tube to
stabilize, the times between two consecutive voltage steps have to be kept
sufficiently long. These times range from 30 seconds at the lowest voltage
to 12 seconds at the highest voltage, the time required for the entire
cycle being about four minutes.
The scanning cycle for a size distribution measurement is controlled by
an internal timing system that also commands a digital read out for the current
before each step to a higher voltage. With a digital voltmeter connected to
the electrometer of the WAA, the size distribution measurement can be fully
automated and the system is then suitable for monitoring time-dependent
aerosol size distributions, A more detailed discussion of the electrical
and timing arrangement for the Minnesota Aerosol Analyzing System (MAAS)
that includes the ¥AA is discussed in part 2 of this section of this report.
Modifications made on the standard commercial instrument
The main components of the WAA were subjected to extensive experimental
and numerical investigations as well as to changes in design and operating
conditions before the instrument was shipped from Minneapolis to Los Angeles
and after its return.
First, in a numerical investigation, ¥hitby et al. (1969) Section IV,
the flow field and particle trajectories in the precipitator section of the WAA
were studied with the ultimate objective of reducing the geometrical and
hydrodynamical distortion effects on the quality of size classification. As
a result of this numerical study, the ratio of the aerosol and clean air
flow rates was reduced from 1/31 to 1/14- This reduced the standard
deviation of the mobility classification from CT = 1.4 to (J~- = 1.18 (for
Dp = 0.5 pi).
In Section V of the same report, Whitby et al. (1969), the results of
experimental studies on the performance of the WAA are discussed, with
particular emphasis on the effect of aerosol flow rate and the aerosol
concentration on the resolution of the instrument. The optimal ratio of
-------
111-16
aerosol/clean air flow was found to be 1/14 at a total flow rate of 2.12 I/sec
(4.5 cfm). The aerosol concentration was found to affect the level and
quality of the charging only slightly, if the total current due to the
deposited charged aerosol was between 2 x 10-12 and 50 x 10 ^ Amp.
In a further series of experiments (unpublished) aerosol losses due
to diffusion and space charge in the charger were investigated. The results
indicated that the losses in the charger depend strongly on the particle
size and on the aerosol flow rate, decreasing in magnitude with increasing
particle size and increasing flow rate. For particle size of O.OS^im for
instance, the aerosol penetration through the original 22 1 charger was
found to be 20$ at an aerosol flow rate of 0.15 I/sec (optimal flow rate for
the classification) and increased to 40$ at a flow rate of 0.5 I/sec.
A significant reduction of the aerosol losses in the system was achieved
by replacing the charger vessel of the standard instrument which had a volume
of 22 liter, with a much smaller, 1.7 liter vessel. The new cylindrical
charger vessel is 7-5 cm in diameter and 34-0 cm long with its axis coinciding
with the jet axis. The aerosol exit is placed radially on the cylindrical
vessel, 3.0 cm from the jet entrance (Fig. III-6). The penetrations for a
particle size of 0.08 jum with the small charger vessel were 58% at an
aerosol flow rate of 0.15 I/sec and 70$ at 0.5 I/sec. The increase in
the aerosol penetration is attributed to the reduction of the aerosol
residence time in the small vessel. In a highly recirculating flow field,
such as the jet-driven recirculation in the larger charging vessel, the aerosol
is forced several times to the vicinity of the walls before it leaves the
vessel. In the small vessel, this contact time is reduced, by that reducing
both the diffusion and space charge losses. The jet entrance and the
primary aerosol mixing section of the charger remained unchanged.
There were also several minor changes and adjustments made on the
electrical circuit of the standard model 3000 TSI instrument.
First, it was found that the stepping of the lower to a higher voltage
on the collecting rod occurred simultaneously with the read command to the
digital voltmeter. This caused a distorted current reading. The interaction
was eliminated by inserting a time delay of about 0.2 seconds between the
read command and the subsequent voltage stepping.
In the intermittent cycling mode, the instrument automatically scans
through the 15 voltages, then returns to zero voltage and awaits the
pulse from the digital clock to start a new cycle. In a further modification,
a switch was installed which permitted the selection of either a zero or
15 Kv voltage, for the time period when the instrument was standing by
awaiting the pulse for a new cycle. With the added choice of 15 Kv in the
stand-by position, the instrument could be used for size distribution
measurement, let's say every hour (for four minutes that is needed for the
scanning) while for the rest of the time it would be classifying the aerosol
on the rod. This mode permits then the correlation of size distribution data
with those obtained by classification.
-------
II1-17
The scanning times between the steps 1 through 6 were found to be too
short, so that the corresponding currents were not stabilized before the
next voltage step occurred. Accordingly, the time between first and second
step was prolonged from 23 seconds to 30 seconds, then the step between 2
and 3 from 20 to 25 and those between 3 and 6 from 15 to 22 seconds. The
stepping times for the remaining steps were unchanged at 12 seconds.
The stepping voltages were adjusted according to the newly calculated
calibration curve discussed in the following paragraphs.
Calibration of the Miitby Aerosol Analyzer
After the main components of the ¥AA had been restudied and the
necessary modifications made, the instrument was calibrated before its
use in L.A. However- the calibration for sizes smaller than about 0.02 jum
was still inadequate primarily because of the lack of good monodisperse
aerosols smaller than 0.02
After the return of the instruments to Minnesota, Mr. Husar made a
more thorough calibration as part of his Ph.D. thesis work. Means were
found to generate adequately monodisperse aerosols smaller than 0.02/^m.
The principal effect of the recalibration was to change the constants for
sizes smaller than 0.03 JJUSL. The new calibration has been used in all data
reduction.
The calibration required the determination of two parameters as a function
of particle size, namely the average number of charges carried by a particle
and secondly the fraction of particles lost in the system.
The complexity of the instruments' geometry and of the flow field did
not permit a sufficiently accurate theoretical prediction of the fraction
of particles lost in instrument. The losses (or the penetration) therefore
had to be determined experimentally.
The average number of charges on a particle as a result of diffusion
charging was studied by several investigators both theoretically and
experimentally and it can be predicted with fair accuracy if the so called
Not product is known, where N is the number concentration of ions and t is
the time the particle spent in the ion cloud. In the presently used jet-
charger. however, neither the ion concentration nor the time the particle
spent in the ion cloud is known with sufficient accuracy to permit accurate
calculation of the NQt product. Furthermore, for particles less than
about 0.05/un in diameter, a fraction of the aerosol escapes charging entirely,
so that the average charge, based on the total number of particles, may be
less than unity. For this latter charging range the available theory
is only applicable for the free molecular regime and experimental data to
test its validity are missing entirely. For the above stated reasons an
experimental determination of the average number of charges seemed most desirable.
In order to arrive at a fair estimate on the size resolution of the WAA,
not only the mean number of charges rip, but also the charge spectrum had
-------
111-18
to be investigated at several different sizes. Such data^for the ion- jet
diffusion charger were not available prior to this study. #
An experimental determination of the penetration P, and of the level of
charging np, for particles below 0.1 /^m presented considerable difficulties
since several of the techniques used here had to be developed for this
particular study.
The experimental procedures to determine the two parameters, P and np,
are discussed in the following section.
The Average Number of Charges np and the Charge Distribution on Particles
The average number of charges and the charge distribution on particles
of different sizes were determined from the distribution of their electrical
mobilities. The electrical mobility Zp is defined as
Zp = np-e-B (HI-1)
where n is the number of elementary charges carried by the particle, e is
the elementary electron charge, 4.8 x lO"-^ stat coul. and B the fluid
dynamical mobility. The electrical mobility of the singly charged spherical
particle Zps depends only on the properties of the fluid surrounding the
particle and on the particle size, Dp.
The electrical mobility of an arbitrarily charged particle is most
easily determined from the deposition distance, x, in an electrostatic
precipitator. For an annular type of precipitator with the inner tube (with
radius r]_) charged oppositely from the charge of the aerosol, and with
particles entering the precipitator at the outer tube radius r2, the following
relatioi holds for the mobility :
Qt In(r2/ri)
~
where Qt is the total flow rate and V is the precipitation voltage. Combining
(III-1) and (III-2), the number of charges may be obtained from the following
relationship:
np = Zp= 3 ^g DP ZP (in_3)
zps C e
where ug is the gas viscosity and C the Cunningham slip correction.
The electrical mobility distribution as a funtion of particle size
was determined by two methods: First, monodisperse Polystyrene Latex (PSL)
was charged and deposited on the central rod and counted on optical micro-
scope and secondly, polydisperse methylene blue particles were classified
*The charge spectrum for diffusion charged particles is presently being
investigated in considerable detail analytically and experimentally in our
laboratory. ^ *
-------
111-19
on the rod and the mobility distribution was obtained from size-selecting
electromicrograph counts.
PSL Classification Runs
The mean electrical mobilities, and the mobility distribution for 0.5/(m
and 0.365 /6m PSL particles were obtained by collecting the PSL particles'on
the black painted rod and then counting the number of particles collected
on a unit area of the rod.
The PSL aerosol, generated by a Collision atomizer, was first passed
through a radioactive neutralizer (Kr-85, 1 me) and then through two diffusion
batteries connected in series. The first in the line was a glass bead
diffusion battery (a 5 cm diameter^ 50 cm long tube filled with 0.3 cm
glass beads) Snd as a second battery, a 1000 cm3 silica-jel air dryer was
used. With the two batteries in line, small particles from the PSL suspending
liquid were eliminated by diffusion and larger agglomerates were precipitated
by interception and impaction.
The concentration distribution on the rod with 16 Kv precipitating voltage
is shown in Fig. III-7. The experimental points, circles for 0.5 Mm and
squares for 0.365, are fitted with log-normally distributed mobilities
(solid line) in order to obtain the mean and standard deviation of the
mobility spread. Both concentration distributions could be closely fitted _
with a logarithmic standard deviation of, CJ~" = 1.18 while the mean mobilities,Zp
were 6.15 x 10~4 cm2/sec, V, and 6.78 x 10~4 Cm2/sec, V, for 0.5 ^m and 0.365
PSL respectively. By eqn. (III-3) the number of elementary charges were
calculated as TTp = 25.0 for 0.5 /^m PSL and n = 19-5 for 0.365/^01 PSL.
In a previous work, Whitby et al. (1969), it was pointed out that the
mobility spread obtained by the above discussed method is the result of two
major factors, namely the charge distribution of particles and hydrodynamic
distortion effects. There is experimental evidence, Whitby et al. (1969),
that the hydrodynamical distortion effects are small compared to the effect
of the charge distribution.
The mean number of charges and charge distribution for 0.075 /^m and
0.125 ^m particle diameter was determined by selective electromicrograph
counts of atomized 0.1% methylene blue particles, collected on the rod.
In order to obtain statistically representative counts, the size interval
for the selective counts was chosen to be between 0.075 /^m and 0.1/tm.
The measured distribution is shown in Fig. III-7- The indicated dispersion
is partly due to the charge distribution. The mobility spread shown in
Fig. III-7 corresponds to (7~ =1.28 but it is estimated that the actual
spread of n for 0.075/^ni gparticles (and not for the size interval 0.075/664H
to 0.1 jUj&) would be at most O~" = 1.25 with a mean charge of n = 3-9.
O S~
The four values obtained for mean number of charges is in close agreement
with previously obtained data, Whitby & Clark (1967), Whitby et al. (1969).
For the charge spectrum of jet-diffusion charged aerosols, however, no data
-------
111-20
are available for comparison.
The experimental points for "n as a function of particle size are shown
in Fig. III-8 along with the fraction of particles charged, discussed in
the following paragraph.
Figure III-9 shows the measured electrical mobilities Zp, as a function
of particle diameter.
Fraction of Particles Charged
When particles of less than 0.1/L-m are charged by a conventional
diffusion (or field) charger, a fraction of the particles will escape^the
charger vessel without acquiring a single charge, i.e. without colliding
with a negative ion.
Since for size distribution measurements the concentration of particles
in a given size range is calculated from the current due to the particles
giving up their charge, it is of basic importance to know the mean number
of charges, with respect to the total number of particles, charged plus
uncharged.
The fraction of particles charged was determined by passing relatively
monodisperse aerosol through the charger and analyzer sections and then
measuring the concentration of particles penetrating to the collecting
filter, Fig. III-6. The concentration was then measured twice, first with
no voltage on the rod so that particles, charged plus uncharged, would
penetrate to the collecting filter and the second time with high (15 Kv)
voltage on the rod such that all the charged particles would be collected
and only the uncharged would penetrate to the collecting filter. The ratio
of the number of particles penetrating the analyzer section with high
voltage on the rod to the total number at zero voltage is the fraction of
particles that are not charged.
The most difficult part of the calibration procedure was to find proper
test aerosols in the size range around 0.01/^m. Fast diluted, combustion
generated aerosols were found to serve surprisingly well for this purpose.
By controlling the aging process in a proper way, the size of the aerosol
could be varied by a factor of ten (0.005 - 0.05) still retaining adequate
monodispersity.
The test aeroeols up to 0.05/am number median diameter (NMD), were
produced by aging propane torch aerosols in a 4.5 m? mylar bag (fig. III-ll).
For .007/^m NMD aerosol, the propane torch was inserted for about 2 seconds
in the glove box, Fig. III-ll, where it was diluted in the first stage and
subsequently transported to the nearly filled mylar bag. After a short
mixing period the aerosol was ready for calibration purposes.
The larger aerosols were obtained by inserting the propane torch for
longer periods in the glove box to grow the particles by reinforced coagulation.
For 0.05^m NMD aerosol, the torch was inserted for 5 minutes. Polydispersity
-------
II1-21
was reduced by aging the aerosol for 30 minutes to eliminate the small
particles. The size distribution of the test aerosol was measured before
every calibration run.
All of the concentrations were measured with the General Electric
condensation nuclei counter, since for the calibration only relative con-
centration measurements were required. For such measurements the linearity
of the instrument's reading was important rather than the absolute con-
centration it indicated. The linearity of the GE CNC was established in a
series of experiments using different type of diluters.
The monodispersity of the aerosol was evaluated by running a current I
vs. voltage V curve before every test. The steepness of the I (V) curve was
taken as a measure for the monodispersity of the aerosol. If the I (V)
curve for a run was not steep enough, the run was rejected and a new aerosol
was generated. The particle size was determined using the median voltage to
calculate the mobility according to formula (III-1) and with the known
mobility from Fig. III-9, the particle size was obtained.
The geometrical mean particle size for D < 0.03/^-m was also measured by
the diffusion battery method described by FucSs, Stech'kina and Storoselskii
(1962) . The mean aerosol sizes obtained by the two methods agreed within
ten percent. Further details on this comparison are given by Husar (1971).
The results of the test for the fraction of particles charged, f, are
shown in Fig. 111-10. At about 0.06^-m practically all the particles
carry at least one elementary charge' but at 0.025 /^-m practically all the
particles carry at least one elementary charge but at 0.025 /-6m only 50$ of
the particles are charged, the fraction decreasing rapidly to 15$ at
The data for the mean number of charges n"p and for the fraction of particles
charged are plotted on the same figure, because it is believed that such a
presentation permits a more useful interpretation of the data. If the mean
number of charges is defined as the mean with respect to the total number
of particles, charged or uncharged, then for f less than about 0.2 } n
and f are equal. This follows from the low probability that there will
be double -charged particles if f {0.2. Using this observation, the n
and f curves may be joined as shown in Fig. III-8. This approach appears
to be useful for obtaining reasonably accurate mean n in the size range
O.OK Dp< 0.05 where the discrete nature of the charge distribution makes
such an estimate otherwise rather difficult.
Figure III-8 also yields some information about the charge distribution
in the size range between 0.01/<.m and 0.04y^tm. Let us consider for instance
the probable charge distribution at 0.02 /tm: the fraction of particles
charged is 0.4 while the mean number of charges n - 0.5. This means that 10$
of the particles must carry double charge in order to have a mean charge
of 0.5- Intuitively, it is suggested that the fraction of particles carrying
three charges is negligible. Accordingly, the charge distribution at 0.02/^m
would be: 40$ uncharged, 50$ singly charged, 10$ double-charged. Sijnilar
estimates are made for 0.03 /tm and 0.04^/cm NMD.
-------
111-22
Figure 111-12 presents the estimated cumulative charge distributions for
several particle sizes between 0.01/6m and 0.075/ 0.075/6^0 and estimated (Dp < 0.075/^-m)
geometric standard deviations of charging are shown in Fig. 111-13. The
fj~a (charging) vs. Dp curve in Fig. 111-13 was used to construct the discrete
charge distributions for Dp > 0.075 /£m in Fig. 111-12. The objective of
these intuitive estimates as well as the charge distribution measurement for
larger sizes is to obtain data which can be used to calculate the size
resolution as a function of particle size.
