U.S. DEPARTMENT OF COMMERCE
National Technical Information Service
PROCEEDINGS OF A SYMPOSIUM ON FINE PARTICLES
HELD IN MINNEAPOLIS, MINNESOTA ON MAY 28-30, 1975
MINNESOTA UNIVERSITY
PREPARED FOR
INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
t
OCTOBER 1975
-------
KEEP UP TO DATE
Between the time you ordered this report—
which is only one of the hundreds of thou-
sands in the NTIS information collection avail-
able to you—and the time you are reading
this message, several new reports relevant to
your interests probably have entered the col-
lection.
Subscribe to the Weekly Government
Abstracts series that will bring you sum-
maries of new reports as soon as they are
received by NTIS from the originators of the
research. The WGA's are an NTIS weekly
newsletter service covering the most recent
research findings in 25 areas of industrial,
technological, and sociological interest—
invaluable information for executives and
professionals who must keep up to date.
The executive and professional informa-
tion service provided by NTIS in the Weekly
Government Abstracts newsletters will give
you thorough and comprehensive coverage
of government-conducted or sponsored re-
search activities. And you'll get this impor-
tant information within two weeks of the time
it's released by originating agencies.
WGA newsletters are computer produced
and electronically photocomposed to slash
the time gap between the release of a report
and its availability. You can learn about
technical innovations immediately—and use
them in the most meaningful and productive
ways possible for your organization. Please
request NTIS-PR-205/PCW for more infor-
mation.
The weekly newsletter series will keep you
current. But /earn what you have missed in
the past by ordering a computer NTISearch
of all the research reports in your area of
interest, dating as far back as 1964, if you
wish. Please request NTIS-PR-186/PCN for
more information.
WRITE: Managing Editor
5285 Port Royal Road
Springfield, VA 22161
Keep Up To Date With SRIM
SRIM (Selected Research in Microfiche)
provides you with regular, automatic distri-
bution of the complete texts of NTIS research
reports only in the subject areas you select.
SRIM covers almost all Government re-
search reports by subject area and/or the
originating Federal or local government
agency. You may subscribe by any category
or subcategory of our WGA (Weekly Govern-
ment Abstracts) or Government Reports
Announcements and Index categories, or to
the reports issued by a particular agency
such as the Department of Defense, Federal
Energy Administration, or Environmental
Protection Agency. Other options that will
give you greater selectivity are available on
request.
The cost of SRIM service is only 450
domestic (60£ foreign) for each complete
microfiched report. Your SRIM service begins
as soon as your order is received and proc-
essed and you will receive biweekly ship-
ments thereafter. If you wish, your service
will be backdated to furnish you microfiche
of reports issued earlier.
Because of contractual arrangements with
several Special Technology Groups, not all
NTIS reports are distributed in the SRIM
program. You will receive a notice in your
microfiche shipments identifying the excep-
tionally priced reports not available through
SRIM.
A deposit account with NTIS is required
before this service can be initiated. If you
have specific questions concerning this serv-
ice, please call (703) 451-1558, or write NTIS,
attention SRIM Product Manager.
This information product distributed by
U.S. DEPARTMENT OF COMMERCE
National Technical Information Service
5285 Port Royal Road
Springfield, Virginia 22161
-------
TECHNICAL REPORT DATA
(Please read Instruction* on the reverse before completing
1. REPORT NO.
EPA-600/2-75-059
2.
3,
4. TITLE AND SUBTITLE
Proceedings: Symposium on Fine Particles-
Minneapolis, Minnesota, May 1975
6. REPORT OAT6
October 1975
6. PERFORMING ORGANIZATION CODE
PB 249 514k
7. AUTHOH(S)
Benjamin Y.H. Liu, Editor
8. PERFORMING ORGANIZATION REPORT NO,
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Particle Technology Laboratory
University of Minnesota
Minneapolis, Minnesota 55455
10. PROGRAM ELEMENT NO.
1AB012; ROAP 21ADL-034
11. CONTRACT/GRANT NO. '
EPA Grant No. R803556-01
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Proceedings: 5/28-30/75
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
These proceedings contain technical papers presented at the Symposium on Fine
Particles held in Minneapolis, Minnesota, May 28-30, 1975. Also contained are
several papers which were not presented at the Symposium because of a lack of
time. The purpose of the Symposium was to review the state of the art and recent
developments in instrumentation and experimental techniques for aerosol studies.
The focus was on fine particles below about 3.5 micrometers in diameter. Topics
covered include aerosol generation, measurement, sampling, and analysis.
K«produc»d by
NATIONAL TECHNICAL
INFORMATION SERVICE
US D«(Mrtm«nl of Commero
Springfi.ld, VA. 22151
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS C. COSATI FieW/GrOUp
Air Pollution
Aerosols
Instruments
Aerosol Generators
Measurement
Sampling
Analyzing
Air Pollution Control
Stationary Sources
Fine Particles
PRICES
13B
07D
14B
20F
EG TO CHANGE
8. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. Q.F-PA0ES,
20. SECURITY CLASS (This page)
Unclassified
EPA Form 2220-1 (9-73)
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development,
U.S. Environmental Protection Agency, have been grouped into
five series. These five broad categories were established to
facilitate further development and application of environmental
technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in
related fields. The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed
to develop and demonstrate instrumentation, equipment and
methodology to repair or prevent environmental degradation from
point and non-point sources of pollution. This work provides the
new or improved technology required for the control and treatment
of pollution sources to meet environmental quality standards.
EPA REVIEW NOTICE
This report has been reviewed by the U.S. Environmental Protection
Agency, and approved for publication. Approval does not signify that
the contents necessarily reflect the views and policies of the Agency, nor
does mention of trade names or commercial products constitute endorse-
ment or recommendation for use.
-------
EPA-600/2-75-059
PROCEEDINGS
SYMPOSIUM ON FINE PARTICLES
MINNEAPOLIS, MINNESOTA, MAY 1975
Benjamin Y.H. Liu, Editor
University of Minnesota
Particle Technology Laboratory
Minneapolis, Minnesota 55455
Grant No. R-803556-01
ROAPNo. 21ADL-034
Program Element No. 1AB012
EPA Project Officer: D. C. Drehmel
Industrial Environmental Researc Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
October 1975
II
-------
PREFACE
This volume contains the technical papers presented at the Symposium
on Fine Particles held in Minneapolis, Minnesota, May 28 - 30, 1975. In
addition, several papers which were not presented at the Symposium due to
the lack of time are also included in this Symposium volume.
The purpose of the Symposium was to review the state of art and re-
cent developments in instrumentation and experimental techniques for
aerosols studies. The focus was on fine particles below about 3.5 ym in
diameter. The topics covered include aerosol generation, measurement,
sampling and analysis.
The importance of fine particles to air pollution control is well
known: While many of the undesirable effects of particulate air pollutants,
such as those on human health and atmospheric visibility, are due to fine
particles, the control of fine particles is considerably more difficult
than the control of coarse particles. In industrial hygiene, mining safety,
fire detection, and other related areas, fine particles have also been
found to play an important role.
In a rapidly growing field of science where there is a diversity of
workers, there is a need for a periodic review of the subject and a con-
venient source of reference. It is hoped that this Symposium volume will
help in part to meet that need.
As the Editor, and General Chairman of the Symposium, I wish to thank
Dr. Dennis C. Drehmel of the Industrial Environmental Research Laboratory
of the Environmental Protection Agency for helping to bring the Symposium
into being. I am grateful also to Mr. Joe Kroll of the Department of Con-
tinuing Education and Extension of the University of Minnesota for helping
to organize the Symposium. In addition, the help of the Program Committee
— J. H. Abbott (EPA), J. A. Dorsey (EPA), D. C. Drehmel (EPA) and K. T.
Whitby (UM) — and that of the Session Chairmen — A. B. Craig (EPA), D. C.
Drehmel (EPA), J. Nader (EPA), K. Willeke (UM) and D. B. Kittelson (UM) —
are also gratefully acknowledged. I would like to thank the staff and the
students of the Particle Technology Laboratory of the University of Minn-
esota for their help during the Symposium. A special word of thanks goes
to David Y. H. Pui for his help in bringing the manuscripts to a form suit-
able for printing. Last, but not the least, the contributions of the authors
and the financial support of EPA are gratefully acknowledged.
B.Y.H. Liu
iii
-------
CONTENTS
PREFACE iii
PART I. GENERAL
European Aerosol Studies
C. N. Davis 3
Research and Development in Japan on Fine Particles
Measurement and New Control Devices
Koiohi linoya 23
Standardization and Calibration of Aerosol Instruments
Benjamin Y.H. Liu 39
PART II. AEROSOL GENERATION
The Generation of Aerosols of Fine Particles
Otto G. Raabe 57
Generation of Monodisperse Subtnicron Aerosols by Ablation
from Transpiration-cooled Porous Matrices
Cesare V. Boffa, Augusto Mazza, Delfino Maria Rosso Ill
Generation of Monodisperse Aerosols of 67Ga-Labeled Aluminosilicate
and 198Au-Labeled Gold Spheres
George J. Newton, 0. G. Raabe, R. I. Yarwood and G.M. Kanapilly . 129
Aerosol Generation for Industrial Research and Product Testing
Eugene E. Grassel 145
Aerosol Generation Using Fluidized Beds
J. C. Guichard 173
Large Flow Rate Redispersion Aerosol Generator
Fred Moreno, Dale Blann 195
Generation of Inorganic Aerosols for Weather Modification
Experimentation
William G. Finnegan and John W. Carroz 219
Preceding page blank
-------
Generation of Aerosols by Bursting of Single Bubbles
Milos Tomaides and K.T. Whitby 235
An Investigation of an Exploding Wire Aerosol
James E. Wegrzyn 253
Aerosol Particle Formation From Photo-oxidation of
Sulfur Dioxide Vapor in Air
Kanji Takahashi and Mikio Kasahara 275
PART III. AEROSOL SAMPLING
Size-Selective Sampling for Inhalation Hazard Evaluations
Morton Lippmann 287
Dichotomous Virtual Impactors for Large Scale Monitoring of
Airborne Particulate Matter
Billy W. Loo, Joseph M. Jdklevio, and Fred S. Goulding . . . 311
Design, Performance and Applications of Spiral Duct
Aerosol Centrifuges
Werner Stober 351
Problems in Stack Sampling and Measurement
D. B. Harris and W. B. Kuykendal 399
Inertial Impactors: Theory, Design and Use
Virgil A. Marple and Klaus Willeke 411
The Cylindrical Aerosol Centrifuge
Mohammad Abed-Navandij Axel Berner and Othmar Preining ... 447
PART IV. AEROSOL MEASUREMENT AND ANALYSIS
Methods for Determination of Aerosol Properties
Rupreoht Jaenidke 467
Aerosol Mass Measurement Using Piezoelectric Crystal Sensors
Dale Lundgren, Lawrence D. Carter^ Peter S. Daley 485
* Measurements of Aerosol Optical Parameters
* A. P. Waggoner and R. J. Charlson 511
A Review of Atmospheric Particulate Mass Measurement Via the
Beta Attenuation Technique
E. S. Mao-las and R.B. Husar 535
vi
-------
Detection of Ultra-fine Particles by Means of a Continuous
* Flux Condensation Nuclei Counter
J J. Brioord, P.Delattre, G. Madelaine and M.Pourprix 565
<
Electrical Measurement of Aerosols
Kenneth T. Whitby 581
Recent Developments Regarding the Use of a Flame lonization
Detector as an Aerosol Monitor
Laurence L. Altpeter, Jr., J. P. Pitney, L. W. Rust, A. J. Senedhal
and D. L. Overland 625
Contact Electrification Applied to Particulate Matter-Monitoring
Walter John 649
Open Cavity Laser "Active" Scattering Particle Spectrometry
From 0.05 to 5 Microns
Robert G. Knollenberg and Robert Luehr 669
Single Particle Optical Counter: Principle and Application
Klaus Willeke and Benjamin Y.H. Liu 697
Comparison of Impaction, Centrifugal Separation and Electron
Microscopy for Sizing Cigarette Smoke
R. F. Phaleny W. C. Cannon and D. Esparza 731
Extended Electric Mobility Method for Measuring Aerosol
Particle Size and Concentration
Earl 0. Knutson 739
Rapid Measurement of Particulate Size Distribution in the
Atmosphere
R. L. Chuan 763
Identification and Measurement of Particulate Transport Properties
Donald L. Fenton 777
Optical Aerosol Size Spectrometry Below and Above the
Wavelength of Light - A Comparison
J. Gebhart, J. Heyder, C. Roth and W. Stahlhofen 793
vii
-------
PART I
GENERAL
-------
EUROPEAN AEROSOL STUDIES
C. N. Davies
Department of Chemistry, University of Essex
Wivenhoe Park, Colchester, Essex
England
ABSTRACT
Over the last 5 years about 60% of publications on aerosol research
originated in Europe. Experimental work can be traced back to Tyndall's
use, in 1869, of an intense beam of light to show the presence of particles,
a method which is still of basic importance.
Recent work is reviewed and documented, the main subjects being
filtration, particle size measurement and the atmospheric aerosol; there
is also discussion of papers on diffusion, light scattering, sampling,
gas-particle conversion, condensation, coagulation, evaporation, electric
charging, inhalation, fluid mechanics, and so on.
Preceding page blank
-------
EUROPEAN AEROSOL STUDIES
by
C. N. Davies
Department of Chemistry, University of Essex,
Wivenhoe Park, Colchester, Essex, England
The Journal of Aerosol Science displays on the inside back cover its
policy to "publish research work and occasional reviews of recent work of
basic scientific value. The intention is also to encourage the presenta-
tion of results in forms which can be generalized as far as possible,
rather than the restricted presentation of most applied research". In
this context it is now an important medium of publication. The first
five volumes (1970-74) have been analyzed with the following results.
Countries of origin Number of papers
published
U.S.A. 66
Britain 50
Germany 25
France 19
U.S.S.R. 14
Canada 8
Czechoslovakia 7
Sweden 7
Italy 5
Holland, Hungary, India 9
3 each
Austria, Israel, Japan 8
Norway; 2 each
Australia 1
TOTAL 219
-------
Of the papers submitted for the 1974 volume, 65% were accepted,
mostly with changes suggested by referees. From the table it will be
seen that 63% of the published papers originated in Europe and 30% in
the U.S.A.
The cover design of the Journal recalls the first experimenter with
aerosols, John Tyndall, who in 18691 showed how a powerful beam of light
could be used to reveal the presence of minute particles in air. With
this useful tool he demonstrated gas to particle conversion; destruction
of organic particles by radiant heat; formation of a fine aerosol by
heating platinum; filtration by cotton wool; putrefaction in ordinary air,
which did not take place if the air was free of particles; the correct-
ness of the germ theory of disease and the principle of sterilization
used so successfully by Lister; the blue color of light, scattered by
fine particles, which gradually becomes white, this being due to coagula-
tion which he did not understand. It is rather extraordinary that he
wrote about "Heat - a mode of motion" but never associated Brownian
motion with thermal energy.
Experimental aerosol science therefore started in Europe with
the Tyndall beam. Applications date from Aitken's work on conden-
sation nuclei in 1880 and his subsequent particle counters . Aerosols
became an integral part of fundamental science with the cloud-chamber of
C.T.R. Wilson (1895) , Townsend's studies of electricity in gases4 and
on clouds (1897) which led to the first aerosol experiments in the U.S.A.
by Millikan-' and his determination of e.
Theoretical aerosol science could be said to start with the
speculation of Leonardo da Vinci that the blue color of the sky was
due to "the most minute imperceptible particles"^. The real take-off
was Einstein's relationship between diffusion and thermal motion in
1905? followed by Smoluchowski's coagulation theory (1916)° which was
applied to aerosols by Whytlaw-Gray who made the first accurate measure-
ments of particle concentration by number (1922)^; Lomax and he construct-
ed the first thermal precipitator which, as developed by Green and
Watson (1935), is still the best instrument for sampling many kinds of
aerosol.
Improvised filters against dust and smoke long preceded applica-
tions of aerosol science which started with World War I. The Tyndall
beam was used extensively to measure smoke and filtration efficiency.
Interest in the optical properties of the atmosphere and in smoke screens
developed. Aerosols of organic arsenicals appeared, the differentation
of nose and chest symptoms being correctly associated with particle size.
Applications to industrial hygiene also started about this time. The
Kotze konimeter (1916) was introduced into the South African gold mines
to assess the risk of silicosis according to concentration by number of
particles instead of by weight, thus emphasising the small particles
-------
which were shown to be important in silicotic lungs.
The experimental study of aerosols was stimulated in a big way
by military interests and occupational health. Particulate air
pollution, due to the action of sunshine on atmospheric SC^, was
studied by Aitken in 19112, although his idea of the wavelength concern-
ed was perhaps erroneous; the stimulus to aerosol science did not come
until after the Donora incident, in 1948, and the British Clean Air Act
in 1956 which followed the London fog of 1952.
It is a peculiarity of aerosol science that, like biochemistry
sixty years ago, it exists because it is needed. The fine early
advances were the product of the enthusiasm, energy and sometimes the
money of a few individuals; then the practical problems took over and
it has since been difficult to pursue consistent lines of experimental
research over the necessarily long periods. There is very little of the
university teaching and fundamental research to which, in other fields,
applied science looks repeatedly for basic knowledge and for trained
staff. This is reflected in the subject matter of this symposium -
generation, sampling, measurement and analysis - and in the skills of
some of the delegates which have to extend beyond aerosol science to an
ability to invent convincing one-to-three year projects and to fill in
grant application forms, the questions on which having been answered,
there seems little need to carry out the research.
Looking back over the last three years of the Journal of Aerosol
Science, the most popular subjects in Europe were filtration theory
(8 papers), particle size (7) and the atmospheric aerosol (6). The
latter subject is overlapped by 5 papers dealing with the growth of
atmospheric particles with rising atmospheric humidity and its effect
on impactor sampling. Five papers deal with diffusion, 4 with light
scattering and 3 each with inhalation, non-spherical particles, sedi-
mentation, spectrometers and sprays, 2 each with coagulation, diffusio-
phoresis, sampling and sulphur dioxide. There are also papers on
condensation nuclei, deposition from moving air, electric charging,
evaporation, fluid mechanics, ions, impaction and visibility.
It is rather noticeable how the early interests persist -
filtration light scattering, atmospheric particles and the effects
of Brownian motion. Because of the nature of its support, experimental
aerosol science has tended to limited objectives and the early interests
are evidently not yet terminated. There are too few fundamental
experimental studies. Theory, like the cactus, can flourish in an
unfavourable environment so there is too much theory, doubly so because
the intrinsic difficulty of handling aerosols encourages theoretical
speculation. At the same time, the theoretical back-up of difficult
-------
or necessarily incomplete experiments is very important.
12
Work on filtration by Fuchs and his colleagues has been backed
with experiments over a period of 15 years. It was shown that the
Kuwabara flow field gave an accurate description of the velocity round
individual fibers and that the theoretical difficulties were not
important for particles of the most penetrating size. Work with model
filters showed that the fan model, consisting of a series of parallel
sheets of parallel fibers, but with each sheet turned radomly in its
own plane, represented ordinary manufactured filters of the same
volume fraction quite well. Filter resistance has been studied at
reduced pressure up to Kn < 0.5 and the effects of fiber arrangement
and inhomogeneity explored. His latest paper deals with resistance
and efficiency of filters made of polydisperse fibers of 0.2 ym mean
radius. Zebel , following Smutek and Pich , contributed to the theory
of membrane filters. The writer" analyzed the results of failure of
particle adhesion. Gebhart-^ and his colleagues have made an exhaustive
experimental study of glass bead filters for particles between 0.1 and
2 ym diameter. An entirely new means of realizing inertial deposition,
the rotary impaction filter, is due to Soole and colleagues.
Naturally a lot of thought has gone into particle size determina-
tion. There has been doubt about polystyrene latices. Fuchs advises
the use of only highly diluted suspension which avoids the formation of
multiplet particles in the aerosol and of "empties" consisting of
particles of dried, non-volatile stabilizer which are the bane of photo-
electric counters; stabilizer shells on polystyrene particles may increase
the size. Direct size measurements at latex spheres were reported by
Bierhuizen and Perron^? who compared Stokes diameters with geometrical
ones using Perrin's method of averaging from the number in a straight
chain. Bexon and Ogden^" give useful tables for finding the original
spherical droplet size from the measured diameter of a droplet spread to
a spherical cap with given angle of contact on a treated glass surface.
This is an accurate and useful way of dealing with sprays provided that
evaporation can be checked. Other methods of measuring particle size
include a laser spectrometer which measures diameters between 0.1 and
0.7 ym and is claimed to measure the equivalent spherical volume diameter,
irrespective of the shape of the particle; a parallel plate mobility
analyzer, described by Maltoni et al. , with 16000 volts between the
plates, spreads the particles out over 50 cm according to size and charge
and, at A 1/min, covers diameters from 0.3 to 3 ym. Owing to the
alternating rise and and fall with particle size of the intensity of
scattered light in a given direction, ambiguities arise in interpreting
measurements of aerosols made with photoelectric particle counters in
the range 0.5 - 5 ym diameter; particles of different sizes may scatter
the same amount of light. Bakhanova and Ivanchenko^O describe a method
of deriving the particle size distribution from the distribution of
-------
scattered light intensities, which are recorded as a series of pulses
from the particles, using a calibration curve obtained for a series of
homogeneous aerosols. Sutugin and colleagues^ suggest that aerosols
with bimodal size distributions may be formed from condensing vapours.
The reason is that collision efficiency between a vapour molecule and
a cluster of molecules decreases rapidly as the number of molecules
in the cluster decreases. This means that condensation is delayed so
that supersaturation is very high and molecule-with-molecule collisons
result first in homogeneous dimerisation. Larger particles may form
heterogeneously at the same time, depending on the vapour concentration
and whether it is maintained. Each mechanism has its own peak on the
size distribution. A computer simulation showed these ideas to be
plausible and they were verified experimentally with highly dispersed
aerosols (r < .05 ym) of NaCl, Ag and Agl.
21
Storebo has made a series of calculations of condensation in
the fireball of a nuclear explosion using the classical equations for
nucleation, condensation and coagulation. The fireball reaches maxi-
mal size at a temperature of 2000 K and is supposed to contain Fe
vapour which condenses to liquid FeO and freezes at 1641 K. The
radioactivity of a particle is taken to be proportional to its mass.
The initial process is one of homogeneous nucleation but this switches
to vapor condensation upon the primary particles. In the case of
ground bursts there are also particles of contaminent, mainly above
1 ym in radius. The particles in air bursts are too small for
sedimentation under gravity to be appreciable; there is no fall-out.
23
E. Meszfiros has taken filter samples in Budapest and shown
correlations of the sulphate/SO~ ratio with the near ultraviolet of
sunlight and with temperature. However, there was no significant
difference between summer and winter levels of sulphate (- 10 yg/ra^)
so that some mechanism other than gas-phase photochemical conversion
seems necessary; it is suggested that sulphate is formed in winter
on the surface of atmospheric aerosol particles since positive correla-
tion was found with the concentration of aerosol by mass and also by
number of particles exceeding 0.15 ym radius.
24
A. Meszaros carried out a membrane filter study of the air over
the ocean of the southern hemisphere which is likely to be less affected
by human contamination than the northern. The particles consisted main-
ly of ammonium sulphate, sea salt and mixtures of the two; some
sulphuric acid was found. A Gardner condensation nucleus counter was
also used. Aitken particles numbered 300-450/cm , compared with
600/cm3 in the North Atlantic; above 0.03 ym radius there were up to
55/cm^. The peak size was near to 0.1 ym radius. A case for sampling
during summer in Greenland, where there is little exchange of air with
lower latitudes, is made out by Megaw and Flyger^ since long term trends
-------
would become evident. A Danish expedition obtained useful data in
1974 and another is going again this summer. The size distribution
of particles in city air is complex in form - broadly bimodal by mass
but showing numerous irregularities which are presumably due to
individual sources; between 0.1 and 1 urn radius the aerosol is stable.
The writer ° gave a simple technique for fitting distributions of atmos
pheric particles by number or mass with compatible sets of lognormal
components.
Cascade impactors, invented by K. R. May, have been used for 25
years for sampling the atmospheric aerosol. It is interesting
that Jaenicke and Blifford^S now conclude that the cut-off of each
stage is sharper than has hitherto been believed from the results of
experimental calibrations and, in fact, is closer to theory. This
may be caused by errors in calibrations with aerosols of broad size
distribution or by the presence of multiplets in nominally monodisperse
aerosols. A new theory has been presented by Pelassy.^'
Another problem with impactor sampling is the effect of humidity.
shows that in atmospheric relative humidities above 75%
(NaCl crystals liquefy) adhesion of particles to the plates is good
but in drier air there is often a loss of particles. He also studied
the growth of various particles with rising humidity, demonstrating the
very important point that particles of mixed constitution as found
in the atmopshere, increase in weight gradually unlike pure crystals
which grow suddenly at the liquefaction value. Another complication
is the effect on cascade impactor performance of the growth of atmos-
pheric particles. Hinel and Gravenhorst-^ study this theoretically
and show that the cut-off radius can nearly double over the whole
range of humidity. The number of particles per stage is not so much
affected and there is no problem below 70% RH.
Mechanisms of sulphur dioxide oxidation have received a good deal
of attention, both theoretically and in the laboratory. Takahashi
at al. deal with the sequence: - photo-oxidation to sulphur trioxide,
combination with water to give sulphuric acid vapor , nucleation by
the combination of several molecules of sulphuric acid with several
water molecules to form critical embryos, growth of the embryo by
condensation of both vapors , coagulation. The critical size of embryo
is about lOc)?. and contains 5-20 molecules of sulphuric acid. The
nucleation rate is strongly dependent on relative humidity.
Stauffer and colleagues-^ start with the assumption that the
concentration of sulphuric acid which is required to nucleate water
vapor is very low and that, during growth by condensation of the
vapors , the droplets receive one molecule of acid for every ten of
water. Heat of condensation is neglected and the condensation
-------
coefficient is taken to be one. The results are compared with
experimental data obtained by Cox who irradiated SC>2 - N£ mixtures
at 290-400 nm and at 185 nm, the former giving very slow oxidation.
The measured concentration of particles of acid changed much less
with relative humidity than the calculated figures; there was rough
agreement at about 20% R.H., enough to suggest the general plausibility
of the heteromolecular condensation theory.
Work on light scattering continues to be mainly theoretical
but there are two useful experimental papers by Proctor et al. " deal-
ing with the extinction coefficients of ground dust particles, and one
by Seaney on the forwards scattering by limestone and coal dusts.
On the theoretical side Bradshaw^" gives a program and results for
total and angular scattering for various refractive indices and values
of a up to 60; Tonna^" has worked out the correction for finite accept-
ance angle (0.2-1°) up to a = 200 for spheres (m = 1.33). The correc-
tion amounts to 6% on the extinction efficiency at 0.2° and 42% at 1°,
when a= 100, thus underlining the problem of designing apparatus for
particles up to 50 ym diameter.
Diffusion is employed for measuring the size of fine particles.
The writer^ reviewed the solution of the diffusion equations for
flow through tubes, giving criteria for avoiding errors due to short or
curved tubes and for the employment of long tubes for transporting
aerosols without appreciable loss by diffusion or sedimentation.
The mathematical solution becomes difficult to evaluate and compute
for large particles in short tubes at high velocities: Inghain* gives
an empirical formula which works well under these conditions and
is also valid under the opposite conditions. A solution is also
deduced with allowance for slip at the tube wall. Diffusion theory
is applied to the experimental problem of evaluating the results of
flow through diffusion batteries, systems of parallel tubes; the
difficulty then arises of interpreting the penetration when the aerosol
is polydisperse. Maigne et al. propose a new method of finding the
size distribution from diffusion battery data, by measuring penetration
at a series of rates of flow, and present an experimental check against
the electron microscope. Another experimental method of using the
diffusion equation is described by Matteson, Sandlin and Preining^*;
they used commercially available collimated hole structures with holes
15.5 um mean diameter and a length to diameter ratio of 32.7. Experi-
mental results at room temperature agreed with theory but at -16 and
-72°C the rate of diffusion of the particles was much increased. It
is suggested that this might be due to gas molecules adsorbing on the
particles. Many cases of evaporation and condensation are diffusion
controlled; Gallily^ discusses the change of shape of a particle
which these processes may cause, so complicating the mathematical
analogy with electrostatic capacity. He shows that an ellipsoidal
10
-------
particle in still air will maintain its shape with constant axial
ratios.
Two papers are concerned with diffusiophoresis. Yalamov and
Gaidukov^J deal with particles in the slip-flow regime using the
Onsager principle; account is taken of diffusion in the surrounding
gas mixture, thermophoresis and evaporation of the particle.
Hochrainer and Zebel^" made an experimental test of the possible signif-
icance of diffusiophoresis on the deposition of aerosol particles in
the lungs during breathing. A single two channel model was made with
a membrane filter separating the channels which contained aerosols
dispersed in different gas phases. It is concluded that diffusio-
phoresis has little effect due to the counter-flow of oxygen and carbon
dioxide.