The main features of the curves, plotted in Fig. 111-12 is the obviously
discrete nature of the charge distributions, particularly for the size
range 0.01 < Dp < 0.1 /Lm and secondly that the number of charges are
nearly log-normally distributed on particles. This latter property of the
charging was used to show the '(quality" of the charging in Fig. 111-13
where the logarithmic standard deviation of the charge distribution CT
is plotted vs. Dp. The (7~K for Dp = 0.02^111 was estimated to have an
equivalent geometric standard deviation of C7~g =1.17 for the discrete, two
point distribution (single and double-charged particles only).
For particles less than O.Ol/ttm the <7g of the charged particles is
unity since the particles either carry a single charge or no charge at
all. Accordingly there is no charge distribution (on the charged particles)
and the electrical resolution of the instrument is perfect. The o~ of
charging increases rapidly to 1.29 at 0.04/^m which corresponds togthe
greatest electrical dispersion of the size'resolution. For Dp >0.04
the ;J~g slowly decreases to about 1.18 at 0.365 //, m and .5 //m. In a later
paragraph on the size resolution of the WAA, Fig. 111-13 will be related
directly to the size-classification ability of the instrument.
With the known mean number of charges on the particles, the particle
mobility vs. Dp may be constructed, using the known values of the singly
charged particles and eqn. (III-3). The derived Z^ vs. D^ curve is shown
in Fig. III-9. P P
Diffusion and Space Charge Losses in the System
Simultaneously with the experiments for the evaluation of the fraction
of particles charged, the aerosol losses were also determined. In a similar
manner as described in the previous paragraph, the concentration of monodisperse
particles was measured at the aerosol entrance to the WAA, 1, Fig. III-6,
after the charger, 2, and then at the collecting filter, 3. The ratio
of the concentrations 1 and 3, corrected by the dilution ratio for the clean
-------
111-23
air core, gave the fraction of particles penetrating through the system,
P, which is plotted in Fig. 111-10. The ratio of the concentrations 1 and 2,
also corrected by the fraction of clean air introduced by the jet, gave the
losses in the charger alone.
The aerosol losses to the walls of the system were caused mainly by
Brownian and turbulent eddy diffusion, space charge and image force on the
uni polar ly charged mixture of aerosol particles and ions. For the separation
of the diffusion and space charge losses the assumption is made that the two
mechanisms are mutually independent. This assumption is not strictly correct,
but for the determination of K itself the separate contributions are not
important as long as the total loss is known and that can be measured accurately.
As a matter of interest, most of the losses occurred in the charger section
where there is high turbulence level and recirculation. The measured pene-
tration curve, along with the fraction of particles charged, is shown in
Fig. 111-10.
Evaluation of the Calibration Constants
With a knowledge of the previously discussed parameters, namely the
electrical mobility, mean number of charges, and the fraction of particles
lost in the system, one can calculate the constants which are used to
transform the current vs. voltage curve into a particle size distribution curve.
The differential current A I between two voltage steps is
AIjj/P, where ^ 1^ is the measured current and P the penetration through
the system. The number of particles & W, in a discrete size range (correspond-
ing to a voltage range) is calculated from the relationship:
W= Al
p Qa ~p e
where qa is the aerosi flow rate, np is the mean number of elementary
charges in the size range under consideration, and e is the unit electron
charge .
In order to set the 15 voltage steps, the size intervals were chosen(with
approximately equal A log Dp) using the mobilities obtained from Fig. 111-9*
and the precipitating voltages were calculated from eqn. (III-2) . The selected
size intervals for the size range between 0.0075 A m anci- O.Sy-tm, the np
and P values as well as the resulting calibration constants are listed in
Table I.
In the course of this work it was found that the size distributions of
coagulating aerosols around 0.01/em NMD, were very narrow. For a typical
size spectrum in that range, the use of the calibration data of Table I
would give only four points in the aerosol size spectrum. In order to
obtain more detailed information for the smaller sizes, a second calibration
table was prepared which covered the size range between 0.004 and 0.033/t-m.
-------
II1-24
The same calculation procedure that was used to prepare the range 0.0075
to 0.6 ftm was applied, except that for the range D£0.070 pm, no experimental
data being available, the data for P and np were extrapolated. The precipi-
tating voltages for this range of operation were set manually.
In Table 2 the corresponding size intervals, voltages, and calibration
constants are listed.
The Resolution of the WAA and Error Estimates
In this paragraph, an attempt is made to interpret the available
calibration data in terms of the WAA's resolution. The particular question
of interest for this study is to what degree the measured size distributions
are distorted as a result of non-ideal classification characteristics of the
instrument.
First, the factors limiting the WAA's resolution, namely the shape of the
mobility curve, the hydrodynamic distortion effects and the charge distri-
bution are discussed and subsequently numbers are given for the errors
involved.
The inherent limit for the use of the WAA as a classifier is given
by the vanishing dependence of the electrical mobility on particle size
for Dp > 1.0/,m (see Fig. III-9). The particle size dependence of the
electrical mobility decreases to essentially zero for Dp > 1.0 because
in eqn. (III-1) the number of elementary charges is np'-'IL, while the fluid
dynamic mobility approaches B-^Dp-1. Accordingly in that aze range and
for diffusion charging (np -" Dp) particles with different sizes cannot be
classified according to their electrical mobility. From these considerations
and further error estimates, the upper size limit for the WAA was set to 0.6 /m.
The second factor affecting the size classification of the instrument is
hydrodynamic distortionl due to: 1) the finite width of the aerosol sheet
entering the precipitation section, and 2) axial non-uniformities of the flow
field in the annulus.
In an earlier work, Whitby et al. (l) it was demonstrated that by
reducing the aerosol/clean air flow rate ratio the hydrodynamic distortion
effects may be reduced to a small fraction of the charging effects. The
effect of flow distorions was neglected.
The third and most significant factor limiting the resolution of the WAA
is the charge distribution on the aerosol, discussed on page HI - !#•
The WAA sizing resolution will now be derived using recently acquired experi-
mental data on the aerosol charge distribution and the aerosol losses in
the instrument. The effect of the instruments imperfect sizing resolution
on the smog aerosol spectra measured in Los Angeles are calculated and
discussed.
-------
111-25
= f (np)dnp (m_5)
and using eqn. (III-1)
f(Z) =f(n).
Attention may now be directed to Fig. 111-14 which shows the mobility
distribution f (Zp) as it changes with particle size. ¥e also recall that the
instrument classifies particles according to their mobility and that between
two voltage steps it cuts out a band (shadowed in Fig. 111-14) from the
Zp-Dp plane. If we follow the shadowed area it is evident that instead of
sharp size cut-out at the mean of the stochastic distribution there is a
particle distribution that satisfies the criterion of constant Zp. The
smaller size particles at a constant Zp are the ones carrying"less than
average1' charge and the larger ones in the collected particle population are
those charged by "higher than average" number of charges.
In the presently used data evaluation a mean charge n is assigned to
each mobility band and the aerosol number spectrum, g(Dp) for unit flow rate
is calculated from the relationship:
AI = K-np-e.g(Dp) aop (III-7)
where A Dp is determined from the voltage step to calculate A z by
eqn. (III-2) and the Zp vs. Dp figure (Fig. III-9) is used to obtain the
corresponding A D .
The accuracy of the presently used data evaluation may be tested by
constructing a mathematical model which takes into consideration the charge
(and mobility) distribution as well as the size dependent losses in the system.
In the following, a numerical experiment is presented which in judgment of
the authors could be easily adapted to the resolution analyses of other size
distribution measuring instruments such as the single particle optical counters,
Coulter counter, etc.
Referring to Fig. II 1-14 the current carried by the particles in a given
mobility interval, dl/dZ , is made up of the contributions from different
sizes, and it can be expressed by the following integral:
||= K(Dp).f(Zp)-np.e.f(Dp)dDp
o
The product f(Zp).np.e denotes the amount of charge carried by a
particle of size Dp and mobility Zp. The number of charges np is obtained
from eqn. (III-1) with known Dp and Z . The true aerosol distribution
function is f (D ) and K accounts for the losses in the system and for the
uncharged fraction as shown in Fig. 111-10.
-------
111-26
We note that eqn. (HI-8) which was used to calculate the "apparent"
size spectrum g(Dp) may be written as
(III-9)
P dZp
Equating the left hand sides of eqns. (III-8) and (III-9) we obtain an
integral equation which relates the apparent distribution g(Dp) and the
true distribution f'~
'"
/TV \ va^j / u,j->v^ \ Y\
g(Dp) = / P !K-f (rip)2E f
K-nD»e J p B
P c
Solution of the above integral equation for f(Dp) would, in principle,
permit the determination of the true size distribution.
A simpler procedure for the error estimate is to insert a known, say
log-normal size distribution for f(D ) and to observe the difference between
f(Dp) and g(Dp). P
This latter method was used to obtain g(Dp) for log-normal f(Dp)
with. $~a = 1.02 and -TO — 1.35 and for mean sifees 0.03 and 0.1 /vm. The
results of the numerical experiment are shown in Fig. (111-15) on a log-
probability plot. Essentially monodisperse aerosols, -Tj, = 1.02 are indicated
by the instrument as size distributions of ^g —1.22 ana <71 =1.21 for 0.03
and 0.1 ^m respectively. Aerosols with 3"g = 1.35 are also distorted on
the lower and the upper ends but their geometric standard deviation is
essentially preserved in the main portion of the distribution. There is
in addition a systematic shift to smaller Dp for the "apparent distributions",
which may be attributed to the polydisperse (^->1.35) small (Dp < .02)
aerosols used for the calibration of the instrument. This is in accordance
with the observation that for T„ ~ 1«35 the number median diameters are
closer to the true NMD than for ^p = 1.02. For
-------
111-27
was placed at the sensor inlet to reduce the particle concentration by a
factor of 100 and to a level acceptable to the system.
The system was operated automatically and was programmed to measure the
aerosol size distribution for a period of 10 minutes at periodic intervals.
At the end of each 10 minute measurement period the data were recorded
automatically on printed and punched tapes by the Teletype printer. The
command signal to begin each measurement cycle was provided by the digital
clock which also started the data acquisition system by which other data
were recorded. Thus a precise time correlation of all recorded data could
be obtained.
The lower sizing limit of the system is approximately 0.33 Am. The
upper sizing limit, corresponding to the saturation level of the preamplifier
in the optical sensor, is approximately 6.8 ^m. Thus no sizing information
was obtained for particles larger than 6.8 ^m, although these particles
were counted by the instrument. The size distribution data were spread
over 58 channels in the multi-channel analyzer, giving ana/erage size
increment of 0.11 /wm per channel.
Optical Sensor
Except for the modification described below, the sensor used, a Royco-x-
model PC 220, is a standard commercial instrument of the 90° scattering type.
The operating principle of the sensor is well known. As each particle
passes through the illuminated sensing volume, it scatters a pulse of light
which is detected by the photomultiplier tube. The output from the sensor
is in the form of voltage pulses, each approximately 100 /-fswide. The
amplitude of a pulse provides a measure for the particle size.
The sensor was operated at the standard flow rate of 2.83 liters per
minute (0.1 cfm). According to the manufacturer, at a particle counting
rate of 5000 particles/sec, the loss of particle count due to coincidence is
approximately 10$. This corresponds to a particle concentration of 106
particles/cc at the standard flow rate of 2.83 liters/min.
In order to improve the resolution of the sensor, the standard inlet
tube of the sensor was replaced with the sheath-air inlet tube shown in
Fig. III-9. In the sheath-air inlet tube, part of the sample drawn in
is by-passed through an absolute filter. The particle-free air thus obtained
is reintroduced as a clean air sheath around the unfiltered sample and the
combined stream then enters the illuminated viewing volume where the particles
are counted. Of the 2.83 liters/min total sample drawn in by the sensor,
2.36 liters/min is by-passed through the filter, while the remaining 470 cc/min
is unfiltered. Thus the effective sampling rate of the instrument, on which
measurement is made, is reduced from 2.83 1pm to 0.470 1pm using the sheath-
air inlet.
#Royco Instruments, Inc., 141 Jefferson Drive, Menlo Park, Calif. 94025
-------
111-28
The sheath-air inlet was used primarily because it was found that the
light intensity in the sensor viewing volume was quite non-uniform and that
by confining the particles to a smaller diameter stream the sensor output,
when sampling a monodisperse aerosol, could be made considerably more uniform,
thus greatly improving the resolution of the sensor. With this modification
the sensor was found to have a resolution of approximately 10$ in terms of
particle diameter (see Fig. 111-15).
In addition to improving the resolution of the optical sensor, the
sheath-air inlet has the following effects upon the operating characteristics
of the sensor: (a) As discussed previously, the effective rate of flow of
the aerosol on which measurement is made is reduced from 2.83 liters/min to
470 cc/min. While a disadvantage for such applications as clean-room
monitoring where the particle concentration is low and a higher counting
rate can produce a given statistical counting accuracy in a shorter time,
this is not a problem for such a concentrated aerosol as the smog. In fact,
with the sheath-air inlet, the maximum particle concentration acceptable to
the sensor is increased. Theoretically, for the same coincidence loss of 10$
the acceptable particle concentration is increased to (2830/470)(106) =
638 particles/cc by means of the sheath-air inlet. Unfortunately, the
Royco 170-1 pulse converter imposes an additional limit on the system and
the maximum particle concentration acceptable to the system is considerably
lower than what the sensor can accept, (b) The calibration of the sensor,
i.e. the relationship between particle size and pulse amplitude, is affected
by the sheath-air inlet. This shift in calibration is apparently due to the
change in the average light intensity in the illuminated area of the viewing
volume as the diameter of the stream containing the particle is reduced.
(d) With the sheath-air inlet, aerosol particles can no longer spill into the
cavity surrounding the viewing volume SB they do with the regular inlet.
This has the effect of reducing the sensor response time from the order of
1 minute to what my be described for all practical purposes as instantaneous
response.
Diluter
A two-stage series diluter was used to reduce the particle concentration
of the smog aerosol by a factor of 100 and to a level acceptable to the
optical counter. Each diluter stage consists of an absolute filter (GeLnan
Model 12103) punctured with a small capillary tube. The capillary tube,
approximately 0.6l mm in diameter and 2.5 cm or 5 cm long, serves as a controlled
"leak". A mixing baffle at the outlet of each diluter stage insures that
the particles are uniformly mixed before entering the next diluter stage or
the optical sensor. The design of the diluter is shown in Fig. III-l.
The series diluter was calibrated with a monodisperse latex aerosol.
The aerosol concentration was first measured without the diluter and then
with the diluter in series with the optical sensor. The dilution factor,
as calculated from the ratio of aerosol concentrations thus determined
was found to be independent of the size of the latex aerosol within the size
-------
II1-29
range of the latex aerosol, 0.5 to 1.97 /•* m, used in the calibration. It
was found also that the dilution factor thus determined was sensibly the same
as the value calcuJated using the flow rates through the filter and the
capillary.
Pulse Converter
The Royco 170-1 pulse converter provided the necessary electronic
interface between the optical sensor and the Hewlett-Packard 5400A multi-
channel analyzer. Its function can best be described with reference to
Fig. 111-10.