Coagulation is very rapid in the early stages of formation of
aerosols. Walter^ has obtained solutions of the integro-differential
equation and shows how sensitive they are to the gas-kinetic correction
of Fuchs to the coagulation constant, which occurs in the kernel of the
equation. The importance of previously existing large particles in
mopping up the newly formed small ones is demonstrated and the dis-
tributions obtained suggest that homogeneous gas reactions are far
more important in forming atmospheric aerosols than ion-molecular
reactions. Commenting on this paper, Storeb^8 points out that the
early, non-gaseous nuclei of radioactive decay or gas phase chemical
reaction in the atmosphere are larger than can be accounted for in
terms of their birth mechanism. He thinks that other gas molecules
are collected on them to create embryos and subsequently evaporated
after coagulation of the embryos. The low coagulation rate due to
the shell adsorbed to nuclei is formulated.
Observations of the mobility of small ions near the ground are
reported by UngethunT* and confirm the existence of discrete mobility
groups with break-points in between. Mobilities up to 11 cm^ V'^sec"^-
were measured. The small ions decay to larger ones by several routes
and the sharpness of the break-points depends on air conditions. It
is considered that no conclusions can yet be drawn about the structure
of the ions in different groups.
New calibrations of two condensation nuceli counters by O'Connor ,
a portable one for use in mines and an automatic one, correlated well
with a Nolan-Pollak instrument using both natural and artificial nuclei
over the range 5000 to 300,000 cnf^.
A good deal of work has been carried out with Stdber spiral
centrifuges. PorstendBrfer-^ obtained size distributions of various
polydisperse aerosols by allowing radom decay products to deposit on
11
-------
the particles, sampling the tagged aerosol with the centrifugal
spectrometer and scanning the deposit for a-activity. An important
factor is the deposition coefficient of the decay atoms upon the
aerosol particles which depends upon their size. Kops and his
colleagues^ have calibrated one of these centrifuges over the range
of aerodynamic diameter from 0.06 to 1.5 um. The aerosols were of
polystyrene and important practical points for users of these centri-
fuges are discussed.
A very much simpler aerosol centrifugal spectrometer was made
by Matteson et al. and calibrated from 0.5 to 1.5 ym diameter, the
agreement with theory, based on air flow and sedimentation, was
reasonable.
Sedimentation under gravity is an important mechanism of aerosol
deposition. Pich-^ developed a comprehensive theory for deposition
on cylinders, with filters particularly in mind, and Heyder-", thinking
of the airways in the human lungs, dealt with settlement upon the walls
of tubes, inclined at any angles, from aerosol flowing through the tubes.
The rotating drum as a container for aerosols is as near as we
are ever likely to come to switching off gravity. Such systems have
been important for studying the viability of bacterial aerosols over
long periods. Frostling^" describes splendid equipment which he
tested on particles up to 3 ym diameter. The advantage over an
ordinary chamber is quantified. He had previously used the rotating
drum for determining the vapor pressure of high boiling organics
dispersed as aerosols
Deposition of particles on surfaces in the open air has been
studied for many years because of radioactive aerosols. Clough^S
described new measurements of the deposition rate in a large wind
tunnel. Particles over a range of diameters from 0.08 to 30 ym were
used with wind speeds up to 6 m/sec. Grass acts as a filter for fine
particles up to 0.5 ym and is also a good collector of coarse particles
which bounce off smooth surfaces.
Sampling in the open air and some industrial situations involves
orifice inertial effect with large particles. A detailed analysis has
been made by Belyaev and Levin , who consider the shape of the nozzle,
effects of departure from isokinetic conditions and the effect of
fluctuating wind velocity. Inertial effect is used in a 35 1/min
sampler by Stevens and Churchill^O to segregate fine particles and
coarse particles. All are collected on a single filter paper disc
but a direct stream, carrying mainly particles above 3 ym diameters,
passes through the periphery of the filter and the rest of the air, with
only fine particles, is directed round sharp bends into the centre of the
disc.
12
-------
Problems concerned with aerodynamic drag of aerosol particles and
filters of ten extend to high values of the Kundsen number, Kn. There
is little experimental data apart from spheres, but Dahneke"^ has in
three papers described procedures for correcting the drag of discs,
cylinders and spheroids for the hydrodynamic slip regime , free molecular
flow, and in between.
Of course, sprays are used a great deal as a source of aerosols
and recent work has involved several varieties. May°^ has published
a full account of the Collison atomizer which has long been a favorite
source of fine particles. Topp ^ obtained some excellent photographs
of the ultrasonic nebulizer which throw a lot of light on how it works.
Philipson"° has illustrated the formation of droplets by the spinning
disc atomizer and presented useful performance data. Dombrowski and
colleaguesb7»68 present some excellent photographs of spray production
and discuss the correct procedure for analysis of photographs for drop
size distribution. They also studied fan spray nozzles and showed
how vibrating the nozzle could result in the production of uniform
droplets.
Low Reynolds number hydrodynamics is represented by a single paper
by Gallily and Mahrer^ who derive expressions for the trajectory of
a sphere approaching a wall and for the slowing down of a sphere pro-
jected towards a wall.
Experiments on human inhalation of aerosols continue to be
important for a variety of applications. The results of early work
were widely scattered and considerable overestimates of alveolar
deposition have been incorporated in safety criteria in health physics
and occupational health. One reason for overestimating alveolar
deposition has been a large theoretical underestimate of the efficiency
of impaction in the airways of the dead space. This error, dating
from 1950, has been explained by Johnston and Muir'O.
A large number of experiments on the human inhalation of aerosols
by Heyder et al.'-*- are in progress. Particle size was measured by
laser spectrometry from 0.2 to 0.8 \im diameter and by higher order
Tyndall spectra above 0.8 \im. The particles of di-2-ethyl hexyl sebacate
showed serious loss by evaporation below 0.2 ym. The results show
little dependence of deposition on particle size between 0.3 and 1.0 urn
diameter with tidal volumes near to 500 cm-* and 15 breaths/min.
72
Fry and Black , using Y -tagged particles from 2 - 10 pm diameter
and collimated crystal counters, studied deposition in the human nose
13
-------
and clearance of deposited particles. They demonstrated the indefinite
holding of particles in the anterior passages which indicates that the
nose can be a critical organ.
An exhaustive analysis, supported by experimental checks, of the
evaporation of saline droplets (radius 1-8 vim-) has been made of
El Golli et al7V Allowance is made for the change in activity with
concentration, as the solvent evaporates.
Horvath has made a number of experimental observations of
visual range in the laboratory, using aqueous suspensions of homo-
geneous latex spheres and other particles between 0.1 and 5 um diameter.
Using white light from above and taking the wavelength as 550 nm,
where eye sensitivity is maximal, he showed that the visibility was
inversely proportional to the extinction efficiency of the particles
and that the aqueous turbid medium simulated the atmopshere satis-
factorily. The writer'^ has studied experimental data on visual
range in fogs and show that the measured ranges are sometimes only
about half the calculated values because the latter were derived from
measured concentrations of water droplets taken with samplers which
could not collect particles below 1 ym diameter. It is shown that
the visual range can be calculated from the sum of the reciprocals
of the ranges due to each size fraction. Since inferences can be
made about particle size from observations of visual range, it is
considered that these should be made routinely when sampling for
atmospheric particles by weight.
CURRENT WORK
Work in progress in Europe, or not yet published can be briefly
mentioned. Interest in the performance of filters at high altitudes
resulted in laboratory experiments being carried out in the U.S. over
a range of pressures; interpretation of these results and their
correlation with theory is being examined. Filters of plastic fibers
seem to perform better than mechanical theory indicates, perhaps because
of electric charge on the fibers. Filtration and analysis of aerosols
containing fibers of asbestos is being studied.
Polishing up diffusion mathematics, particularly with reference
to radon decay products is in progress. Turbulent diffusion in jets
of aerosol is being studied theoretically. Development of the Stflber
centrifuge continues with a piezo electric method of weighing the
deposit at intervals along its length. This instrument is in use for
sorting out the aerodynamic diameters of various types of aggregated
particles; studies with DOP aerosols indicate difficulties due to
coagulation and evaporation as well as loss by deposition in the intake
14
-------
of the centrifuge. The use of two lognormal distributions is suggested,
one for singlet and one for doublet particles each, of course, being
spherical.
K. R. May will shortly be publishing a review of his work on
aerosol irapaction jets which has extended over many years.
New generators are being tried out, one using a welding torch to
evaporate refractory substances and another, developed by Tarroni
and his colleagues at Bologna, using a 5 megaherz oscillator of up
to 12 kilowatts to actuate an argon plasma torch at over 10,000°C
through which the aerosol material is blown. Everything vaporizes
at this temperature and aerosols have been generated down to 5 ran
diameter with Og = 1.3.
Theoretical work on La Mer-Sinclair generators is in progress.
A device for recording short term output fluctuations of powder and
liquid aerosol dispensers is in use; it works by tagging the particles
with a fluorescent substance and impacting them on to a rotating drum.
The wide size range of atmospheric particles makes it necessary
to use simultaneously several different kinds of sampler; hence
there is interest in techniques for combining the measurements to pro-
duce a single size distribution. One of the most recent ideas is an
inversion method which minimizas the deviation between the measured
data and the ordinates of a specified type of size distribution.
The data come from an integrating nephelometer, a condensation nucleus
counter and a Royco counter; this work is being carried out by
J. Heintzenberg at Mainz.
Atmospheric pollution stimulates research on S02> NH3 and other
gas-particle reactions. Botanists are studying the role of vegetation
in producing aerosols as well as adverse effects on plant growth of
lead and salt aerosols. Interest in the atmospheric aerosol is
associated with work on light-scattering methods of size analysis,
including laser holography ( above 1 pm). Interesting new methods of
presenting the results of air pollution surveys are being developed
by M. Benarie of Vert-le-Petit, France. In France, also, comprehensive
surveys of Aitken nuclei of human and natural origin are in progress.
A balloon-carried nucleus counter has been used by KMselau, Lindan,
Harz up to a height of 27 km??.
Aerosols are frequently connected with corrosion. Impingement
of salt containing droplets in boiler superheater tubes is enhanced
by a force of complicated origin (evaporation, local surface tension
variation on the droplet, motion of adjacent liquid and vapor phases,
force on droplet opposite to the temperature gradient) which is being
15
-------
studied by Gardner at the Central Electricity Research Laboratories,
Leatherhead. Corrosion, also, has motivated current studies of
the size distribution and concentration of salt particles over the
North Atlantic which are being conducted from weather ships and
reduced at the Heriot-Watt University, Edinburgh.
Current meteorological work overlaps with air pollution studies
in being concerned with the physical and chemical nature of atmos-
pheric particles, their sources and sinks. There is much work on
water droplet clouds, including the extent to which they are
influenced by radiation from the sky and from the ground. The nature
of freezing nuclei continues to offer a challenge.
More accurate measurements of the deposition of aerosols in the
human respiratory tract are being obtained at the Gesellschaft ftlr
Strahlen- und Umweltforschung, Frankfurt, and at the University of
Essex. Ranges of particle size and breathing pattern are being
covered, nasal deposition is being evaluated and the causes of subject-
to-subject variations are being thought about. At Saarbrllcken, Stauffer
is comparing different mammalian species by a dimensional scaling
argument.
There is much interest in aerosol therapy which is gradually
acquiring the sound scientific basis which was formerly lacking. Drugs
administered as aerosols by pressure pack generators depend for their
effectiveness on particle size. In some cases the Freon propellant
may not fully evaporate with two serious possibilities - the dose of
propellant droplets absorbed through the mucous membranes of the upper
respiratory tract may be dangerous, and the drug will fail to reach
the alveolated regions of the lungs and so be ineffective. Various
drugs are used for enhancing mucociliary clearance and for broncho-
dilation, in the treatment of bronchitis and asthma. The effects can
be gauged by scanning the subjects chest with crystal counters after
depositing in his airways a 5 ym diameter y-active insoluble aerosol.
Some research on aerosols has industrial applications. This
includes the generation of metal particles by flash heating a metal
in a cooler atmosphere of argon. Buckle, at Sheffield University,
has produced unusual shapes of particles, including dendrites, in this
way?°. Graham and Homer (Shell Research, Chester)'^ have worked on
the coagulation of lead aerosols on a millisecond time scale. The
light scattered per particle is proportional to the square of the
particle volume; assuming the size distributions of the coagulating
aerosol to be self-preserving enables the whole scattered light from
all particles to be taken as proportional to the mean particle volume.
The aerosols were produced by shock heating of tetramethyl lead
16
-------
vapor in argon at about 940 K.
Studies of radiation front flames (A. R. Jones, Dept. of Chemical
Engineering, Imperial College, Longon)"^ give information about the
nature of soot particles which are present and their effect on the
emissive properties of the flame. Optical properties of metal oxides
in flames have also been studied*^. Particles in flames are
charged electrically so that they can be manipulated by electric fields
at any stage of growth; this holds out the possibility of improving
combustion efficiency and diminishing pollution. ^
A powerful instrument for measuring rapidly changing distributions
of particle size has been developed by Carabine and Moore.®^ Light
scattered from a laser beam by the aerosol particles is collected at
angles from 8° to 172° in 3.75° steps by a system of multiple fixed
and a single rotating mirror wuich flashes each signal in turn to a
photomultiplier. Angular intensity distributions were calculated for
lognormal aerosols having values of Og between 1.1 and 1.65 and the
measured angular distribution of light intensity was inverted by a
computer program to the representative sum of the weighted log-
normal distributions. The system should operate in the size range
0.1 to 1.0 urn radius and has been used to demonstrate coagulation.
There are obvious problems related to the effects on the scattered
light of aggregates of particles.
In 1972 Hinds and Reist described the measurement of particle
size by Laser Doppler Spectrometry based on the Brownian motion of
the particles for diameters between 0.3 and 3 ym. An alternative
method, based on particles moving across a grid of light fringes at a
constant velocity has been discussed by Fristrom et al. Such
systems have been used for liquid suspensions with some success but
the setting up of the apparatus is very difficult. Hopefully,
developments in the aerosol field cannot be too far away. Such a
tool for studying particle motion in fluid mechanical problems would
have many applications, apart from measuring particle size. Laser
holography, also, has a long way to go, particularly in the concise
handling of the information which can be obtained.
Finally, continued interest is maintained in the evaporation of
aerosol particles. The writer^ described an easy and direct method
of calculating the lifetime of particles in which both the Kelvin
effect and the defect of vapor concentration near the surface were
of importance. It was based on an evaporation coefficient of unity;
in the discussion this was questioned and a brief review of the
literature was given. New measurements are in progress. The
evaporation of particles in mono and heterodisperse aerosols of
infinite extent is also being studied, including the effects of small
17
-------
scale variations of particle concentration due to turbulence.
Clearly, some current projects must have been missed from this
review of research on aerosols which is now going on in Europe. It
is hoped that those who are engaged in them will accept my apologies,
and that readers of this article, both in Europe and elsewhere, will
find some new information about work which is related to their own
interests. Perhaps, too, some valuable personal contacts may be
encouraged.
18
-------
REFERENCES
1. J. Tyndall. Essays on the floating matter of the air.
Longmans Green, London, 1881.
2. J. Aitken. Collected Scientific Papers. Ed. by C. G. Knott.
Cambridge, 1923.
3. C. T. R. Wilson, Nature 52, 144, 1895.
4. J. S. Townsend. Proc. Camb. Philosoph. Soc. 9, 244, 1897.
5. R. A. Millikan. The electron. Chicago, 1917.
6. The notebooks of Leonardo da Vinci. Ed. by E. MacCurdy 1^, 418.
Cape, London, 1948.
7. A. Einstein. Ann. d. Physik. 17, 549, 1905.
8. M. von Smoluchowski. Phys. Zeits. 17_ 557, 583, 1916.
9. R. Whytlaw Gray & H. S. Patterson. Smoke. Arnold, London 1932.
10. H. L. Green & W. H. Watson. Med. Res. council Spec. Rep. Ser.
No. 199, HMSO, London, 1935.
11. N. A. Fuchs, A. A. Kirsch & I. B. Stechkina. Symp. of Faraday
Soc. No. 7, 143, 1973.
12. A. A. Kirsch, I. B. Stechkina & N. A. Fuchs. J. Aerosol Sci. 6_,
119. 1975; 5_, 39, 1974; 4_, 287, 1973.
13. G. Zebel, ib. 5, 473, 1975.
14. J. Gebhart, C. Roth & W. Stahlhofen, ib. 4^ 355, 1973.
15. B. W. Soole, H. C. W. Meyer & K. Pannell. ib. 4_, 59, 1973;
B. W. Soole, ib. 4_, 171, 1973.
16. N. A. Fuchs. ib. 4_, 405, 1973.
17. H. W. J. Bierhuizen & G. A. Perron ib. 6, 19, 1975.
19
-------
18. R. Bexon &. T. L. Ogden. ib. 5, 509, 1974.
19. G. G. Maltoni, et al. ib. 4_, 447, 1973.
20. R. A. Bakhanova & L. V. Ivanchenko,ib. 4., 485, 1973.
21. P. B. Storebo, ib J>, 557, 1974.
22. A. G. Sutugin, A. A. Lushnikov & G. A. Chernyaeva, ib. 4_, 295, 1973.
23. E. Me'sza'ros, ib. 4_, 429, 1973.
24. A. M^sza'ros & K. Vissy. ib. _5, 101, 1974.
25. W. J. Megaw & H. Flyger. ib. 4_, 179, 1973.
26. J. Heyder & J. PorstendBrfer. ib. _5, 387, 1974.
27. P.Pelassyib. _5, 531, 1974.
28. C. N. Davies. ib. _5, 293, 1974.
29. R. Jaenicke & I. H. Blifford. ib. _5, 457, 1974.
30. P. Winkler, ib. 5., 235, 1974.
31. P. Winkler, ib. _4, 373, 1973.
32. G. Ha*nel, & G. Gravenhorst. ib. _5, 47, 1974.
33. K. Takahashi, M. Kasabara & M. Itoh. ib. 6., 45, 1975.
34. D. Stauffer, V. A. Mohnen & C. S. Kiang, ib. 4_, 461, 1973.
35. R. A. Cox. ib. 4_, 473, 1973.
36. T. D. Proctor & G. W. Harris, ib. _5, 81, 91, 1974.
37. R. J. Seaney. ib. 4_, 245, 1973.
38. P. Bradshaw. ib. 6_, 147, 1975.
39. G. Tonna. ib. J5, 579, 1974.
40. C. N. Davies. ib. 4_, 317, 1973.
41. D. B. Ingham. ib. j>, 125, 1975.
20
-------
42. J. P. Maigne1, P. -Y. Turpin, G. Madelaine & J. Bricard.
ib. J>, 339, 1974.
43. I. Gallily, ib. 4>, 457, 1973.
44. M. J. Matteson, I. W. Sandlin, & 0. Preining. ib. 4, 307, 1973.
45. Y. I. Yalamov & M. N. Gaidukov. ib. 4_, 65, 1973.
46. D. Hochrainer & G. Zebel. ib. .5, 525, 1974.
47. W. Walter, ib. 4_, 1, 1973.
48. P. B. Storebtf, ib. 4_, 337, 1973.
49. E. UngethUm. ib. 5>, 25, 1974.
50. D. T. O'Connor, ib. £, 23, 1975.
51. J. PorstendBrfer. ib. 4_, 345, 1973.
52. J. Kops, L. Hermans & J. F. Van de Vate. ib. 5_, 379, 1974.
53. M. J. Matteson, G. F. Boscoe & 0. Preining. ib. .5, 71, 1974.
54. J. Pich. ib. 4_, 217, 1973.
55. J. Heyder, ib. 6, 133, 1975.
56. H. Frostling. ib. 4., 411, 1973.
57. H. Frostling. ib. I, 341, 1970.
58. W. S. Clough, ib. 4_, 227, 1973.
59. S. P. Belyaev & L. M. Levin, ib. _5, 325, 1974.
60. D. C. Stevens & W. L. Churchill, ib. 4_, 85, 1973.
61. M. Smutek & J. Pich. ib. 5_, 17, 1974.
62. B. E. Dahneke. ib. 4_, 139, 147, 163, 1973.
63. D. Hochrainer &. G. HHnel. ib. 6_, 97, 1975.
64. K. R. May. ib. 4_, 235, 1973.
21
-------
65. M. N. Topp. ib. 4_, 17, 1973.
66. K. Philipson. ib. 4_, 51, 1973.
67. C. J. Clarke & N. Dombrowski. ib. 4., 27, 1973.
68. N. Dombrowski & N. D. Neale. ib. _5, 551, 1974.
69. I. Gallily & Mahrer. ib. 4_, 253, 1973.
70. J. R. Johnston & D. C. F. Muir. ib. 4_, 269, 1973.
71. J. Heyder, J. Gebhart, G. Heigwer, C. Roth & W. Stahlhofen.
ib. .4, 191, 1973.
72. F. A. Fry & A. Black, ib. 4_, 113. 1973.
73. C. N. Davies. ib. _5, 487, 1974.
74. S. El Golli, J. Bricard, P. -Y. Turbin & C. Treiner. ib.
.5, 273, 1974.
75. H. Horvath. ib. 6., 73, 1975.
76. C. N. Davies. ib. 6_, 1975.
77. Report of 2nd Annyal Meeting of Ges. fur Aerosol Forschung.
ib. 6 (4) 1975.
78. E. R. Buckle. Symp. Faraday Soc. No. 7, 17, 1973.
79. S. C. Graham & J. B. Homer, ib. 85, 1973.
80. P. A. Tesner, ib. 104, 1973.
81. F. J. Weinberg. ib. 120, 1973.
82. A. R. Jones, J. Phy. D. Appl Physics. 6^, 417, 1973.
7_, 1369, 1974.
83. M. D. Carabine & A. P. Moore. Symp. Faraday. Soc. No. 7, 176, 1973.
84. W. Hinds & P. C. Reist. J. Aerosol Sci. 3_, 501, 515, 1972.
85. R. M. Fristrom, A. R. Jones, M. J. R. Schwar and F. J. Weinberg.
Symp. Fraday Soc. No. 7, 183, 1973.
86. C. N. Davies. Symp. Faraday Soc. No. 7, 34, 1973.
22
-------
RESEARCH AND DEVELOPMENT IN JAPAN ON FINE PARTICLES
MEASUREMENT AND NEW CONTROL DEVICES
Koichi linoya
Kyoto University
Kyoto, Japan
ABSTRACT
Fundamental research on the measurement and sampling of fine dust par-
ticulates has been conducted both analytically and experimentally in
several laboratories of Japanese universities and institutes. The
research efforts include work on anisokinetic sampling errors, particle
settling in a sampling line, light scattering calculations, cascade
impactors, and special analytical methods for the determination of par-
ticle size distributions. Some of the work is introduced in detail
along with the titles of the research papers which have appeared in
Japanese journals. Several new instruments and apparatuses, which have
been developed in Japan for dust generation and for the measurement of
particle size, dust concentration and other particle characteristics,
are also introduced along with the name of their manufacturer, A few
new control devices for fine particulates are also referred to briefly.
23
-------
RESEARCH AND DEVELOPMENT IN JAPAN ON FINE PARTICLE
MEASUREMENT AND NEW CONTROL DEVICES
Koichi linoya
Kyoto University, Kyoto, Japan
INTRODUCTION
This paper introduces some of the recent fundamental and applied work
which has been done in Japan on the development of instruments and new
control techniques for fine dust. Fundamental research has been con-
ducted both analytically and experimentally in Japanese universities
and institutes. A few industrial developments have also taken place in
Japanese industry.
FUNDAMENTAL RESEARCH
Several laboratories in Japanese institutes are devoted to the study of
techniques for the measurement of fine dust particles. This includes,
e.g., anisokinetic sampling error, particle loss in a sampling line,
light-scattering phenomena, cascade impactors, and various analytical
methods for the determination of particle size distribution. A few
examples are given as follows:
Figure 1 shows a typical calculated result of the anisokinetic sampling
error for fine particles, and Figure 2 is the comparison between the
experimental values and theoretical curves for the sampling error. The
inertia theory for fine particle sampling has been confirmed through
comparison with experiments. However, the amount of particle adhesion
in a sampling probe and a sampling tube is quite significant and leads
to erroneous results if neglected.
Yuu and linoya have studied the separation mechanism of a cascade
impactor for several years. The experimental fractional efficiency
curves are in agreement with their analytical solution as shown in
Figure 3, when the velocity distribution caused by the boundary layer
at a nozzle outlet is taken into account in the gas flow model. linoya
and Makino are trying to develop an electric dust-dislodging technique
for fabric filter cleaning. They have found the optimal arrangement of
electrodes, and the optimum frequency and phase of the applied electric-
ity.
24
-------
-. _ Jf-. — JL? I" 1il" I -U -.V*V
~ Pavies Eq. c.- u lp-hO.5/1 ^ p +Q.5
0.5
Velocity ratio -jj(-)
Fig. 1 Calculated results of anisokinetic sampling errors
for particle concentration
—Theoretical curve
R=0.52Cm p=0.05H
°U.=8m/s R=0.52Cm P=0.12(-)
*U.=6.5m/s R=0.42Cm P=0.12(-) A.28lTn
R=O.A2Cm P=0.29(-)
methylene blue +uranine
Velocity ratio
Errors in particle concentration due to anisokinetic
sampling
25
-------
1.0
0.9
0.8
0.7
0.6
T
£0.5
& 0.4
c
Q)
U 0.3
•H
IW
^ 0.2
I 0.1
n
(0
•po\ DC Q085(cm)
i cal. curve
io considering
boundary layer
[exb results
Vuu etal
Ranzetal
cal. curve
based on ,
potential flow -j
0.2 0.3 0.4 Q5 0.6 07 0.8
•(__} Inertia parameter
Fig.
3 Comparison of experimental and calculated target
collection efficiency in an impactor
nozzle head
nozzle inlet
inner
static
hole
middle thimble
part paper
filter
outer
static inner static
hole pressure
outer stati
pressure
Fig. 4 New equilibrium type of dust sampling probe
26
-------
9—22
Takahashi of Kyoto University, has studied the aerosol size distri-
bution and the electrical charge on particles by use of numerical cal-
culations and various experimental methods. Kanagawa23-30 Of Nagoya
University, has analytically studied the light scattering method for
particle size analysis and for the measurement of aerosol concentration.
He recommends a scattering angle of 60 degrees from the incident rays.
Yoshida31~33 Of Osaka Prefectural University, has developed a new tech-
nique for aerosol size analysis based on the measurement of settling
velocities using an ultra-microscope and video system.
34-35
Kitani of the Japan Atomic Energy Research Institute, also has
developed a method for the direct measurement of aerosol concentration
in a fast reactor chamber using a light scattering photometer. He has
also studied the thermal precipitation of aerosol particles.
Ikebe of Nagoya University, has also used a unique method of diffu-
sion response for submicron particle size analysis. Suganuma"-38 of
Tokyo University, has studied the effect of humidity on airborne dust
generation.
39_A3
Tamori of the National Research Institute for Pollution and Re-
sources has studied a new equilibrium type of dust sampling probe which
has an enlarged inner inlet, as shown in Figure 4. Isokinetic sampling
can be achieved by using this type of probe, when the inner static
pressure is controlled so as to be equal to the outer static pressure.
44
Hashimoto of Keio University is using a neutron activation analysis
for the, study of aerosol size.
Masuda of Tokyo University has developed the electric curtain for
dust particles and a new type of electric precipitator.
Fourteen kinds of industrial test dust have been authorized by Japa-
nese Industrial Standard JIS Z 8901. Their size distributions are
fixed as shown in Figure 5. These test dusts are sold through the
Association of Powder Process Industry and Engineering, Japan. (See
Table I). The quoted Japanese research papers reflect only some of
the recent research work in Japan.
INDUSTRIAL DEVELOPMENT
I would like to introduce several recent develpments in the field of
fine particle measurement and new control devices.
Hosokawa Micromeritics Laboratory has manufactured a powder character-
istics tester which is based on Carr's research work. The powder
27
-------
/ I I I I I
//. *>,
<#>
o
•H
4J
O
(0
5
10
20
30
40
50
60
70
80
90
0)
d
95
96
97
98
CO
0)
99.5
99.<
100 = io~b
-------
tester can give flowability and floodability indices for any fine dust.
Figure 6 shows an overall schematic of the device.
Shitnazu Seisakusho Ltd. has developed many kinds of dust measuring
instruments such as: two types of cascade impactor for the size
analysis of stack and ambient dust particles, a dust concentration meter
for the measurement of suspended particle mass in a building and in the
environment, and a smoke meter measuring the opacity of exhaust gas
from a diesel engine. The company is also manufacturing a powder dis-
perser for test purposes, which is shown in Figure 7, and a micro
multicyclone as a dust control device in which the unti cyclone is one
inch in diameter as shown in Figure 8.
Kony Company Ltd. has developed an automatic, continuous, dust concen-
tration meter, Konytest, under a technical agreement with Dr. Prochazka
of West Germany. It is based on the contact electrification principle.
Figure 9 shows a general view of it.
Shibata Chemical Apparatus Manufacturing Company manufactures a light
scattering type of dust concentration meter, Digital Dust Counter, which
has a 45 degree scattering angle. A multishelves type of gravitational
classifier is attached to the meter in order to remove the fraction
above 10 micrometers by environmental regulation.
Aichi Watch Company has also developed an air volume sampler with a
centrifugal classifier for 10 micrometers separation.
Dan Industry Company, under Professor Kanagawa's instruction, has de-
veloped an aerosol size photo-counter based on a 60 degree scattering
angle.
Ishikawajima-Harima Heavy Industry Company has developed a new two-stage
type electrostatic precipitator, which comprises a conventional one-
stage type ESP of fairly small size in its charging zone and a newly
developed channel electrode system called "electrostatic screen" in its
particle collection zone.
Onoda Cement Manufacturing Company has developed a hybrid type electro-
static precipitator consisting of a dry and wet stage combined inside
a single casing as an integral unit and a slurry processing unit coupled
to the casing.
Sumitomo Heavy Machinery Company has manufactured an electrostatic pre-
cipitator mounted directly on the roof of a workshop, as shown in
Figure 10. This precipitator is light in weight because the dust col-
lecting electrodes are made from reinforced plastic or synthetic resin.
A suction blower is not used.
29
-------
DISPERSIBILITY
MEASURINO UNIT
SPATULA ASSEMBLY
ARROW SWITCH
START BUTTON
Fig. 6 Powder tester (Hosokawa)
2SO
Fig. 7 Powder disperser for test purposes (Shimadzu)
30
-------
Fig. 8 Micro nulticyclone (Shimadzu)
31
-------
Signal processing
unit
Sampling line connector
Dust sensor
Recorder.
Control
pannel
Blower
Heating jacket
High-sensitivity
amplifier
Differential pressure
gauge for flow control
Sensor valve
Differential pressure
gauge for calibration
Calibrator
Solenoid valve for
soot blow
Flow control valve
Fig. 9 Kon ytest (Electrification dust concentration meter,
Kony)
electricity duct
insulator
compartment
'washing nozzle
surface
electrode
discharge
'electrode
roof
fl
hopper
rectifier
Fig.10 Natural convection type of electrostatic precipitator
mounted on roof
32
-------
Kimoto Electronics Company has also developed a four-stage annular slit
no7.7.1 f» twne> nf r-staraAa •itimart-ni-_
nozzle type of cascade impactor
Table I
Dusts and Aerosols for Industrial Testing (1974)
(JIS Z 8901)
No Materials mmd (ym) geo. st. dev. a
g
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Silica sand
it
n
Talc
Flyash
Portland cement
Kan to Loam
M
Talc
Flyash
Kan to Loam
Carbon black
Aerosol
n
Mixed dust
195
30
8
1
15
26
30
8
4.2
5.1
2
-
0.3
0.8
(72% No. 8+25% No.
linter)
2.14
4.00
3.63
2.43
2.24
3.65
4.00
3.63
2.95
2.40
2.15
-
1.0
1.8
12+cotton
33
-------
REFERENCES
1 linoya K., S. Yuu, K. Makino and K. Nakano: On Measurement of Par-
ticle Size Distribution by Cascade Impactors - In Case of Setting
the Clearance Ratio Three for Round Nozzle, Kagaku Kogaku 33, 689
(1969).
2 Yuu, S. and K. linoya: On Separation Mechanism of Two Dimensional
Cascade Impactor, Kagaku Kogaku, 33, 1265 (1969).
3 Yuu, S. and K. linoya: Particle Precipitation on Two Dimensional
Nozzle Wall, Kagaku Kogaku, 35^ 1251 (1971).
4 Yuu, S., T. Yukawa and K. linoya: Effect of Gravitation on a Round
Nozzle Cascade Impactor, J. Chem. Eng. Japan, 5_, 285 (1972).
5 Yuu, S., N. Miyake and K. linoya: Effect of the Velocity Boundary
Layer at the Nozzle Outlet on the Target Collection Efficiency of an
Impactor, Kagaku Kogaku Ronbunshu, _!, 115 (1975).
6 linoya, K. and H. Yamanaka: Experiments on Anisokinetic Sampling
Errors for Solid-Liquid Two Phase Flow, Kagaku Kogaku 34, 69 (1970).
7 linoya, K., Z. Tanaka and H. Takai: Particle Size Analysis With a
Gas Centrifuge at Reduced Pressures, Kagaku Kogaku 35, 1041 (1971).
8 linoya, K., K. Makino, S. Toyama and A. Goto: Development of Mist
Size Analyser, Kagaku Kogaku, 37, 858 (1973).
9 Takahashi, K. and S. Iwai: Estimation of Size Distribution of Small
Aerosol Particles by Light Scattering Measurement, J. Colloid Interf.
Sci. 2_3, 113 (1967).
10 Takahashi, K.: Determination of Number Concentration of Polydis-
persed Small Aerosol Particles by Turbidity Measurement, J. Colloid
Interf. Sci. 24_, 159 (1967).
11 Takahashi, K.: Application of Diffusion Tube Method to Character-
ization of Gas-Particle Mixture, Hoken Butsuri (J. Japan Health Phys.
Soc.) 2, 115 (1967)
12 Takahashi, K. and M. Kasahara: A Theoretical Study of the Equilib-
rium Particle Size Distribution of Aerosols, Atmos. Environ. 2_, 441
(1968).
13 Takahashi, K. and M. Kasahara: Numerical Calculation of Light Scat-
34
-------
tering from Polydispersed Small Aerosol Particles, Techn. Kept. Eng.
Res. Inst., Kyoto Univ., No. 143 (1968).
14 Kudo, A. and K. Takahashi: Numerical Calculation for Electrical
Charge on Aerosol Particles, Pt. 1, Techn. Rept. Eng. Res. Inst.,
Kyoto Univ., No. 147 (1969).
15 Takahashi, K.: Changes in Particle Size Distribution of Aerosols
Flowing Through Vessels, Techn. Rept. Eng. Res. Inst., Kyoto Univ.,
No. 149 (1970).
16 Kudo, A., K. Takahashi and T. Kojima: Numerical Calculation for
Electrical Charge on Aerosol Particles, Pt. 2 Estimation Method of
Particle Size Distribution, Techn. Rept. Eng. Res. Inst., Kyoto
Univ., No. 151 (1971).
17 Takahashi, K.: Numerical Verification of Boltsmann's Distribution
for Electrical Charge of Aerosol Particles, J. Colloid Interf. Sci.
35^, 508 (1971).
18 Takahashi, K.: Concentration Change of Aerosols in Closed Vessel,
Nihon-Genshiryoku-Gakkaishi (J. Atomic Energy Soc. Japan), 13, 632
(1971).
19 Kudo, A. and K. Takahashi: A Method Determining Aerosol Particle
Size Distribution Applying Boltzmann's Law, Atmos. Environ. 6^, 543
(1972).
20 Takahashi, K. and T. Tamachi: Monitoring of Atmospheric Aerosol
Particles in Uji-City, Kukiseijo (J. Japan Air Cleaning Assoc.)
1£, 72 (1972).
21 Takahashi, K. and A. Kudo: Electrical Charging of Aerosol Particles
by Bipolar Ions in Flow Type Charging Vessels, J. Aerosol Sci. 4^
209 (1973).
22 Kasahara, M. and K. Takahashi: Numerical Calculation of Aerosol
Particle Concentration Passing Through Diffusion Tube, Techn. Rept.
Inst. Atomic Energy, Kyoto Univ., No. 165 (1974).
23 Kanagawa, A.: Response Calculations for Sideways Light Scattering
Aerosol Counters, Kagaku Kogaku, 34, 521, (1970).
24 Kanagawa, A.: Response Calculations for Forward Light Scattering
Aerosol Particle Counters, ibid., 34_, 991 (1970).
25 Yokochi, A. and A. Kanagawa: Improvement in Sideways Light Scat-
35
-------
tering Particle Counter-Rearrangement of Optical Lens Geometry,
ibid, 34_, 997, (1970).
26 Kanagawa, A.: Size Determination of Aerosol Particles by Owl Spec-
trometer-Measurable Particle Size and Dependency on Refractive
Index, ibid, 36, 91 (1972).
27 Kanagawa, A.: Size Determination of Aerosol Particles by Owl Spec-
trometer-Influence on Dispersion of Aerosol Particles, ibid, 36,
97 (1972).
28 Kanagawa, A.: Size Determination of Aerosol Particles by Owl-Type
Photometer-Effect of Wavelength Ratio, ibid, 36, 647 (1972).
29 Yokochi, A. and A. Kanagawa: Problems in Generation of Monodis-
perse Latex Aerosol, ibid, 36, 676 (1972).
30 Kanagawa, A.: Response Calculation of Light Scattering Photometer
for Measurement of Aerosol Concentration, ibid., 38, 513 (1974).
31 Yoshida, T., Y. Kousaka and K. Okuyama: A New Technique of Particle
Size Analysis of Aerosols and Fine Powders Using an Ultramicro-
scope, I. & E.G. Fund. L4, 47 (1975).
32 Yoshida, T., Y. Kousaka, K. Okuyama and S. Nishio: Effect of Brown-
ian Coagulation and Brownian Diffusion on Gravitational Settling
of Polydisperse Aerosols, J. Chem. Eng. Japan, _8_, 137 (1975).
33 Yoshida, T., Y. Kousaka, S. Inage, and S. Nakai: Pressure Drop and
Collection Efficiency of an Irrigated Bag Filter, I. & E.G. Process
Design and Develop., 14, (2)(1975).
34 Kitani, S., S. Uno and J. Takada: Direct Measurement of Variable
Aerosol Concentration in a Fast Reactor Chamber,Kuki Seijo (J. Japan
Air Cleaning Association) 10, (2) 34 (1972).
35 Nishio, G., S. Kitani and K. Takahashi: Thermophoretic Deposition
of Aerosol Particles in a Heat Exchanger Pipe, I. & B.C. Process
Design Develop., 13, 408 (1974).
36 Ikebe, Y.: Determination of the Size Distribution of Heterogeneous
Submicron Aerosols by Response Matrix Method, Pure Appl. Geophs.,
9_8, 197 (.1972).
37 Mori, Y., A. Suganuma and K. Ishibashi: Influence of Air Humidity
on the Degree of Airborne Dust Agglomeration, J. Res. Assoc. Powder
Tech. Japan, 7, 29 (1970).
36
-------
38 Mori, Y., A. Suganuma, M. Oka and Y. Kaida: Influence of Air Humid-
ity on Airborne Dust Generation Phenomena, Kagaku Kogaku, 34, 198
(1970).
39 Tamori, I. and Tadao Shirasawa: On Collection Efficiency of Dust
Tube Sampler, J. Res. Assoc. Powder Tech. JJ, 8 (1971).
40 Tamori, I., N. Kogure and K. Imagami: Proposal of Cylindrical Fil-
ter Paper Method for Stack Dust Sampling, J. Environ. Pollut. Con-
trol, _?, 435 (1971).
41 Tamori, I., N. Kogure and K. Imagami: Measuring of Stack Dust Con-
tents by Cylindrical Filter Paper Method, Kogai 6^ 289 (1971).
42 Imagami, K., N. Kogure and I. Tamori: Development of New Isokinetic
Sampler by Comparison between Two Kinetic Pressures, Kogai, 8^, 362
(1973).
43 Shirasawa, T., A. Ito, I. Tamori and T. Ohyanagi: Filtration of
Combustion Gas by Coal Packed Bed, Kogai, _5, 10 (1970).
44 Fujimura, M. and Y. Hashimoto: Analysis of Size Distribution Data
of Particulate Matters by Andersen Sampler, Bunseki Kagaku, 24,
36 (1975).
45 Shibuya, A. and S. Masuda: EP-ES Type Electrostatic Precipitator,
Proc. 1975-General Conf. Institute of Electr. Engrs. Japan, No. 920.
46 Masuda, S., S. Ago, T. Itoh and H. Saito: Hybrid-Type Electrostatic
Precipitator, Proc. 1975-General Conf. Inst. Electr. Engrs. Japan,
No. 921.
37
-------
STANDARDIZATION AND CALIBRATION
OF AEROSOL INSTRUMENTS
BENJAMIN Y.H. LIU
Particle Technology Laboratory
Mechanical Engineering Department
University of Minnesota
Minneapolis, Minnesota 55455
ABSTRACT
This paper reviews the aerosol standards development work at the
Particle Technology Laboratory, University of Minnesota. The operating
principle and the performance characteristics of the vibrating orifice
monodisperse aerosol generator and those of the electrical aerosol gen-
erator are described. It is shown that with the use of these generators,
monodisperse aerosols of a known particle size and concentration can be
generated from about 0.01 ym to 50 pm in particle diameter. In addition,
calibration studies on condensation nuclei counters and the diffusion
battery have also been briefly reviewed.
39
Preceding page blank
-------
STANDARDIZATION AND CALIBRATION
OF AEROSOL INSTRUMENTS
BENJAMIN Y.H. LIU
Particle Technology Laboratory
Mechanical Engineering Department
University of Minnesota
Minneapolis, Minnesota 55455
INTRODUCTION
The calibration of aerosol measuring and sampling devices is faci-
litated by the use of monodisperse aerosols. When the size or concen-
tration of the aerosol, or both, are known to a sufficiently high degree
of accuracy, the aerosol can be referred to as an aerosol standard.
During the past few years, the Particle Technology Laboratory, Uni-
versity of Minnesota has devoted much effort to the development of appa-
ratus and procedure for generating monodisperse aerosols of a primary
standard quality. Monodisperse aerosols can now be generated from 0.01
ym to over 50 ym in diameter at concentration levels up to 106 particles
/cc in certain size ranges. In addition, both solid and liquid particles
can be generated and the particle size can be calculated from the opera-
ting conditions of the aerosol generators to a high degree of accuracy
(better than 1%). And the need to measure particle size by the tedious
and often inaccurate microscopic method has largely been eliminated.
In this paper we shall briefly review the aerosol standards develop-
ment work at the Particle Technology Laboratory and the application of
these standards to instrument evaluation and standardization.
GENERATION OF MONODISPERSE AEROSOL STANDARDS -
Vibrating Orifice Monodisperse Aerosol Generator
For generating monodisperse aerosols in the 0.5 to 50 ym diameter
range, the vibrating orifice monodisperse aerosol generator1 has been
developed. The generator, shown schematically in Figure 1, is comprised
of a droplet generation and dispersion system and an aerosol dilution and
transport system. In addition, a radioactive source of Krypton 85 (10 mCi
activity) is placed in the generator to neutralize the particle electro-
static charge incurred during the droplet generation process.
Briefly, the operation of the generator is as follows. The material
to be aerosolized is first dissolved in a suitable solvent, such as water
or alcohol. The solution is then forced by a syringe pump through a small
(5 to 20 ym diameter) orifice, which is vibrated by a piezoelectric ceramic.
40
-------
Monodiwwrso
Aereaoi Oyt
Piezoelectric
Ceramic
Holder
Porous Piste
Dispersion1
Air
Absolute
Filter
Compressed
Air
(15 PSD
Eleotricol
Drgplet Generotor Detail
o""''-.;V3
PytTsp
Figure 1 Schematic diagram of the vibrating orifice i?ei:,odisterse aerosol
generaLor,
Figure 2 Electron micrograph of methylene blue particles (3.7 ym diameter)
produced by the vibrating orifice generator,
41
-------
The liquid jet is broken up by this vibration into uniform droplets, which
are then injected along the center of a turbulent air jet to disperse the
droplets. The dispersed droplets are then mixed with a much larger volume
of clean dry air to evaporate the solvent and to transport the particles
to where they are needed. Table 1 summarized the operating conditions of
the generator and Figures 2 through 5 show some typical particles generat-
ed by the device.
The fact that particles of a known size can be generated by the vi-
brating orifice principle is based on the fact that in the region of uni-
form droplet production, the rate of droplet production is equal to the
frequency of oscillation of the piezoelectric ceramic. Consequently, the
volumetric rate of flow of the liquid, Q.. (cc/sec), through the orifice
is given by
Q! - vd f (i)
where V, (.cc) is the volume of the individual droplets and f (Hz) is the
vibrating frequency. It follows from Equation (1) that the droplet dia-
meter is given by
Dd = ( 6 Qx / TT f )1/3 (2)
Further, if C is the volumetric concentration of the non-volatile solute
in the solution, then following the evaporation of the solvent from the
solution droplets, an aerosol of the diameter, D , is obtained where
d (3)
By the use of Equations (2) and (3), the particle diameter, D , can be
calculated if the quantities Q , f and C are known. p
Assuming a 1% error in the measurement of Q , negligible error in
the measurement of f and a 2% error in the measurement of C, the overall
error in the calculated particle size is about 1%. However, to achieve
such a small error, the following conditions must be met:
(1) There must be no leak in the liquid flow system.
(2) The solvent used must contain a negligible amount of non-vola-
tile impurity.
(3) The solvent must be completely evaporated from the solution
droplets and
(4) the density of particles must be the same as the intrinsic den-
sity of the material.
Condition (1) stated above can be easily met through careful operat-
ion of the droplet generator.
42
-------
g
•H
*J *
n) *
M *
4-1 O
C o
01 ^
u co
C 0)
O rH
O O
CO
t~.
CM
r-
CO
rH
CO
Oi
4J
g
o
4-1
cd
cd
co
a-
(U
M
U)
M
0)
4-J
Q) CO
60
14H T3
•H (3
^ cd
4J U
td
VI O
43 O
•H in
> CN
cu .
X •
•u u
C
U-l M
o
to G
O o)
•iH 4J
w tn
co >.
•H C/l
^ I
0) O
4-J S
cd ^
S-i 01
cd jc
je H
u ^
•3
H
0)
4J
td
•H *
Q 6
a,
o
O
C
oi
cr
01
n)
C
O
53
•H
3
O OJ
O
M -H
a) >4-i
4J -iH
01 )-l
a o
cd
•H
Q
i-H
I
I
o
I
oo
C--1
O
m
cs
CM
o
CM
ao
C
•H
m
CM
O
O
OH
3
O
o
in
O
M
01
R)
V Cfl
•rl C
C -H
01
4-1
n) O o
B 0) C
•H 4-1
X 0)
o C ,c
t-l O 4-1
CX -H
CX 4J C
C O T3
cd ft Q)
td 03
n > td
0) CU f>
O 4J C
0 O
0) CU -H
rH > 4J
^ *"< r~H cd
td O M
4J CO 4-1
co C
3 0) 0)
•r-! ,n CJ
T3 4J C
ta o
p^ a
co
o
3
4J
C 4J
H O
3 «H
Cd 4-)
C a)
•rl M
td o
a)
O H
*
*
*
43
-------
Figure 3 Optical micrograph of '.aouodisperse BOI' pjriieles (905 \im diameter)
produced by tl;e -(Hbraf i.ig orifice generat >r ,
Figure 4 Electrpn wicrcg, apb of sodium chloride particle," of 27.4 ym volume
produced bv the '/Ibrgt Jtip orifice generato1*,
" 44
-------
Figure 5 Electron micrograph of hollow NaCi particles (125 ym total solid
volume) produced by the vibrating orifice generator.
i.o
i .e
£
g£
II
o£
ZQ
8s
o
*E
iS
O MASS METHOD
• COUNT METHOD
_L_L_J«J,_uL™__
5 10
20
40
PARTICLE DIAMETER,
Figure 6 Output aerosol concentration of the vibrating orifice generator
as a function of particle diameter,,
45
-------
To insure that condition (2) is obtained, it is necessary that C,
the solution concentration used, be large compared to the non-volatile
impurity content of the solvent. For instance, for a diameter reduction
OC 20, i.e. D./D = 20, the solution concentration is C = (1/20)3 » 125
x 10~6 or 125 ppm. If the non-volatile impurity concentration, I, of the
solvent is 10 ppm, this impurity will contribute an error of 10/125 = 8% :;,.
to the value of C. This corresponds to a particle diameter error of about • '•" .
2.6%. However, by measuring 1 and calculating the particle diameter as
follows
1/3
Dp = ( C + I )1/J Dp (4)
the impurity contribution to the particle diameter error can be corrected ..
for. One of the simplest methods of measuring I is to spray the solvent
directly through the droplet generator and measuring the residue particle
size with an optical particle counter.
To insure that condition (3) is obtained, the dilution air used must
be dry and sufficient time must be allowed for the solvent to completely
evaporate from the solution droplets. ,-':',. ;"•'
In the case of condition (4), there is generally no difficulty when
the aerosol material is a liquid. The droplets are spherical (see Figure
3) and the particle density is the same as the bulk liquid density. In the
case of solids, experience shows that amorphous solids, such as methylene
blue, will ususally dry from a solution state to form spherical solid
particles (see Figure 2) and the particle density will be the same as that
of the parent material. However, the behavior of crystaline solids, such
as NaCl, is considerably more complex and the solution droplets may cry-
stalize to form particles such as those shown in Figures 4 and 5. Con-
sequently, Equation (4) must be used with caution under such circumstances. /'
The ability of the vibrating orifice aerosol generator to produce
aerosols of a known particle concentration is based on the fact that the
rate of droplet production is equal to the vibrating frequency of the • ;'.;
piezoelectric ceramic. When the droplets are mixed with a total-air flow
of Q (cc/sec), this will give rise to an aerosol having the theoretical
concentration of
Nth - f > Qa (5) ; '
Unfortunately, there are losses in the system and the actual aerosol con-
centration at the generator output is somewhat less than the theoretical
concentration calculated above. Figure 62 shows the measured output of
the generator, expressed as a percentage of the theoretical output, as
a function of the particle size. With the use of Equation (5) and Figure
6 the generator can be used as a secondary standard for aerosol concen-
46
-------
tration in the particle diameter range below 20 vim.
Generation of Monodisperse Aerosols by the Electrical Mobility Classifier
For generating monodisperse aerosols below 0.5 pm, the system shown
in Figure 7 has been developed3. The system is comprised of a Collison
atomizer, a diffusion dryer, a Kr-85 bipolar charger and a mobility class-
ifier. The operation of the system is based on the fact that for particle
diameters smaller than about 0.1 pm an aerosol in charge equilibrium with
bipolar ions will acquire a bipolar charge distribution with most of the
particles either electrically neutral or carrying only ±1 elementary unit
of charge. Because of the small particle size, only a small fraction of
the particles will be multiply charged. Consequently, by passing this
aerosol through a mobility classifier, a monodisperse fraction can be
extracted according to electrical mobility. Since the particles are singly
charged, the electrical mobility is related to particle size as follows,
2
Z =eC/3TrpD 300, cm /volt-sec (6)
where e = 4.8 x 10 esu is the elementary unit of charge, C is the slip
correction, \i (poise) is the viscosity of the gas and D (cm) is the part-
icle diameter.
In the system shown in Figure 7, a polydisperse aerosol is first
produced by spraying a liquid solution with the Collison atomizer and
drying the particles with the diffusion dryer. The aerosol is then passed
through the Kr-85 bipolar charger in which the particles are brought to
an equilibrium state with the bipolar ions produced by the radioactive
source. This charge-equilibrated, polydisperse aerosol is then classified
electrostatically.
In the mobility classifier shown in Figure 7, the aerosol is intro-
duced along with the clean air in two concentric, laminar streams in the
annular gap region between the concentric metal cylinders. Because of
the voltage applied on the inner cylinder, charged particles are deflected
through the clean air stream. Those that have the appropriate electrical
mobility would arrive at the slit near the lower end of the electrode and
be swept out by the small airstream flowing through the slit. The mobility
of these particles can be calculated by the following equation,
Z = [ q + (1/2) (qe - q J ] In (r /r,) / 2 TT V L (7)
p O S3. £. J-
where q (cc/sec) is the clean air flow in the mobility classifier, q
(cc/sec; is the aerosol flow at the inlet, q (cc/sec) is the aerosol flow
through the slit, r, and r (cm) are the inner and outer radii of the an-
nular gap region between the cylindrical electrodes, L is the length of
the precipitating region and V (volt) is the applied voltage on the inner
47
-------
Figure 7 Schematic diagram of apparatus for generating submicron aerosol
standards by the electrical classifier.
Figure 8 Electron micrograph of inonodisperse NaCl particles produced by
the electrical aerosol generator.
48
-------
electrode, the outer electrode being grounded. By the use of Equations
(6) and (7)> the size of the particles can be calculated when the flow
rates, the applied voltage and the dimensions of the device are known.
In the experiment reported by Liu and Pui3, the uncertainty in the cal-
culated particle size is estimated to be ±2%, based on the measurement
accuracies for the various quantities that enter into the calculation.
Figure 8 is an electronmicroscope picture of the particles generated by
this method.
An additional advantage of the electrical classification method of
monodisperse aerosol generation is the fact that the aerosol concentra-
tion can be easily measured to a high degree of accuracy. This can be
accomplished by collecting the aerosol onto an absolute filter in a Faraday
cage and measuring the collected particle charge with an electrometer.
Since the particles are singly charged, the electrometer current, I (amp)
is related to the aerosol concentration N (particles/cc) as follows,
I - q£ e N (8)
-19
where e = 1.61 x 10 coulomb is the elementary unit of charge and q
(cc/sec) is the sampling flow rate into the electrometer current sensor.
More details concerning the method including the procedure for correcting
for the effect of the small percentage of doubly charged particles in the
output aerosol stream is given in reference 3.
INSTRUMENT CALIBRATION AND STANDARDIZATION
The monodisperse aerosol generators described above have been used
to calibrate a variety of aerosol measuring and sampling devices. The
work on optical particle counters has been described by Liu, Berglund and
Agarwal2 and is reviewed elsewhere in this symposium volume (see the paper
on optical particle counters by Willeke and Liu). The work on the elec-
trical aerosol analyzer has been reported by Liu and Pui4 and reviewed by
Whitby in the paper on electrical measurements, also in this volume. The
following review is limited to the calibration study on condensation nuclei
counter and the diffusion battery, two devices that have found widespread
use in aerosol studies.
The condensation nuclei counter is an aerosol concentration measuring
device based on the growth of particles to a visible size by condensing
vapor on the particles. Particles as small as 20 A can theoretically be
detected by this method. Such particles are usually referred to as con-
densation nuclei.
Figure 9 shows the result of calibration studies reported by Liu,
Pui, Hogan and Rich5 on the Pollak counter, a manually operated conden-
sation nuclei counter. The Pollak counter has been carefully calibrated
49
-------
i—rrp
~ O 0.05/im NaCI
~ • 0.068/imNaCI
A 0.032/xm NaCI
0.025/im NICHROME
a 0.03Z/im NICHROME
• 0.11/im NaCI
o 0.15am NoCI
IDEAL
RELATIONSHIP
ACTUAL
CALIBRATION
°"*
^ _.. . 10 100
AEROSOL CONCENTRATION (ELECTRICAL). N£ , I03 PARTICLES/cc
Figure 9 Calibration of the Pollak counter with the electrical aerosol
generator.
140
120
100-
O
2
I I I I I I I I
CNC NO. I (ENVIRONMENT/ONE, SERIAL NO 149)
50 100 150 200 350 300 350 400 450 500 550 600
ACTUAL NUCLEI CONCENTRATION, I03( PARTICLES/CC)
Figure 10 Calibration of the Environment/One, Rich 100 condensation nuclei
counter with the electrical aerosol generator.
50
-------
BATTERY NUMBER
EQUIVALENT LENGTH-KILOMETERS
Figure 11 Penetration of monodisperse NaCl aerosol through a portable
diffusion battery measured with a GE CN counter at 6 1pm,
A 0.110 urn diam., ^ 0.075 urn, O 0.055 urn, X 0.0375 urn,
D 0.024 ym, solid lines ( ) calculated from the theory.
51
-------
by Pollak and Metnieks6 and has been used as a secondary standard for
condensation nuclei measurement. The calibration study of Liu, Pui, Hogan
and Rich5, made with monodisperse aerosol generated by the electrical
mobility classifier described in the preceeding section, shows that the
original calibration of the Pollak counter is in good agreement with the
present calibration, and that the maximum difference of these two calib-
ration is only about 18%. Further, the study shows that within the part-
icle size range studied, viz. from 0.025 to 0.5 pm diameter, the counter
response is essentially independent of the particle size. However, a
similar calibration study of a commercially available automatic conden-
sation nuclei counter3 shows that there is a substantial discrepancy be-
tween the factory calibration and the present calibration (see Figure 10).
The calibration study on the diffusion battery was made with the
portable diffusion battery described by Sinclair7. The diffusion battery
is comprised of a series of porous metal plates with approximately 50%
porosity and holes having diameters of approximately 230 ym. In the cal-
ibration study, which was recently reported by Sinclair, Countess, Liu
and Pui8, monodisperse aerosols of various diameters were generated by
mobility classification as previously described and the penetration of
the aerosols through the diffusion battery was measured with a conden-
sation nuclei counter. The measured penetration was compared with the
theoretically predicted penetration. Some sample results are shown in
Figure 11. The agreement is seen to be very good. Such good agreement
between theory and experiment constitutes a direct verification of the
theory of diffusion battery.