The sensor output, shown as A in Fig. III-l?, is in the form of a
voltage pulse approximately 100 ^s wide. The amplitude of the pulse is a
measure of the particle size. This pulse is amplified and held at its
peak by the pulse converter. Curve B shows the pulse converter output, which
is applied to the input of the multi-channel analyzer. The pulse converter,
after detecting an incoming pulse at its input, and after a suitable interval
of delay, then issues a "gating" pulse to the multi-channel analyzer, shown
as curve C, causing the latter to read the voltage- present at its input at
that particular instant. The delay time is variable and can be set by a
front panel control to one of the following values: 50, 100, 200, 400
and 800 ^s. The delay time used for a specific optical sensor must be
sufficientJy long to insure that the incoming pulse has already reached
its peak at the time when the gating pulse is issued. However, it was
discovered, and for reasons still unknown to us, that the pulse converter
would occasionally issue two gating pulses after it was triggered. The problem
was especially severe when the delay time used was short. At a delay time
of 400 /
-------
Ill- 30
In using the pulse converter in an optical counting system such as
the present one, two additional effects must be considered which further limit
the maximum rate at which particles can be counted. One such effect is the
increased coincidence loss due to the dead time in the converter. As the
converter is triggered by an incoming pulse, there is a certain period, the
dead period, tH, during which the arrival of another pulse at the converter
input would not be regarded by the converter as a separate pulse. Since the
arrival of pulses at the input of the converter is a random phenomenon,
the loss of particle count due to this finite dead time can be calculated
by the Poisson law
Pfx) = Exp (-f trf)(f td)x (111-11)
x
where P(x) is the probability that x pulses will arrive at the converter
during the time interval, t^ and f td. is the average number of pulses arriving
during this same time interval, td. Here, f is the mean rate at which pulses
arrive at the converter input, or the average rate at which particles are
counted by the system. Since there is no coincidence loss only when x = 0,
the fraction loss of particle count due to coincidence is therefore,
cf= 1 - exp (-f td) (111-12)
For the pulse converter, the dead time, t,, is approximately equal to the
delay time, which was set to 400 jxs in these experiments. Therefore, at a
coincidence loss level of 10$, the maximum particle counting rate would be
250 particles/sec, or an equivalent particle concentration of 32 particles/cc
at the aerosol flow rate of 470 cc/min.
A second, and a more serious effect, is one that could have been avoided,
perhaps with a better converter circuit design. It was discovered that as
the count rate exceeded approximately 10 counts/sec, the dc level of the
converter output was depressed, causing the apparent pulse amplitude to
decrease, since the pulse amplitude was measured from the ground level. This
had the curious effect that as the count rate exceeded approximately 10
counts/sec there would be a shift of particle count from a higher to a lower
channel, in the multi-channel analyzer, corresponding to an apparent decrease
in the particle size. With the system operating under the standard conditions
noiTually no counts should appear below channel 10 since this is the first
channel above the minimum trigger level of the pulse converter. This was
found to be indeed the case when the count rate was below approximately
10 counts/sec. However, as the count rate exceeded this limit counts would begin
to appear, first in channel 9, and then in channels 8, 7, etc. as the count
rate continued to increase. For this reason, data in channels 7, 8 and 9
were also recorded during these experiments. They serve as a useful index
for determining whether the count rate limit of the converter was exceeded.
Although there were a few instances when the smog was very heavy and this count rate
limit was exceeded, the effect on the overall accuracy of the data was not
considered to be serious The particle concentration corresponding to a
count rate of 10 counts/sec is 1.28 particles/cc for an aerosol flow rate
-------
111-31
of 470 cc/min.
The input-output (transfer) characteristics of the pulse converter
were determined by calibration with simulated pulses of a trapesoidal shape
produced with a Tektronix Model 115 pulse generator. The repetition rate
of the pulses were kept low to insure that the output dc level of the pulse
converter was not depressed. Input and output pulse amplitudes were measured
with the multi-channel analyzer. The results are shown in Figures III-18
and III-l9 which also include a detailed specification of the shape of the
calibrating pulses used.
D. Multi-Channel Analyzer
A Hewlett-Packard Model 5400A multi-channel analyzer was used to measure
and record the amplitude distribution of the pulses from the converter.
Its operation will not be described here as this informiion is available in
the technical literature published by the company.*
For the present application, the instrument was operated on the ''sampled
voltage analysis"modalithis particular mode of operation the instrument
measures the instantaneous voltage at its input and registers one count in
the appropriate channel upon receiving a gating pulse at the input gate. It
was not possible to measure the pulse amplitude directly in the "pulse height
analysis" mode because the maximum rise time permitted for this particular
mode of operation is 12.6 jis, which is considerably shorter than the rise
time of the pulses produced by the optical sensor.
E. System Calibration and Performance
A prime system calibration is needed in order to determine accurately
the relationship between the size of the particles and the corresponding
channels in which the counts are registered. The calibrating procedure
consists of feeding a monodisperse latex aerosol of a specific size to the
optical sensor and recording the corresponding pulses with the multi-channel
analyzer- for a specific interval of time. Results of such a calibration
performed in Minneapolis are shown in Figure III-2D . A complete system cal-
ibration was performed again after the system was installed in the Keck
Laboratory in Cal. Tech. The results are shown in Figure III-23, If these
results are compared it will be seen that they do not completely agree,
indicating that there is a shift in the system calibration. This shift in
system calibration was found to result from the slight damage the optical
sensor suffered during its shipment from Minneapolis to Pasadena. The optical
system in the sensor was realigned upon its arrival in Pasadena.
In operating the system, it was decided to chose a specific set of conditions
as standard and to operate the system under these standard conditions over
the entire experimental period. The standard conditions chosen are as follows.
-^Hewlett-Packard Co., 1101 Embarcadero Road, Palo Alto, Calif. 91604.
-------
111-32
Optical Sensor: 2.83 liters/ min total flow
470 cc/ min aerosol flow
Pulse Converter: Coarse Gain = 2
Fine Gain = 0
Baseline = 0
Delay time = 400 us
Multi-Channel Analyzer: Input sensitivity = 10 volts into 128 channels
Baseline = 0
Data output: channel 7 to 68
The system calibration was checked twice with monodisperse latex aerosols
of 0.5 )jm during the entire experimental period. The system was found to
retain its calibration well during this period.
The prime system calibration was performed only with monodisperse latex
aerosols of five different sizes, 0.36, 0.5, 0.79, 1-305 and 1.97 urn. The
manufacturer provides a calibration curve relating particle size and pulse
amplitude for the 200 sensor up to a diameter of 5 urn. However, this curve,
shown in Figure 111-14, does not agree with our calibration of the sensor after
the sheath-air inlet modification. If we assume that the ratio of the pulse
amplitudes for two different sized particles is given correctly by Figure 111-14
for our sensor, then a calibration curve can be constructed for our system
up to a particle diameter of 5 jum. Curve B in Figure III-22 is constructed
on this basis using the actual system calibration point at 1.97 >mi as a base, and
the pulse amplitudes ratio as determined from Figure 111-14. However^ if we
assume that the pulse amplitude is proportional to the square of the particle
diameter ibr particles of 1.97 pm and larger, then curve C is obtained. Since
the true calibration of the system is not known above 1.97 fan, it was decided
that data reduction would be done on the basis of the straight line relation-
ship, shown as curve A in Figure 111-22. This straight line is given by the
equation
Dp = 0.11 (N - 7) (111-13)
where Dp is the particle diameter in urn and N is the corresponding channel
number. It is hoped that this question will be resolved in the future as we
make further calibration studies on the optical counting system. However, it
should be noted that the difference in particle diameter as determined from
these three curves is less than 28$ in the range, 1.97 urn to 6.8 urn, where
actual calibration data are not available.
The size distribution data in the first channel, and the last few channels
of the system are not reliable for the following reasons. According to
equation (3), the first channel, i.e. channel 10 has a nominal particle size
of 0.33 urn, and a nominal interval width of 0.11 urn. On a relative percentage
-------
111-33
basis, both of these values could be in substantial error. Further, as the
signal to noise ratio is decreased with reducing particle size, the pulse
amplitudes resulting from a monodisperse aerosol would show an increased
spread as the noise voltage becomes increasingly significant compared to the
signal voltage. Thus some of the pulses would fall below the trigger level
of the pulse converter, causing only a fraction of the particles to be counted.
Therefore the counting efficiency should be less than 100$ for this first
channel. The data in the last few channels of the system (channel 68 is the
last channel in which counts were registered) are difficult to interpret
because all particles larger than the minimum size needed to saturate the
preamplifier are registered there.
E. Effect of the Refractive Index of Particles
Since the current practice in calibrating optical counters is to use
monodisperse latex spheres as the standards of calibration, we wish to propose
that the size of a particle measured with an optical counter should strictly be referred to
only as an"equivalat latex sphere diameter". Here, the equivalent latex sphere
diameter of a particle is defined as the diameter of a latex sphere that
produces the same response in the optical counter as the particle under
consideration. The use of equivalent diameters in the science and technology
of aerosols is quite common. Familiar examples include the Stokes diameter, the
aerodynamic diameter, etc.
The equivalent latex sphere diameter of a particle generally depends on
the optical design of the optical counter and for counters of a specific
optical design, it is also a function of the size, the shape and the optical
properties of the particle. Since the majority of the smog particles are
probably liquid drops, they are probably all spherical in shape. Therefore, the equivalent
latex sphere diameter for a spherical particle with various indices of re-
fraction is of interest in interpreting the data obtained in the present study.
Figure 111-15 shows the relationship between the geometrical size of
a spherical particle and its equivalent latex sphere diameter for various
refractive indices. The curves are constructed according to theoretical
results presented by Hodkinson-"- for the Royco PC 200 sensor whose optical
design is similar to that of the 220 sensor. The refractive index of the smog
aerosol is not known. However, if it is within the range of 1.4 to 1.8,
then the ratio of the geometrical size to the equivalent latex sphere size
varies from a maximum of 1.5 to a minimum of 0.9 , with the mean ratio being
on the order of 1 over the size range of our measurements.
^Hodkinson, J.R. and J.R. Greenfield, "Response calculations for light
scattering aerosol counters and photometers", Applied Optics 4:1463-1474 (1965)
-------
UJ
ID
3
UJ
_l
z
z
o
4
(E
uj g
5 2
j o:
o >-
in o:
o
o
2
UJ IO
1 u
5 a: N
m £ «
ld*
s « <
ui 52
$ u 5
s\
L
S
&
\
— •
y^y^Vv^v^T^vy^w y S/\/VSO>^O^O<^X>^(\/^^
^ \ ^ J
fcll^XX^A^^L^^F
|^vS(Vy><>9<^o"
O
o
(E K
UJ UJ
_i
-I O
o to
I Q)
C CQ
H S
I
H
H
H
bO
•H
H
CD-P
H Pi
O 3
0 0
P! o
FROM WAA
NOT IN USE
FROM THE
MAIN SAM-
PLING TUBE
NOT IN USE
;-:
COOLING AIR
^-SOUNDPROOF BOX | |
1 T iLML!'
I^ VACUUM / 1'
£•"••1 ' PUMP y^ J
•-..-4 l.J \. i 1 r-Jn-1
L J I ) \ J 1 ^^
X U w U 1 ^n,1,^
^^ MUFFLER- MUFFLER _^ i !
FILTER (S\ ' j
HI-VOLUME GAGEE EL'° Vl/"""""-
y SAMPLER umru.t
\n ' SURGE TANK — j j
^ - Q/)— :..!
- JT-'^^ (N>-_---,
ORIFICE FLOWMETER , ^ — | ,„]
ORIFICE
.
COMPRESSED AIR
SIIPPIY— TO WAA
WASTE OUT
NOT IN USE
WASTE OUT
Figure Ii.1-2 Schematic of the air handling
box used for the hAAS instruments.
-------
N
Channel
Number Equipment
1 Pyroheliometer
2-4 Wind Instruments
5 Temperature
6 Temperature
7 Temperature
8 Relative Humidity
9 open
10 Date Code
11-13 open
14 Atlas - Oo
15 Mast - 03
16 Atlas - NO
17 Atlas - M02
18 open
19-22 Univ. of Washington
Nephelometer
23 MRI Nephelometer
24 G.E. Nuclei Counter
25 Whitby Aerosol Analyzer
• Data Signal
"Control Signal
W
E
Pumps
25
7
24
6 & 8 .,-•'/'
19
to
22
-' 1 to 5
O' Stack
Impactor
Equipment
(Hood)
Figure III-3 Sketch of Room #10, W.M. Keck Laboratory
(not to scale)
-------
AUG. 22, LA RUN 72-77
SEPT. 4, LA RUN 325-335
LAB. EXPERIMENT NICHROME WIRE AEROSOL
LAB. EXPERIMENT PROPANE TORCH AEROSOL
CNC = 1.75-WA A
10
20
CNC XIO"3,
Fig. III-5 Comparison of total numbers measured by the GE Condensation Nuclei Counter and calculated
from the number spectra obtained by the WAA.
-------
3.5 KV (-}
0-15 KV(+)
COMPRESSED
AIR
-UNCHARGED
AEROSOL
CHARGER
DETAIL
CORONA
DISCHARGE
JET OF
NEGATIVE IONS
MIXING ZONE
(AEROSOLS IONS)
CHARGED
AEROSOL
AMP
! i n
ELECTRONICS
TO
DATA <
AQUIS.
SYSTEM
SLG_NA_LJ
READ
COMMAND'
EXT. TRIGGER
AIR IN1
CHARGED
AEROSOL-
NOT COLLECTED
CLEAN
AIRS
COLLECTED
AEROSOL
MOBILITY
ANALYZER
COLLECTIO
CTION/
ROD^
Q.
T2
^b
TOTAL
FLOW
VACUUM
PUMP
CURRENT
COLLECTING
FILTER
Fig.III-6. Schematics of the Whitby Aerosol Analyzer
-------
CO
or
i,
00
UJ
o
o
°0
I I I I I
O 0.5/tm PSL, o-g = 1.18, np= 25.0
D 0.365/j.m PSL, Og = |.l8, np= 19.5
A O.O75—O.l/im MB, crg = l.28,
np* 3.9-5.2
.2
.3 .4 .5 .6 .7
DIMENSIONLESS DISTANCE X / L
.8
1.0
Fig. III-7. Deposition pattern of Polystyrene Latex and Methylene Blue
particles on the rod of the ¥AA.
-------
10
£
>
a:
<
| 1.0
UJ
UJ
li.
o
z
<
£ .1
_ I I i I ii| 1—I i i i i ii| 1—r—r
I I 111
FRACTION CHARGED, f
1 I
J I I I I I M
0.01
O.I
1.0
D,
Fig. III-8. The mean number of charges n_ and the fraction of particles
charged, f as a function of particle size.
-------
OJ Hsl I I I I
o
LJ
0)
.01
Q.
IN
m
o
o
cr
I-
o
Ld
_l
LJ
<
LJ
.001
,0001
1—I I I I III
1 I I i M m
I I I I
O APRIL, 1969
D MAY, 1970
SINGLE CHARGED
PARTICLE
I I I I I I I I
I 111 1 IN
.004
.01
O.I
1.0
D ,
Fig. III-9. The mean electrical mobility Z as a function of particle size.
-------
I~\ I I I 1
A t>_
I I
PENETRATION, p
O FRACTION CHARGED, f
I III
0.001
0.01 O.I
PARTICLE DIAMETER , Dp ,
Fig. 111-10. The fraction of particles charged and the aerosol penetration
through the WAA.
1.0
-------
FAN
ABSOLUTE
FILTER
SAMPLING LINE
DIFFUSION BATTERY
AEROSOL'
INPUT
ABS. FILTER
HEATED
NICHROME WIRE
NQCL
^•AEROSOL
GENERATOR
ROYCQ'
Fig. III-H. Experimental set-up for the calibration of the WAA showing
the method i>r generating the small aerosols.
-------
24 10 20
NUMBER OF ELEMENTARY CHARGES, np
Fig. Ill-12. Cumulative change distributions on particles of different sizes.
40
-------
1 1 I 1 I 111
1 I I 1 I 111
1 I I I I 1 I
1.3
1.2
CHARGING
O M
H Q
1.0 —
o—o-
I I _L I I LL
I III
I L
0.01
O.I
D,
1.0
Fig. m-13. The logarithmic standard deviation of charging as a function
of particle size.
-------
O.I |Z
o
LJ
CO
2
O
a.
N
I-
CQ
O
O
CE
h-
o
LU
_J
LU
0.01
0.001
0.01
Fig. 111-14. The mobility spectrum of particles charged with the
jet charger.
-------
0.01
Fig. 111-15. Comparison between the true size spectra, f(Dp) and the
indicated spectra g(D_). The dashed lines show f(Dp) and
g(Dp) for a
-------
4-" O.D. COPPER TUBE
STANDARD £ GYKOLOK
3 SQUALID SPACED
HOLES "
\
N
^"DISMOUNTING FLANGE
x-g" i.b STAINLESS
0.07S" O.Z>. x - J. D. STAINLESS TUBE
FILTER
M-4S DZ 78001 HOLDER
Figure 111-36. Sheath-air inlet for Royco 220 optical sensor
-------
LJ
-I
<
O
(/)
>
a:
<
o:
CD
cc
LJ
O
§
§
(A) INPUT SIGNAL
(B) OUTPUT SIGNAL
(C) GATING SIGNAL
100 200 300 400
TIME,
Figiire III-l? Input and output signals from Royco 170-1
pulse converter
-------
140
120
0123
Figure 111-18 rtoyco 170-1 pulse converter c
4 5
alibration w , VOLTS
-------
o
I
Figure III-19 Royco 170-1
3 4
pulse converter calibration V,, VOLTS
8
IO
-------
15
20 25 ~ " 30 35
igure 111-20 Orvticql counter systen ca libra-
ticn shewing the instrument resolution.