CONCLUSION
In this paper we have briefly reviewed the aerosol standards develop-
ment and instrument calibration studies made at the Particle Technology
Laboratory, University of Minnesota. We have shown that the use of the
vibrating orifice monodisperse aerosol generator and the electrical aerosol
generator has led to the development of monodisperse aerosol standards
in the 0.01 to 50 pm diameter range. Through the use of these aerosol
standards we have performed accurate calibration studies on such devices
as the condensation nuclei counter, the electrical aerosol analyzer, the
optical particle counter and the diffusion battery. Such instrument stand-
ardization work should lead to improved measurement accuracies for aerosols,
both in the laboratory and in the open atmosphere.
REFERENCES
1. Berglund, R.N. and B.Y.H. Liu. Generation of Monodisperse Aerosol
Standards. Env. Sci. Tech. 7:147-153, 1973.
2. Liu, B.Y.H., R.N. Berglund and J.K. Agarwal. Experimental Studies
of Optical Particle Counters. Atm. Env. 8:717-732, 1974.
52
-------
3. Liu, B.Y.H. and D.Y.H. Put. A Submicron Aerosol Standard and the
Primary Absolute Calibration of the Condensation Nuclei Counter.
J. Colloid and Interface Sci. 47:155-171, 1974.
4. Liu, B.Y.H. and D.Y.H. Pui. On the Performance of the Electrical
Aerosol Analyzer. Aerosol Sci. 6:249-264, 1975.
5. Liu, B.Y.H., D.Y.H. Pui, A.W. Hogan and T.A. Rich. Calibration of
the Pollak Counter with Monodisperse Aerosols. J. Appl. Meteor.
14:46-51, 1975.
6. Pollak, L.W. and A.L. Metnieks. Intrinsic Calibration of the Photo-
electric Nucleus Counter Model 1957, with Convergent Light Beams.
Tech. Note No. 9, Contract AF61 (052)-26, School of Cosmic Physics,
Dublin Institute for Advanced Studies, 1960.
7. Sinclair, D. A Portable Diffusion Battery. Am. Ind. Hyg. J. 33:
729, 1972.
8. Sinclair, D., R.J. Countess, B.Y.H. Liu and D.Y.H. Pui. Experi-
mental Verification of Diffusion Battery Theory. Presented at the
68th Annual Meeting of the Air Pollution Control Association, Boston,
Mass. June 15-20, 1975.
53
-------
PART II
AEROSOL GENERATION
Preceding page blank
-------
THE GENERATION OF AEROSOLS OF FINE PARTICLES*
Otto G. Raabe
Inhalation Toxicology Research Institute
P. 0. Box 5890
Albuquerque, New Mexico 87115
ABSTRACT
Aerosols of fine particles required for experimental applications vary
widely in necessary characteristics and require means of aerosol produc-
tion that are consistent with the scientific goals of an investigation.
In this report, the common techniques for generation of aerosols in the
respirable size range are reviewed with emphasis upon experimental, non-
therapeutic applications. Methods described include dispersion of dry
powders and the atomization of solutions and suspensions with various
compressed air and ultrasonic nebulizers. Special emphasis is placed
on the generation of aerosols of insoluble forms under controlled and
reproducible conditions. Heat treatment is described as an ancillary
step to speed the drying of droplets or to change the chemical nature
of aerosols after they are produced. Particular attention is devoted
to the various methods for producing monodisperse aerosols including
nebulization of suspensions of monodisperse particles, spinning disk
droplet generation, electrostatic dispersal, controlled condensation
and periodic dispersion of liquid jets. Reduction of the electrostatic
charge of aerosols as an important part of aerosol generation is dis-
cussed.
NOMENCLATURE
A = liquid aerosol concentration at time of droplet formation ex-
pressed as volume of liquid aerosolized per unit volume of air
at ambient conditions; a measure of nebulizer efficiency (yl/s,
ml/s, ul/nvin or ml/min, depending on equation units).
c = solute concentration in solution being nebulized (g/ml or
g/cm3).
c = initial solute concentration in solution to be nebulized
(g/ml or g/cm3).
D = geometric diameter of a spherical particle (cm or urn).
D, = geometric diameter of a spherical droplet (cm or urn).
* Prepared under U. S. Energy Research and Development Administration
Contract E(29-2J-1013.
57
Preceding page blank
-------
D = geometric diameter of a cylindrical liquid stream (cm or ym)
" emitted from a circular orifice.
D . = geometric diameter of droplets generated at the optimum fre-
^ quency for periodic dispersion of liquid jets.
e = either the base of Naperian logarithms equal to about 2.718
(unitless) or the basic electrostatic unit of charge equal in
absolute value to the charge of an electron.
F = fraction by volume of monodisperse particles in a stock suspen-
sion (unitless).
f = frequency of periodic vibrations or disturbance for dispersal
of cylindrical liquid streams (Hz).
f , = optimum value of f (Hz).
f = maximum value of f defined as the lowest frequency required to
produce the smallest possible droplets by periodic dispersal
of a cylindrical liquid stream (Hz).
L = length of a separated segment of a cylindrical stream of liquid
(cm or ym).
L . = minimum value of L.
In = natural logarithm.
L . = optimum value of L given by fQDt-
MMD = mass median diameter of an aerosol size distribution (cm or ym).
MNC = maximum number concentration of particle per milliter of water
for nebulization of an aqueous suspension of monodisperse par-
ticle to produce an aerosol with no less than 95% single par-
ticles.
MNC = MNC expressed in units of radioactivity concentration in neb-
ulizer suspension (Ci/ml) for radioactive particles.
n = integer number.
2
P = ambient barometric pressure (dynes/cm , cm Hg, psi).
2
P = minimum pressure in nebulizer jet (dynes/cm , cm Hg, psi).
58
-------
o
Q = volumetric flow rate (cm /s, ml/s or 1/min).
R = the ratio of droplets with single particles produced by a
nebulizer from a suspension of monodisperse particles to all
droplets produced that contain particles, i.e., the fraction
of singlets in an aerosol produced by nebulization of a mono-
disperse suspension.
r = droplet radius (cm or ym).
r = initial droplet radius (cm or ym).
s = specific activity of radioactive material (Ci/g).
t = time (s or min).
VMD = volume median diameter of a droplet distribution (cm or ym).
o
V = initial volume of solution to be nebulized (ml or cm ).
W = volume of water evaporated from a nebulizer per unit volume of
aerosol-containing air generated at ambient conditions (yl/s,
ml/s, yl/min or ml/min, depending on equation units).
y = liquid dilution ratio required for preparing monodisperse
suspensions (unitless).
TT = ratio of circumference to diameter for a circle = 3.14159
(unitless).
o
p = physical density of solid material (g/cm).
a = geometric standard deviation of log-normal distribution (unit-
^ less); minimum value is unity.
T = surface tension (dynes/cm).
u = speed of angular rotation (radians/s).
o
AP = gage pressure above Pa (dynes/cm , cm Hg, psig).
a
59
-------
THE GENERATION OF AEROSOLS OF FINE PARTICLES
Otto G. Raabe
Inhalation Toxicology Research Institute
Lovelace Foundation
P. 0. Box 5890
Albuquerque, New Mexico 87115
INTRODUCTION
Aerosols of fine particles can be defined as relatively stable suspen-
sions in a gaseous medium of liquid droplets or solid particles with
settling speeds under the influence of gravity less than the settling
speed of a unit density sphere of 3.5 m geometric diameter. Such aer-
osols are encountered in all phases of life. In fact, it appears to be
a fundamental property of the earth's atmosphere to contain aerosols.
Even in thoroughly filtered air, aerosols will spontaneously form from
trace gases under the influence of radiant energy.1-6 There is great
scientific interest in aerosols of fine particles not only because of
their use in medicine, industry and research and their collection by
the respiratory tree during inhalation, but also because of their inti-
mate involvement in the workings of nature. It is because of natural
aerosols that the sky appears blue and the sunset red.
Fundamental to the use, study or understanding of aerosols of fine par-
ticles is the ability to produce suitable forms for experimental or
applied purposes. The extensive use of aerosols in medical therapy has
led to the design of many types of aerosol generators which fall into
three basic groups: (1) compressed air nebulizers, (2) ultrasonic neb-
ulizers and (3) dust blowers. In addition to medical nebulizers and
dust blowers, other specialized nebulizers, dust blowers, spray nozzles,
spray cans and aerosol generation devices are widely used in experimen- -
tal research. Among these specialized instruments are some designed to
produce uniform droplets or particles. Aerosols of fine particles can
also be produced by photochemical, radiolytic and condensation processes.
The characteristics of the ultimate form of aerosols produced by various
means depend not only upon the generation method but also upon the phys-
ical and chemical nature of the source material and the temperature and
humidity conditions or treatment received. On this basis, aerosols can
be produced that are heterodisperse, polydisperse, or monodisperse. The
term heterodisperse is used to describe aerosols of particles which vary
widely not only in size but also in physical and chemical characteristics.
Polydisperse usually implies homogeneity of basic physical and chemical
characteristics but with widely different particle sizes and differences
in those aerodynamic properties which depend in part upon particle size.
Monodisperse is a term which implies both homogeneity of basic chemical
60
-------
and physical characteristics including composition and physical density
and relative uniformity of particle size and aerodynamic properties.
Monodisperse is a relative term, since a distribution of particles which
is satisfactorily monodisperse in one application may be considered
polydisperse in another. However, the definition suggested by Fuchs and
Sutugin' provides a guideline for defining "practical monodispersity".
Their definition of monodisperse particle size distribution is one having
a coefficient of variation less than 0.2. Nevertheless, a distribution
of particles which extends from 1 to 5 urn and has a coefficient of
variation much bigger than 0.2 may be clearly polydisperse to most, but
a distribution that extends from 0.01 to 0.05 pm may be considered es-
sentially monodisperse by some investigators even though the coefficient
of variation may be much greater than 0.2. In addition, the phenomenon
being studied may affect the definition. Particles may be numerically
monodisperse with respect to geometric size, but because of variations
in physical density, they may be polydisperse with respect to aerodynamic
properties such as settling speed.
This review summarizes a variety of methods commonly used to produce
aerosols of fine particles from the perspective of the investigator who
is interested in aerosol behavior or the biological effects of aerosols
inhaled by experimental animals or man.
POLYDISPERSE AEROSOLS
NEBULIZERS
A nebulizer is an atomizer used to produce aerosols of fine particles by
atomization of liquids to produce droplets, the majority of which are
less than 10 urn in diameter, as opposed to those from a spray atomizer
which may range up to 100 urn or larger. Both compressed air and ultra-
sonic nebulizers produce aerosols from liquids. Often, the particles of
residue remaining after the droplets evaporate form the desired aerosol
of fine particles. Either solutions or suspensions may be aerosolized
with nebulizers.
Important factors in describing the operation of nebulizers include (a)
the output rate and concentration of usable aerosol, (b) the volumetric
rate of air, (c) the evaporation losses of liquid which are independent
of usable aerosol, (d) the droplet size distribution, (e) the volume of
liquid required for proper operation and (f) the maximum unattended op-
erating time.8 Cognizance of these factors in relation to a particular
application determines the choice of a nebulizer.
The output rate of usable aerosol is usually described as volume of
liquid at initial formation associated with droplets that leave the
61
-------
c
co
r v0 -I
V0-(A+W)Qt
w
A+W
generator and may be expressed as ml/min or ul/min. The concentration
of droplets from most nebulizers is usually between 1C)6 and lO^/cm^
of air; however, since the volumetric rate of air may be several liters
per minute, the droplet output rate may be well in excess of 10^/min.
Evaporation losses increase the concentration of the solute in solution
or particles suspended in the aerosolized liquid. This can be described
as given by Mercer, et al.9
(1)
with the c0 and c solute concentrations at time t = 0 and t(min), respec-
tively, and W and A are the evaporated and aerosolized liquid output con-
centrations in air (ml/1), respectively, V0 is the initial volume of
solutions and Q is the volumetric output of the nebulizer (1/min). This
evaporation occurs primarily from the surface of the liquid and from the
droplets which evaporate slightly and then hit the wall of the nebulizer
to be returned to the reservoir. This change in concentration causes an
increase in the sizes of the particles that are formed when the droplets
dry. Since this is undesirable, it is advantageous to keep the ratio
W/(A + W) as small as possible. This is accomplished by high aerosoliza-
tion efficiency (as given by A) and by keeping the nebulizer solution
cool. Using a nebulizer with a small reservoir reduces W by the natural
cooling from evaporation and air expansion in compressed air operated
devices. An effective way to reduce evaporation is to operate the nebu-
lizer with the reservoir submerged in an ice-water bath.10 Other methods
of maintaining a more uniform concentration of liquid to be nebulized
that have been used include continuous feed of liquidll and humidifica-
tion pretreatment of the compressed air.12
Nebulizers produce droplets of many sizes and resultant aerosol parti-
cles after evaporation are concomitantly polydisperse. ^ The droplet
distributions described for nebulizers are the initial distributions at
the instant of formation; droplet evaporation begins immediately even at
saturation humidity since the vapor pressure on a curved surface is
elevated.14 The rate of evaporation depends upon many factors including
surface tension, energy availability, degree of saturation of the air,
the solute concentration,15 the hygroscopicity of the solute,^"~^° the
presence of immiscible liquids or evaporation inhibitors!9-20 and the
size of the droplets (smaller droplets have higher vapor pressures and
dry fasterl4).
62
-------
The distribution of droplets produced by nebulizers has been described
satisfactorily in many ways.13,21-23 Most useful is the assumption that
the logarithms of droplet size are normally distributed. This log-normal
distribution of sizes allows for simple mathematical transformations24-25
and usually satisfactorily describes droplet volume distributions.23
The characteristic parameters of a log-normal distribution are the me-
dian (or geometric mean) and the geometric standard deviation (on).
The median of a distribution of droplet diameters is called the aroplet
count median diameter (CMD); the median of the mass or volume distribu-
tion of the droplets is called either the mass median diameter (MMD) or
the volume median diameter (VMD). These are related by:
(2)
in which In refers to the natural logarithm.25
Aerosols produced from aqueous solutions (and some other methods) are
electrostatically charged by the imbalance of ions in the droplets as
they form.26-27 After evaporation, aerosol particles can be relatively
highly charged; this may cause a small evaporating droplet to break up
if the Rayleigh Iimit28 is reached due to the repelling forces of the
electrostatic charges overcoming the liquid surface tension.29-30 jhe
Rayleigh limit, the minimum stable diameter, Dj, of a droplet with n
electronic units of charge, e, can be expressed as:
2lTT
(3)
with T the surface tension. In some cases the net charge on a particle
may be tens or even hundreds of electronic charge units, which may af-
fect both aerosol stability and behavior. Therefore, a reduction in the
net charge of aerosol particles produced by nebulization either by mix-
ing with bipolar ions31 or by passing through a highly ionized volume
near a radioactive source32-33 is desirable and may be imperative in
some experiments.
Compressed air nebulizers generate droplets by shattering a liquid
stream in fast moving air.9»34-37 jne liquid is usually drawn into the
air flow by the natural reduction in pressure that occurs at right angles
to the fast moving air stream (Venturi effect given by Bernoulli's
Theorem).
63
-------
When such a nebulizer is operated at gage pressures greater than the
ambient barometric pressure (which is usually the case), the air jet op-
erates under sonic conditions, that is, the maximum velocity of the air
jet is the speed of sound at the temperature of the air stream at a given
location. (Supersonic atomizers require special configurations.38) fhe
flow rate, Q, at a given barometric pressure, Pa, and gage pressure, AP,
is:39
= Q'
(4)
with Q1 the known flow rate at some ambient barometric pressure Pa' when
operated at gage pressure AP1.
Adiabatic expansion occurs at the appropriate sonic velocities until the
sum of the pressure head and velocity head equal the ambient pressure,
Pa, at which point the pressure is a minimum P^ given by:39
Pm = 0.83 P.
m
a
(5)
If it is assumed that liquid is aerosolized at a rate proportional to
Pa-Pm and that the total flow rate is given by equation 4, the nebulizer
output concentration, A, is given by:
A=A'
a
p /
in
-P
(6)
with A1, AP1 and Pa' a set of known operating conditions.
This equation can be used to correct reported or measured nebulizer
characteristics for the effect of change in barometric pressure since
these characteristics are customarily reported in terms of the gage
pressures (AP) at which they are operated under local conditions. Much
of the available nebulizer data were collected by fiercer and his
colleagues at a barometric pressure of 62 cm Hg.™>">26,40-42 /\n attempt
has been made in this report to correct nebulizer operation data to
21 °C and barometric pressure 76 cm Hg.
64
-------
Auxiliary Air Flow
The air flow exiting from a nebulizer creates a negative pressure
at the back side of the nebulizer jet. An opening placed nearby will
allow auxiliary air to be drawn into the device and mixed with the out-
flow. Such auxiliary flow can not only increase the effective volumetric
output of a nebulizer, but also markedly increase aerosol droplet output
frequently without markedly affecting aerosol output concentration.
Many atomizers are designed to incorporate auxiliary air flow into nor-
mal operation.42
DeVilbiss No. 40
The DeVilbiss No. 40 (DeVilbiss Co., Somerset, Pa.) is one of the sim-
plest compressed air nebulizers. This glass nebulizer (Fig. 1) has a
vertical jet and a separate capillary through which the liquid to be
aerosolized is drawn into the air stream. Its reservoir holds only
about 10 ml. About 10 liters of air and 0.22 ml of liquid droplets are
released per minute at an operating pressure of 20 psig with VMD of 3.2
ym and ag 1.8.9
Dautrebande D-30
The Dautrebande D-30 nebulizer (Fig. 2) also uses a vertical jet with
separate feed capillary.43-44 jhe aerosol droplets, however, are re-
quired to follow tortuous paths through baffle holes so that most of the
larger drops are unable to negotiate this scrubbing action and are col-
lected and returned to the liquid reservoir. Consequently, droplets
which leave the Dautrebande generator as useful aerosol are much smaller
than those produced by other nebulizers and the aerosol output concen-
tration is low. The D-30 releases about 25 liters of air and 60 yl of
liquid droplets per minute at an operating pressure of 20 psig with VMD
of 1.4 um and og 1.7.9 it i$ usually constructed of plastic.
Lauterbach Nebul i zer
The Lauterbach nebulizer45 uses a jet consisting of a metal tube sealed
at one end with a small hole drilled near the sealed end. This jet is
operated by attaching a compressed air line to the open end of the tube
with the orifice positioned very near the surface of the liquid so that
the air stream is emitted parallel to the surface. The liquid touching
the tip of the metal tube is drawn directly into the air stream because
of the reduction in pressure at right angles to the stream. The Lauter-
bach improved nebulization by using a recirculating reservoir system,
consisting of the generator reservoir and a larger volume supply reser-
voir. Liquid from a 200 ml supply reservoir continually flows into the
generation reservoir, which is maintained at a constant level with a
65
-------
Figure 1. Tue ;\-vi !b
40 glass i-->rv! :;>.. r; +
pressed air irler iv
bottom and the >iG."o':r-
is at the upper r^cht
auxiliary a ir ve-~t •;•-
trie
OJt:Ot
The
-. !x -~,
Figure 2, Exploded /iew of a
Dautrebande D-3C nebulizer
showing the vertical jet, ori-
fice, neck and baffle holes.
This unit has a separate liquid
feed to allow the addition of
liquid during operation.
66
-------
fixed overflow tube that allows excess liquid to be pumped back to the
main reservoir. Since the recirculation system works independently of
the jet operation, the liquid is continually mixed and concentration
changes are minimized. The reported output is about 2.4 liters of air
and 17 ul of liquid droplets per minute at an operating pressure of 20
psig, with VMD of 2.4 urn and ag 2.0.45
A schematic of the glass Lauterbach nebulizer is shown in Figure 3.
Mercer et al9 report a plastic modification40 of the Lauterbach genera-
tor that is somewhat different in design.
COMPRESSED
AIR AT 20 RS.I.
CONSTANT LEVEL
OVERFLOW
RESERVOIR
200 ML.
CAPILLARY
LIQUID FLOW
LIQUID FLOW
Figure 3. Schematic drawing of the Lauterbach
glass nebulizer (from Lauterbach et al45).
Collison Nebulizer
One of
is the
wide'y
The Col
e.g. 0.
within
ing of
larger
ul/min
aerosol
urn and
the early medical nebulizer modifications, which is still in use,
nebulizer introduced in 1932 by W. E. Collison.45 It has been
used in Great Britian for both medical and experimental purposes.'''7
lison jet consists of one or more orifices of various diameters,
035 cm, set at right angles to a vertical liquid feed channel
a cylindrical metallic rod. A simple baffle arrangement consist™
a hollow cylinder surrounding the jet is used to collect the
droplets by impaction. Output concentrations are low with 55
of aerosol in 7.1 1/min at 20 psig; water evaporation exceeds the
output and is equal to 90 yl/min at 20 psig with VMD about 2.0
9
on 2.0.47
67
-------
Laskin Nebulizer
The Laskin nozzle (designed by Dr. Sidney Laskin, Department of Environ-
mental Medicine, New York University Medical Center, New York, New York)
consists of an orifice in a sealed metal tube. The basic arrangement is
similar to the Collison nebulizer. The orifice is positioned at the top
of a metal capillary through which the liquid is drawn into the air
stream. The bottom of the feed tube can be submerged into the liquid at
various levels without markedly affecting the aerosol output. Although
the efficiency of the Laskin nozzle is not remarkable, it is useful for
producing large quantities of aerosol when large volumes of air are ei-
ther desirable or not prohibitive. Operating characteristics are similar
to those of the Collison nebulizer.
Wright Nebulizer
B. M. Wright introduced the principle
of the jet baffle in 1958 with the de-
sign of the Wright Nebulizer.48 By
placing a collection surface close to
the point of disintegration of a nebu-
lizer jet, the output concentration of
usable aerosol is markedly increased.
The basic arrangement is shown in Fig-
ure 4. The outlet of the liquid feed
tube is placed near the point of mini-
mum static pressure of the air jet
emitted from a small orifice (0.074 cm
ID) so that liquid is drawn into the
jet flow within a small compartment and
sprayed through a larger hole (0.16 cm)
and impinged upon an open-sided flat
collector at short distance (0.116 cm).
Only the smaller droplets can negotiate
the turn at the baffle. The liquid
from the large droplets that collects
on the baffle is forced to the edges by
the airstream where it is extensively
aerosolized enhancing the output of
small droplets. If the jet baffle is
too close to the point of disintegration
of the liquid stream, it will interrupt
the formation of small droplets and if
it is too far away, the secondary aero-
solization of small droplets will be
less because of the lower speed of the
air passing the baffle. By optimizing
AIR INLET
BAFFLE (OMem)
TUBE
Figure 4. Schematic draw-
ing of the Wright Nebulizer
showing the air jet, liquid
feed tube, secondary flow
hole and jet baffle.
68
-------
the position of the boffU.^ thf1 concentration of small droplets (< 10 urn
in diameter) in the output ';s greatly enhanced. Use of the Wright jet
baffle principle is a common feature of modern nebulizers. The Wright
nebulizer jet assembly is usually constructed of plastic. Typically,
the reservoir holds about 500 til. About 12 liters of air and 350 ul of
water droplets are released per minute at operating pressure^of.^24 psig.
The VMD has been reported to be 2.6 urn with a
og about 2.0.49-50
Lovelace
Nebulizer
The ''Lovelace" (Lovelace Foundation for Medi
Albuquerque, i\ew Mexico) Nebulizer (figures
erator which usas, the Wngr;t jet baffle
principle but with a small rounded baffle
rather than a flat one. A small adjusta-
ble nylon or teflon screw with rounded end
mounted naar ana in direct opposition
to the air jet. serves as this primary
baffle in the Lovelace design. For its
size and simplicity, it has an outstanding
efficiency in the generation of usable aer-
osols at a small volumetric rate. It op-
erates with a small liquid volume of about
4 ml and an air flow of
20 psig jet pressure.
provide up to 70 iJ/min
with VMD about 5,8 i_,m
Newton et a! .51 cescribe the basic design
of the Lovelace nebulizer. The construc-
tion is of lucite for the liquid cup,
epoxy for the top, stainless steel for the
outlet tube anc scainless steal for the
tube supplying compressed air w the lu-
cite jet assembly. The jet assembly cori-
t a i n s a siua 1 1 orifice (0.25 mrn ID), a
cylindrical capillary through which liquid
is fed to the jet and a support for tne
baffle screw. Mercer et al§ demonstrated
the importance of positioning the jet
baffle by measuring the output fov various
settings of distances between the end of
the screw ar.d the on' rice (Figure 7); they
also report a very favoraole output-- to-
evaporation ratio.
cal Education and Research,
5 and 6) is a miniature gen-
about 1,4 "i/min at
This device can
n of liquid droplets
and og about 1,8.
Figure 5. Exploded sche-
matic of the Lovelace
Nebulizer which uses a
small screw as the jet
baffle.
-------
Retec X-7Q/N Nebulizer
The Retec X-70/N Nebulizer (Retec Development Laboratory, Portland, Ore-
gon) is another small, efficient nebulizer using the Wright jet-baffle
principle (Figure 6). The jet orifice is 0.5 mm ID; the secondary ori-
fice is 0.75 mm. Clever design of the components allows all parts to
be molded of plastic for easy assembly and disassembly while the op-
erating dimensions are maintained within narrow limits. The jet baffle
is a plastic sphere 6.0 mm in diameter A double orifice system as
with the Wright Nebulizer (Figure 4) allows use of a liquid flow tube.
By recessing the secondary orifice, the speed of airflow passing the
jet baffle increases so that the secondary aerosolization is enhanced.
The reservoir volume is normally 10 ml, but larger reservoirs can be
used. Burns52 determined the aerosol output to be 230 ml of liquid per
minute with a flow rate of 5 1/min when operated et 20 psig.
Babington Nebulizer
A new class of compressed
Babington principle.53-55
usually a slit at a right
air nebulizer^ is possible based upon the
Liquid is allowed to flow over an orifice,
angle to the flowing liquid, at a rate which
Figure 6. Photograph cf AM
-------
)2 !8 24 30 36 42 48 54
BAFFLE-OftSFSCE SEPARATION in MILS
iv /, Lffect of baffle screw position on output
of useful aerosol from the Lovelace Nebulizer showing
the Wright jet-baffle effect which provides enhanced
output by proper jet-to-baffle spacing. The output at
optimum position is about six times the output of the
unbaffled jet. The data squares are based on mass
output measurements and the data circles are based on
light-scattering measurements. (From Mercer et al^).
maintains an untorn film over the orifice even as compressed air is
forced through the liquid film. Small droplets are generated at the
surface of the liquid and carried away in the air stream without the
intermediate Rayleigh stream disintegration which is characteristic of
conventional nebulizers. The usual design is to have several small slit
orifices on the surface of a hollow sphere. The inside of the sphere
is pressurized and liquid to be aerosolized is pumped from the reservoir
to the outside top of the sphere so that it flows over the surface of
the sphere at a volumetric rate sufficient to maintain a closed film
over each orifice.
The Hydrosphere Nebulizer (Owens-Illinois, Toledo, Ohio) is a commercial
nebulizer using the Babington principle. The jets are slits about 0.2
mm by 1.2 mm cut in two hollow glass spheres 1.5 cm in diameter. Two
slits in one sphere and one in the other provide three orifices, The
unit is designed for auxiliary air flow on demand to enhance output.
It also incorporates jet baffles based on the Wright principle; these
71
-------
are metal spheres 4 mm in diameter. Large auxiliary reservoirs are re-
quired with the Hydrosphere Nebulizer. At 20 psig with auxiliary air,
the volumetric output rate is about 110 1/min with about 3.3 ml of
liquid aerosolized per minute. Of the 3.3 ml of liquid aerosolized per
minute about 2.7 ml represents a usable droplet aerosol with VMD equal
to about 4 um and ag about 2.3.56
Ill trasonic Nebu 1 i zers
Ultrasonic nebulizers operate differently from compressed air nebu-
lizers. 57-62 Although a flow of air is used to carry off the aerosol
droplets, the air is not involved in the initial formation of these
droplets which are formed by ultrasonic vibrations. A sectional view
of an experimental ultrasonic generator (designed by Mr. G. J. Newton,
Lovelace Foundation, Albuquerque, New Mexico) is shown in Figure 8. A
110-volt, 60-cycle line current is converted to a high-frequency signal
(in this model approximately 800 kHz) and transmitted to the transducer
A through the shielded cable B^. The transducer, a cylindrical piezo-
electric device, transforms the high-frequency electrical current to
mechanical oscillations. Because of the shape and nature of the crystal,
these mechanical vibrations are highly directional and create an intense
acoustic field in the coupling fluid C. The energy is carried through
Air Intel (F
^^ Aerosol Outlet (H)
>f
.Aerosol Particles (S)
Figure 8. Sectional schematic
view of an operating Ultrasonic
Aerosol Generator showing trans-
ducer assembly (A) receiving
power through shielded cable (B)
generating an acoustic field in
the coupling fluid (C) creating
an ultrasonic geyser (D) in the
generator reservoir (E) and air
entering at (F) carrying away
aerosol (G) through the aerosol
outlet (H) (from RaabeS).
72
-------
the liquid by the motion of the molecules along the direction of propa-
gation. Intense adiabatic compressions and rarefactions occur with cor-
responding density and temperature changes. The actual amplitude of the
motion is only about 0.2 urn in an 800 kHz field, but the acceleration
attained is in excess of 500,000 g_. This turbulence creates a pressure
gradient along the cylindrical axis of the transducer and results in a
water geyser Jj. The high-frequency turbulence in the geyser produces
the aerosol droplets.
The sizes of the droplets formed depend on the frequency of the acous-
tical field and the physical-chemical character of the liquid. The sizes
of the droplets carried by the air stream out of the generator depend up-
on the rate at which the droplets are carried away from the geyser since
coagulation is rapid at the high concentrations of droplets initially
formed. For example, with an air flow of 1 1/min, the droplet distribu-
tion had a VMD of 10 urn while at a flow of 10 1/min it was only 3 um.63
Ultrasonic nebulizers impart intense energy to the liquids being nebu-
lized which causes heating with an increase in evaporation. Also, mate-
rials in solution or suspension can be denatured, degraded or deformed.
Ultrasonic nebulizers are also sensitive to the chemical properties of
the aqueous solutions or suspension to be nebulized and may malfunction
because of this sensitivity. Operating characteristics of several ultra-
sonic nebulizers have been reported.41,64-65
Summary
The operating characteristics with water of several compressed air and
ultrasonic nebulizers are summarized in Table 1. The output concentra-
tion, A(yl/l), is a measure of aerosolization efficiency. Only selected
operating pressures have been described for each compressed air nebulizer,
although most may be operated at various pressures with associated var-
iations in operating characteristics. Other conditions being equal,
compressed air nebulizers which utilize the Wright jet baffle principle
are about 5 times as efficient as those that do not.
AEROSOL GENERATION WITH NEBULIZERS
Fine aerosols of both soluble and insoluble* materials may be produced
by nebulization of solutions or suspensions. For example, spherical
insoluble particles of plastics may be made from solutions with suitable
organic solvents,66 or soluble forms may be made from aqueous solutions
of electrolytes.67 Changes may be made in the chemical state of the
particles formed after solvent evaporation by heating, and in the size
distribution by selective collection of a portion of the particles.
The terms 'soluble" and "insoluble" are relative terms especially
when used to describe the character of inhaled aerosols.
-------
Table 1
Typical Operating Characteristics of Selected Nebulizers With Reference, Gage
Pressure (AP), Output Concentration (A), Evaporation Output (W), Volumetric
Flow Rate (Q) and Droplet Distribution (VMD and ag)
Corrected, as Required, to Operation at 76 cm Hg and 21 °C; Values in
Parentheses are my Estimates.
Nebulizer
Dautrebande D-30
(9)
Lauterbach
(9)
Collison
(3 jet model)
(47)
DeVilbiss #40
(9)
Lovelace
(Baffle screw set
for optimum operation
at 20 psig)
(63, 10)
Retec X-70/N
(52, 10)
DeVilbiss Ultrasonic
(setting #4)
(Somerset, Pa.) (41)
AP
(psig)
10
20
30
10
20
30
20
25
30
40
10
15
20
30
15
20
30
40
50
20
20
30
40
50
A
1.6
2.3
2.4
3.9
5.7
5.9
7.7
6.7
5.9
5.0
16
15.5
14
12
27
40
31
21
27
56
53
54
53
49
W
(yl/D
9.6
8.6
8.2
(12)
12.7
12.6
12.6
12.6
(10)
8.6
7.0
7.2
(10)
10
11
9
11
20
12
11
7
9
Q
VMD
(1/min) (ym)
17.9
25.4
32.7
(1.7)
(2.4)
(3.2)
7.1
8.2
9.4
11.4
10.8
13.5
15.8
20.5
1.3
1.5
1.6
2.0
2.3
5.0
5.4
7.4
8.6
10.1
1.7
1.4
1.3
3.8
2.4
2.4
(2.0)
2.0
4.2
3.5
3.2
2.8
5.8
4.7
3.1
2.6
5.1
5.7
3.6
3.7
3.2
A_
1.7
1.7
1.7
2.0
2.0
2.0
(2.0)
2.0
1.8
1.8
1.8
1.8
1.8
1.9
2.2
2.3
2.0
1.8
2.0
2.1
2.2
150
33
41
6.9 1.6
74
-------
Both crystalline and amorphous forms of aerosol particles may be produced
from aqueous solutions. Often the type of particles produced will depend
upon the conditions of drying as well as the chemical nature of the ma-
terials. If drying is too rapid low density particles, which are essen-
tially shells, may be formed by the encrustation of the surfaces of the
drying droplets.6' More often, drying is seriously hindered by the
hygroscopicity of the solute,16,17 the presence of immiscible liquids or
evaporation inhibitors,19 the insufficiency of energy available to the
droplet, the saturation of the air, and other factors. Adequate drying
usually requires mixing the primary droplet aerosol with filtered, dry
air. Warming of an aerosol to speed drying, by passing it though a
heated tube, may be necessary in some experiments. Reduction of electro-
static charge to Boltzmann equilibrium is usually essential.33
When droplets evaporate, the residue particles become the aerosol.
Since nebulizers produce droplets of irany sizes, the aerosols formed
after evaporation are pclydisperse with geometric standard deviation
about equal to the value for the droplet distribution produced by the
nebulizer. The size of a given particle depends upon the solution con-
centration, c, and the croplet diameter, D^. These are related for
spherical particles by:
D3p = Dd3c
with D the geometric diameter of the resultant spherical particle of
density, p. Since aerosol particles vary in density and shape, it is
useful to relate various size particles in terms of their dynamic char-
acteristics. For example, the aerodynamic equivalent diameter for a
particle may be der,:ribed es the diameter of a unit density sphere with
the same falling speed as the particle.18 The aerodynamic equivalent
diameters of sol-id spheres of various densities produced from a 5-ym
droplet for various solution concentrations are shown in Figure 9.
Nebulization of an aqueous colloid;! suspension forms insoluble parti-
cles of the aggregates of the colloidal micelles. This method has the
advantage of requiring no organic solvents. If the micelles are small
and in high concentration,the resultant particles will be nearly spher-
ical in shapr- and their size distribution may be predicted by assuming
that the suspension behaves as a solution. For example, a colloidal
suspension of ferric hydroxide mi/ be aerosolized to produce relatively
insoluble and spherical particle- of ferric oxide for inhalation experi-
ments. 68-69 Water physically tripped or chemically bonded in such
particles may be removed by heat ,ng. If the colloidal micelles are com-
parable in size to the droplets produced by the nebulizer1,70 or if the
concentration is small, physical factors and the statistics of the ran-
dom pic! up of micelles by droplets will determine the size distribution
75
-------
Figure 9. The aerodynamic
(equivalent) diameter of
solid spherical particles
formed by the evaporation
of a liquid droplet of 5 ym
geometric diameter for var-
ious solute concentrations
in the droplet (milligrams
per cubic centimeter or
per milliliter) plotted
with respect to the phys-
ical density (grams per
cubic centimeter) of the
resultant solid particles.
579
PARTICLE DENSITY
of the resultant aerosol. Unfortunately, aerosols produced from colloids
sometimes have inherent porosity?! because of the interstices between
micellular components. Aerosol of bacteria and viruses can also be pro-
duced by nebulization of suitable suspensions.72
Changing the chemical nature of aerosols by heat treatment after they
are produced is a useful way of creating particles with desirable chem-
ical and physical characteristics.73-74 Kanapilly, et a1°7,75 describe
the generation of spherical particles of insoluble oxides from aqueous
solutions with heat treatment of the aerosols with temperatures up to
1400 °C. This procedure involves (a) nebulizing an appropriate aqueous
preparation containing the desired metal cation, (b) drying the drop-
lets, (c) passing the aerosol through a high temperature heating column
to produce the spherical oxide particles and (d) cooling the aerosol
with the addition of diluting air. Figure 10 shows a ZrO£ aerosol pro-
duced by nebulizing zirconium oxalate with heat treatment at 1150 °C.
Thermal degradation of aerosols to produce desired chemical forms has
also been used by Horstman, et a!77 and by Arvik and Zimdahl.78 Another
example of aerosol alteration is the production of spherical aluminosil-
icate particles with entrapped radionuclides by heat fusion of clay
aerosols.76 This method involves (a) ion exchange of the desired radio-
nuclide cation into clay in aqueous suspension and washing away of the
unexchanged fraction, (b) nebulization of the clay suspension yielding a
clay aerosol as shown in Figure 11 and (c) heat fusion of clay aerosol
removing water and forming an aerosol of smooth solid spheres of density
2.3 g/cm^ as shown in Figure 12. Freeze-drying of droplets has also
been reported as a useful technique for preparing certain types of aer-
osols.79
76
-------
Figure 10. Electronmicrograph
of an aerosol sample of
zirconium dioxide (shadowed
with chromium vapor) produced
by nebulization of an aqueous
zirconium oxalate mixture and
degraded at 1100 °C. (from
Kanapilly et al75).
Figure 11. Electronmicrograph
of an aerosol sample of mont-
morillonite clay (shadowed
with chromium vapor) generated
by nebulization of an aqueous
clay suspension.
Figure I?.. Electronmicrograph
of a sample of an aerosol of
aluminosilicate spheres
(shadowed with chromium vapor)
generated by heat fusion at
1150 °C of the clay aerosol
shown in Figure 11. (from
Raabe et al/6).
77
-------
SPRAY CANS
Commercial "aerosol" containers
operate on a very different prin-
ciple than nebulizers.80-81 They
use various mixtures of the liq-
uid to be atomized with a suit-
able volatile liquid (usually a
fluorinated hydrocarbon such as
dichloro-difluoro-methane
(freon)). Numerous formulations
have been tested to evaluate the
applicability to different pur-
poses. 82 The pressure in the
sealed container caused by the
volatile liquid forces the fluid
mixture through a feed tube to
the nozzle orifice when the
nozzle is depressed (Figure 13).
The rapid evaporation of the
volatile liquid causes numerous
bubbles to form in the liquid
mixture as it approaches the
orifice and shatters the released
liquid stream into droplets which
usually have a broad range of
sizes, often up to 100 urn in
diameter. However, the aerosols
produced by these pressure con-
tainers can be widely varied by
orifice nozzle design and chemical
formulation.83
Figure 13. Schematic illustration
of the basic components of a spray
can which uses liquid fluorinated
hydrocarbons or other hydrocarbons
as propellant.
A major portion of the aerosols produced by spray cans may, in many
cases, be smaller than 3.5 urn in aerodynamic equivalent size, and,
therefore, qualify as fine particles. Spray cans have been in use to
deliver respirable aerosols of drugs.°^ They also may produce respir-
able aerosols of fine particles as undesirable side-products. Vos and
Thompson determined an average of 57% by weight of the nonvolatile aer-
osols produced by commercial deodorant spray cans were in the fine par-
ticle range.85 in another study, they observed mass fractions in the
fine particle range of 6.6% for a hair spray, 11.3% for a frying pan
spray, 9.3% for disinfectants, 7.5% for an insect killer, 31.4% for an
air freshener and 3.2% for another deodorant.86
Some investigators may object to the use of the term aerosol in describ-
ing spray products.87 Many books and articles that contain the word
78
-------
aerosol in their titles are devoid of aerosol information in the scien-
tific sense. The word is certainly misused when applied to cans that
dispense such commodities as shaving cream, tooth paste or whipped
potatoes.
DRY PARTICLE AEROSOLIZATION
Dust generators such as the DeVilbiss dust blower #175 (Figure 14) (De-
Vilbiss Co., Somerset, Pa.) are designed to turbulently suspend dry
dusts and carry the resultant aerosol into an air stream. The DeVilbiss
unit uses turbulent air flow in the glass dust container to stir the
dust. Many methods may be used to generate dry dust: however, all in-
volve two factors: getting the dust in motion by shaking, grinding,
stirring, placing in fluidized bed, etc., and moving the proper amount
into the desired air volume. Some methods may not yield reproducible or
unvariable concentrations or size distributions. One unusual approach
to dust generation involves the loading of dry powder into a capsule and
propelling it with an air gun through a sharp cutting device into an aer-
osol chamber.88 jnis method has even been used for animal exposures.89-90
Figure 14. The DeVilbiss No. 175 dry-dust blower used to aerosolize
powders.
79
-------
One of the most popular dust generating devices has been the B. M. Wright
Dust Feed91 which allows dust to be ground off of a packed plug. A tim-
ing and gearing mechanism provides a constant rate for this dust feeding
device. Another somewhat similar device, providing a narrower size
range of particles, has been described by Dimmick.92 Variations on dry
dust feed systems have been described by Ebens and Vos93 and by Fuchs
and Murashkevich.94
An elaboration of a kitchen blender for producing aerosols of dry powders
and fibers has been described by Drew and Laskin.95 jn this system a
preground dust is mixed in a container having a four-bladed, high-speed
fan at the bottom. The air-dust mixture is aspirated into a separate
baffling chamber mounted at the top of the fluidizing container. From
the baffling chamber, the dry aerosol is conducted to its intended des-
tination. Drew and Laskin showed the fluidizing dust chamber to be of
practical value in inhalation studies with fiberglass dust with concen-
trations as high as 200 mg/m^ and coefficient of variation of concentra-
tion only 6.7% for 12 samples taken at half-hour intervals.
Fluidized beds of solid particles such as bronze spheres kept in motion
by upward air flow have been used for aerosolization of dry dust.96-97
Dusts are injected into the fluidized bed which serves to break up the
dust particles and readily releases them as an aerosol.
PHOTOCHEMICAL AND RADIOLYTIC AEROSOLS
Chemical reaction can occur at the molecular level in air under the in-
fluence of light or other radiation to produce a variety of aerosols of
ultrafine and fine particles. The production of condensation nuclei by
the irradiation of filtered nuclei-free air by X-rays, alpha particles
and beta particles has been described by Megaw and Wiffen,! Madelaine^
and Mercer and Tillery.99 Aerosols produced by alpha, beta and gamma
(X-ray) radiation are usually called radiolytic aerosols.
The topic of photochemical production of aerosols of fine particles is
beyond the scope of this report. Needless to say, it has and will con-
tinue to receive considerable attention with respect to its role in air
pollution.
MISCELLANEOUS
Useful aerosols of dry particles of metal and metal oxides have also
been produced with electrically heatedlOO-104 or exploded wires.105-107
These techniques have some disadvantages because of the broad size
distributions of the resulting particles and because of the tendency of
particles to coalesce. There are applications, however, for this type
of aerosol since they tend to be in the ultrafine range (less than 0.1
urn). It is possible with a technique such as Couchman1 siOO wire heating
80
-------
method to produce spherical particles of many different metals or their
oxides. Aerosols of ultrafine particles have also been produced by arc
vaporization.108-110
Spurny and Lodge have described a method for producing radiolabeled aer-
osols of ultrafine particles by vaporization of radiolabeled metals and
metallic compounds.111 Lasers have also been employed to provide heat
for vaporization and combustion of metals under conditions which lead to
condensation aerosols of ultrafine particles or chain aggregates.112-113
Sinclair and Henchliffe used an induction furnace to vaporize silver to
yield ultrafine spherical particles of metallic silver.li4 Many other
vaporization and combustion methods can be designed on the basis of
these techniques.115-118
MONODISPERSE AEROSOLS
GENERAL
The generation of monodisperse aerosols with particles of similar size
and physical-chemical characteristics is desirable in many experimental
applications since the effects of particle size can be evaluated. Many
of the methods for producing monodisperse aerosols have been reviewed by
Fuchs and Sutugin,? Raabe,8 Liu11^ and Mercer.12° These methods include
the growth of uniform aerosol particles or droplets by controlled con-
densation, the formation of uniform droplets (which may dry to solid
particles) by controlled dispersion of liquids electrostatically, by dis-
persion of liquid jets with periodic vibration, or with a spinning disk
or top. Another popular method for producing monodisperse aerosols is
the nebulization of a suspension of monodisperse particles, but involves
some inherent difficulties. It is not a simple matter to produce suit-
able monodisperse aerosols for particular applications. The variety of
physical-chemical types of such aerosols is still somewhat limited and
the specialized equipment usually requires careful operation.
The definition for monodispersity suggested by Fuchs and Sutugin? pro-
vides a guide for evaluation of monodispersity of particle size distri-
butions. If the coefficient of variation (ratio of standard deviations
to the mean) of the distribution of sizes is less than 0.2 (20%), the
aerosol may be satisfactorily described as having "practical monodis-
persity". For a log-normal distribution, this is about equivalent to a
0g < 1.2. In general this criterion is not very stringent, and many
investigators may prefer to achieve greater size uniformity. However,
even with very effective devices for producing uniform particles or
droplets, it is sometimes necessary to tolerate a small fraction of odd-
sized particles or doublets (two primary particles which have coalesced).
81
-------
NEBULIZATION OF MONODISPERSE SUSPENSIONS
General
The nebulization of suspensions of uniform particles which have been
chemically grown or separated from polydisperse particles has been a
simple and useful means of generating monodisperse aerosols. Suspen-
sions of polystyrene latex spheres of fairly uniform size grown by emul-
sion polymerization as developed by Bradford and Vanderhoffl21-122 have
been commonly used for this purpose.123-126 other types of monodisperse
hydrosols have also been made.127-128 Polystyrene latex spheres!29-132
and othersl33 can be labeled with radionuclides.
It is not possible at reasonable dilutions to guarantee that only indi-
vidual particles will be aerosolized and droplets containing more than
one of the suspended particles will become undesirable aggregates upon
evaporation. Raabel26 has calculated the dilution, y, for polystyrene
latex suspensions required to generate an aerosol with singlet ratio R
defined as the ratio of droplets with but one monodisperse particle to
all droplets that contain one or more monodisperse particle:
y =
F(VMD)3e4-5(ln^)2[l-0.5e(ln(I/] (a)
(I-R)D3
for R > 0.9 and Og < 2.1 with F the fraction by volume of the particles
in the original stock suspension, D the diameter of the monodisperse
spheres and with VMD and Og the volume median diameter and geometric
standard deviation of the Hroplet distribution from the nebulizer.
Equation 8 can be used to calculate the maximum number concentration,
MNC* of monodisperse particles of any type in a liquid suspension to be
nebulized so that the singlet ratio is not less than R:
MNC. _
~
ir(VMD)3[l-0.5ellno')2]
When VMD is the nebulizer droplet median diameter in centimeter units,
MNC* is number of monodisperse particles per milliliter of the suspension.
82
-------
Raabe also showed that more than 90% and probably more than 99% of the
droplets produced at these low suspension concentrations contain no
spheres when the singlet ratio, R, is taken at 0.95. These empty drop-
lets dry to form ultrafine particles of residue of the impurities in the
liquid; these secondary aerosols can be undesirable since they are con-
siderably more numerous than the monodisperse particles themselves. 126
Polystyrene and Polyvinyl Toluene Latex
Monodisperse suspensions of polystyrene and polyvinyl toluene latex
spheres (Dow Diagnostics, Indianapolis, Indiana) have been available in
sizes from 0.088 urn to 3.5 urn and have been widely used for preparing
monodisperse aerosols. 12^- 126 Their monodispersity of size is excellent
with an < 1.05. However, many problems have been encountered in their
use. 134- 136 The density of polystyrene latex has been measured to aver-
age 1.05 g/cm3 and the density of polyvinyl toluene latex at 1.027
g/cm3.137 These particles are grown by emulsion polymerization and are
stabilized in aqueous suspensions with an anionic surfactant having an
active sulfonate radical. These particles carry a negative charge in
aqueous suspensions, which contributes to their stability. The composi-
tions of some typical latex suspensions are given in Table 2. Solids
normally represent 10% of the aqueous suspension as supplied by the
manufacturer. Using equation 8, the dilutions of 10% by volume stock
suspensions of monodisperse spheres required to generate 95% single par-
ticles are shown in Figure 15 for various log-normal droplet distrib-
utions and sphere diameters.
Table 2
Relative Concentrations of Solids in Latex Suspensions
138
Reference
Number
LS-040-A
LS-052-A
LS-057-A
LS-63-A
LS-449-E
LS-464-E
Particle
Pi ameter
0.088
0.126
0.264
0.557
0.796
1.305
Composition of So1ids,%
Emulsifier Inorganics
92.59
97.09
98.23
98.78
99.04
99.15
5.56
2.43
1.13
0.26
0.25
0.29
1.85
0.48
0.64
0.95
0.70
0.56
If the 0.264 urn particles, (batch LS-057-A in Table 2) were generated
with a Lovelace nebulizer (VMD = 5.8 ym, ag = 1.8) with no dilution,
the 0.177% emulsifier and inorganics in the suspension would increase
the diameter of a single particle which happens to be aerosolized in a
5.8 ym droplet from 0.264 pm to 0.714 ym. Proper dilution to 30,000 to
1 to generate an aerosol with 95% singles (Equation 8 and Figure 15)
-------
IO.Q
K)
10'
10* I04
DILUTION RATIO, y
10*
10*
Figure 15. The dilution ratio, y, required to generate a
singlet ratio (R) equal to 0.95 vs. the sphere diameter
from stock of 10% spheres by volume for various values of
the volume median diameter (VMD) and geometric standard
deviation (ag) of the droplet distribution. The lines were
derived by the empirical equations and the points calculated
numerically by a theoretical equation (from Raabel26).
makes a 0.264 \im particle aerosolized in a 5.8 pm droplet yield a latex
particle essentially unchanged in diameter. For the larger particles,
it is convenient to separate the particles from the liquid by centrifu-
gation and resuspend them in purified and filtered water before use.
Monodisperse aerosols of polystyrene and polyvinyl toluene particles are
highly charged when generated by nebulization.139-140 A reduction of
charge to Boltzmann equilibrium with a suitable aerosol neutralizer is
absolutely essential in studies using these aerosols.33 The presence of
a background aerosol produced from the numerous empty droplets must be
given appropriate consideration as well.134-136
Monodisperse Particles Separated by Centrifugation
Another method for obtaining a suspension of monodisperse particles for
nebulization is to generate a suitable polydisperse aerosol of insoluble
particles and aerodynamically separate the particles into monodisperse
size groups using a spiral-duct aerosol centrifuge. Kotrappa and Mossl^
introduced this technique and Kotrappa et aU42 u$ed it to prepare
monodisperse particles of 239puo2. Raabe et all43 produced 238puo2 with
84
-------
this method and described an elaborate system for separating monodisperse
particles of Pu02 which simultaneously used four Lovelace Aerosol Par-
ticle Separators.144 Particles are collected on stainless-steel foils
which are cut into small segments for suspension in water. Dilute solu-
tions of various surface active agents and dilute NH40H have been used
to stabilize aqueous suspensions. Samples of the polydisperse primary
aerosol and three monodisperse size groups of Pu02 are shown in Figure
16. A monodisperse aerosol of Pu02 generated by nebulization of an
aqueous suspension of monodisperse particles is shown in Figure 17.
Note the presence of some doublets and the ultrafine background aerosols;
cascade impactor samples demonstrate that these background particles do
not contain plutonium and that the alpha activity distribution is in
fact monodisperse with regard to aerodynamic size.
' ' '
* * . •
Figure 16. Composite of electronmicrographs of four aerosol
samples of Pu02 particles: (A) primary polydisperse aerosol
of Pu02 collected with an electrostatic precipitator and (B),
(C) and (D) monodisperse particles collected at three posi-
tions on the foil of a spiral-duct centrifuge (Lovelace Aer-
osol Particle Separator )(from Raabe et al!43).
85
-------
Figure 17. Electronmicrograph of
a sample of an aerosol of Pu02
(shadowed with chromium vapor)
generated by nebulization of an
aqueous suspension of separated
monodisperse particles as shown
in Figure 16. Note the presence
of some doublets and the back-
ground aerosol of ultrafine par-
ticles which are not Pu02 but are
formed from trace impurities in
"empty" water droplets (from
Raabe et a
j*m
CONTROLLED CONDENSATION
The isothermal growth of droplets under supersaturated conditions by va-
por diffusion and condensation causes the droplet surface area to in-
crease linearly with time as given by Wilson and LaMer.146
r2 = r
bt
do)
with r the droplet radius at the time t, r0 the initial radius and b may
be maintained nearly constant under controlled conditions. Hence, if
droplets are grown by controlled condensation onto two small nuclei of
widely different size, say radii 0.02 ym and 0.1 ym, with bt = 0.164
ym2 the droplets formed will be almost of identical size with radii of
0.40 ym and 0.41 ym, respectively. Sinclair and LaMerl47 used this
principle to design an apparatus for generating monodisperse droplet aer-
osols of such materials as oleic acid, stearic acid, lubricating oils,
menthol, dibutyl phthalate, dioctyl phthalate and tri-o-cresyl phos-
phate. Many researchers have used and improved on the basic Sinclair-
LaMer generator*^-159 an(j tne basic technique has also been used to
produce solid aerosols by sublimation.1^0-161 Heyder et al,162 Muir and
Daviesl63 and Davies et al!64 have used this method to produce monodis-
perse aerosols of di-2-ethylhexyl sebacate for inhalation studies with
humans. Giacomelli-Maitorn" et al have made monodisperse carnauba wax
particles for human deposition studies.165
86
-------
GENERATION OF UNIFORM DROPLETS
General
Monodisperse aerosols of both soluble and insoluble forms can be pro-
duced from solutions or suspensions if small uniform droplets can be
dispersed. These droplets can then be dried and, if desired, altered in
character by heat treatment to yield the required aerosols. Many devices
have been designed to produce small uniform droplets. Two of these, the
vibrating reed!66 and vibrating thread,!67 have created only moderate
interest because of the low concentrations available. Liu and associates
have developed a method for separating a chosen monodisperse aerosol
fraction of a polydisperse aerosol using an electrostatic mobility
analyzer.168-170
Spinning Disk
The most popular device for dispersing uniform droplets has been the
spinning disk aerosol generator first described by Walton and Prewett.171
They observed that primary droplets thrown off at the perimeter of a
spinning disk were of uniform size. Liquid is fed to the center of the
disk and flows to the edge by centrifugal forces where it accumulates
until the centrifugal force, which increases with increasing liquid at
the edge, overcomes the surface tension and disperses the liquid. This
dispersion also produces some secondary fragments (satellite droplets)
which are easily separated dynamically from the larger primary droplets.
This is usually done by a separate flow of air near the disk, into which
the satellites move but beyond which the larger primary droplets are
thrown. A spinning disk generator, shown schematically in Figure 18,
can attain disk speeds up to 100,000 rpm. The drop diameter, Dd, pro-
duced by a spinning disk is given theoretically by:
with T the surface tension, p the fluid density, d the disk diameter, w
the speed of angular rotation of the disk and K a constant given theo-
retically by / 12 which varies in practice from 2 to 7. Many investiga
tors have developed and successfully used spinning disk monodisperse
aerosol generators for a variety of experimental applications including
aerosol studies and inhalation experiments.172-179 yne spinning top
generator introduced by K. R. May,180-181 works on essentially the same
principle as the Walton and Prewitt spinning disk. It, too, has been
successfully used by many investigators.182-194
87
-------
Liquid Feed
'quid
Figure 18. Schematic drawing
of a Spinning Disk Generator
used to produce monodisperse
aerosols of both soluble and
insoluble forms from solutions
of suspensions. Air flow into
the satellite collector is
adjusted so that the inertia
of the primary particles allows
them to enter the main air flow.
Spinning Disk
/(IOO.OOO rpm)
Primary
'Droplets
Satellite
Droplets
Satellite
Air Flow
Although the spinning disk and top have probably been the most generally
successful methods for producing monodisperse aerosols, the production
of particles smaller than 0.5 ym geometric diameter is not practical
because droplets produced with these devices are generally in the 20 to
30 ym diameter range and even purified water has trace impurities which
may yield resultant particles as big as 0.5 ym. Monodisperse aerosols
produced with spinning disk and spinning top devices are usually near or
larger than 3.5 ym in aerodynamic equivalent diameter. As with other
methods, the resulting aerosols are highly charged and must be passed
through a suitable neutralizer soon after being produced.
Electrostatic Dispersion of Liquids
Electrostatic dispersion1^ provides another method for generating uni-
form droolets for certain solutions. Vonnegut and Neubauer1^ an(j
othersl9/-19° have studied this approach using a filament of electri-
cally charged liquid released from a small capilllary. The filament
breaks up into fragments, forms a conical spray and further breaks up
into small droplets of almost uniform charge and size. These droplets
must be discharged soon after generation and are difficult to control
because of electrostatic behavior.