40
45
50
55
60
-------
\SL
4_
O.L
M
012345
figure 111-21 Royco 220 sensor calibration according to the manufacturer.
-------
PARTICLE DIAMETER,/!,
Figure 111-22 System calibration for the optical counter under the
standard ooerating conditions
-------
EQUIVALENT LATEX-SPHERE DIAMETER OF
PARTICLE INDICATED BY OPTICAL COUNTER, >L
Figure 111-23 Effect of refractive index on the response of the optical counter.
-------
Table III-l. Calibration constants for WAA, particle
size range 0.0075 - 0.6/4m.
Date
WHITBT AEROSOL ANALYZER, Calibration May 1970, RBH
Aerosol Flow Rate: 0.32 cfm Test
n
DP
-0075
.01
.015
.02
.03
.04
.06
.08
.1
.125
.15
.2
.3
• 4
.6
A n
A Dp
.0025
.005
.005
.01
.01
.02
.02
.02
.025
.025
.05
.1
.1
.2
n
DPi
.00875
.0125
.0175
.025
.035
.05
.07
.09
.1125
.1375
.175
.25
.35
.5
4 N
AI
1,250,000
637,000
343,000
163,000
89,200
47,700
26 , 800
18,300
13,70C
10,700
8,340
5,790
4,100
2,800
A N
4 1 ^Dp
500,000,000
124,000,000
68,600,000
16,300,000
8,920,000
2,380,000
1,340,000
915,000
548,000
428,000
166,000
57,900
41,000
14,000
Volts
225
400
920
1400
2400
3200
4450
5800
7200
8300
9200
10,600
12,300
13,300
14,000
St
0
1
2
3
4
5
6
7
8
9
a
b
c
d
e
I
A I
4 N
AN
A Dp
-------
Table III-2. Calibration constants for ¥AA, particle
size range 0.004 - 0.035 ^m.
WHITBY AEROSOL ANALYZER
Calibration May 1970, RBH
Aerosol Flow Rate: 0.32 cfm
Date
Test"
Dp
.004
.005
.006
.007
.008
.009
.011
.013
.015
.017
.019
.023
.027
.031
.035
^DP
.001
.001
.001
.001
.001
.002
.002
.002
.002
.002
.004
.004
.004
.004
DPi
.0045
.0055
.0065
.0075
.0085
.010
i.012
.014
.016
.018
.021
.025
.029
.033
AN
Al
7,800,000
4,000,000
2,500,000
1,750,000
1,400,000
1,100,000
700,000
510,000
400,000
320,000
240,000
170,000
130,000
102,000
Al dVp
7.8 E+09
4.0 E+09
2.5 E+09
1.75 E+09
1.4 E+09
5.5 E+08
3.5 E+08
2.55 E+08
2.0 E+08
1.6 E+08
6.0 E+07
4.25 E+07
3.25 E+07
2.55 E+07
Volts
65
100
145
200
260
330
500
680
880
1080
1280
1680
2080
2450
2750
St
0
1
2
3
4
5
6
7
8
9
a
b
c
d
e
I
Al
AN
AN
a°p
-------
IV-1
BECT1GN IV
Aerosol Sampling for Determination
of Particulate Mass Concentration,
Chemical Composition and
Size Distribution
by
Dale A. Lundgren
Staff Scientist
Environmental Research Corporation
Introduction
This paper describes the use of a Lundgren Impactor together
with a total particulate filter, to determine the size distribu-
tion, concentration and chemical composition of particulate matter
in air.
Determination of the above information involves:
1) Obtaining size fractionated samples of the
aerosol over desired time periods.
2) Collecting out the particulate matter in
such a way that it is amenable to analysis.
3) Analyzing the particulate samples by tech-
niques capable of providing the required
information.
These criteria are interrelated; therefore, the aerosol
sampling method was based on knowledge of both the sampling and
analysis limitations and on knowledge of the type of aerosol prop-
erty data desired.
Aerosol sampling normally involved simultaneous use of the
impactor (with its after filter) and a total particulate filter.
Both the impactor and total filter are of the same size, use the
same filter media and sample at about the same flow rate. This pro-
vided an excellent check on the total particulate concentration and
particle losses within the impactor. All samples are to be analyzed
chemically; this, in turn, provides a similar quantitative check on
the chemical results. Oa several occasions two impactors were run
simultaneously, each using different collection substrates and filter
media. Many additional total filter samples were also obtained when
the impactor was not run.
-------
IV- 2
Method Description
The developed sampling-analysis method is primarily based on a
fairly high flow rate four-stage impactor (called the Lundgren Im-
pactor) originally developed for the National Air Pollution Control
Administration to characterize urban particulate matter by size dis-
tribution over day-long time periods. The unit is a compact field-
operational sampler having well defined operating characteristics and
great versatility. A cross-section schematic of the impactor is
shown in Figure IV-J. Its four impaction stages were designed so as to
fractionate particulate matter at diameters of 10, 3, 1 and 0.3
microns (y), for spherical particles of density two, at a flow rate
of 4 cfm. The impactor ife followed by an after filter-
Impactor flow rate is not fixed, therefore, air can be sampled
at any rate from less than 0.5 cfm to over 5 cfm. A useable range is
from 0.5 cfm (to prevent gravitational loss of very large particles)
up to 5 cfm (to prevent excessive pressure drop in the unit and high
impactor wall loss of medium size particles). This flow rate range
enables classification of particles from about 0.2y to about 40y dia-
meter--depending on particle density, impactor stage and flow rate--
as seen in FigureIV-2. Collection characteristics of the impactor were
*
determined experimentally. Stage 50% cut points shown in FigureIV-2
were calculated from the calibration data. The diameters are actual
diameters for spherical particles and are aerodynamic or Stokes dia-
meter for other than spherical particles.
The total filter holder and the impactor filter holder were alike
and both accommodate 90 mm diameter filter media. Total particulate
concentration was determined with the total filter, while the impactor
filter served to determine particulate concentration of a size too
small to be collected by the impactor. Filter media for both filter
holders was handled identically; therefore, the total filter sampling
procedure is not discussed separately as all comments made regarding
the impactor filter apply equally to the total filter-
*
Lundgren, D. A., "An Aerosol Sampler for Determination of Particle
Concentration as a Function of Size and Time". APCA Journal,
Vol. 17, No. 4, April 1967.
-------
IV- 3
The operation.of all impaction devices is based on the fact
that a particle's inertia is a function of particle velocity and
mass-or particle size, density and velocity. This principle ex-
plains the separation of particles into two fractions: those
having sufficient inertia to leave the airstream, and those with
less inertia which follow the airstream lines. Air drawn into the
impactor is accelerated in the first stage converging nozzle and
then forced to flow around the first stage collection drum. Large
particles have sufficient inertia to leave the airstream and impact
out against the first stage collection surface. Remaining particles
are carried by the airstream to the second nozzle where the air is
accelerated to a higher velocity, allowing somewhat similar particles
to be impacted out. This process is repeated in the third and fourth
stages and then all remaining particles are collected out by an
efficient after filter.
All four impactor nozzles are two inches high but of different
widths, giving a fixed collection deposit height of two inches and
a deposit length which is proportional to collection drum rotational
speed and sampling time. The one and one-half inch diameter collec-
tion drums have a circumference of about 4.7 inches, thereby pro-
viding almost 9.5 square inches of collection area per stage. Because
of this great sampling area, this unit can sample normal atmospheric
air over 24-hour time periods, or sample very dusty air over short
time periods—without the normal problem of collection surface build-
up and blowoff. The collection drum drive can be set to provide the
proper rotational speed for the desired sampling time. In all cases,
a chronological collection deposit is produced with very good time
resolution.
In the referenced paper describing impactor design and calibra-
tion, data on particle losses within the impactor as well as the
effects of collection surface coating and deposit density on collection
efficiency are given. Particles such as silica or glass beads do not.
readily stick to the impaction surface; therefore, a viscous oil or
grease should be used to coat the collection surface when sampling
particles of that type.
-------
IV- 4
Sampling Procedure
Any sampling procedure should be mated to the analysis method
or methods. Because several analysis methods may be chosen, the
sampling procedure should be as versatile as possible without being
overly involved. The following procedure resulted from over a year
of trial and is considered satisfactory in most respects.
It was stated that particles to be collected are impacted onto
the collection drum. Normally the drum is coated with a thin film
material and the particulate matter collected onto this removable
drum coating material. Materials such as aluminum foil, stainless
steel shim stock, sticky tapes, and various plastic films have been
used with some success, but none have proven more suitable or
desirable than a film of Teflon, about 0.001 inch or so thick. In
general, the sticky tapes (such as two-sided sticky tape) were the
least desirable method and are not recommended.
Teflon film has the advantages of being inert, having a low
chemical background, and having a low affinity for water vapor. It
is also very weight stable, is not affected by most acids or bases,
and is quite transparent under the microscope. It can easily be cut
into time based segments for chemical analysis.
Because of the inert nature and very low hygroscopicity of
Teflon, its weight reproducibility is excellent and quite independ-
ent of humidity. This weight stability is essential if atmospheric
aerosol weight distributions are to be determined accurately. Col-
lection stage weight changes (particulate collection) of less than
0.05 mg have been determined by this method (using a Sartorius semi-
micro analytical balance with an obtainable weight accuracy and
reproducibility of 0.01 mg) . Static charge on Teflon is often a
problem in weight determination; therefore a static charge
*
eliminator (a radioactive ion source) is routinely used. If small
It
Nuclear Products Co., El Monte, Ca. ModeT 2U500 (uses SOOyc
of PO 210).
-------
IV- 5
weight determinations are to be made, it is essential that the film
be held onto the collection drum with a mechanical type clamp; ad-
hesives or tapes should not be used if weight changes much less than
1 mg are to be accurately determined. If only a microscopic or chemi-
cal analysis is to be performed, the Teflon film can be held to the
drum with cellophane type tape; the taped end can then be cut off
before chemical analysis to prevent contamination from the tape
itself or from dirt it may have picked up.
As mentioned, an after filter must be used to collect particu-
late matter passing the impactor fourth stage. Desirable charac-
teristics of an after filter include high collection efficiency, low
pressure drop, good weight stability, low chemical background, and
good particulate loading characteristics (filter does not plug up).
A variety of filter materials have been tested to determine their
overall usefulness. Based on weight stability and low chemical back-
ground, Teflon filters (sold by Millipore Corp.) ranked highest and
were normally used. These filters can be pre-cleaned by washing in
acid and water to reduce their chemical background to near zero.
Teflon filters are expensive; if a low-cost filter media is
desired, the standard glass fiber filter media should be considered.
It has reasonably good weight stability and a fairly low chemical
background, with a few notable exceptions (such as iron). Standard
type membrane filters have low chemical backgrounds but are not
sufficiently weight stable for small weight change determinations.
The standard impactor after filter holds 90mm diameter filters.
This large area is necessary to minimize pressure drop through Teflon
and other membrane filters. Even with a large size filter, problems
will be encountered because of filter plugging unless large pore
size filters are used. Ten micron poire size Teflon filters were
tested and found to perform adequately but five micron pore size
filters were found to be borderline because of pressure drop in-
creases with time (filter plugging). Both pore size filters were
-------
IV-6
found to be about 99 percent efficient by weight on atmospheric
particulate matter at a test flow rate of 3 cfm. This is con-
sidered quite satisfactory and the ten micron pore Teflon is rec-
ommended for general use whenever low chemical background is im-
portant. In analysis, the particulate matter is actually dissolved
out from the Teflon filter, leaving the filter intact. If the
Teflon filter is pre-cleaned it will contribute almost no chemical
background to the analysis.
Assuming a complete analysis of an aerosol is to be made, a
stepwise procedure is outlined below. Explanations are given for
several of the step procedures.
1. Drum Rotational Speed Selection
Based on a desired sampling time, sample time resolution, dust
concentration and deposit density limitations, a suitable rotational
speed is selected and set for the colle.ction drums (all drums rotate
at the same rate). One revolution in 24 hours is adequate for nor-
mal atmospheric sampling, one revolution per hour for factory air
sampling, and one revolution per minute or less for sampling a dirty
source .
2. Air Flow Rate Selection
Based on the desired particle size fractionation (which in turn
is based on particle density), volume of air to be sampled, dust
concentration, and amount of sample needed for analysis, a suitable
flow rate is selected and set. Normally, the flow rate is set reason-
ably high (2 to 5 cfm) to enable collection of as much particulate
material as possible and to allow particle size fractionation at the
smallest size possible. The high flow rate obtainable with the
impactor is often a great advantage.
-------
IV- 7
3-A• Collection Surface Preparation - For Weight
Determination
a) Cut pieces of film to proper
size.
b) Wash film and drum in clean acetone to remove
any grease or dirt.
c) Weigh film and after filter.
d) Stretch film over drum surface and fasten
in place.
e) Install drums in impactor and filter in
filter holder.
f) Sample for desired time.
g) Remove drums from impactor.
h) Remove films and filter (containing collected
particle matter) and weigh. Weights should
be made at a reference humidity because the
particle matter does pick up and lose water-
i) Examine film under microscope and photograph
if desired. Film (or film stripl is then ready
for chemical analysis.
j) Analyze.
If weight determinations are not required, the collection
surface preparation is as follows:
3.B. Collection Surface Preparation - No Weight
Determinations
a) Cut pieces of film to size.
b) Wash film and drum in clean acetone.
c) Stretch film over drum surface and tape
down film ends.
d) Sample as desired.
e) Remove film, cut off taped ends, examine
or analyze as desired.
-------
IV-8
Analysis Procedure
Once a sample has been obtained, it is ready for analysis. If
weights are to be determined, this is done first. The film is re-
moved from the collection drums and the filter from the filter
holder; both are allowed to equilibrate to some reference tempera-
ture and humidity before weighing. Microscopic viewing of a sample
is done after weighing (or done first if weights are not to be made),
Changes in particulate deposit density with time can often be seen
directly on the film; these changes can be determined by particle
count as a function of deposit position (or time) or by cutting the
collection surface film into time based strips and analyzing. The
large collection surface area greatly facilitates division of the
collected samples. All impactor nozzles are two inches high and
produce a uniform collection deposit over this height. If two dif-
ferent analyses or sample extraction procedures are to be run on
the same sample, or if analysis replication is desired, a simple
method for obtaining duplicate samples is to divide the two-inch
deposit into halves — or even into fourths. Weight determinations
cannot be made on the cut film sections because the initial weight
cannot be determined accurately.
Heavy film deposits can be examined by techniques such as
infrared absorption spectroscopy, X-ray diffraction, etc. In this
way, deposit density changes or composition changes can be read
directly off from the film, thereby giving the composition or con-
centration changes as a function of time within each of the various
particle size ranges.
Normally the particulate is removed (or dissolved) from the
film for analysis; using water, for example, to determine sulfate
or nitrate, or acid to determine elements such as iron or lead (the
Teflon film is not dissolved). Background levels for analyses such
as these were found to be very low; therefore, very small quantities
of material can be detected. Many methods of chemical analysis can
be used to analyze the particulate coated Teflon films.
-------
IV-9
Electron microscope grids can also be attached directly to
the drum for subsequent viewing or for analysis of individual
particulates by electron microprobe or particulate crystals by
electron diffraction. Two-sided sticky tape has worked well for
fastening grids onto the drums.
All collection-analysis methods have limits. These limits
are determined by: 1) the sensitivity of the analysis procedure;
2) the background level of the component to be detected; 3) the
concentration of that component in the aerosol; and 4) the amount
of aerosol sampled.
Example: If ambient air containing 100 yg/M suspended dust
is sampled for a 4-hour period at a 3 cfm flow rate, then about
2 mg of particulate will have been collected out on the four impact-
or stages plus the after filter, giving an average of 0.4 mg (20%
of 2 mg) per collection surface. The actual amount per section
will normally vary widely. Atmospheric air sampling results have
shown that the cleanest film may have less than 1% of the total dirt
and the dirtiest film have over 50% of the total dirt collected.
Results
Although results are not part of this method writeup, the
type of information obtainable with the described sampling instru-
ment is best illustrated by example. During November 1968, the
Lundgren Impactor was used to obtain 10 samples of atmospheric
particulate matter. These samples, all obtained on the Riverside
campus of the University of California, were analyzed as follows:
1) Particulate weight distribution determined (based on the
dry particle weight obtained by desiccating the samples for 6 hours
at room temperature and pressure in a desiccator containing drierite,
or CaSO ).
2) Water soluble fraction extracted and analyzed for sulfate
and nitrate by standard wet chemical methods.
-------
IV-10
3) Nitric acid fraction extracted and analyzed (in addition
to part of the water soluble fraction) by atomic absorption spectro-
photometry for lead and iron.
Results for these ten tests are plotted in Figure IV-3as yg of the
particulate, nitrate, etc., per M of air sampled for each of the
four impactor stages plus the impactor after filter. In all cases,
samples were run from 4 p.m. of one day until 8 a.m. of the next.
Air sampling rate was held constant at 2.9 cfm for all runs except
number one which had a flow of 4.0 cfm. Based on a density one
spherical particle and a 2.9 cfm flow rate, the impactor 50% cut
point diameters are: greater than 17y on stage one, 5.2 to 17 on
stage two, 1.7 to 5.2 on stage three, 0.5 to 1.7 on stage four and
less than 0.5 on the after filter. None of these tests run were
divided into time fractions; therefore, the numbers given represent
averages over the 16-hour sampling periods.
The data for these ten runs were used to obtain an average
distribution for a total particulate weight, sulfate, nitrate, lead
and iron. FigurelVWj.is a log probability plot of these weight ave-
rages vs. particle diameter. Again, the diameter is based on an
assumed density one spherical particle. Microscopic examination
of the collected particulates indicates this assumption was reason-
ably good for photochemical aerosols around one micron diameter- In
noting the differences in the mass median diameters and distributions
shown in FigurelV^, it is important to remember that they are all
based on the same samples — not different samples taken at different
times.
Summary
The Lundgren Impactor was described as well as the procedure
for using it to obtain samples of time-separated, size^-classified
particulate matter. This procedure was outlined in a detailed step-
by-step manner. Various methods by which the obtained samples can
be analyzed were mentioned. Results from the analysis of ten samples
were presented to illustrate the type of data that can be obtained
using the described sampling instrument.
-------
Fig. iv-i Schematic of Impactor
-------
10.0
4.0
o
i
LJ
H-
<
or
1.0-
0.3
STAGE #2
\ \ \
\ \ \
\ \ \
STAGE#1
\ \ \
\ \ \
\ \ \
i i i i i
I
i i i i i I
O.I
0.3
1.0 3.0 10.0
PARTICLE DIAMETER - microns
30.0
100.0
Figiv-2 Lundgren Impactor Calibration Giving The 50% Cut Size
-------
SAMPLE#
#8
10
-M 10 v1;
/><
— CM 10 *u-
coco
— CM rot ^
coco coco <
— CJ tO <•
COCO CO CO <
coco coco<
CO COCOCO<
co co co co <
—CM 10 ^r i»-
coco coco <
-1 0
<-
t- <
Q.
LU
2
_l
^
-------
rr
LJ
LU
5
Q
Z
<
o:
LU
LU
tr
o
LU
>-
CD
LU
U
(T
LU
Q.
2
5
10
15
20
30
40
50
60 -
70 -
80
851-
90
O.I
SULFATE
NITRATE
LEAD
TOTAL PARTICLE
IRON
L-A IRON
H—O TOTAL PARTICLE
D NITRATE
V LEAD
SULFATE
l i
MASS MEAN
DIAM.-MICRONS
2.2
0.9
0.8
0.5
~0.3
i I i I i i
DIAM. 16% SIZE
DIAM.50%SIZE
8
I I
3
7
I i I I I I
0.3
3O.O
IOO.O
Fig
1.0 3.0 10.0
PARTICLE DIAMETER - microns
xv-^Average size distributions for IO impactor samples. From ises Riverside study
-------
References
Wiitby, K.T., R.B. Husar, A.R. McFarland and M. Tomaides (1969), "Generation
and Decay of Small Ions", U. Minnesota Particle Technology Lab, Publ. No. 137.
Whitby, K.T. and W.E. Clark, "Electric Aerosol Particle Counting and Size
Distribution Measuring System for 0.015 to 1 ^m size range", Tellus XVIII,
(1966) pp. 573-586.
Fuchs, M.A., I.E. Stechkina and V.I. Staroselskii (1962), "On the
Determination of Particle Size Distribution in Polydisperse Aerosols by the
Diffusion Method", Brit. Journ. Appl. Phys. 1J, pp. 280-281.
Husar; R.B., "The Size Distribution of Coagulating Submicron Aerosols",
Thesis, (1971), University of Minnesota
-------
V-i
SECTION V
Short Summary
University of Washington
1969 SMOG EXPERIMENT
R. J. Charlson and N. C. Ahlquist
Water and Air Resources Division
Department of Civil Engineering
1. Description
The University of Washington experiment, conducted by Professor
Robert J. Charlson and Mr. Norman C. Ahlquist, consisted mainly of light
scattering measurements made with three integrating nephelometers (Ahlquist
and Charlson, 1968, 1969). Dew point and instrument temperature were
also recorded. The instruments were:
1) a four channel instrument operating in four very narrow wavelength
bands located at 360, 436, 546 and 675 nm.
2) a broad-band device covering the wavelength range from 420 to
550 nm.
3) a device with a medium wavelength band located at 550 nm
approximating the response of the human eye.
2. Purpose
The initial purpose of this portion of the experiment was to gather
data on real L.A. smog. These data necessarily must be compared and
correlated with the measurements by the other groups in order to put the
results into proper perspective. The goals fall into three classes:
1) Experiments with 3 nephelometers and one hygrometer alone:
a) Correlation of broad-band and narrow band light scattering
coefficient with wavelength dependence as a parameter.
-------
V-ii
L. A. provides a sufficiently variable aerosol for such a
study.
b) Relation of light scattering to humidity.
c) Relation of light scattering to visibility utilizing Weather
Bureau data for visibility.
2) Experiments relating nephelometers and hygrometer data to that
from other experimenters.
a) Relationship of wavelength dependence of light scattering
(Angstrom exponent, a) to size distribution (Junge exponent, 3).
(University of Minnesota)
b) Mass concentration - light scattering correlation with a as
parameter (Dale Lungren).
c) Extinction due to light scattering compared to that by NO-
(State of California).
d) Correlation of light scattering with gaseous pollutants
(State of California).
3) Experiments or interpretations which will arise as a result of
observations.
3. Initial Data Summary
The instruments were operated continuously for the period from 13 August
to 3 September. No failures occurred except for the dew point hygrometer
which began to behave erratically about halfway through the period. The
light scattering levels covered a range (at 546 nm) from below 1 x 10~ m~
to over 10 x 10 m . A study of the first weeks' data show an extremely
high correlation of broadband and monochromatic light scattering coefficient.
Perhaps the most dramatic and interesting effect is the occasional high
-------
V-iii
correlation of light scattering and dew point. At other times the correlation
is nil. Data reduction is currently in progress, and more results can be
anticipated.
-------
V-l
Summary - 1969
Los Angeles Smog Measurements
University of Washington Experiments
R. J. Charlson and N. C. Ahlquist
Water and Air Resources Division
Department of Civil Engineering
University of Washington
Seattle, Washington 98105
I. INTRODUCTION
Four separate instruments, all located at the end of about 5 meters
of approximately 3.5 cm diameter plastic tube, were operated for the
period of 13 August to 3 September 1969, inclusive. These instruments
were:
1) A multi-wavelength integrating nephelometer.
2) A broad-band integrating nephelometer.
3) A narrow band, photopic-characteristic integrating nephelometer
(MRI).
4) A dew point hygrometer.
The following discussion will consider these in sequence, including
calibration and accuracy information, data format and the recorder interface
arrangement. Finally, the basic purpose of the University of Washington
participation will be outlined.
II. MULTI-WAVELENGTH INTEGRATING NEPHELOMETER
A. Description
A detailed description of the instrument was published by Ahlquist
and Charlson (1969). Briefly, it measures the extinction coefficient due
to light scattering, b , at four separate wavelengths; 360 + 15 nm,
s cat
436 + 5nm, 546 + 5nm, and 675 +_ 15nm. FigureV-lis a curve of the relative
-------
V-2
response of the four channels as set up for the 1969 SMOG Experiment.
The instrument includes electronic analog devices which transform the
signal to a logarithmic form, i.e. so that the logarithm of bscat is
recorded rather than b itself. There are two reasons for this.
scat
1) The logarithmic format increases the usable range of the
instrument.
2) The logarithms facilitate further analog computation.
Specifically, the Angstrom exponent, a, defined by
b = cra
scat
for wavelength A and a constant C, can be given as
b
_
01 ~ dlogA
This in turn can be approximated by the finite difference form
^ Alogb
a = - . "..
AlogA
which is rigorously correct if the original power-law holds, and is only
approximate and dependent on the appropriately small magnitude of Alogb
and AlogA. if the power law is not valid.
The instrument produces voltage signals proportional to logb for
scat
each of the four wavelengths, and the differences between them thus provide
three independent values for a. Each of the three values of a applies to
a given spectral range between bands in the instrument.
Air for this (and the other devices) was pumped through the instruments
with a Rotron blower (Model 250 AS) throttled at its exhaust to reduce the
flow rate and pressure drop. The pressure in the instruments ran approximately
7 cm H20 below atmospheric. The temperature of the air was typically 5° to
-------
V -3
10°C above that outside, due to heat dissipation by the large amount of
electronic equipment in the room. Temperature and dew point were therefore
recorded.
B. Calibration
The multiwavelength instrument (and the other nephelometers as well)
were calibrated with particle free gases (filtered through Gelman type E
filters) primarily air and Freon-12 (CC12F ). Figure V-2is a graph of the
light scattering coefficient of these gases as a function of wavelength,
with the points for the specific wavelengths noted. In addition to the
gas calibration, a wire of approximately 0.5 mm diameter coated with MgO
from freshly burned magnesium was used to provide a large scattering
coefficient with a = 0, i.e. with the same scattering at all wavelength.
Thus, a separate calibration for the recorded Angstrom exponent was
possible, with a = 4 for Freon 12 (a Rayleigh scatterer) and a = 0 for
MgO. The only other factor effecting the independent measurement of a
(that is, independent of the absolute accuracy of the log b calibration)
SC3. t
is the quality of the logarithmic function generator. These devices were
checked and were consistently good to within 1% over three or more decades
of input signal. Thus, the value of a is best obtained from the separate
recording of the quantity based on this calibration rather than from
calculations made on measured light scattering coefficients.
C. Accuracy
There are several necessary accuracy statements for the multiwave-
length integrating nephelometer. First is the absolute accuracy of the
instrument independent of strip chart reading errors or the adjustment
of the position of the trace on the strip chart. This quantity is pertinent
as the controlling factor for data going to the University of Minnesota
-------
V-4
data system. The absolute accuracy depends on
1) The Rayleigh scattering of the gases.
2) The angular truncation error.
3) The other "instrument constants" such as the non-ideality of
the cosine source.
These have been evaluated in a preliminary study and amount to +5%, -10%
of b for most cases. This error is constant for a wide range of
scat
atmospheric aerosols.
The next accuracy figure relates to the reproduceability of the reading
for a given scatterer with fixed properties such as a gas, i.e. the
relative accuracy. This figure is controlled by drift and signal/noise ratio,
and is less than + 5% of b in magnitude.
— scat 6
Next is the reading accuracy for b on the logarithmic scale. This
scat
figure is pertinent for data read from the University of Washington strip
chart recorder but not for data on the data system. This accuracy is
determined to be about + 15%, including the error in adjustment of signal
position during Freon calibration. Last is the accuracy of the recorded
and tabulated values for the Angstrom exponents, of which two were recorded
in the experiment. The first value of a was obtained over the spectral
range from 436 to 546 nm, and the second a up between 436 and 675 nm.
(The latter difference was chosen to improve the signal to noise ratio.)
Both were separately calibrated with Freon-12 (a = 4) and MgO (a = 0).
Using the reproduceability of both MgO and Freon-12 signals as a measure
of system accuracy (i.e. assuming the log generators remained stable) an
error of less than + 0.1 (dimensionless) unit of a is obtained. The
reading accuracy for tabulating this quantity is very good because the
scale of 0 to 4 units spans eight inches of chart, so the accuracy of + 0.1
unit is appropriate.
-------
V-5
The Angstrom exponent can also be determined from the digitally recorded
values of log ^>scat on the data system, and depends on the existence of good
calibration information on the magnitude of signals for Freon-12 and MgO
in the data system, or on a post-computed value from the strip chart.
Since a is obtained by the subtraction of voltages which represent
Io8 bscat> the value is independent of the absolute accuracy of the measure-
ment of bgcat for either the strip chart or the data-system case. In the
latter case, the accuracy depends on the calibration of the data system in
a units. A good calibration was obtained and the data is shown in figure V~4«
Calibration checks of the multiwavelength nephelometer were performed
frequently during the experiment. The times and operations accomplished
are listed in Table I for comparison with the data system.
III. BROAD BAND INTEGRATING NEPHELOMETER
A. Description
The broad band device has identical optical geometry to that of the
multiwavelength version, the only difference arising in the wavelength
characteristic. This instrument has been used for a large number of
published experiments, for instance relating aerosol mass concentration
to light scattering (Charlson, Ahlquist and Horvath, 1969). As such, it
is a more or less "standard" instrument against which others can be compared.
The relationship between the outputs of these instruments is a present
subject of study. Particle free air and Freon-12 were again used for
calibrations, and the same sort of accuracy figures result as for the
multiwavelength device. The only question on calibration is the effective
wavelength of the instrument. Wavelength definition is controlled by a
Wratten 2A filter (minus UV) the spectral characteristic of the light
-------
V-6
source and the S-ll characteristic of the photocathode. Since the effective
center of this broad band is determined by the wavelength dependence of
the intensity of scattered light, and since a varies from perhaps 1 to 2
for aerosol and is 4 for calibration gas, some variation exists in the
effective wavelength of the instrument. Preliminary empirical studies
have shown that the shift in the effective wavelength for these different
values of a is small; however, one purpose of the experiment is to more
-4 -1
fully study this problem. The value of 3.6 x 10 m used (in the broad-band
device) for Freon-12 is based on empirical studies which resulted in an
effective wavelength of approximately 500 nm. This value includes a
correction for angular truncation error. The strip chart recorder
-4 -1
sensitivity used for this instrument was 0 to 20 x 10 m full scale.
B. Accuracy
As a result of these considerations, it is possible to conclude that
the overall absolute accuracy of the broad band device is about + 10% for
scattering centered at approximately 500 nm. The relative accuracy or
reproduceability is better than + 5% for most scattering levels encountered.
The reading accuracy on the strip chart is somewhat better than that of
the logarithmic scale of the multiwavelength instrument, and amounts to 1%
of full scale, or approximately 0.2 x 10 m
IV. NARROW BAND, PHOTOPIC FILTER, INTEGRATING NEPHELOMETER (MRI MODEL 1550)
A. Description
The Meteorology Research Incorporated Model 1550 prototype nephelometer,
provided by MRI, was fitted with a Kodak Wratten #106 filter on 19 August
1969 in order to more closely match the instrument to the response of the
human eye. The calibration was performed in a fashion identical to that
of the other two nephelometers. Due to a design flaw in the prototype
-------
V-7
(corrected in production) the angular range over which integration occurs
was somewhat decreased, with a resultant higher negative truncation error.
As a result, the calibration necessarily had to take account of the extra
truncation. The simplest calibration is to utilize the monochromatic scattering
at 546 nm as determined by the multiwavelength device. FigureA?-8 is a graph
of percent full scale deflection for the MRI instrument on both the
University of Washington and the MRI strip chart recorders versus the 546 nm
scattering for several periods early in the experiment (following 19 August).
Since the University of Minnesota data system was operating during the
-4 -1
morning of 20 August the value of 13 x 10 m at 13:15 hours can be used
as a reference point.
The recorder sensitivity is equivalent to 0-30 full scale with zero
light scatter at 2% of full scale. The accuracy of the instrument appears
to be about + 7%.
V. DEW POINT HYGROMETER
A Cambridge Instrument Model 880 thermoelectric dew point hygrometer
was installed with its air sample coming from the air ducts immediately
in front of the nephelometers. The calibration and accuracy figures
provided by the manufacturer suggest a + 1°C absolute accuracy and a
precision of + 0.3°C. However, this instrument began to malfunction on
31 August and data from 0000 31 August to 1200 2 September are suspect.
The instrument was usable from 1200 PDT 2 September until 2000 PDT 3
September when experiment x^as terminated.