88
-------
Periodjc Dispersion of Liquid Jets
If a thin liquid stream is emitted
from an orifice under pressure,
this stream is by nature unstable
and will soon disintegrate into
droplets by the action of any ex-
ternal forces as described by
Rayleigh.34 Tne collapse of such
a stream into very uniform drop-
lets is easily attainable with
the application to the stream of
a periodic vibration of suitable
amplitude and frequency. In
1931 Castleman35 demonstrated
this principle experimentally
using stroboscopic and spark pic-
tures (Figure 19). Castleman
principle that the
cylindrical stream
"be made so beauti-
that it may be
by stroboscopic
described the
collapse of a
of liquid can
fully regular
readily viewed
means if the chamber from which
the jet issues be influenced by
Figure 19. Early observation
of the principle of periodic
dispersion of a liquid jet in-
to uniform droplets as photo-
graphed by Castleman35 in 1930
using a spark illumination
technique (from Castleman35).
a periodic vibration of proper
amplitude and frequency".35
Although sometimes called a vib-
rating orifice generator, this
description is too specific since
the disintegration occurs in the
same manner irrespective of where the oscillatory vibrations are imparted
to the liquid, and may even be accomplished at a distant reservoir.199
Although the rate of disruption of a liquid stream depends upon both
the viscosity and surface tension of the liquid, the characteristics of
the droplets formed depend only upon the diameter of the stream and the
frequency of the vibrations. This is true because the minimum length
of such a stream that can be separated into a separate droplet is equal
to the circumference of the stream; a shorter length cannot be separated
since the surface tension along the circumference of the stream will
tend to elongate a shorter segment and prevent separation. Hence the
smallest segment of a stream which can be separated has a length, Lm-jn,
given by:200-201
•mm
= irD/
(12)
89
-------
with Dg the diameter of the stream. For most liquids, it is satisfacto-
ry to assume that the stream diameter is equivalent to the orifice diam-
eter. If the liquid is emitted with a flow rate, Q, and the frequency
of periodic disturbance is f, then a spherical droplet formed of each
separated cylindrical segment will have a diameter Dj given by:
(13)
irf
The frequency required to produce the minimum droplet size is given by
a maximum frequency, fmax> for disruption of the stream:
. 4Q
Any frequency of disturbance less than fmax will disrupt the stream into
uniform droplets. However, Castleman35 showed that the optimum disturb-
ance cuts off a length of stream given by:
L
opt
and therefore the optimum frequency of disruption is
fopt =
4Q
and the minimum diameter droplet is given by:
D
mn
l.77D
90
-------
and the optimum diameter is given by:
D,
'opt
1.88 a
(18)
Hence the minimum or optimum diameters of droplets that can be generated
in practice depend only on the stream diameter (orifice diameter).
Measurement of volumetric flow rate from the orifice, Q, can be made for
any combination of orifice diameter and reservoir pressure, so that
droplet sizes can be calculated.
Since Castleman's work was reported in 1931, this principle has been
rediscovered and studied by many investigators.200-212 Fulwyler 202
reported the use of a device of this type as a cell separator. Fulwyler
and Raabe,200 Raabe^ and Raabe and Newton201 described the application
of the Fulwyler droplet generator for producing monodisperse aerosols.
In the Fulwyler device an audio oscillator and amplifier provide a high
frequency power signal which is converted to mechanical vibrations by an
ultrasonic transducer linked by a
coupling rod to a small liquid reser- STROBE
voir. The reservoir is pressurized LIGHT .
(= 30 psig) to emit the liquid through
a small orifice (- 10 ym) as a fine
stream. This stream is uniformly dis-
rupted by the ultrasonic vibrations
into droplets that vary less than 1%
in volume. To produce an aerosol of
these uniform droplets, they must be
disturbed from their forward direction
of motion so that their forward speed
is reduced without coalescence. This
can be accomplished by concentric
mixing or cross flow of air streams.
TELESCOPE
|-*|STROBE
-LIQUID
PIEZO-
f ELECTRIC
CRYSTAL
A block diagram of the operation of the
Fulwyler droplet generator is shown in
Figure 20 and the experimental arrange-
ment used by Raabe and Fulwyler200 anc|
Raabe and Newton201 for dispersing the
aerosol is shown in Figure 21. Monodis-
perse particles of Zr02 produced from
monodisperse droplets containing Zr
oxalate and heat treated at 1150 °C are
shown in Figure 22.76
Figure 20. Block diagram of
equipment used for operation
of the Fulwyler Droplet Gen-
erator for producing monodis-
perse droplets by periodic
dispersion of aliquid jet
(from Fulwyler and Raabe200).
91
-------
Atrosol
Uniform
Ittt
Air
\-~-Uquld F»»d
(SOfflg)
Pimeottictric
Cryttal
IOO Klu Signal
Figure 21. Schematic drawing
of the Fulwyler Droplet Gen-
erator used to produce mono-
disperse droplets to yield
monodisperse aerosols of both
soluble and insoluble materials.
Air is directed at the stream
of droplets to perturb them out
of alignment and create the
droplet cloud (from Fulwyler and
Raabe200).
Figure 22. Uniform spherical
aerosol particles of Zr02
produced with the Fulwyler
Droplet Generator from solu-
tions of zirconium oxalate
degraded to the oxide with
a heating column by the
method of Kanapilly et al&7
(from Fulwyler and Raabe200).
92
-------
Generation of monodisperse aerosols with similar devices using electro-
constructive elements around the orifice rather than an ultrasonic trans-
ducer has been described by Strom204 and by Berglund and Liu.211
One operational problem that exists with the periodic dispersion method
of producing aerosols of fine particles is that the small orifices
required an subject to clogging if small particles are present in the
liquid; membrane filtration of the liquid is essential but doesn't corn-
Since the minimum practical orifice diameter is in the range 8 ym to 15
urn, the smallest droplets that can be produced by periodic dispersion
of liquid jets are in the range 14 ym to 27 ym, which is about the same
range of droplet sizes as produced with spinning disks and spinning tops.
Hence, as with the spinning disk, it is difficult to produce solid par-
ticles smaller than 0.5 ym geometric diameter with periodic dispersion
devices, and most useful aerosols will be at the high end and above the
upper limit of the definition of fine particles.
ACKNOWLEDGEMENT
The author is indebted to Dr. Brian Mokler and Mr. George Newton for
beneficial advice, to Mr. Emerson Goff and Mr. Michael Rios for prepara-
tion of illustrations, to Mr. Fred Rupprecht for editorial and composi-
tion assistance, to Mr. David Velasquez for proofreading and to Mrs.
Judith Miller for typing the manuscript. This report was prepared under
U. S. Energy Research and Development Administration Contract E(29-2)-
1013.
REFERENCES
1. Megaw, W. J. and R. D. Wiffen. The Generation of Condensation
Nuclei by Ionising Radiation. Geofis Pura Appl 50:118-128, 1961.
2. Renzetti, N. A. and G. J. Doyle. Photochemical Aerosol Formation
in Sulfur Dioxide-Hydrocarbon Systems. Intern J Air Water Pollu-
tion. 2:327-345, 1960.
3. Madelaine, G. J. Formation et e'volution des aerosols dans 1'air
filtre' et dans Tair nature! Action de la radioactivite', Doctoral
Thesis, Commissariat a 1'Energie Atomique (France) CEA-R-3614, 1968.
4. Bricard, J., F. Billard and G. Madelaine. Formation and Evolution
of Nuclei of Condensation that Appear in Air Initially Free of Aer-
osols. J Geophys Res. 73:4487-4496, 1968.
5. Knott, M. J. and M. Malanchuk. Analysis of Foreign Aerosol Produced
in N02-Rich Atmospheres of Animal Exposure Chambers. Am Ind Hyg
Assoc J. 30:147-152, 1969.
93
-------
6. Pfefferkorn, G. Photochemische Bildung SchwerflUchtiger Trbpfchen
aus organischen Dampfen in Luft. Staub Reinhaltung Luft. 27:138-
140, 1967.
7. Fuchs, N. A. and A. G. Sutugin. Generation and Use of Monodisperse
Aerosols. In: Aerosol Science, Davies, C. N. (ed.). New York,
Academic Press, Inc., 1966, pp. 1-30.
8. Raabe, Otto G. Generation and Characterization of Aerosols. Inha-
lation Carcinogenesis, pp. 123-172, U.S.A.E.C. Division of Technical
Information, Oak Ridge, Tennessee, April 1970.
9. Mercer, T. T., M. I. Tillery and H. Y. Chow. Operating Character-
istics of Some Compressed Air Nebulizers. Am Ind Hyg Assoc J.
29:66-78, 1968.
10. Raabe, 0. G. Operating Characteristics of Two Compressed Air Nebu-
lizers Used in Inhalation Experiments. Fission Product Inhalation
Program Annual Report, 1971-1972, LF-45, Lovelace Foundation,
Albuquerque, New Mexico, pp. 1-6, November 1972.
11. Liu, Benjamin Y. H. and K. W. Lee. An Aerosol Generator of High
Stability. Amer Ind Hyg Assoc J. In press, 1976.
12. Berke, Harry L. and T. E. Hull. An Improved Aerosol Generator.
Amer Ind Hyg Assoc J. 36:43-48, 1975.
13. Mugele, R. A. and H. D. Evans. Droplet Size Distribution in Sprays.
Ind Eng Chem. 43:1317-1324, 1951.
14. LaMer, V. K. and R. Gruen. A Direct Test of Kelvin's Equation
Connecting Vapour Pressure and Radius of Curvature. Trans Faraday
Soc. 48:410-415, 1952.
15. Straubel, Harold. Condensation, Evaporation, Crystallization and
Charge Changes of Solution Droplets. Staub Reinhaltung der Luft in
English. 33:179-182, April 1973.
16. Orr, C., F. K. Hurt and W. J. Corbett. Aerosol Size and Relative
Humidity. J Colloid Sci. 13:472-482, 1958.
17. Pilat, M. J. and R. J. Charlson. Theoretical and Optical Studies
of Humidity Effects on the Size Distribution of Hygroscopic Aerosol.
J Rech Atmospheriques. 2:165-170, 1966.
18. Morrow, P. E. (Chairman, Task Group on Lung Dynamics). Deposition
and Retention Models for Internal Dosimetry of the Human Respira-
tory Tract. Health Phys. 12:173-208, 1966.
94
-------
19. Snead, C. C. and J. T. Zung. The Effects of Insoluble Films Upon
the Evaporation Kinetic of Liquid Droplets. J Colloid Interface
Sci. 27:25-31, 1968.
20. Anshus, Byron E. The Effect of Surfactants on the Breakup of Cyl-
inders and Jets. J Colloid and Interface Sci. 43:113-121, April
1973.
21. Nelson, P. A. and W. F. Stevens. Size Distribution of Droplets
from Centrifugal Spray Nozzles. A I Ch E Journal. 7:80-86, 1961.
22. Lewis, H. C. D. G. Edwards, M. J. Goglia, R. I. Rice and L. W.
Smith. Atomization of Liquids in High Velocity Gas Streams. Ind
Eng Chem. 40:67-74, 1948.
23. Mercer, T. T., R. F. Goddard and R. L. Flores. Output Character-
istics of Several Commercial Nebulizers. Ann Allergy. 23:314-326,
1965.
24. Hatch, T. and S. P. Choate. Statistical Description of the Size
Properties of Non-Uniform Particulate Substances. J Franklin Inst.
207:369-387, 1929.
25. Raabe, Otto G. Particle Size Analysis Utilizing Grouped Data and
the Log-Normal Distribution. Aerosol Sci 2:289-303, 1971.
26. Mercer, T. T. Aerosol Production and Characterization - Some Con-
siderations for Improving Correlation of Field and Laboratory
Derived Data. Health Phys. 10:873-887, 1964.
27. Matteson, Michael J. The Separation of Charge at the Gas-Liquid
Interface by Dispersion of Various Electrolyte Solutions. J Colloid
and Interface Sci. 37:879-890, December 1970.
28. Rayleigh, Lord. On the Equilibrium of Liquid Conducting Masses
Charged with Electricity. Phil Mag. 14:184-186, 1882.
29. Whitby, K. T. and B. Y. H. Liu. The Electrical Behavior of Aero-
sols. In: Aerosol Science, Davies, C. N. (ed.). New York, Aca-
demic Press, Inc., 1966, pp. 59-86.
30. Doyle, Arnold, D. Read Moffett and Bernard Vonnegut. Behavior of
Evaporating Electrically Charged Droplets. J Colloid Sci. 19:136-
143, 1964.
31. Whitby, K. T. Generator for Producing High Concentrations of Small
Ions. Rev Sci Instr. 32:361-355, 1961.
95
-------
32. Soong, A-L. The Charge on Latex Particles Aerosolized from Suspen-
sions and Their Neutralization in a Tritium De-Ionizer. M. S.
Thesis, University of Rochester, Rochester, New York, 1968.
33. Liu, Benjamin Y. H. and David Y. H. Pui. Electrical Neutralization
of Aerosols. Aerosol Sci. 5:465-472, 1974.
34. Rayleigh, Lord. On the Instability of Jets. London Math Soc Proc.
10:4-13, 1878-1879.
35. Castleman, R. A. The Mechanism of the Atomization of Liquids. Bur
Std J Res. 6:369-376, 1931.
36. Schweitzer, P. H. Mechanism of Disintegration of Liquid Jets. J
Appl Phys. 8:513-521, 1937.
37. Merrington, A. C. and E. G. Richardson. The Break-Up of Liquid
Jets. Proc Phys Soc (London). 59:1-13, 1947.
38. Bitron, Mosche D. Atomization of Liquids by Supersonic Air Jets.
Ind and Eng Chem. 47:23-28, 1955.
39. Raabe, Otto G. Analysis of the Operation and Behavior of Compressed
Air Nebulizers. Unpublished, 1975.
40. Mercer, T. T., M. I. Tillery and M. A. Flores. Operating Charac-
teristics of the Lauterbach and Dautrebande Aerosol Generators.
AEC Research and Development Report LF-6, Lovelace Foundation for
Medical Education and Research, Albuquerque, New Mexico, 1963.
41. Mercer, T. T., R. F. Goddard and R. L. Flores. Output Characteris-
tics of Three Ultrasonic Nebulizers. Ann Allergy. 26:18-27, 1968.
42. Mercer. T. T., R. F. Goddard and R. L. Flores. Effect of Auxiliary
Air Flow on the Output Characteristics of Compressed-Air Nebulizers.
Ann Allergy. 27:211-217, 1969.
43. Dautrebande, L. Microaerosols. New York, Academic Press, Inc.,
1962. pp. 7-22.
44. Dautrebande, L., H. Beckmann and W. Walkenhorst. New Studies on
Aerosols. III. Production of Solid, Small-Sized Aerosols. Arch
Int Pharmacodyn. 66:170-186, 1958.
45. Lauterbach, K. E., A. D. Hayes and M. A. Coelho. An Improved
Aerosol Generator. A M A Arch Ind Health. 13;156-160, 1956.
96
-------
46. Collison, W. E. Inhalation Therapy Techniques, Heinemann, London,
1935.
47. May, K. R. The Collison Nebulizer: Description, Performance and
Application. J Aerosol Sci. 4:235-243, 1973.
48. Wright, B. M. A New Nebulizer. Lancet. 2:24-25, July 5, 1958.
49. Geoffrion, Louis A. Evaluation of Sodium Chloride Aerosols Gener-
ated for Filter and Respirator Studies. American Industrial Hygiene
Conference, Miami Beach, Florida, May 12-17, 1974.
50. Williams, Truman. Personal Communication from Frontier Enterprises,
Albuquerque, New Mexico.
51. Newton, G. J., J. E. Bennick and S. Posner. A Miniature High Out-
put Aerosol Generator. In: Selected Summary of Studies on the
Fission Product Inhalation Program from July 1965 Through June
1966. AEC Research and Development Report LF-33, Lovelace Founda-
tion for Medical Education and Research, Albuquerque, New Mexico.
1966. pp. 29-35.
52. Burns, H. L. Nebulizer Evaluation by the Dew Point Method. Re-
spiratory Therapy. Sept./Oct. 1972.
53. Babington, R. S., W. R. Slivka and A. A. Yetman. Method of Atom-
izing Liquids in a Mono-Dispersed Spray. U. S. Patent 3421692.
January 14, 1969.
54. Babington, R. S., W. R. Slivka and A. A. Yetman. Apparatus for
Spraying Liquids in Monodispersed Form. U. S. Patent 3421699.
June 14, 1969.
55. Babington, R. S. Fuel Burner. U. S. Patent 3425058. January 28,
1969.
56. Raabe, Otto G. Report on Prototype Model of Owens-Illinois Hydro-
sphere Nebulizer. Unpublished Report to Owens-Illinois, March 1973.
57. Sollner, K. The Mechanism of the Formation of Fogs by Ultrasonic
Waves. Trans Faraday Soc. 32:1532, 1936.
58. Antonevich, J. N. Ultrasonic Atomization of Liquids. In: Proceed-
ings of the National Electronics Conference. Chicago, Illinois,
October 7-9, 1957, pp. 798-807.
97
-------
59. Muromstev, S. N. and V. P. Nenashev. The Study of Aerosols III.
An Ultrasonic Aerosol Atomizer. J Microbiol Epidemiol Immunobiol
(USSR) (English Trans!.). 31:1840-1846, 1960.
60. II'in, B. I. and 0. K. Eknadiosyants. Nature of the Atomization of
Liquids in an Ultrasonic Fountain. Soviet Phys Acoust (English
Trans!.). 12:269-275, 1967.
61. Boucher, R. M. G. and J. Kreuter. The Fundamentals of the Ultra-
sonic Atomization of Medicated Solutions. Ann Allergy. 26:591-600,
1968.
62. Denton, M. B. and D. B. Swartz. An Improved Ultrasonic Nebulizer
System for the Generation of High Density Aerosol Dispersion. Rev
Sci Instrum. 45:81-83, January 1974.
63. Newton, G. J. Lovelace Foundation for Medical Education and Re-
search Unpublished Observation. 1968.
64. Goddard, Roy F., T. T. Mercer, Peter X. F. O'Neill, Richard L.
Flores and Ray Sanchez. Output Characteristics and Clinical Effi-
cacy of Ultrasonic Nebulizers. J Asthma Research. 5:355-368, June
1968.
65. Reif, Arnold E. and Margot Holcomb. Operating Characteristics of
Commercial Nebulizers and Their Adaptation to Produce Closely Sized
Aerosols. Ann Allergy. 16:626-638, 1958.
66. Raabe, Otto G. The Adsorption of Radon Daughters to Some Polydis-
perse Submicron Polystyrene Aerosols. Health Phys. 14:397-416,
1968.
67. Kanapilly, G. M., 0. G. Raabe and G. J. Newton. A New Method for
the Generation of Aerosols of Insoluble Particles. Aerosol Sci.
1:313-323, 1970.
68. Gibb, F. R. and P. E. Morrow. Alveolar Clearance in Dogs After
Inhalation of an Iron-59 Oxide Aerosol. J. Appl Physiol. 17:429-
432, 1962.
69. Ramsden, D. and D. A. Waite. Inhalation of Insoluble Iron-Oxide
Particles in the Submicron Range, Assessment of Radioactive Contam-
ination in Man. International Atomic Energy Agency, Vienna, 1972,
pp. 65-81.
70. McKnight, M. E. and R. W. E. Norgan. A Study of the Exchange Char-
acteristics of Montmorillonite Clay for Fission Product Cations for
98
-------
Use in the Generation of Insoluble Aerosols. AEC Research and De-
velopment Report LF-37, Lovelace Foundation for Medical Education
and Research, Albuquerque, New Mexico, 1967.
71. Craig, D. K. An Investigation of the Interactions that Occur Be-
tween Radionuclides and Aerosols in the Respirable Size Range.
Health Phys. 12:1047-1067, 1966.
72. Stern, S. C., J. S. Baumstark, A. I. Schekman and R. K. Olson. A
Simple Technique for Generation of Homogeneous Millimicron Aerosols.
J Appl Phys. 30:952-953, 1959.
73. Willard, D. H., W. J. Bair, L. A. Temple and V. H. Smith. Tech-
niques for Exposure of Mice to Aerosols of Radioactive Particles,
HW-52368, Hanford Atomic Product Operation, Richland, Washington,
November 1958.
74. Walkenhorst, Willhelm. Uber die Herstellung eines Testaerosols aus
Glaskugeln. Staub 24:97-98, 1964.
75. Kanapilly, G. M., Otto G. Raabe and G. J. Newton. Physical and
Chemical Characteristics of Aerosols Produced from Solutions of
Polyvalent Metallic Oxides. Proceedings of the Health Physics
Society Midyear Topical Symposium, Idaho Falls, Idaho, 1971. pp.
578-589.
76. Raabe, Otto G., George M. Kanapilly and George J. Newton. New
Methods for the Generation of Aerosols of Insoluble Particles for
Use in Inhalation Studies. Inhaled Particles III. Surrey, England,
Unwin Brothers, 1971. pp. 3-17.
77. Horstman, Sanford, William Barkley, Edwin Larson and Eula Bingham.
Aerosols of Lead, Nickel and Cadmium. Archives Environ Health.
26:75-77, 1973.
78. Arvik, Jon H. and Zimdahl, R. L. An Apparatus for the Exposure of
Plants to Selected Particulate Aerosols. Rev Sci Instrum. 44:800-
802, 1973.
79. Werly, Emil F. and Elven K. Bauman. Production of Submicron Powder
by Spray-Freezing. Archives Environ Health. 9:567-571, 1964.
80. Sanders, Paul A. Principles of Aerosol Technology, New York, Van
Nostrand Reinhold Co., 170. 418 p.
81. Shepherd, H. R. Aerosols: Science Technology. New York, Intersci-
ence Publishers, 1961. 548 p.
-------
82. Sciarra, John J. and Leonard Stoller. The Science and Technology
of Aerosol Packaging. New York, John Wiley and Sons, 1974. 710 p.
83. Polli, Gerald P., Wayne M. Grim, Frederick A. Bacher and Martin H.
Yanker. Influence of Formulation on Aerosol Particle Size. J
Pharm Sci. 58:484-486, 1969.
84. Nelson, Sidney W. and Anthimos J. Christoforidis. An Automatic In-
halation-Actuated Aerosol Anesthesia Unit: A New Method of Applying
Topical Anesthesia to the Oropharynx and Tracheobronchial Tree.
Radiology. 82:226-293, 1964.
85. Vos, Kenneth D. and David B. Thompson. Particle Size Measurement
of Three Commercial Pressurized Products. Aerosol Age. 20:31-32,
34, 52, 54, 56, 58, 1975.
86. Vos, Kenneth D. and David B. Thompson. Particle Size Measurement
of Eight Commercial Pressurized Products. Powder Technology. 10:
103-109, 1974.
87. Davies, C. N. Book Review of Principles of Aerosol Technology.
Aerosol Science. 2:125, 1971.
88. Gordieveff, V. A. Studies on Dispersion of Solids as Dust Aerosols.
A M A Arch Ind Health. 15:510-515, 1957.
89. Bair, W. J. and B. J. McClanahan. Plutonium Inhalation Studies.
Arch Environ Health. 2:48-55, 1961.
90. Bair, W. J., D. H. Willard, J. P. Herring and L. A. George II. Re-
tention, Translocation and Excretion of Inhaled 239pu02- Health
Phys. 8:639-649, 1962.
91. Wright, B. M. A New Dust-Feed Mechanism. J Sci Instrum. 27:12-15,
1950.
92. Dimmick, R. L. Jet Disperser for Compacted Powders in the One-to-
Ten Micron Range. A M A Arch Ind Health. 20:8-14, 1959.
93. Ebens, Rinze and Maria Vos. A Device for the Continuous Metering
of Small Dust Quantities. Staub-Reinhalt Luft in English. 28:24-
25, 1968.
94. Fuchs, N. A. and F. I. Murashkevich. Laboratory Powder Dispenser
(Dust Generator). Staub-Reinhalt Luft in English. 30:1-3, 1970.
95. Drew, Robert T. and Sidney Laskin. A New Dust-Generated System for
Inhalation Studies. Amer Ind Hyg Assoc J. 32:327-330, 1971.
100
-------
96. Willeke, K., C, S. K. Lo and K. 1. Whltby. Dispersion Character-
istics of a Fluidized Bed. Aerosol Sci 5:449-455, 1974.
97 GUI chard, J. C, and J. I, Magru-' Application du lit Fluidise en
M'il-'eu Gazeux a la Generation vies Aerosols. Note Interieure No, 5i
IRCHA, VERT-le-Petit, France, 1967.
98. Pfe^ferkorn, Gerhard. Photochemsche Bildung Scnwerfliiohtiger
Troofchen aus Organischer, Dh'mpfer! in Luft. Staub-ReinhaTt. Lufr,
27-138-140, 1967.
99. Mercer, 1, T. and M, 1, Tiller/. Coagulation Rates of Particles
Produced in Air by Thoron. J Colloid Interface Sci. 37:785-792,
1971.
100. Coachman, J. C. Metallic Hicrosphere Generation, ilG&G., Inc.
S.jnta Barbara, Ca1ifornia4 1966.
101. Goldsmith, P., A. C. Wells and L. C, Cox. Aerosol Emission from
Chromium Alloys wh<>n Heated in Air and in Carbon Dioxide/1% Carton
Monoxide. AERE-R5426, Atomic Energy Research Establishment,
Harwell, U.K., 1968. 20 p.
IC2, Goldsmith, P., F. &. May and R, 0. Viiffen. Chromium Trioxide Acre-
so"! from Heated 80:20 Nickel Chromium Wire. Nature. 210:475-477,
April 30, 1966.
103. Malwald, Elenor. Ein ^olframoxld - Testaerosol. Staub. 25:535-
537, 1965.
i04. Spurny, Kvetoslav and James P. Lodge, Jr. Production of Highly
Dispersed Model Aerosols. Staub-Reinhalt Luft in English. 33:171-
173, 1973.
:05. Karioris and 8. R. Fish. An Exploding Wire Aerosol Generator. J
Colloid Sci. 17:155-161, 1962.'
106. Phdlen, Robert F. Evaluation of an Exploded-Wire Aerosol Generator
for Use in inhalation Studies. Aerosol Sci. 3:395-406, 1972.
107. Chace, W. G. and H. K. Moore (eds). Exploding Wires. New York,
Plenum Press, (2 volumes), 1962.
108. Holmgren, J. D., J. 0. Gibson and C. Sheer. Some Characteristics
of Ate Vaporized iubmicron Particulates. J Elecfrochem Soc. III.
362-369, 1964.
101
-------
109. Pfender, E. and C. V. Boffa Generation of Hitrafme Aerosols Uitn
a Transpiration Labeled Anode in ? Hioh Intensity Arc. Rev Sci
Instrum. 4:655-657, 1970.
110. Pfefferkorn, Gerhard and Hans Pesk-r, Investigating Smoke? Geierv -..•,
in Electric-Arc Weldinq, Staub-Rtinhall; i.uft in FnQlish. r7:i-:~,
1967.
111. Spurny, Kvetoslav R. and James P. Lodge, Jr. Padioactively Lflbe'ec;
Aerosols. Atmos Environ. 2.-429-44G, 1968,
112. Nelson, L. S. Explosion of Burning /ircormni Droplets Caused by
Nitrogen. Science. 148:1594-1595, 1965.
113. Nelson, Lloyd 5., Herman S. Levine, Daniel I. Po-.ner and Shelby r.
Kurzius. The Combustion of Zirconium Droi/lor,: ir Oxygen -'"Pare C«s
Mixtures, High Temp. Sci, 2::c4J-37h, 1970.
114. Sinclair, David and Lawrence Hinc'i^f fe Product '.or- and
of Submicron Aerosols. Irr A^.ess^ent of f/irbc-'ne Particles,
Mercer, 7. T. , P. E. Morrow and V. Sibber (ed<;. /. Sp^i'igf ieM, 11
Charles C. Thomas, 197?, pp. 18?- 1.9^
115. Chatfield, E. J. Some Studies of the Aerosol:, Produced by the
Combustion or Vaporization of Huroni jm-Aika! i Mstal Mixtures (..
and II). J Nuclear Materials. J^:228~?67, L969.
116, Polishchuk, D. I., V, L. Vel ikano'.c, :mJ I, ^ Necin'tailo. Co^bus •
tion of Mul ti component Alui'-nr 'j.i'-.'.oiiibubi :o:i Systems. li\: A<1vo;-. <:i,
in Aerosol Physics No. /L Feoo-.fe;," , V, fl. >, ed.}, Nev«r York, -ehn
Wiley and Sons, 1973.
117. Selivanov, S. E. Oxidation of Ku;,,t: :iui!' in Nitvo^ 'i/.idt-. In:
Advances in Aerosol Physics No. 4, '''-n^so^v ,. x', A (ed.). ''h.-H
York, John Wiley and SOPS, l^f'j, p; .
118. Fedoseev, V. A. Combustion jf /% '•o-ir:'!-; , I,i. A-Jv-jncvs in A^eso"'
Physics No. 4, Fedoseev, '••'. A :,-. -f ) '<:..jw Yo^k , John Wilev snd
Sons, 1973. pp. 55-C: .
119. Liu, B. Y. H. Methods of ijeneratio-s uf Mcr-odisperse Atrosois.
COO-1248-10, Department of Mechfinical Enginyi-r'nq, University of
Minnesota, Minneapolis, Mi fines jf :i, l?^"7,
120. Mercer, T. T. .-'eresoi Technoicg- in H?7=i>'
-------
121. Bradford, E. 8., J. W. Vanderhoff and T. Alfrey, Jr. The Use of
Monodisperse Latexes in an Electron Microscope Investigation of the
Mechanism of Emulsion Polymerization. J Colloid Sci. 11:135-149,
1956.
122. Bradford, E. B. and 0. W. Vanderhoff. Electron Microscopy of Mono-
disperse Latexes. J Appl Phys. 26:854-871, 1955.
123. Reist, P. C. and W. A. Burgess, Atomization of Aqueous Suspensions
of Polystyrene Latex Particles, J Colloid Interface Sci. 24:271-
273, 1967.
124. Langer, G. and J. H. p-ierrad. Anomalous Behavior of Aeroso'is Pro-
duced by Atomization of Monodisperse Polystyrene Latex, J Colloid
Sri. 18:95-97, 1963.
12j. Lichtenbelt, J. W. Th., C. Pathmamanoharah and P. H. Wiersema,
Kapt'd Coagulation of Polystyrene Latex in a Stopped-Flow Spectropho-
tometer. 0 Colloid Tnte;-face Sci. 49:281-285, 1974.
12G. Raabe, Otto G. The Dilution of Monodisperse Suspensions for Aero-
sol ization. Am Ind Hyg Assoc G. 29:439-443., 1968.
U77, Ottewill, R. H. and R. F. Woodbridge. Studies on the Electrokif-.t^
Behdvior of Monodisperse Silver Halifie Sols. J Colloid Sci. 19:
606-620, 1964,
128. Stober, Werner and Arthur Fink. Controlled Growth of Monodisperse
Silica Spheres in the Micron Size Range. J Colloid Interface Sri.
26:62-69, 1968.
129. Singer, Jacques, M., Care! J. Van Oss and John W. Vanderhoff.
Radioiodination of Latex Particles. J Reticuloendothelial Soc, 6:
281-286, 1969.
130. Wilkins, D. J. Fluorescent Labelling of Polystyrene Latex for
Tracing in Biological Systems. Nature. 202:798-799, 1964
i3i. Black, A. and M. Walsh. The Preparation of Bromine-82 and Iodine-
131 Labeled Polystyrene Microspheres with Diameters From 0 1 to 3C
Microns. Ann Occup Hyg. 13;87-100, 1970.
122. Bogen, Donald D. Preparation of RadioactiveLabeled Polystyrene
Latex Monodisperse Submicron Aerosols. Am Ind Hyg Assoc J, 31:
319-352, 1970.
103
-------
133. Flachsbart, H. and W. Stober, Preparation of Radicactively Labeled
Monodisperse Silica Spheres of Colloidal Size, J Colloid Interface
Sci. 30:568-573, 1969,
134. Fuchs, H, A. Latex Aerosols - Caution! Aerosol Sci. 4:405-41Q1
1973.
135. Langer, G. and A. Lieberman. Anomalous Behavior of Aerosols Pro-
duced by Atomization of Monodisperse Polystyrene Latex, J Colloid
Sci. 15:357-360, 1960.
136. Rimberg, D., R. Knuth and L. Hinchliffe. Detection of Emulsifiori
Particles in Latex Aerosols, J Colloid Interface Sci. 40:315-336,
1972.
137. Pugh, Thomas L. and Wilfried Heller. Density of "o'lysty-ene and
Polyvinyltoluene Latex. Particles. J Colloid ScL 1^:173-180,
1957.
138. Lippie, L. J. Letter to Aerosol Group at University of Rochester,
New York. The Dow Chemical Company, Midland, Michigan, January i7,
1966.
139. Whitby, Kenneth T. and Benjamin Y. H. Liu. Polystyrene Aerosol' -
Electrical Charge and Residue Size Distribjtion, Atmos Environ.
2:103-116, 1968.
140, Thomas, Jess W. and David Rimberg. Ein einfache Methode ?ur Mecsm^
der Mittlereri elektrischen Ladung eins Monodi -.p^rsen Aercso'i.
Staub-Reinhalt Luft. 27:354-'j?7. 1967.
14]. Kotrappa, P. and Owen R, Mor,-->. S-; oduction of Relatively Monool-:-
perse Aerosols for Inhalation Experiments by Aerosol Centrifugetior;.
Health Phys. 21:531-^35, 1971.
142, Kotrappa, P., C. J. Wilkinson ana H. A Ro/d. Technology for tnr-
Production ot Monodisperse Aeroso-s of Oxides of Transuranic Ele-
ments for Inhalation experiment: '^dltt i>^ys. ??:837-843, 197Z
143. Raabe, Otto G., Howard A. Boyn, George M. Kanapilly Charles J.
Wilkinson and George J. Newton. Development and Use^of a System
for Routine Production of Monod-coerse Parti',les of ?1&Pu()2 and
Evaluation of Gamma Emitting labels. Health Phys, In Press, 1975.
144. Kotrappa, P. and M. E. Light. O'-^ign anu Performance of the Love-
lace Aerosol Particle Separator. Rev Sci Ifitrum, 43:1106-1112,
1972.
-------
145. Raahe, 0. G.s ri. A. Boyd, J. A. Mewhinney, 0. -.I. Migllo and C. J,
W-'Ikinson. Production of Hone-disperse Aerosols of 239puO£ Labeled
with 51o. Fission Product Inhalation Program Annual Report, 19?)-
197?, LF-45, Lovelace Foundation for Medical Education and Research,
Albuquerque, New Mexico, pp. 21-28, 1972.
146, Wilson, Irwin B. and Victor K. LaMer. The Retention of Aerosol
Particles in the Human Respiratory Tract as a Function of Particle
Radius. J !nd Hyg Toxicc1. 30:265-280, 1948.
147. Sinclair, David and Victor K, LaMer. Light Scattering as a Measure
of Particle Size in Aerosols. Chem Rev. 44:245-267/1949.
148. LaMer, Victor K., Edward C. Y, Inn and Irwin B. Wilson. The Methods
of Forming, Detecting and Measuring the Size and Concentration of
Liquid Aerosols in the Size Range of 0.01 to O.H5 Microns Diameter.
J Colloid Sci. 5:471-497, 1950,
149, Rapaport, E. and S. E. Weinstock. A Generator for Homogeneous Aer-
osols. Experlentia. 11-363-364, 1955.
150. Lassen, Lars. A Simple Generator for the Production of Monodis-
perse Aerosols in the Size Range of 0.15 to 0.70 Microns (Parf^e
Radius). Z Anqew Phys. 4:1.57-159, 1960.
i5t. Movilliat, P. Production et observation d'aerosols Honodisperses
Application a S'etude de quelques proprietes physiques d'rtats
disperses, Ann Occupational Hyg. 4:275-294, 1962.
152. Matijevic, E.-, S. Kitani and M. Kerker. Aerosol Studies by Light
Scattering. II. Preparation of Octanoic Acid Aerosols of Narrow
and Reproducible Size Distributions. J Colloid Sci. 19:223-237,
1964.
153. Swift, D. L. A Study of the Size and Monodispersity of Aerosols
Produced in a Sinelair-LaMer Generator. Ann Occupational Hyg.
10:337-348, 1967.
154. Liu, Benjamin, Kenneth T. Whitby and Henry H. S. Yu. A Condensa-
tion Aerosol Generator for Producing Monodispersed Aerosols in the
Size Ranae, 0.036 u to 1.3 p. J Rech Atmospheriques, 397-406,
1966.
155. Edwards. J. and B. J. Brinkworth, A Device for the Control fo
Particle Size in the LaMer Aerosol Generator. J Sci Instrum, Ser
2. 1:636-537, 1963.
105
-------
156. Sutugin, A. G. A Simple Monodisperse Aerosol Generator, In: Ad-
vances in Aerosol Physics No. 4, Fedoseev, V. A. (ed.). New York,
John Wiley and Sons, 1973. pp. 36-41.
157. Nicolaon, G., D. D. Cooke, M. Kerker and F. Matijevic. A New Liq-
uid Aerosol Generator. J Colloid Interface Sci. 34:534-544, 1970.
158. Nicolaon, G., D. D. Cooke, E. J. Davis, M. Kerker and E. Matijevic.
A New Liquid Aerosol Generator, II. The Effect of Reheating and
Studies on the Condensation Zone. J Colloid Interface Sci. 35:
490-501, 1971.
159. Davis, E. James and G. Nicolaon. A New Liquid Aerosol Generator,
III. Operating Characteristics and Theoretical Analysis. J
Colloid Interface Sci. 37:768-778, 1971.
160. Matijevic, E., W. F. Espenscheid and M. Kerker. Aerosols Consist-
ing of Spherical Particles of Sodium Chloride. J Colloid Sci.
18:91-93, 1963.
161. Kitani, Susumu and Sohei Ouchi. Preparation of Monodisperse Aero-
sols of Sodium Chloride. J Colloid Interface Sci. 23:200-20?,
1967.
162. Heyder, J., J. Gebhart, G. Heigwer, C. Roth and W. Stahlhofen.
Experimental Studies of the Total Deposition of Aerosol Particles
in the Human Respiratory Tract. Aerosol Sci. 4:191-208, 1973.
163. Muir, D. C. F. and C. N. Davies. The Deposition of 0.5 ^ Diameter
Aerosols in the Lungs of Man. Ann Occup Hyg. 10:161-174, 1967.
164. Davies, C. N., J. Heyder and M. C. Subba Ramu. Breathing of Half-
Micron Aerosols, I. Experimental. J Applied Ph/siol. 32:591-611,
1972.
165. Giacomelli-Maltoni, Carlo Melandri5 Vittorio Prodi and Giuseppe
Tarroni. Deposition Efficiency of Monodisperse Particles in Human
Respiratory Tract. Amer Ind Hyq Assoc J. 1?:603-610, 19?2^
166. Wolf, W. R. Study of the Vibrating Reed in the Production of Small
Droplets and Solid Particles of Uniform Size, Rev Sci Instrum.
32:1124-1129, 1961.
167. Binek, Bedrich and Blanka Dohnalova. Ein Generator zjr Herstellung
monodisperser Aerosole aus der flussigeri Phase. Staub Reinhaltung
Luft. 27:49?-494, 1967.
106
-------
li:L. I lu, ^enjaniir Y. H. anr* f-fc-, ij • H. Pul. A Submicron Aerosol Stand*
9-fn and the Primary Absolut;, f^nbration of the Condensation Nuclc-i
Counter, 0 Colloid Interface L;i, 47:155-171, 1974.
169. Liu, B. V. H., V. A. Marple, K. T. Whitby and N .J. Barsie. Size
Distribution Measurement of Airborne Coal Dust by Optical Particle
Counters. Amer Ind Hyg Assoc .J, 35:443-451, 1974.
170. Uu, Benjamin V. H., David v, H, s-'ui, Austin W, Hocjan and Ted A.
Rich. Calibration of the Pollek Counter with Monodisperse Aerosols.
J Applied Meteor. 1.4:46-51, 1975.
i71. Wai LOO, W. h. and 'A. C. Prewett. The Production of Sprays and Mists
of Jr.iform Drop Size bv Mean:: of Spinning Disc Type Sprayers. Proc
Phys Soc (Lender). 62":34]-350, 1949.
172. Schwendiman, L. C., A. K, Pcr-ima end L. F, Coleman. A Spinning Disc
Aeroso! Generator. Halth Phys. 10:947-953, 1964
1/3. Albeit, R. E. , H. G. Petrow, A. S. "~aiam and J. R. Spiegelman.
Fabrication of Monodisperse Lucite and Iron Oxide Particles with a
Spinning Disk Generator, Health Phys. 10:933-940, 1964.
,74. Sehmel, &. A, The Oensity of U^anine Particles Produced by a Spin-
ping Disc Aerosol Generator. Am Ind Hyg Assoc J. ?8:431-492, 1967.
175. Lippmann, Morton and Roy t. Albert- A Compact Electric-Motor Driven
Spinning Disc Aerosol Generator. Am Indus Hyg Assoc J. 28:501-506,
1967.
j76. Partridge, Jennings E. and Harry 0. Ettinger. Calibration of a
Spinning-Disc Aerosol Generator and Two-Stage Air Samplers. AEC
Research and Development Report LA-4066, Los Alamos Scientific Lab-
oratory, Los Alamos, New Mexico, 1969.
177. Albert, Roy E., Morton Lippmann. Jack Spiegelman, Anthony Liuzzi
and Norton Nelson. The Deposition and Clearance of Radioactive
Particles in the Human Lung. Arch Fr>viron Health. 14:10-15, 1967.
178. Tarroni, G. hisipioi et caraccerlstiques D'un Generataur D'Aefosols
Moncdisperse a Disque Tournant. Aerosol Sci, ?:257~260, 1971.
179. l.cui-enco, Ray V., Mary F. Kliinek and Claudia J. Borowski. Deposi-
tion and Clearance of 2 v Partirles in the Tracheobronchial Tree
of Normal Subjects • Smokers and Non Smokers. J Clinical Invest.
5C.1411-11?0, 1971.
-------
180. May, K. R. An Improved Spinning Top Homogeneous Spray Apparatus,
J Appl Phys. 20:932-938, 1949.
181. May, K. R. Spinning-Top Homogeneous Aerosol Generator with Shock-
Proof Mounting. J Sci Instrum. 43:841-842, 1966.
182. Ellison, John McK. Adaptation of the Spinning Top Generator to
Provide Aerosols in the Respirable Range, Ann Occup. Hyg. 10:365-
367, 1967.
183. Kajland, A., M. Edford, L. Friberg and B. Holma. Radioactive Mono-
disperse Test Aerosols and Lung Clearance Studies. Health Phys,
10:941-945, 1964.
184. Booker, D. V., A. C. Chamberlain, J. Rundo, D. C. f. Muir and M. L,
Thompson. Elimination of 5p Particles from the Human Lung. Na-.:urp,
215:30-33, 1967.
185. Muir, D. C. F. and C. N. Davies, The Deposition of 0.5y Diameter
Aerosols in the Lungs of Man. Ann Occupational Hyg. 10:161-174,
1967.
186. Holma, Bo. Lung Clearance of Mono- and Di-Disperse Aerosols De-
termined by Profile Scanning and Whole-Body Counting. A Study on
Normal and $03 Exposed Rabbits. From the Department of Hygiene,
Karolinska Institute and the Department of Environmental Hygiene,
National Institute of Public Health, Stockholm, Sweden, 1967.
187. Holma, Bo. The Initial lung clearance o+ Di-Disperse (2u and 7i)
Polystyrene Particles in Rabbits. Arch Environ Health, 17:871-
373, 1968.
188. Black, A. and M. Walsh. Production of Uniform Radioactive Parti-
cles for Aerosol Inhalation Experiments. Part I. Particles in the
Size Range 1-10 microns. United Kingdom Atomic Energy Authority
Research Group Report, AERE-R 5716, Health Physics and Medical
Division, Atomic Energy Research Establishment, Harwell, Berkshire,
1968.
189, Philipson, Klas, Per Camner and John Svedberg. fagging 1\.\ Monodis-
perse Polystyrene Particles with He. J Appl RacJ arH Isotopes.
21:639-642, 1970.
190. Camner, P. and K. Philipson. Production jf 7u Moriodisoerse Fluoro-
caroon Resin Particles Tagaed with "l8F, J Applied Rad and Isotopes.
22:349-353, 1971
108
-------
191. Camner, P., K. Phllipsor, and L. Linnman, A Simple Method for
Nuclidio Tagging of Monodisperse Fluorocarbon Resin Particles. J
Applied Rad and Isotopes. P?:7jl-734, 1971.
",92. Camner, P. Per-Aki hellstrom and Margot Lundborg. Coating 5(i Par-
ticles with Carbon and Metals fcr Lung Clearance Studies. Arch
Environ Health. 27:331-333, 1973.
193. Philipson, Klas, On the Production of Monodisperse Particles With
a Spinring Disc, Aerosol Sci. 4:51-57, 1973.
194. Aldrich, John E. and J. R. Johnston. Use of the Spinning Disk Tech-
nique to Product Monodisperse Microspheres of Human Serum Albumin
for Labeling with Radlonuclides. J Applied Rad and Isotopes.
25:15-18, 1974.
195. Drozin, VadSm G. The Electrical Dispersion of Liquids as Aerosols.
J Colloid SCT. 10:158-164, 195*.
196. Vonnegut, B. and R. Neubau*^-. Deduction of Moncdisperse Liquid
Particles by Electrical Atemization. ..! Colloid Sci. 7:616-621,
1952.
197. Yurkstas, Edward P. and Charles ,]. Meisenzahl. Solid Homogeneous
Aercsol ProdijCtion by El?ctr-;cal Atomization. AEC Research and
Development Report UR-652, University of Rochester Atomic Energy
Project, Rochester, New York, 1964.
198. Huebner, A. L, Disintegration of Charged Liquid Jets: Results
With Isopropyl Alcohol. Science. 168:118-119, April 3, 1970.
199. Fulwyler, M. J., J. D. Perrings and L. S. Cram. Production of
Uniform Microspheres. Rev Sci Instrum. 44:204-206, 1973.
200. Fulwyler, M. J. and 0. G. Raabe, The Ultrasonic Generation of
Monodisperse Aerosols. Presented at the American Industrial Hy-
giene Conference, Detroit, Michigan, May 11-15, 1970.
201. Raabe, 0. G. and G. J. Newton. Development of Techniques for Gen-
erating Monodisperse Aerosols with the Fulwyler Droplet Generator.
Fission Product Inhalation Program Annual Report, 1969-1970, LF-
43, Lovelace Foundation for Medical Education and Research,
Albuquerque, New Mexico, pp. 13-17, 1970.
202. Fulwyler, M. J. Electronic Separation of Biological Cells by
Yrl'j'me. Science. 150:910-91;, 1965.
109
-------
203. Fulwyler, M. J., R. B. Glascock and R. B. Hiebert. Device Which
Separates Minute Particles According to Electronically Sensed
Volume. Rev Sci Instrum. 40:42-48, 1969.
204. Strom, L. The Generation of Monodisperse Aerosols by a Means of 3
Disintegrated Jet of Liquid. Rev Sci Instrum. 40:778-782, 1969.
205. Crane, L., S. Birch and P. D. McCormack. The ,Effect of Mechanical
Vibration on the Break-Up of a Cylindrical Water Jet in Air.
Brit J Applied Phys. 15:743-750, 1964.
206. Schneider, J. M., N. R. Lindblad and C. D. Hendricks. An Apparatus
to Study the Collision and Coalescence of Liquid Aerosols. J
Colloid Sci. 20:610-616, 1965.
207. Lindblad, N. R. and J. M. Schneider. Production of Uniform-Sized
Liquid Droplets. J Sci Instrum. 42:635-638, 1965.
208. McCormack, P. D., L. Crane and S. Brich. An Experimental and
Theoretical Analysis of Cylindrical Liquid Jets Subjected to
Vibration. Brit J Appl Phys. 16:395-408, 1965.
209. Dabora, E. K. Production of Monodisperse Sprays. Rev Sci Instrum,
38:502-506, 1967.
210. Lindblad, N. R. and J. M. Schneider. Method of Producing and
Measuring Charged Single Droplets. Rev Sci Instrum. 38:325-327,
1967.
211. Berglund, Richard N. and Benjamin Y. H. Liu. Generation of Mono-
disperse Aerosol Standards. Environ. Sci ana Technol. 7:147-153,
1973.
212. Liu, Benjamin Y. H. Laboratory Generation of Particulates With
Emphasis on Submicron Aerosols. APCAJ. 24:1170-1172, 1974.
110
-------
GENERATION OF MONODlSPERbE SbhMICRON AEROSOLS BY ABLATION
FROM TRANSPIRATION-COOLED POROUS MATRICES00
by
Cesare V.Boifa,* Augusto Maz/a,""* Delfino Maria Rosso
Istituto di Fisica Tecnica - Politecnico di Torino - Italy
ABSTRACT
In this paper the generation is described of monodisper-se
submicron particles of conl.ro] Led size by ablation from
transpiration-cooled porous matrices,
A aew experimental apparatus is i 1 lust rated which has beeit
realized in the Isti.tut» di Fisica Tecnica - PoJ itecnico
di Torino - Italy, and which xs based on previous works
in cooperation with the University of Minnesota, USA and
the Sonderforschungsbereich ^O, Univer-itat Karlsruhe,
We s t Ge r ma n y.
In the new version of the apparatus an ewternai heat source
(a. pi asm* jet) heats a porous matrix up to ablation
temperatures arid particles art- generated by ablation of
the porous material itselt. The temperature of the ablated
surface of the porous matrix is controlled by a flow cf
f^as transpiring through the matri.x, whirh also quenches
and dilutes the particles fojmed in the ablation process.
Particles can be generated from any porous material of
sufficiently fine structure and uniform porosaty, in
particular i'^'oni graphite, tungsten, ceivimic.s, etc.
f°°) This work has been sponsored by the Italian National
Research Council, Centro Nazionale delle Ricerche.
(' ) Cesare V.Boffa, Processor.
{* *) Augusto Mazza, Graduate student.
(*"*) Delfiao Maria Rosso, Laboratory technician.
U 1
-------
The experimental apparatus is described as well as the
particle measuring techniques, based on the utilization ot?
a calibrated T.S.I. Model 3030 Whitby Electrical Aerosol
Analyzer.
Experimental results are given referring to the generation
of submicron graphite particles. The measured number,
surface, and volume distributions of these particles are
given, together with the logarithmic standard deviation
The results show that, for the generated particles
: 1.32.
112
-------
GENERATION OF MONODISPERSE ^Xt'MICRON AERO SO I S BY ABLATION
FROM TKANSPIRATl'JN'-CGoLfclJ POROUS MATRICES °
Cesar r V.Bofi'a,* Au gusto i**a ' - i, *"* DeLfino Maria Rosso AA*
Istituio di Fisica Teenier - Politecnico dl Tor-ino - Italy
INTROOnCTTOX
Chi;- report describes n -iew inproach i or the generation of
jnorvdi spcrse aerosols whl-.-b ...re- presently cf interest for-
.'ill \inds of mediVaJ and aii" pollution re.<~-.ea"ch, including
f^valuiii irip; aerosol sampJi-.Ej an«* 'iieasurlno; enuipment, air
cieaner- e valviat i.oti, ir.hal a (,1 • n -\udies, and for interr*. 1 •*-
ted studies. Although numerous research-type aerosol
^OMC-i'ators utilising elect ri.- arcs lia're been developed or
oropo^rrl i'or i?eneratiny motif>H i.spei'se aerosols ranging fr'on
0.02 to 20/u.n>, there is --j t pr'es<;nt no rel iable method
availa^jle fox- generating ultrarine, monodlspei se aerosols
'..'or a wide spec! rurr, of materials, and i !> particular of
refractory materials. bas-"d on the utilisation of electric
•arc:- „
Aerosol generation by means of electric arcs between solid
eJec^rodes relies on the high temperature of the attachment-
spots at the electrodes vvhicl' Causes an evaporation of some
electrode material. This metal vapor, if cooled rapidly,
condenses into an aerosol. The properties of the aerosol
particless, including theii size, depend on a number of
parameters, particularly on the temperature distribution at
( '-' ) 1'hJs work has been sponsored by the ItaJ ian National
Research Council, C.N.R.
(*! Cesare V'.Bct'fa, Protessur.
( * * ) lu^jsto Maz.za. Graduate student.
' * ' " ) Delfino Ma^ia Rosse. Laboratory beclmitian.
113
-------
and in the vicinity of the arc attachment spot. It has not
been possible to generate particles with a nearly uniform
size or monodisperse particles under these conditions bueau_
se of the temperature variations and the varying condition:-.
at the location at which individual particles are generated»
The time needed to collect a sufficient number of particles
is large compared to the characteristic time during which
changes on the surface occur, i.e., the relevant parameters
cannot be held constant during this period. The main reason
for these varying conditions lies in the nature of the
attachment spot of the arc which moves more or less quickly
over the electrodes, even if current and voltage do lot
fluctuate appreciably.
But these variations, in addition to the explosive character
of the vaporization process, change the conditions fo*1 the
aerosol production so rapidly that for any useful 3.engt! or
time the produced aerosol has a wide distribution of itt;
particle size.
Various experimental configurations have been suggested for
the production of aerosols with these techniques (Chace et
als^, Reichelt 3 Holmgren et als^, Kuhn4) employing both
high and low intensity arcs. Solid electrodes have been
used by all investigators«, The electrodes have frequent iv
been cooled by a conventional water cooling system, espe-
cially in conjunction with high intensity arcs. With low
intensity arcs coagulated and chaiiilike parti clots are produ
ced. The size distribution of tnese particles covers a wide-.
range, from approximately 100 to 1000 X. (Raiehelt **) ana it-
poorly controlled. Better results nave been obtained by
using high intensity arcs (Holmgren et als^j. The feed mala-
rial is incorporated in the body of the anode which consumes
a large fraction of the arc power. In most cases the so i ±ti
electrodes are water-cooled and dilution gas can be blown
through the arc. By varying the cooling rate of the electro
des and the dilution flow rate through the plasma in the
region of maximum condensation, the particle size may to a
certain degree be controlled. Monodisperse spheroidal shaped
particles with a diameter ranging from approximately 100 to
1000 A have been obtained.
114
-------
The results of the above-discussed in vest.L gat ions indicate
that electric arcs are a poor .source for generating aerosoJs,
It seems to be impossible to obtain small monodisperse
particles of uniform si*",e because coagulaticn occurs as
scon as the particles are produced. In addition, the partic_
le generation process is characterized by strong non-unifo£
ml ties which are difficult to control in conventional arc
arrangements with solid electrodes.
During the past two years, a new snethcd for the generation
of ultrafine, monodisperse aerosol has been successfully
explored (Boffa et als> ! . This method makes use of a
transpj ration-cooled anode In a high intensity arc. In
this situation, the aerosol is generated continuously
rather than ir an explosive manner at the steady attachment
spot which is the hottest, pail of the anode surface. The
transpiring gas passing through the anode provides
quenching of the vapor ariJ au immediate dilution of the
generated particles in the vicinity of the anode surface
avoiding in this way coagulation and formation of agglome-
rations.
The method as such has been very successful, but the partic_
le generator itself is a rather complicated devi.cn which
requires skilled personnel for its proper operation. The
knowledge acquired during prc , Lous work has been, however,,
instrumental for the success of 1 he new approach described
ir this report,,
In the new approach here described the arc plasma source
is entirely separated from the body frcra which the aerosol
is generatedj but the basic features for the aerosol
production process inherent to che previously-developed devi^
ce are retained. Instead of a transpiration cooled anode
which is an integral part of the previously-developed plasma
torch, a porous, transpiration-cooled body will be utiLi/ed
heated by the plasma jet emanating fron; a conventional arc
plasma torch. Since transpiration cooling t'ullfilla two
tasks simultaneously; namely, quenching of the generated
vapor- and dilution of the particles formed in the que.iching
process, utilization of this mechanism is considered as the
basic principle which also governs this approach.
-------
The separation of the heat source from the particle provi-
ding surface has at least two important advantages over the
previous method:
I) The resulting particle generator is simple and straight^
forward to operate, and easily movable, as required by
the experiments which will be performed in this research
program.
II) The useful parameter range for operation of this device
and the associated flexibility is substantially enhan-
ced.
EXPERIMENTAL APPARATUS
The experimental apparatus consists essentially of a•comme£
cial power supply and control unit, which enable the
operation of a plasma torch, which in turn represents the
heat source, necessary for the operation of the aerosol
generator.
The plasma torch and the aerosol generator are schematically
represented in fig. 1.
The plasma torch consists essentially of a water cooled,
thoriated tungsten cathode, with a conical tip, and a nozz-
le shaped water cooled copper anode, coaxial with the
cathode. Argon is injected tangentially at the base of the
catode, thus providing a swirl flow around the cathode tip,
which stabilizes the arc. This gas, passing through the arc,
emerges from the anode nozzle in the form of a highly ener-
gized plasma jet.
The cooling of anode and cathode is provided by a water
cooling system, as indicated in fig. 2.
For an overall energy balance the cooling water flow is
measured by means of flowrators, and the temperature increa
se of the anode and cathode cooling water is measured by
means of four thermocouples located in the water circuit
immediately upstream and downstream from the electrodes.
In this way the energy losses to the electrodes, that is the
heat transferred from the arc to the anode and cathode, can
be determined. The power in the emerging plasma jet can be
116
-------
AEROSOL A PARTICLES
• Ar
trantpirttion
THORIATED TUNGSTEN
PLEXIGLASS
Fig. 1 Schematic of the plasms torch with aerosol generator.
117
-------
HOWER SUPPLY-CONTROL UNIT
L.V.
H.V.
PLENUM
CHAMBER
AEROSOL <*y
PARTICLES I
PLASMA'""V
TORCH
HJ
REACTION
CHAMBER
Ar transpiration
FLOWRATORS
Ar
PECULATOR
PARTICLE
ANALYZER
FLQWRATOR (?f I H20
t
.1 1
i
^ 1
T i,
TC
RECORDER
Fif.. 2 Schematic of the experimental set-up.
118
-------
calculated as the difference between the electrical power
input and the power losses to the electrodes, carried away
by the cooling water.
As before mentioned,the plasma jet emerging from the anode
nozzle enters the ,%erosot generator located on top of the
plasma torch and coaxial with it.
Th'.- aerosol generator itself consists essentially of a
transpiration cooled porous carbon nozzle, concentrical
with the plasma jetj inbedded in a steel and plexiglass
structure which, supports the porous nozzle and acts as
pJenum chamber feeding the cooling gas to the nozzle itself,
The cooling gas transpires through the nozzle and emerges
into the plasma id, diluting th«- aerosol particles, v/hich
are continuously produce*] by ablation from the poroi'.s
carbon nozzle inside surraee facing the plasma.