VI. DATA SYSTEM INTERFACE
Some, but not all, of the University of Washington data were stored
by the University of Minnesota data system. All the University of Wash-
-------
V-8
ington data were, however , recorded on a Leads & Northern Model W multichannel
strip chart recorder. These data are being read and will eventually be
available on punch cards for the other participants use. The data entered
on the University of Minnesota system were:
Channel of U.M.
Data System Input
19 Light scattering 675 nm
20 Light scattering 546 nm
21 Light scattering 436 nm
22 Light scattering 360 nm
23 MRI nephelometer 550 nm after 19 August
FigureV-4is a graph of the voltage fed to the U.M. data system as a function
of light scattering coefficient for the four monochromatic channels only,
(i.e. channels 19, 20, 21, and 22). The Angstrom exponent, which was
recorded directly on the U.W. strip chart recorder, can be computed from
the difference signals of the log b signals. The difference voltages
S Celt
for a = 0 are zero, and the difference voltages for Rayleigh scatter
(i.e. a = 4) are given in the figure. Figure'V-S gives the voltage for the
MRI Nephelometer as a function of light scattering at 546 nm.
Punch cards with hourly readings (5 minute averages) of all variables
for the whole period will be prepared. In addition, periods of special
interest will be tabulated on a shorter time base. Xerox or ozolid prints
of the strip charts can be prepared if necessary for selected periods. The
variables recorded on the strip chart recorder at a speed of two inches per
hour were:
Channel of Strip
Chart Recorder Variable
bscat' 5^6 nm> log scale
------- blank
-------
V -9
3 bscat' 436 nm>
4 bscat' 36° nm>
5 a (Alogb 436 to 546 nm), linear scale
S C3. U
6 a (Alogb 436 to 675 nm), linear
S Co. C
7 bscat' 675 nm' lo§
& bscat7 broad b°nd, linear scale
9 b MRI, 550 nm photopic, linear
S Celt
10 Dew point
11 Instrument temperature
12 blank
Instruments response times of '^1 minute were maintained throughout the
experiment.
VII. PURPOSE OF UNIVERSITY OF WASHINGTON EXPERIMENTS
The discussions of the purpose of this effort is, of course, of primary
importance since it governs the conduct of the work. In most cases, the
discussion of purposes and goals precedes the technical description.
However, the measurements and instrumental capabilities outlined above
dictated the type of experiments which-could be done in a cooperative
program. As a result, the technical capability is presented first and the
purpose later. It should be noted, of course, that these measurement
techniques were developed expressly for the purpose of making measurements
of important atmospheric properties; it is in the cooperative context that
technique precedes purpose.
The boundary condition for this whole effort is that several groups of
atmospheric aerosol and atmospheric chemistry researchers should study the
same air in a true Los Angeles smog situation. Nonetheless, each group
-------
V-10
still had the capability for making its own independent observations
exclusive of the cooperative aspect. As a result, a variety of experiments
were performed. Those participated in by the University of Washington
group fall into three distinct classes:
a) Experiments done by the U. W. workers alone.
b) Experiments done and pre-planned with other workers.
c) Experiments or data evaluation arising as a result of observations
made during the data-taking period.
These can be further described:
a. Experiments with three nephelometers and one hygrometer alone.
1. Correlation of broad-band and narrow band light scattering
coefficient, with wavelength dependence (a) as a parameter.
Los Angeles smog provides a sufficiently variable aerosol for
such a study. Preliminary data indicate a high correlation
without stratification with respect to a. Further data
reduction should show how important a is to the measurement
problem.
2. Relation of monochromatic and broad-band light scattering
coefficient to humidity. Previous experiments with maritime
(sea-salt) aerosol have shown effects only above 60 or 70%
Relative humidity. Preliminary L.A. data indicate an occasional
strong dependence at lower values of R.H.
3. Relation of light scattering to visibility, utilizing Weather
Bureau data for visual range. No preliminary data analysis was
made in this regard.
b. Experiments relating nephelometer and dew-point hygrometer data to
that from other experimenters.
-------
V- 11
1. Relationship of wavelength dependence of light scattering
(Angstrom exponent, a) to the size distribution (Junge exponent,
3). A simple relationship between these has been suggested
by Van de Hulst and others (Ahlquist and Charlson, 1969) namely
that:
a = 3-2
when a power law size distribution holds. For the more complex
case of a non-power-law distribution, a relationship can be
obtained via Mie scattering calculations (Quenzel, 1969). The
combined experiments of University of Washington and the
University of Minnesota groups should provide an excellent
test of the theories. Preliminary observations suggest that
the simple theory may be correct if a power-law size distribution
exists.
2. Correlation of light scattering coefficients and mass concentra-
tion. It has been suggested by Charlson, Ahlquist, and Horvath
(1968) and by Horvath and Charlson (1969) that there exists a
sufficiently high correlation between these two variables to
permit the use of light scattering as an index of the atmospheric
aerosol concentration. Heretofore, experiments were limited to
gross measurements of broad-band scattering and total mass.
This experiment has a more refined mass determination as a
function of particle size, thereby testing directly the importance
of size distribution. The main collaborators in this connection
would be the University of Washington group, Dale Lundgren and
the University of Minnesota group.
3. Extinction coefficient due to scattering compared to that by NCL.
-------
V-12
Charlson and Ahlquist (1969) suggested that the brown color of
some smogs might be largely due to the wavelength dependent
extinction due to scattering rather than that of NO,,. The
University of Washington data in comparison to the State of
California data (Dr. Mueller et.al.) should help to clarify
this point.
4. Correlation of light scattering with other gaseous pollutants.
These data (University of Washington and State of California)
should shed some light on the causes of visibility reduction.
5. Experiments or data evaluation (interpretations) which developed
as a result of observations made during and following the
experimental period. One possible study is that of the apparent
increase in particle count in the Royco counter correlated
to the nephelometer indication. The full development of this
portion of the experiment can only be discussed after the data
on processed and thoroughly discussed by all participants.
REFERENCES
1. Ahlquist, N. C. and Charlson, R. J. (1969), Atmospheric Environment,
_3, Number 5, September.
2. Charlson, R. J., Ahlquist, N. C., and Horvath, H. , (1968), Atmospheric
Environment, 2_, 455-64.
3. Quenzel, H., (1969), presented orally at the CACR meeting, Heidelberg.
To be published in JGR.
4. Horvath, H. and Charlson, R. J., (1969), J. American Industrial Hygien_e
Assoc., in press.
5. Charlson, R. J., and Ahlquist, N. C., (1969), Atmospheric Environment.
in press.
-------
TABLE V-l
Periods when Multiwavelength Nephelometer was being calibrated,
MgO
14 Aug.
15 Aug.
17 Aug.
18 Aug.
18 Aug.
19 Aug.
20 Aug.
21 Aug.
22 Aug.
27 Aug.
27 Aug.
31 Aug.
2 Sept.
3 Sept.
1215-1315 1415-1515
1745-1815 1630-1730
1325-1345 1245-1305
1145-1330
2115-2245 2100-2115
1500-1630
0900-0915
1725-1745
2145-2200
0800-0815
1915-1935
1445-1450
1900-1935 1700-1730
1825-1850
-------
Chaw. 3
4-36
70/00-
o
j
u>
a
300
£00
700
Figure V-l Relative response of multiwavelength. integrating
nephelometer as a function of wavelength.
-------
UJ
(J
E
UJ
8
2
UJ
1
7
I
S
2
io4
9
a
T
c
4
i
2
s.
S
V
s
•V
\
-,(
^
k
x^
v
\
*
^
s
s
\
^
\
\
^*.
\
X
>
\
k, — ,
s
^
s
\
*,
^v
\
V
S
S
'X
\
\
s
V
\
x.
s
s
X,
\
"v
.
\
'•'x:
\
Vy
\
•N.
\
^
V
\
\
x^
\
\
V
^
\
x,
\
^x.
\
^ ^
^
x^_
s,
^
~x^
yx
\
"X^
^
s
\
X.
\
v
•^
S
s,
"X
"X,
\
X
*x^
^^
*v
s
\
X,,'
s
s
\
s.
X
V
*v
"•x
\
*>
s
'
v
•^
^
\!
^
^
Wi,
ff
^
\
-^
^
s,
^x,
s
s
^
s
^
eJ ^
"^
\
\
v
«x
\
^
*v
^
*v
S
.
v
\
x
v
s,
~^
^
V
\
N
\
\
s
v-
s
x^
\
•x,
\
s
^x.
AIR COLOR DI/J&RAM
V
X
s
S
X,
\^
"x
s^
x>
\
\
V
S
s
"X
X
x^
\
^
x^
\
"\^
n
rn
n
n
n
FT
1 1 KL
T
TO"
N, i i i IM i i i i i i i i i vi i i iixJ i i i i i M
Figure V-2 Light scattering coefficient
as a function of wavelength for the
calibration gases Air, CO, and Freon 12
(filtered) . i
FM"1^ 1 ' 1 T r\ 1 1 1 — t-^i — I — 1 — I — T I'M ' 1" r T 1 T»«. 1 1 1 rl^%
v
s
N
^v
V,
s
k
s,
^
v
s^
S
v,
s
•v
^
\
s
LOG- LOG, ONE CYCLE ON A^CISSA IS
FOUR TIMES THE LENGTH OF A\YCLE
ON THE ORDINATE. RAYLEIGH SCiXjER
HAS SLOPE OF -1 . N.
*-x.
\
s
^
V
s.
\
'v.
\
^
1 —
is
'-s
\
•^x
s
k,
x^
s
'»
s
X
s
x
s
^
•^
^
^
N
S
s
*x
s
N
X
^
^
V
V
^
^
s
s
v
\
\
V
.
V
t
^
^
\l
X,
^
^
V
X
\
s
u,
^
V
^
Xk,
'
X^
4
s
**•
V
\
\
L^.
s
s,
*s
^
^
k^
V
X,
^
'
^
V
^
s
s
\
^
s
s
s
^
%-
*v
s
X,
s
^
x^
^
"X,
S
^
^
^
s,
^
\
s
k^
k
**
^
'
s
^.
k,
^
^
^
s
^
k
s^
v
^
s.
S
S
s
N
*s
V
K,
V
Is
^v,
^
fl
V
^
s
^
'
^
V
v
^
s
fc"
^4
x.
s
v^
^
v
^
•v,
\
•v.
s
s
V
^
s^
v
'
*
\
S
\
^
X
s,
1
^
v
\
x
's
u
s
-
^
\
V
_
K
\
•U_
N
k
^
-
^v
F
S
3
S
v
\
s
t
1
1
s
*^
1
"X
^
k
'
X,
s^
X
^
"X
s^
's,
s
X
.
.
\
.
h
s
s
's,
s
X
y
X
k,
X
^
">
s
v
y
s
X
s
X
s
s
V
s,
•>
^
s
*
s,
s^
s
s
N,
s
v
S
S
X,
s
X
^
X
X
V
x
V
X,
s
ii>
X,
>
s
.
S
s
S
X
'v
k,
*•
K
^*
1
\
s
x
s
N s,
\
^
ta
. ^
.
sv
S
CCI.F,
400
aoo
WAVELENGTH
'too
no
•00
-------
)| scae
U«£W
/OOC?
£5-
Figure V-J Percent full scale deflection from
MRI nephelometer on strip chart recorders versus
bscat at 546 nm from multiwavelength integrating
nephelometer, demonstrating high correlation
between broad band and monochromatic scattering.
-------
PatA
n
CCI2F4
&7S" nm ^5-
5*0-
4-36
365T
'-or
Scatter
345"-436
0,93?
/J75"
•2.O—s
1.0
i
o
I
1,0
^
4.0
\/o/ts =
b
7.0
scttf
s
»cT*m-'
Figure V-4- Signal to U.M. data system from
multiwavelength integrating nephelometer
a function of b (IQ-^m-l).
-------
7oo
---+
i i
U 4
±:i±
IT.
-IT-
Wf
_ | [x
H:
pr
I j
--
H-
400
-~rr-
i i_! i
-—r— — —i—
-fl-H-
444-
:t
•Joo
Ste
' 11:.
bt
±T
;_ I :_,
-it
i
+
Figure V-5 Signal to U.M. data system from MRI
integrating nephelometer (photopic, 550 nm filter)
as a function of b
scat
H+-
hfti-
3
O L-^-
yir
f~Tf
12.
18
-------
DESCRIPTION OF CHEMICAL DETERMINATIONS
IN THE LOS ANGELES SMOG AEROSOL STUDY
(Reporting Phase I)
AIHL REPORT NO. 78
Parts prepared by:
P.K. Mueller and Y. Tokiwa
Chief and Associate Public Health Chemist
February 1970
-------
VI-I
DESCRIPTION OF CHEMICAL DETERMINATIONS
IN THE LOS ANGELES SMOG AEROSOL STUDY
I. Purpose
This paper is a portion of a collaborative study conducted to determine
the physical and chemical properties of Los Angeles smog aerosol. As a
part of this study, the levels of oxidant, ozone, N09 and PAN were continuously
monitored with various gas analyzers. Particles were collected for chemical
analysis and for electron microscopy. Visibility was measured by a nephelo-
meter. Wind speed and direction were also monitored continuously.
The purpose of this part in the study is to provide information on gas
concentrations in the aerosol sample under study and to provide auxiliary
information on particles in the form of mass and visibility.
The second purpose of this part in the study is to determine and compare
the response of four (4) different oxidant analyzers when sampling smoggy
atmospheres. Because of unresolved uncertainties in performance of these
analyzers, the results of this comparison will enable making the best
possible estimate of the actual ozone concentration.
The reports of this study are to be prepared in phases. The results
and interpretation of the data will be subjects of other reports. There are
several applications of this data. One already well under way in collaboration
with Dr. Charlson's group concerns the interaction of particles and NO
as factors causing visibility reduction.
II. General Description
A. Location and Period of Study
The Los Angeles smog aerosol was sampled at the California Institute
of Technology in Pasadena, California. All gas analyzers and equipment were
located in a large air conditioned laboratory in the basement of the Keck
Environmental Sciences building. The continuous gas analyzers were operated
in conjunction with the study from August 19 to September 19.
In order to gain additional data concerning the comparability in
performance between the various oxidant analyzers, this portion of the study
was extended until the end of October. In addition to the analyzers listed
on Table VI-I, a two-stage particle sampler for collecting size segregated
samples on glass fiber filters was added to the study. The pollutants
monitored, analyzer reagent and detector as well as the major interferences
for each analyzer are summarized in Table VI-I.
In addition t-o the oxidant analyzers, the known interferences NO?,
PAN and S02 were also continuously monitored. In addition, integrated
samples for carbon monoxide and C]_ through Cj hydrocarbons were collected
intermittantly during the study period. Using cascade impactors, cyclones and
filters, size segregated and total particulate samples were collected on
filters for chemical analysis. Continuous visibility readings by nephelometry
were alsotaken. Single particles collected with the U of M electrostatic
precipitator were studied by electron microscopy. Information on wind speed
and direction were obtained continuously during the entire period.
-------
VI-2
B. Sampling System
All analyzers were located in a large air conditioned laboratory
in the basement of the Keck Environmental Sciences Building. The air sample
was transported from an inlet 6.7 m above the roof down through a vertical
20.5 m by 7 m internal diameter PVC pipe to the sample distribution system
as shown in Figure VI-VI. The analyzers were conncected to a manifold
constructed from sections of glass tubing butt-joined with Tygon sleeves.
The analyzers were connected to the line with 1/4 inch O.D. Teflon tubing.
Standard 12/5 ground glass ball joint connections were used at each part to
facilitate making and breaking connections. Flow velocities and tubing sizes
for each instrument were chosen so losses of the gas being measured were
minimal.
Actual losses were determined by simultaneous manual samples of the
atmosphere on the roof for oxidant and N02 in the vicinity of the sample
inlet line and from the NOo analyzer port in the sampling manifold in the
laboratory. The results are shown in Table VI-III. The data indicate the
average less than 3% which is within the expected analytical error.
C. Analyzer Operation and Calibration
All analyzers were prepared and placed in operation as
designated by the manufacturer.
All analyzers except the chemiluminescent unit were calibrated prior
to shipment to Pasadena. The calibrations of all analyzers were verified
upon installation and periodically thereafter. Dynamic calibrations were
performed with mixtures of the pollutant gas in filtered air. The gas was
measured simultaneously by the analyzer and by a standard referee procedure.
(AIHL Recommended Methods #1A, 2, and 3.) A tabulation of dates and changes
in response is to be made available later.
D. Performance Factors
Analyzer performance and data validity is a function of several
factors defined as follows:
1. Calibration: The process of determining the output of a
measuring instrument by comparing it to several values of
a primary, secondary or working standard input.
2. a. Minimum detectable change: the smallest change of input
concentration which can be detected for outputs at mid-
and full scale. The magnitude of change may be different
for positive and negative displacements.
b. Minimum detectable sensitivity: the smallest amount of input
concentration which can be detected as the concentration
approaches zero.
3. Response times: The output of most continuous air analyzers
is not instantaneous. Between the input sample and output
data is a definite time delay. The delay is defined by two
terms as:
a. Lag time: The time interval from a change in the input
concentration to a readable change in recorded output.
b. Readibility: Half of the smallest graduated interval on
the chart readout.
-------
VI-3
5. Reproducibility: The variation in response with repeated
inputs of the same concentration.
6. Accuracy: is the degree of agreement between a measured and
the true value which is known or assumed; usually expressed
as +$ of full scale.
7- Precision: is the degree of agreement of repeated measurements
of the same concentration and is expressed as the average deviation
of the single results from the mean.
Data on items 2, 3, and 4 are given in Table VI-II.
III. Measurement of Oxidant
A. General Principles
Continuous analysis of oxidant is usually performed by scrubbing
sample air with an iodide absorbing solution (KI or Nal). Oxidant reacts
with the iodide salt to produce iodine (I2) and triiodide (13"). Their
concentration is measured by a colorimeter, amperometric cell or an electro-
chemical cell and recorded on a strip chart recorder as ppm ozone.
The iodide reagent responds to N02, peroxyacetylnitrate (PAN) and sulfur
dioxide (802) as well as to ozone. The degree of response may also depend
upon contactor efficiency and detection principle. In colorimetric instruments
2,10 and 20$ KI solutions buffered at pH 6.8 to 7.0 are or have been used by
various California monitoring agencies. Laboratory data with NOp and ozone
indicate the approximate response levels are as follows: for 2% KI, 6% of
the N025 for 10$ KI, about 20$ fo the N02 and for 20$ KI, 30 to 40$ of the N02-
Amperometric analyzers which use 2$ KI register about 6$ of the N0£ as ozone
whereas one using 2.5$ Nal registers about 20$. SC>2> on the other hand,
reduces the oxidant readings by 100$ of the S02 present.
More recently a chemiluminescent detector specific for ozone has been
developed. Sample air is aspirated in the dark across the surface of a
chemiluminescent substance (rhodamine B adsorbed on silica gel) which is
scanned by a photomultiplier tube. N02, S02 and PAN do not interfere.
The foregoing facts in addition to the performance factors common to
most instruments imply that data gathered by analyzers with different reagent
formulations and principles of detection may not be comparable. In order to
assess the magnitude of these differences in response, four oxidant analyzers
were used. Data from the N02, S02 and PAN analyzers are to be used to correct
for interferences. Descriptions of a colorimetric, two types of amperometric
and one chemiluminscent analyzer follows.
B. Beckman 77 Oxidant Analyzer (colorimetric)
As shown in Figure VI-I, sample air is drawn by an air pump through
the air flowmeter and enters the contactor where oxidants react with the
absorbing reagent. The airflow is counter-current to the flow of reagent.
The effluent air exits from the upper end of the contactor through an air
control valve to the pump.
-------
VI-4
The absorbing solution metered by the solution pump enters the top of
the contact column and flows in a thin film down the internal helix. The
reagent reacts with the sample air and flows into the sample optical cell
of the photometer. The iodine is measured at 365 nm wavelength and indicated
on a strip chart recorder. The signal is logarithmic with respect to
concentration. The reagent then passes into the reservoir. The reacted
reagent is pumped from the reservoir through the carbon column by the solution
metering pump. The carbon column contains activated charcoal which removes
the free iodine to regenerate the reagent. The regenerated solution again
enters the top of the contactor to react with the sample air.
Beckman oxidant analyzers are available in two models — a portable unit
for field use and a larger, stationary unit for long term monitoring. The
absorber column design is different for each type. The absorber on the
portable unit is a 2 turn 4 inch diameter coil of 7 nan glass tubing mounted
with the axis vertical. On the stationary models a vertici 14 inch long
section of 7 mm I.D. glass tube with a glass helix insert is used. The use
of the portable type analyzer is limited; most oxidant data in California
has been collected by analyzers using the vertical tube scrubber. For this
reason, the Beckman 77 portable was modified to accept a vertical tube
scrubber (9 inches long) in place of the two-turn helical absorber.
C. Mast Ozone Meter (amperometric)
The Mast amperometric oxidant meter shown in Figure VI-II consists
of a solution pump, a contactor-sensor composed of a plastic electrode
support (about 75$ of its length is wound with many turns of a fine wire
serving as the cathode and a single turn wire is the anode), a solution
reservoir and a sample air pump. The absorbing reagent is metered by the
solution pumped from the reagent reservoir in a fine film down over the
electrode and deposited in the waste reservoir. The sample air enters at
the upper end of the sensor and flows through the annulus past the cathode
which is covered with the thin layer of the scrubbing solution. The air and
liquid are separated at the lower end.
About 0.25 volts applied to the electrodes generates a layer of
hydrogen gas which polarizes the cathode. Iodine produced by the reaction
between oxidant or ozone and KI immediately reacts with the Ho to depolarize
the cathode. Removal of the hydrogen allows the current to flow until
polarization is reestablished. The current is proportional to concentration
of ozone. The reaction at the electrodes gives an electron yield which is less
than 100$ of theoretical and must, therefore, be calibrated with a known
ozone source.
D. Atlas Ozone/Sulfur Dioxide Analyzer (amperometric)
The Atlas unit essentially uses the same amperometric principle
as the Mast to measure oxidant as well as S02- The reagent formulation,
however, is different as shown in Table VI-I. In the oxidant or ozone
mode, a set of polarized platinum wire electrodes measure the amount of
iodine produced. In the S02 mode, a second set of electrodes generate a
constant level of triiodide from the reagent by electrolysis. Reaction with
S02 decreases the level of triiodide which is detected by the sensing
electrodes. In both systems, the generator is used for an internal
standardization of the signal transducing part of the analyzer.
-------
VI-5
_ As shown in Figure VI-III, the absorbing solution in the reagent reser-
voir is metered by a peristaltic solution pump through a nylon wool filter
which absorbs any residual iodine in the reagent to a shallow well at the
base of the absorber or contactor (made from a one liter balloon flask).
Sample air is admitted through a 0.5 to 1.0 mm orifice submerged under the
solution. The solution is picked up and sprayed against the absorber walls
by the jet to condense on the walls and flow down to the entrance of the
sensor electrode in the arm of a glass U-tube. The reagent flow, controlled
by a capillary sect ion, emulate past the sensor electrode across the bottom
of the "U" and up past the generator electrode back to the solution inlet
and sample spray jet. The excess spent reagent and air are removed through
a standpipe adjusted to maintain a constant head above the sensor electrode
and insures constant solution flow past the electrodes. The spent reagent
is returned to the reagent reservoir.
Approximately 25 mV is used to polarize the sensor electrode and
currents between 20 and 300 Ma. is used to generate the constant iodine level
for standardization and in the S02 mode.
Depending on the monitoring mode (03 or 802) both ozone and S02
interfere in this method. The sample air is passed through scrubbers to
remove the interference. In the ozone mode, S02 is removed with boiling
chips coated with a mixture of chromium trioxide (CrO^) and phosphoric acid.
This, however, is equally effective in converting nitric oxide (NO) in
the sample to N02- In the S02 mode, ozone and other oxidants are removed
with column filled with ferrous sulfate (FeSO/J crystals. The 2.5$ Nal
absorbing reagent used by the Atlas is not buffered. According to the
manufacturer, the sodium hydroxide produced from the electrolysis and
oxidation by ozone, reacts with the carbon dioxide in the air to form
sodium bicarbonate and adequately buffers the solution. However, for use
in high W02 atmospheres, buffering at pH 5-5 to 6 with disodium citrate is
recommended.
Moreover, the need for buffering was borne out in our investigations
to determine the Atlas analyzer response to N02- Drifting response
indicated continual changes in pH. Further; when iodine generated with
oeone from 2% KI solution containing phosphate buffers were titrated with an
electrometrlc endpoint detector (Mast titrator) the apparent endpoint varied
with buffer concentration. Discussions with Dr. Ferdinand Schulze indicated
the effect of acidity and buffer concentration on the response of the
citrate buffered system due to N02 was largely unknown. At our suggestion
and with his concurrence, the absorbing solution was buffered at pH 6.8
with 0.1 M sodium and potassium buffer.
E. Research Triangle Institute Ozone Monitor(chemiluminescent)
The sample air is aspirated in the dark across the surface^of a
chemiluminescent or fluorescent substance (rhodamine E adsorbed on silica
gel) which is scanned by a photomultiplier tube. The chemiluminescent reaction
between ozone and rhodamine B is said to be specific, the response to N02,
S02 and PAN being smaller than the response to ozone by a factor of at
least 5000. The unit is equipped with a built-in stable ozone source for
-------
VI-6
dynamic calibration since the reaction between ozone and the chemiluminescent
material is not stoichiometric. The ozone source is calibrated occasionally
by a referee procedure. The chemiluminescent substance is also subject
to temporary loss in sensitivity unless exposed periodically to ozone. To
maintain sensitivity the unit may be programmed to check its calibration
with ozone at four minutes, six hour and 12 tour intervals with the calibrated
ozone source.
IV. Measurement of Other Gases
A. Nitrogen Dioxide
Analysis of NC>2 is most commonly performed by scrubbing sample air
with an azo dye forming reagent. Sulfanilic acid reacts with N02 to form
a diazotized sulfanilic acid. This is then further reacted with a napthylamine
derivative to produce a pink dye which is measured photometrically at 550 nm
and displayed on a strip chart recorder.
Atlas M02 Analyzer (colorimetric): The Atlas analyzer reagent contains
1.5/S 2-aminobenzenedisulfonic acid (ABDS) as the diazotizing component and
0.1$ N(2-napthyl) ethylenediamine dihydrochloride (NEDA) as the coupling agent.
For longer reagent life the ABDS and NEDA solutions are stored in separate
reservoirs. The ABDS solution contains, in addition, 15% ethylene glycol
and 0.05$ sodium benzoate or sodium carbonate as a preservative. Figure VI-IV
shows the ABDS solution is first drawn by the main solution pump through a
bed of activated charcoal to remove any existing color and coupling agent
(NEDA) to a tee. The flow of ABDS is joined with a much smaller flow (about
1/100) of NEDA by means of a second solution pump. The main pump then passes
the combined solution through the photometer reference cell to a tee at the
top of the spiral contactor to join and mix with the sample air as they
descend together down the absorber. Reaction of the solution with N0£
forms dye which then passes through the photometer reference cell and then
returns to the ABDS reservoir for reuse. Only the coupling agent must be
replenished; the 300 ml supply is sufficient for at least a weeks opeation.
B. PAN
Panalyzer*-, Wilkins Instrument (gas chromatographic) : Sample air, petioduaHy
collected in a two ml sample loop, is flushed through a 1/8 inch diameter by
nine inch long Teflon chromatographic column, with nitrogen at 20 to 40 ml/min.
The column is packed with 5$ carbowax on 60 to 70 mesh chromosorb G
treated with DMCS and operated at 25 °C. The emerging
constituents are detected by electron capture and the
signal displayed on a strip chart recorder. The unit was programmed to
sample every fifteen minutes.
A stainless steel hexaport valve with an external two ml stainless
steel sample loop was used instead of the glass sample valve described by
Darley et al.*-"- The rate of flow of dry nitrogen carrier gas was increased
from 25 mis per minute to 40 mis per minute. With this system there was
no detectable decomposition of PAN.
-"-Automatic Chromatographic Measurement of PAN. O.C. Taylor, E.R. Stephens
and E.A. Cardiff. Statewide Air Pollution Research Center U.C. Riverside,
California 92502. Presented at the June 1968 meeting of the Air Pollution
Control Association, St. Paul, Minn.
*-*Darley, E.F., K.A. Kettner, and E.R. Stephens, Analysis of peroxyacetyl nitrates
by gas chromatography with electron capture detection. Anal. Chem. 35(4):
589-591 (1963)
-------
Vl-7
The automatic sampling mechanism attached to the outside of the chromato-
graph case, Figure VI-VII, consisted of: 1) a timing mechanism; 2) sample
valve; 3) solenoid activator; and 4) time delay relay. The timing mechanism
consisted of a microswitch, a cam to operate the micro switch, and a standard
Synchrony RPH motor. The cam was designed to energize the solenoid for 90
seconds during each 15 minute period. The shaft of the hexaport valve was
attached rigidly to the solenoid shaft and positioned so that the air sample
would flow through the two ml sample loop during the time the solenoid was
energized. A coil spring of proper tension installed between the solenoid
and sample valve returned the valve to injection position when the solenoid
was de-energized.
Dry nitrogen carrier gas was flushed continuously through the two ml
sample loop of the hexaport valve for 13.5 minutes of each period when the
solenoid was de-energized. When the solenoid was energized the sample loop
was open to the atmosphere to be sampled and a Neptune pump Model 4-K
sucked a continuous flow of air through the loop for 90 seconds. When the
solenoid was again de-energized, the carrier gas flushed the two ml sample
into the column.
The time delay relay, Figure YI-VI, energized the chart drive motor
on the strip chart recorder at the end of the 90 second sample period when
the solenoid was de-energized. The chart drive motor is allowed to run
for 90 seconds which is ample time to record the chromatogram. The column
is too short to separate peroxypropionyl nitrate (PPN) and peroxybutyryl
nitrate (PEN), although a shoulder on the PAN peak, assumed to be PPN,
appeared occasionally.
PAN was synthesized and purified-"- and stored in 34 liter stainless
steel cylinders for use in the dynamic calibration system for the
chromatograph.*-* The cylinders, pressurized with nitrogen to 100 psig,
contained 500 to 1,000 ppm PAN and were stored at 60°F. An infrared spectro-
photometer with a 10 cm cell was used to determine concentration of PAN in
the cylinders using absorptivities reported by Stephens.-»-**- in the first
dilution of the dynamic system, -^B;-;;- one part PAN from the storage cylinder
was diluted with 100 parts activated charcoal filtered air. A similar 1 to 100
dilution with filtered air in the second step of the dilution system reduced
the concentration of PAN in a constant flow of gas to the ppb range.
Calibration curves were plotted from calculated concentration of PAN and
peak height. Calculated concentration of PAN and peak height showed a
linear relationship in the range from 1 to about 50 ppb; therefore,concentra-
tion within this range could be calculated by multiplying peak height by a
constant. The constant varied from one instrument to another and varied
inversely with changes in standing current of a particular instrument.
^•Stephens, E.R., F.R. Burleson and E.A. Cardiff. The production of pure
peroxyacyl nitrates. J. Air Poll. Control Assoc. 15(3):87-89 (1965).
**Plata, R.L. Calibration and comparison of Coulometric and Flame iomzation
for monitoring PAN in experimental atmospheres. Ninth Conference on
Methods in Air Pollution and Industrial Hygiene Studies, Huntington-
Sheraton Hotel, Pasadena, California (1968).
•>BH;-stephens, E.R. Absorptivities for infrared determination of peroxyacyi
nitrates. Anal. Chem. 36(4): 928-929 (1964)
' See this page for #*• item
-------
VI-S
The automatic system has operated continuously 24 hours a day for 11
months with only brief interruptions to clean the electron capture detector.
Contamination of the detector caused a slow decline in standing current and
a concomitant reduction in sensitivity. It was desirable to calibrate the
instrument about once a week to compensate for the gradual reduction in
sensitivity. When the detector was thoroughly cleaned the original sensitivity
was regained.
Sensitivity of the detector was increased and retention time for PAW
and water was decreased by modification of the procedures described by
Barley.-"- These modifications included: 1) reduction of column lenth from
three feet to nine inches 5 2) increase of nitrogen carrier gas flow from
25 to 40 ml/min-l;and 3) reduction of oven temperature from 35 to 25°C.
Retention time for PAN was reduced from two minutes 10 seconds to 60 seconds.
More frequent sample periods may be feasible with this system, but care must
be taken to avoid interference from the water peak which follows that of PAN.
Back flushing of the column has not been necessary since there has been no
indication of accumulation of contaminants.
C. Carbon Monoxide##
Integrated air samples were collected in aluminized Scotchpak
bags and analyzed with a nondispersive infrared analyzer sensitized to
carbon monoxide. Drying tubes containing Anhydrone and Ascarite connected
between the bag and analyzer remove water and carbon dioxide and a membrane
filter placed in the line following the Ascarite tube removes particles.
The amount of carbon monoxide is ascertained by comparing the instrument reading
obtained on the sample to a calibration curve of instrument reading vs.
carbon monoxide concentration prepared from standardized carbon monoxide
samples.
D. Hydrocarbons (C]_ through
Integrated samples were collected in aluminized Scotchpak bags.
The samples delivered to our Berke^r Laboratory were C]_ through C0
hydrocarbons. They were identified and quantiated down to 0.1 ppb by
concentrating 100 ml of the air sample in a liquid oxygen freeze trap.
The samples were then separated on /3 , /3 ' oxydipropionitrate on activated
alumina gas chromatographic column and detected by flame ionization.
V. Particle Measurements
A. Nephelometer; Meteorology Research Inc.
The integrating nephelometer determines the atmospheric extinction
coefficient due to scatter. Sample air is drawn through a chamber where it is
illuminated by a pulsed flash lamp. The light scattered by particles and
aerosols is detected by a photomultiplier tube looking at an illuminated
volume. The phototube output is averaged and compared with a reference from
another phototube looking at the flash lamp. Further details are to be
given by P.J. Charlson.
*See p. 6 for ## item
-"-"-Recommended Method No. 19, AIHL; Method of Analysis for Carbon Monoxide
in Air
-x-:;-*ASTM Method D2820-69T. Tentative Method of Test for C]_ through C$ Hydro-
carbons in the Atmosphere by Gas Chromatography.
-------
Vl-9
B. Size Selective Particle Sampler (two-stage)*
Size selective samples of particles in the atmosphere were collected
with a single-stage and a two-stage sampler side by side. The single-stage
sampler which collects total suspended particles consists of an open-face
filter holder equipped with a 2.54 cm diameter tared glass fiber filter. In
the two-stage sampler the 2.54 cm filter is preceded by a 1.3 cm diameter
stainless steel cyclone collector. At a sampling rate of 18 liters/minute
the cyclone permits only particles smaller than two microns in diameter and
about one-half of the 3.5 micron particles to enter the second stage. The
mass collected on the single stage filter is used to determine the total
particle loading.
C. Total Suspended Particles
Samples for total suspended particles in the atmosphere were
collected throughout the study on 10 micron pore diameter Teflon filters.
The particles were collected by passing 35 to 101 cubic meters of sample air
through tared 3.5 inch diameter Teflon filters at approximately 0.062 m?
per minute. Details of this phase of the study will be described by
D. Lundgren.
D. Electron Microscopy
This part of the report will include some preliminary quantitative
data on grids exposed for two hours in the ESP on two days. Table VI-IV
is a summary of the particle count on the two samples. The technique
description follows.
1. Preparation of grids
a. Nickel grids 300 mesh were coated with a thin layer of
collodion
b. A thin layer of solicon monoxide was evaporated onto the
collodion surface using a vacuum evaporator.
The SiO surface was used to stabilize the coatings in the
electron beam and to provide a resistant surface to possible
corrosive properties of some of the smog particles e.g.,
H2S04 droplets.
2. Electrostatic precipitator sampling for electron microscopy
Its principle,, performance characteristic and operational
aspects will have to be described by Dr. Ben Liu. The grids
were placed in the ESP as shown below:
0
1
0
4
0
3
0
2
0
5
At the end of the sampling period the grid instructions were
given to treat the grids as follows:
^Mueller, P.K. and M. Imada, AIHL Report No. 72. Two stage Particulate Matter
Sampling Procedure.
-------
VI-10
a. Remove grids and place on double sticky tape mounted on microscope
slide (frosted area on the left). The tape is to be put onto the center of
the slide.
b. Place grids on the bottom edge of the tape.
c. Identify run on upper left ahnd corner of frosted area with pencil.
Date on lower end of frosted area.
d. Grids are to be arranged from left to right, the number of the
grid corresponding to the numbers in the diagram above.
e. Place slide in petri dish.
f. Identify experiment on top of petri dish with waterproof marking pen.
g. The grids in the petri dish can not go to the electron microscopist.
h. After shadowing with Pt-Pd or if not to be shadowed place slide in
petri dish and tape the slide in the bottom dish with, magic mending tape.
i. Tape the edge of the petri dish with masking tape.
j. If a grid is to be removed, take number 2, 3 or 4 and when the
grid is replaced, put it into the equivalent place on the top edge of the
sticky tape.
standard convention
of placing grids
No.
Date
I'd
1
0 0
2 3
f)
k
0 '
5
No.
Date
0
1 n n
123
0 0
4 5
if grid is removed
and replaced
3. Electron Microscopic Observations
After sampling the specimens were characterized and counted in a
Siemens Elmiskop I electron microscope. Grids from two samples were examined:
LN 90790, August 20, 1300-1500 hours
LN 90786 VIII, August 22, 1300-1500 hours
LN 90786 VIII, August 22, control grid (unexposed)
Ten fields each were counted on grids 3 and 5 and nine fields were counted
on grid 1 of sample 90790 (August 20). Ten fields were counted on grid 1
of sample 90786 VIII (August 22). Forty-five fields were examined on the
control grid (sample 90786 VIII clean grid). Electron micrographs were taken
of each field counted. No electron micrographs were taken of the control grid.
The magnification used for counting was 2000X. At this magnification, one
field in the grid is approximately 50m x 50m and corresponds to a hole in
the 300 mesh grid. The area of the field is therefore 2.5 x 10" ' cm2 + ca
The image of this area is projected onto the viewing screen onto a circle
of 9-0 cm diameter. The area counted is not precise because the openings
in the grid are not precise.
-------
VI-11
The photographic plates of each field were taken at 2000X magnification
and they coyer an area of 6.5 cm x 9 cm of the viewing screen (9.0 cm diameter),
These conditions were used for counting. The morphology of the particles
was studied at 20,,000 mag.
The count data was reduced to the form of count/cm2 grid surface.
4. Special Samples
Dr.^Barton E. Dahneke. of CIT obtained particle deposits on EM
grids we furnished placed at various positions in his particle beam apparatus
when sampling smog aerosol. Description and results are subject of a separate
memorandum from P.K. Mueller to S.K. Friedlander dated 2/26/70.
Results
The data show some differences in deposition among grids and some differ-
ence in deposition in parts of the same grid. However,, the count difference
in the two samples (August 20 and August 22) was significant.
Table VI-I shows that there is approximately a four-fold decrease in
particle count, by electron microscopy of ESP samples from 3/20 to 8/22/69.
The samples were taken over a two-hour period from 1300 to 1500, on both days.
This correlates roughly with concurrent gas analyses as follows:
Date N02(ppm) 03(ppm)
8/20 0.2 0.55 @ 1245
0.67 @ 1315
0.43 @ 1530
(triple peaks)
8/22 0.12 0.40 § 1245
decreasing 0.15 @ 1500 (plateau)
in time
A subjective report of physiological effect, namely that of severe
eye irritation, correlated with the smog on 8/20/69 was received by telephone
from staff on site in Pasadena on 8/20/69- Less eye irriation occurred
on 8/22/69.
The electron micrographs which accompany this report show a possible
difference in morphology of the particles obtained on the two days.
On 8/20/69 there were more particles than on 8/22/69. It is possible
that the seeding particles were smaller in 8/20/69 samples (8/20/69 was noted
for an eye burning day) than on 8/22/69 which wasn't so much an eye burner.
The particles obtained on 8/22/69 seem more uniform than those of 8/2U/69.
More work needs to be done on the morphology of the particles hour by hour
and day by day during the smoggy weather. Collection of particles by methods
other than by ESP should be tried. It is possible that collection oi
particles on Nucleopore filters would provide valuable information. Ine
collections are more homogeneous than one would^expect.
-------
VI-12
VI. Chemical Analysis
A, Carbonate and Non-carbonate Carbon in Atmospheric Particles
The carbon content of atmospheric aerosol may be an important
criterion of its origin. The carbon content of particles is in the form of
the element, organic compounds and carbonates. In urban aerosol the non-
carbonate carbon is probably of anthropogenic origin. Carbonates more likely-
result from surface erosion. To obtain information on these matters we have
developed a method for measuring 10 /fg or more of carbon in size-segregated
collections.
The airborne particulate matter is collected on aluminum foil or other
carbon free substrates covering the stages of a cascade impactor. A glass
fiber filter is the final stage. The strips of the foil and filter are placed
into porcelain combustion boats. A boat is inserted into a train designed
to generate carbon dioxide from carbonates in a single sample by first
acidifying with 1% H^PO/,, to form carbon dioxide and water. The carbon dioxide
is swept with an oxygen stream through a silica gel trap for water and then
to a freeze-out trap. The concentrated carbon dioxide is moved to a 1/8
inch x 8 foot gas chromatographic column packed with 100/120 mesh Porapak Q
for separating residual oxygen, water and sulfur dioxide and the carbonate-
carbon is subsequently quantitated by means of a thermal conductivity
detector.
The boat and water trap is then heated to 105°C to remove the water from
the sample and trap. The system is purged with 50 /fl/min. oxygen. The organic
carbon in the sample is then combusted catalytically (cupric oxide) at
900°C to C02 and water. The C02 formed is then quantitated by GLC as
indicated above.
-------
TABLE VI-I
ANALYZER DFSCRIPTIONS
Pollutant Designation Analyzer
0 /S02 Atlas
Detection
Model Principle
0 Mast
x
PAW
Eeckman
RTI
Atlas
PAN
Particle ,MRI
Atlas Elec. Dev. Co. 1120 Ampeiometric
Chicago,, 111.
Particle Total
Hydro- HC
carbon
Carbon CO
Monoxide
Mast Dev. Co.
Davenport^ Iowa
Beckrnan Inst.
Fullerton, Calif.
725-11 Amperometric
77 Coloiirnetric
Reagent
2.5$ Nal
0.1 M Ka2H?04 , p
0.1 MKH2P04 PH °'8
2.0% KI
5.01/, K3r
0.018 M HaH2P04
0.025 M IIaaHP04 pH f-
Contactor/ Major
Collector Interferences
Spray jet Ii02; S02
0.1 M KH2F04
Research Triangle Inst. - Chemiluminescent Rhodamine B
Raleigh^ N.C.
Atlas Elec. Dev. Co. 1300 Colorimetri
Chicago. 111.
Wilkens Inst.
Walnut Creek, Calif.
Meteorology Research Inc
Altadena, Calif.
Gas
on silica gel
SDA
cartowax on
Particle 2-stage AIHL, 12 hr samples
AIHL, 12 hr samples
AIHL; Integrated Sampler
Chroir ntography chrornosorb G
Nephelometry none
Mass
Mass
Gas
Chroir: at ography
none
none
AIKL, Integrated Sampler WDIR
none
Wire helix Iv02,
Vertical tube 7-TO
with
glass helix
Disc
Glass helix
1/8 in dia x 9 in
long Teflon tube
Glass fiber
filters
Teflon filter
Mylar bag
Mylar bag
-------
X
Analyzer
Atlas. 0
' y
Beckman, 0
Mast, 0
RTI, 03
Atlas, N02
Altas, S02
PAN
MR1
HC
CO
2-stage filter
Total filter
Flow Rates
Samp
1/min
3.0
1.7
0.150
0.100
3.25
3-0
0.002^*
280
0.078
0.078
13 M3*
35 to 80 M3^
Reagent
ml/min
U.o
1.63
0.021
2.2
2.0
TABLE 71-II
ANALYZER PERFORMANCE FACTORS
Sensitivity
RpSDor.se
Lag
min
0.5
1
0.5
O.h
3.5
0.5
Times
90%
min
12
10
2
0.5
5-3
15
Recorder
Size
inches
2-1/U
11
2-lA
5
2-lA
2-1/U
5
11
Min
Chart
Div
ppm
0.01
0.01
0.005
0.005
0.01
0.002
0.001
Min
Detectable
Cone
ppm
0.02
0.01
0.01
0.02
0.02
o.ook
Min
Detectable
change
ppm
+ 0.01 to 0.02
+ 10$
+ 0.01
+ 0.01 0.0
+ 0.01 1.0
+ 0.01
4- 0.00k-
ppra
ppin
0.1 ppb
1.0
10 to 20 ug/M3
2
integrated 12 hr samples
"batch system
-------
TABLE VI-
TEST OF SAMPLE LINE LOSSES
(10 minute bubbler samples-"-)
Starting Time Pollutant ppm at Location
Date/Time Lab Roof
8/19 1200 Oxidant 0.2? 0.26
1215 " 0.33 0.34
1420 " 0.33 0.37
1450 " 0.31 0.31
Average 0.31 0.32
8/20 0830 N02 0.06 0.0?
-"-Corresponding analyzer readings in the lab confirm the bubbler sample values.
TABLE VI-IV
PARTICLE COUNT ON ESP PLATE
NUMBER OF PARTICLES/CM2 SURFACE
i~)
Count/Cnr
Sample LN ESP Surface
August 20 90790 37.8x 105
1300 - 1500
grids 1,3,5
average
August 22 90786 10.24 x 10
1300 - 1500 VIII
-------
FIG "21-1
OXIDANT ANALYZER-COLOR!METRIC
CARBON
COLUMN
ReA&ENT
PUMP
AIR
METERING
VALVE
D
ft
o
-o
CONTACT
COLUMN
AIRPUJ
VACUUM
PUMP
AIR
FLOWMETER
REAGENT STORAGE
[10% K.I]
PMOTOM ETE R
"Ol^3 ->\>
'{4 * ->dr
I
1
1
L
3
<
s
\
p
AIR SAMPLE /N
W.
x
-s
•?
^SCRUBBER
FOR sog.
\
RECORDER
FLOW DIAGRAM
C50PH - AIHL
NOV. 1968
-------
OXIEANT ANALYZED - COULOMETRIC
REAGENT
£Hu<0
BATTERY
fr
k
RECORDER
3
0
AIR IN
CATHODE
o
VACUUM
PUMP
TO
REAGENT
WASTE
FLOW DIAGRAM
CSOPH - AIHL.
NOV. 1966
-------
SZ '3
~~~X r-'RL
r... \ JMCrEariG
GAS \ | flU'A?
,. te.,_m~^
i! i'iO-j\\ F*'°"* /^'llii ,81
*1 V CONTROL /? .-•/;,*
Lj^f^'fcJ
Flow diagram of ozone-sul-
fur dioxide analyzer
UH MB RCHOCNT
MITLIT ITtNOPIPE
FT CLEcritooe
PLOW CILL
Gas absorber and elec-
trode assembly
-------
l/l-*/
NITROGEN DIOXIDE ANALVZER-CQinRlMETRif
1
AIR
FLOW-
METER
AIR .OUT
VACUUM '
PUMP
SPIRAL
CONTACT
COLUMN
TO NO ANALYZER
*
(?£F£RENC£ SAMPLE
CELL CELL
NED*.
REAOENT
PUMP
MAIN
REAGENT
PUMP
f
CARBO
COLUMN
REAGENT SUPPLY
CNE.DAI
REAGENT STORA6-E-
CABPSI
-------
CHEMILUMIHCSCINT OZOMK DCTCCTDR.
UVUwnp
-------
h-a
PYROHELIOMETER
CHARLSON AHLQUIST/i
NEPHELOMETER
: AEROSOL INSTRUMENTS
AEROSOL FLOW
CHEMICAL INSTRUMENTS
-RESIDENCE TIME
AEROSOL IN THE
SYSTEM , SEC.
-PIPE DIAMETER,
-DISTANCE FROM
LAST MARKING
SEC.
OF THE
PIPE
CM.
THE
CM.
GAS METERS
Figure
1-2 Schematic of the piping system used to transport
aerosol to the various experiments
-------
1 Timing system
2 Sample valve
3 Solenoid
4 Time delay relay
FigureVI- Automatic sampling system consisting of: 1) standard timing system;
7
2) hexaport valve and 2 ml sample loop; 3) solenoid to activate valve;
and A) time delay relay to control recorder chart drive motor.
------- |