A cylindrical plexiglass window surrounds the plenum cham-
ber and seals it, by ine^uc .*"" ''()" rings, Through this
window pyiumetrle readings of I"he outer surface temperature
of the porous nozzle are possible. The plenum chamber is
equipped with an inlet for the cooling gas and with a pres-
sure tap. The steel portion of the strrcr.ure supporting the
porous nozzle is water cooled.
The aerosol ercoi-ging from the generator, in suspension in
the argon gas, is collected into a 2 m3 collection chamber,
where it can be additionally diluted.
A bior.k diagram of the experi men! al setup is shown iu
t-i-g. 2e
The aerosol s5/e distribution measuring .systen. consists
mainly of a calibrated T.S.T. Model 3030 WnJtby Electrical
Aerosol Analizer described e '.sewliere (Liu et als1).
EXPERIMENTAL RESULTS AND CONCLUSIONS
The results referring to carbon particles are summarized in
fig. 3 to ^e
In fig. 3 the normalised number distributions of the produ-
ced aerosols are depicted, corresponding to different
values of the argon flowr,:v^e transpiring through the ablat-
L.19
-------
10'
AN
N A log Dp
10'
10'
o
i
QC
10"
10
.-3
10T4
.001
I' '"I ' ' 'I' '"I
I I I I 11
TRANSPIRING ARGON FLQWRATE
2 Nl/min. o
1 • „ A
0,5 ., n
0,25 .. •
. . . I. ...i
.01
i i i I § 111
.1 1
PARTICLE DIAMETER Op[|i]
Fi«. 3 Norroalized number distributions of the generated aerosols,
120
-------
ing porous matrix. Figures 4 and 5 show the corresponding
normalized surface and volume distributions.
All the data are taken with an arc current of 100 Amp.
The results show that, for a given arc current value,the
mean diameter of the generated aerosol decreases with
increasing of the argon flowrate transpiring through the
ablating matrix.
This finding can be qualitatively understood considering
the influence of the transpiring gas flow-rate on the tempe_
rature of the. ablating surface of the porous nozzle. By
increasing the transpiring mas;-; flow-rate, the temperature
of the ablating surface and the associated rate of ablation
decreases, yielding smaller particle densities, lower
coagulation rates and, therefore, smaller particle diameters,
At the same time, an increase in transpiring mass flow-rate
enhances the dilution of t*e ablated particles in front of
the surface, which decrease-; the coagulation rate and,
consequently, the particle size also.
In fig. 6 the logaritmic standard deviation is depicted as
a function of the transpiring argon flowrate for an arc
current of 100 Amp.
It can be seen that for the generated particles 1.13
-------
10'
AS
S A log Dp
10'
10
€/3
10
-2
10
-3
10
-4
.001
TRANSPIRING ARGON FLOWRATE
2 Nl/min o
1 „ A
0,5 .. a
0,25 • •
I 1 ._!__!..! I 1
.01
.1 1
PARTICLE DIAMETER Op [|i]
Fi°.A Normalized surface distributions of the generated aerosols.
122
-------
ID1
V A log Dp
10
1C"
10
CD
WU
Zj
O
O
X
oc
en
10
-3
10
i i i 1 i
.001
.01
I' "M
r T
TRANSPIRING ARGON FUMRATE
2 Nl/min c
1 „ A
0,5 , n
0,25 •
•4
\
i t I i i 1 1 j i i i i
.1 1
PARTICLE DIAMETER Dp[|ii
Fig, 5 Normalized volume d is 11 \biir ions of the generated aerosols.
123
-------
esi
ca
NOI1VIA30
L24
-------
ess
S313!i»Vd Q31V18V JO X1ISM3Q
[SS31MOSSN3WIQ]
TWD1 40
to
-H
Pu
125
-------
The total concentration values N^-0^-, corresponding 1 n
given\mass flowrate, are plotted in dimonrdonless font*
according to the following relation:
N. . - N* AN
tot tot tot
N-x- N-"-
tot tot
_ 3~7
where N.x~ = 8.736,000 /particle/cm __/ is the refer or-re
value and represents the total concentration corresponding-
to a transpiring argon flowrate of 3.5 Nl/min.
The results given in fig. 7 are i" agreement with the
qualitative explanation of the aerosol production mc-chan L-:-ii;
given above.
During the experiments consumption of the iratrjx taker-
place, and hence, for a given value of the arc current and
of the pressure in the plenum chamber feeding the transpi-
ring gas through the matrix, the temperature of the matrix
ablating surface decreases. This is due io the lot/er tenipr
rature attained by the ablating surface which can be atn 1
buted to two main causes. One is related to the ablation
of the matrix which enlarges the inner diairetre of the
matrix nozzle, decreasing the heat transfer1 rate froi'. >.;,••
plasma to the unit area of the nozzle waMs. The o thr-
one is related to the xncrease of the trvnsp tri ng .''gor*
flowrate throug!i the mati'ix.
For a given value of the relative pressure in the
chamber and a given matrix temperature di^tr i but i or, t'.i- 3 •
gen flow rate transpiring through the matrix is inve :"•?'_ Lv
proportiona I to the matrix thickness.
The ablation of the iratrix reduces its thickrie-,& and h.~<\c c >--
foi'p the mass flowrate increases. This effect is euro<^< (:•..<
by the lower teiriperature attained by t)to lu-itrix, cause.; a;,
che decrease in the heat transfer rate from the plasma per
unit area uf the matrix nozzle, related t,c the incr-ease in
the inner diameter of the ablating nozzle , as above jner;ti :>
ned .
Therefore during a lone operation of the particle generator
the mean diameter tends to decrease. This ca1: be easilv
126
-------
compensated for by Increasing the arc current and decrea-
sing the relative pressure in the plenum chamber for the
feeding of the transpiring gf-s through the matrix.
further experiments will be performed with arc current up
bo JOO Amp', various materials will be investigated such as
tungsten, ceramic am! stainless steel, which are available
as porous matrices. Particles from other materials, which
arc not available as porous matrices, will be generated by
mixing the material to "be studied with graphite. Since the
evaporation temperature of graphite is in most cases higher
than that of other elements, it is expected that the admix
ture predominately evaporates, producing aerosols of the
des Lred mate rial.
BIBL IOGRAPH 1C REFER EN CES
l) Chace, W.G., Moore, U.K., _Explo^n_^ Wires, Vol.1 (1959),
Vol. 2 (lQ6?,)j Vol.3 (1Q64), Plenum Press, New York.
2) H. Reichelt, " \n Aerosol Generated 1 -\ EJectric Arcs
Between Metal Electrodes", Aerosol Research at the
Fio.-'st Physics Institute, Wien (1968),
3) J,G* Holmgren, J.O. Gibson, €„ Sheer, "Some Characteri-
stics of Arc Vaporized Suhmicron Part Lcul atcs" ,
Llll^J^iicJ^s* John Wiley & Sons, New York (1963).
4) W.£. Kuhri, "The Formation of Silicon Carbide in the
E.ie>.tric Arc", Ultra f i ne P at' 1 1 c ie s , John Wiley & Sons,
'^••K \ork (1063).
5) BofCa, Co, Pfender, E., "Controlled generation of
"Kmodisperse aerosols in the submicron range", Journal
o'C Aerosol Science^, v~»l. 4 (
• j !. i u , B . 'I , M . , IVlii tb v- , K . T . , P ,.' i , D . Y .H . , "A portab ie
oioctr : r-a 1 analyze"' for size distribution measurement
of subnicron aerosols" Journal of the Air Pollution
Control Association, ''01.2.4., n*Il (i974)o
177
-------
GENERATION OF MONODISPERSE AEROSOL :> OF 67Ga-LABELED ALUMINOSILICATE
1QR *
AND iy°Au- LABELED GOLD SPHERES
George J. Newton, 0. G. Raabe, R. L. Yarwocd and G. M. Kanapilly
Inhalation Toxicology Research institute
Lovelace Foundation
P. 0. Box 5890
Albuquerque, New Mexico 87115
ABSTRACT
A system has been designed, construe ted, tested and used for producing
roonodisperse (eg < 1.2} submicro'.seter particles of both 198Au-labeled
yold and 67Ga-labeled aluminos incite spheres for inhalation deposition
studies in man, animals and in other types of experiments where short-
lived radioactive, monodisperse aerosols are useful. Aerosol generation
and particle separation equipment housed in a system of five stainless
steel glove boxes produces a primary polydlspcrse aerosol tohich is
passed via a quartz tube throug!' a high temperature furnace (1200 °C)
which converts the clay into insoluble al-jrrinorilicate spheres entrap-
ng the
in the particles fletaiiic. gold particles are produced by
nebulizing an aqueous suspension of "fulminating gold", a mixture of
gold-ammonium-acetate compounds, which is reduce*! to metallic gold in
the high temperature heating column. Two Lovelace Aerosol Particle
Separators are used to separate the prlydisperse aerosol into monodis-
r?erse fractions onto stainless steel foils. Gold particles are removed
trom_the collection foil and subjected to neutron
the iSB^y label Monodisperse particles are then
irradiation to form
uspended in water and
nebulized to provide monodisperse aerosols in the respirable size range
oetween U.5 urn and 3.5 urn aerodynamic diameter. Ihe particle densities
av~e 2.3 g/cm3 for aluninosilicate and about 19 g/cnv for gold.
* Research supported by the National Institute
Sciences via F.RUA Contract F(29-2;-1013.
)f Environmental Health
Preceding pap bianlc
-------
GENERATION OF MONODISPERSE AEROSOL
1QR
iy°
AND
OF 57Ga-LABELED ALUMINOSILICATE
Au-LABELED GOLD SPHERES
George J
Newton, 0. G. Raabe, R. L. Yarwood and G. M. Kanapilly
Inhalation Toxicology Research Institute
Lovelace Foundation
P. 0, Box 5890
Albuquerque, New Mexico 87115
INTROD
CTION
Data from inhalation deposition and particle clearance studies in labo-
ratory animals are used to help evaluate the inhalation toxicity of a
variety of potentially toxic aerosols to v/hi^h mar could be exposed.
Since the extrapolation of experimental animal data to man is not pre-
cise, a useful comparison is made m studies of the same aerosol in
laboratory animals and man. Ideally, en aerosol for human deposition
studies should be monodisperse, insoluble, easily detected, of low tox-
icity and reproducible in the size ranges of major interest. To this
end, we have developed systems for generating monodisperse, insoluble
aerosols of aluninosil icate spheres labeled with 67Qa and gold spheres
labeled with 198/\u. The aluminos'i licate particles have biological re-
tention half-times in the lung on the order of one year or more attest-
ing to their insolubility in body fluids. Gallium- 67 decays with a
78.1 hr. half-life by electron capture to stable 6?Zn and has no par-
ti culate emissions (excepting Auger electrons), •".-.] lium-67 decay is
accompanied by emission of three maje-- gar^ia photo- groups at 9?s 18L
and 300 keV with 69, 24 and 72.':- yields respectively. These gamma
photons have an average entity of lf-0.5 kev and e:»;id deposit a maximum
dose to a 1000 g human lupr,> of 3b mrad/V,:i Gallium- 57 is made D.V a
68.7n(p, 2n) °7ga reaction ir en dCce'-.Ttf r.x.
l rk
ab^'.
The 198Au-gc!d aerosol
other standard aerosols. The radiol
accompanied by 412 KeV gamma ohotonr I
ideal for deposition and early clearance
for repeated exposure of 'he same an hii.f :.
sizes. The density of gold (n ~ 19.3 T'
ture of this aerosol. Mar.y of the
such as the oxides of the actir.ides
and since the important deposition "-
independent of density, a need for o
an example, consider tv,o particles b
aters with one particle having a den
density of 11.0 g/crr-3, These two p-i
ters of 0.40 urn and 0.031 ^, r^spc-c
irable features not found in
^^Au, decays by beta emis •> ;'
half-;ifi -,.f 64.75 hours '\ -,
tudies in animals ind allows
to a variety of partible
'. ^) is crother deslraole fea-
nvi'
and
-r::-:ental aerusolr. of i merer. t,
ec
ci
cher heavy mQta'!s are dense,
s
sm of Bruwnian diffusion
deni ity aercsoi exists. >:\5
viny 0 5 --m aeroaynanric dldm-
r i.o q/c;p3 dn(j *^e other d
wc-jid have geometric diame-
however1 ;, the smeller particle
1JO
-------
would have a diffusion coefficient ten times greater than the larger
particle. Gold aerosols are also insoluble, a requisite for conducting
inhalation deposition and particle clearance studies.
The production of monodisperse particles involves collecting the poly-
disperse aerosol wit'" the LovoKice Aerosol Particle Separator1 (LAPS)l
which separates the particles with 'respect to Aerodynamic size and de-
posits the particles on e. 46 urn lo'.n by 0.015 en thick stainless steel
foil. The foil is then cut i,(to harrow segments, e?c^ containing a
narrow size distribution of particles (on <" 1.1). Particles are then
'-esuspended from the foil ^eqments and aerosolized.
HC FHODS
PREPARATION OF GALL1UM-&7 LABELED /LUMINOSiLICATE SPHERES
Using procedures developed it. the Inh^lati
tute, monttnorillonit.e clay was p-'epj^ed ds
Norgon.2 This procedure is ill us traced by
Basically, the procedure consists jf treat
concentrated Hp02 until no reaction occurs
clay suspension,""packing" the exchange si
running water to remove the excess sodium,
est to exchange with the sodium ions,
filtering and washing the labeled clay.
The 67Ga-labeled clay suspension i;: t,hen
aerosolized and heat treated to fuse the,
clay particles as described by Rsa'oe,
et al.3 Preliminary tests using '0 ;;«Ci
h7fiaCl3 (Amersham Searle, Chicago,
Illionis) yielded a 3?' activity exchange
into 200 mg clay. Since the excnange
capacity of the clay Is 1,3 mEq/^, ?00
r.ia of clay should have a cau-cHy ^*"
5^81 mg of 67Ga. The soecific dcti-'iiy
was listed as 10^ Ci/q of Ga(IlI),
therefore other cations competing *'or
the clay exchange sites ^ere assuiiied to
be present indicating the necessity for
separating 67G<\(T!_i) from the z;nc tur-
get material and other possible impuri-
ties.
on Toxicology Research Insti-
described by McKnight and
the flow chart in Mgure 1.
ing a raw clay sample with
, decanting the finely divided
tes with sodium, dialysis with
a 11 owi no t hf3 ca t i on of i nter-
Ga'i ;!ji:i-67 was extracted from 6 M !-Cl
solution into -«s,opropyl ether AS tne
ciilorocjmplex H^GaCl^as described by
Morrison and Freiser/' ijnder those
^n(il) ^a^ noc e-,t? ;ctffj.
RESUSPENO a
EXCHANGE (ISOTOPE!
FILTER + WASH
!
RESUSPENO 'N WATER AEROSOLIZE
I
HEAT MOO°C
LZ: L___
I""" SPHERICAL AEROSOL
^ 1, rlow Chart of
rreix-'ation and Exchange.
-------
The 6?Ga from the organic phase was stripped by shaking with 0,04 M
HNO-j. The ^Ga obtained was carrier free and was used directly for
exchange into clay.
GENERATION OF 67Ga-LABELED CLAY AEROSOL,
After preparation of the labeled clay, the clay suspension is placed in
a Lovelace nebulizer in an aerosol production system housed in a series^
of five connected stainless steel glove boxes described by Raabe, et aP
and illustrated in Figure 2. The first glove box in the assembly line
houses the generator and a low temperature (- 350 JC) heating column
used to enhance drying of the aerosol droplets. 1 he second box contains
a high temperature tube furnace (1200 ""'C) to fuse the clay aerosol into
aluminosilicate spheres entrapping the 6''Ca. The third glove box con-
tains the sampling chamber trcm whuh the LAPS i-amDle^ and other sample-
are drawn as needed. Routine samples taken include Doint-to-plane elec-
trostatic precipitator samples for electron i)nc»-oscop> . cascade impactor
samples for aerodynamic size dott-nnination and a continuous filter sam-
ple for activity concentration assays. * lead-shielded conconcentric
electrostatic precipitator collects all particles not sampled. The
fourth glove box houses two LAPS units that sample the resultant Dolydis-
perse aerosol and collect particles on stainless steel foils according
to their aerodynamic properties. Ihe fifth box is used for cutting the
collection foils into segments and preparing the seqinents for individual
storage and is not shown in t'u3 schematic.
_' f'~~ ~"~l \' rs
-------
PRODUCTION OF GOLD AFRUSOL
Generator So 1 u ti on
It was assumed that solutions of Au(III) could be aerosolized and sub-
jected to thenral degradation at 1200 "C which should Deduce the Au(III)
to metallic gold since Au(III) is relatively unstable and gold melts at
1063 °C. All chemical systems using Au(III) in the presence of chlorine
to yic^d spherical particles. One system using chlorine free
yielded spherical metallic gold partir'^.s at unable concentra-
M-issive gold (197Aii)
heat. This solution was
ther- continuously boiled and diluted with 1 M HC 1 for several hours to
remove n^rate. The resultant chloroaunc acid, KAuC.14, contained 190
nq Au per n-i -und pH of 1-2. This sloc> solution was diluted with fil-
iered deionized water to yield a ch'oroouric solution of ~ 1.0 mg Au per
ml and 4 < pH < 6. N!1-}OH (1 M) vws fKen added until 8.0 < pH < 8.5 is
obtained. This precipitated an dmmociuin complox. ''fulr.inating gold",0
a mixture of two compounds: Ai^Op/NH^ and NH(AuNri2Cl )2 -n an average
ratio of 1:2.
failed
tions and ij shown in the flow chart, Figure 3.
wss dissolved in aqua regia with gentle (50 °C)
+ 4 NHflOH + H?0
.j-NH3 +• NH (Au'iH^Cl }, ^ H,>0 + HC1 (1)
Au-!97 I
AQUA REGIA 1 *"
1 „ J
_L
DISSOLVE
PPT !N
| ACETIC ADD
' ,
— •
i F Ah'D IMH40H fO
BOIL ADDING HC1 | f DRECIPiTATE.
TO REMOVE NO^ J "\ "FULMINATING GOLD"
" j i 8.0»
pH-8.5 FILTER (NH40H) FILTERING
6 WASH WASH 3 TIMES j
—
FILTER S
WASH PPT.
SUSFEr4D GOLD
COMPLEX IN WATER
GOLDr!.0 mq/mi
AEROSOLIZE 8
HEAT TREAT
igure 3. Gold Aerosol Production How Chart.
This orecipitatu was vacuum filtered using 0.4b i,m po^ii sized membrane
filters and washed witn deionized vidter. Trie wet precipitate was then
dissolved in 25 ml of 1 M OhCOUH The gold complex wis reprec':nitated
1 M NH/iOH at pH = 8.5, filter-Co and washed vnth deionized water.
of dissolution, p^ec v/natian, filtc"ing aid washing was
times. The "'inal p^ecipitjt.e wos ""esuspended in deion-
ul tf--3:->or-ic agitation to \ield a concentra-
ized, f
tion of
;_omDle;<
H40H
cedure
t^ree
ltered water using
1.0 mg Au/ml . This Sur.per-sion was a
hH.ririe ions.
ni urn-acetate
*'ec of
-------
Generation of Gold Aeroso1
The acetate washed fulminating gold suspension was aerosolized using a
Lovelace type nebulizer'' aspirated with compressed air. The same aero-
sol production system described for the ^7Ga-labe1ed aluminosilicate
aerosol was used to separate the polydisperse gold particles into rela-
tively monodisperse fractions according to aerodynamic size.
Gold particles collected on a LAPS foil were resuspended from a segment
of the foil, 1.02 < ag < 1.2, in aqueous ammonia, pH - 10.1 ± 0.1), cen-
trifuged, placed in quartz vials, vacuum dried, sealed and neutron
irradiated to produce the radiolabel ^°^Au. The production of the
labeled gold spheres is illustrated schematically in F:icu*-e 4.
FU MINATING GOLD
SUSPENSION
(Au-197)
NEUTRON
IRRADIATION
|
ULTRASONIC
SUSPENS!ON,fAu-!98)
NH4OH (pH^IO.Ol
— "
_*
NEBULA
(AIR)
TRANSFER
Au-i97 TO
"RABBIT"
NEBULIZE
SUSPENSION
H
r
! HEATING
COLUMNS
2000 C a !200°C
CENTRIFUGE
a WASH
HEAT
~ 200t!C
i
H
LAPS
COLLECTION
™'J
1
ULTRASONIC j
SUSPENSION
NH40H(pH-iO.O) |
SAMPLER,
TEST DEVICE,
ANIMAL ETC
Figure 4. flonod i sperse uolu Particle Prenaralicn
Neutron Activation Flow Chart.
Generator suspensions t^ produce I'lonod'i
suspending a monocr ".;)erse fraction CM t
segment in ;';,U01 f" N!-!;:OH us^ng ul ri-asnr
concentration (f-'NC; oi-1" the particle sus
singlets using « Lovelace^ nebulizer c.s
sol was passed through a S-'Kr -i-;sc no *•']*:•
c;'-r,rgc on tne car* icier a*id *o (..'.liaiice
dRrosol was u.en nriX
exoosure apparatus,
sper'.c Jtr-jzol-. were prepared o
h"1 C-PI'D-.O! ccllected OH a foil
« agitatior, 1 ne piaximoni mnt-e-
i-fftsion w.^s -'id ', t'Sf.ed to vieu: 9[>
^rev'oir". 1 v nec.criDed.3 The aert
•• hfatfi r" 70 "'C tc rcri'r:*-. tr,:}
dropietc, . The
.. . , .
v/ith clean, dry, >1'!ur;ng air an^ ua^sod into
an
"ESUTS AN!}
GAI.L1UM-67 ALiJMINO-ILTCA!T:
Since the procuction of usable quanta
tides would require Hd^dling 100 ,r'C !
solvent extract ion arid <:!;.•/ exchange ,-•
i* '-'7Ga-ciay moiiodisperse pd»'-
'(\> .:!•;•' OT 6"Ga, the system of
:fieci ts 'ermit inanipu idtor
-------
cell operations. Figure 5 is a schematic diagram of the procedures.
The entire production procedure was tested using 10 mCi ^'Ga plus 0.5 mg
* carrier, Ga(IH), to simulate a 100 mCi to 200 mCi prcduction run. The
following list indicates the relative losses in the solvent extraction
and clay exchange processes:
Aqueous phase after organic solvent extraction
Organic phase after back extraction into HN03
Filtrate after cation exchange and filtration
Amount exchanged into clay
The total slightly exceeds 100" due to counting statistics.
'
0.6%
0.2%
16.7%
82.6%
AQUEOUS PHASE
AQUEOUS
PHASE SAMPLE
ADD 10 mi
iSOPf
-------
Preliminary tests indicate that 100 to 200 mCi 67Qa can be exchanged
Into 40 mg of clay. The clay suspension could be aerosolized in 1 to 4
hours and the subsequent resuspension would require approximately 2
hours. Total production of the monodisperse particles would require fi
to 12 hours. Figure 6 is a composite electron micrograph of four sh: .,
of monodisperse aluminosilicate particles labeled with Ga»67. The uec--
metric standard deviation of these particles ranged from 1.02 to 1.10.
F i''jure 6. r.ornposite F'lacLron Micruyraph of Four
Monodisperse Si^es of 673a-Labeled AluminosilicatP
Particles from Collection coil: 'a) 0.86 ;m;
(b; 0.64 ,;m; fc} 0.32 w and (d) O.i2 \>m.
INHALATION EXPOSURLS
Consider the mass aic! activity conceniration of 67ria.]aj_,e]9,;j alur.nno'i :-
icate spheres of 1 urn geometric diameter with a specific activity of 'J.5
Ci/g clay and a maximum number concentration of 2.5 x 10^ particles/ml
in an aqueous suspension.
-------
with
V = volume of each sphere
p = density of the aluminosilicate -2.3 g/cnT
mass/ml = V x r, x 3,5 x 10 particles/ml
TrD3 x '2.3 x 3.5 x 108
- 4.215 x 10"4 g/m
Activity/ml
mass a-_tj.vvty
ml mass
x "-£-: :-e^ = 4.215 x 10 g/ml x 0.5 Ci/g
= 2.107 x ICT'Ci/m!
Using the above calculations, mass and/or activity concentrations for
any particle size of interest can he obtained by multiplying the results
by the cube of the diameter in micrometer units. Table i lists parame-
ters of interest for various particle sizes for both dog and human
exposures.
67
Tab^e 1
Ga-labeled, Monodisperse Aluminosilicate Spheres for
Inhalation Studies With Dogs and Humans
Generator
Concentration R
uCi/ml at 3.5 x 10B
Particle/ml
1400
383
211
18
6
Aerodynamic
Diameter3
3.00
2.00
1.66
0.8C
0.60
Geometric
(Real)
Diameter
1.88
1.22
1.00
0.44
0.31
Expected
Pulmonary
Deposition
in a Dog
uCi/min
2.5
0.67
0.37
0 . 03
C.01
Expected
Pulmonary
Deposition
in a Human
yCi/rrrin
5.3
1.4
0.79
0.07
0.02
Lovelace Aerodynamic Diameter as discussed by Raaoe, et al where
Daer " Dreal
137
-------
IQfi
^%-GOLD AEROSOLS
Neutron activation of stable gold-197 was the method used to produce the
|98Au tag on the particle. Consider an 8-hr irradiation time for 1 g of
197A.u. Then the specific radioactivity (A) is given by
-At, -xt,
e 1 1 . i 'v _ f-
A - ^__L '-V-L.- . c i
' in
3.7 x 101U
where
a - thermal neutron cross section activation area
- 9,8 x 10"2* cm3
13 ?
f - neutron flux - 8.0 x 10 n/c.n /s (Omega West reactor at
21
n = number of target nuclei per gram of gold - 3.06 x 10
t-i - irradiation time = 8 hr
i
tp ~ time f i cm removal from reactor to use (-- 34 hr)
X = ±j_-L = t-jr-.0 - gold-i93 decay constant
1 1 / P OH , / b h
'Consider the rrass and activity concentration of 1.0 \in\ diameter- ^^
gt,-td particles it thtj specific attivi:,- o; 3-5-9-^ Ci/j1 and a co- -eri u
tiO'-, of 3.5 x 1C)8 uart,icles/ril in an .-q^e'ius su^^ensi
•wher^ .-. - density of Au - 19. S c. ci?;"'
S
Total mass/n.l -• -~ -A--~- x .* x ';.'c x 10 particles/ml
p ir ' t < C ( c
- :; - x 19.3 f 2 y x 198 = 3.54 x 10" J
Activity/ml - ™- x ~-- = 3.54 x 10~3 g/ml x 36. 9^1 Ci,
iTi ! ilia •• :>
-- C.I31 Ci/ml
Usiisq tne calculation al-ov-fj the na,.s •:«• ditivity for any particle s
of interest, w« bt computed by rn^'s tv;.iy ;r.o the results obtained h. t;
;.ut)e o*' th? diarr.f »••;>(-* iri msc
-------
Consider the following hypothetical dog exposure. Wanted: the activity/
mass concentration required to yield an aerosol of ^^Au-gold that will
deposit in the lungs of a Beagle dog - 1.0 iiCi initial lung burden (ILL-)
in 10 min.
Using a dog exposure apparatus previously described^ and assuming 25%
pulmonary deposition and a respiratory minute volume of 2.8 liters, the
activity concentration of the particle suspension in a Lovelace nebu-
lizer (0.05 ml/min output) with 20 1/min total aerosol volume was:
___ 1 uCi X 20 1/min '
10 min x 0.7 1/min x 0.05 ml/min
- 57.1 uCi/ml
(6)
Table 2 lists parameters of interest, for various particle sizes for both
dog and human inhalation exposures.
TabU- 2
198,
Monodisperse ij Au-Gold Aerosols for Inhalation
Studies With Dogs and Humans
Generator
Concentration
uCi/ml at 3.5 x 10
Particle/ml
8
131000
6636
2302
442
174
28
Aerodynamic
Diameter3
(um)
4.81
2.00
1.50
1.00
0.80
0.55
Lovelace Aerodynamic Diameter - D
Geometric
(Real)
Diameter
1.00
0.37
0.26
0.15
0.11
0.06
real
(PC)
Expected
Pulmonary
Deposition
in a Dog
Expected
Pulmonary
Deposition
in a Human
uCi/miri
230
11.6
4.0
0.8
0.3
0.05
* 478
24.2
8,4
1.7
0.6
0.1.
Figures 7 and 8 show electron micrographs of monodisperse gold particles
collected on electron microscope grids placed on the LAPS sampling foil.
The grids contain a replica of a diffraction grating with a line spacing
of 0.83 ym. Figure 8 contains monodisperse gold particles and reference
particles of polystyrene latex, p = 1.05. Both populations of particle
sizes have the same aerodynamic diameter.
139
-------
Figure 7. Electron Micrograph
of Manodisper.se Gold Particles
from Collect''on Foil
(0.83 >M Ruling;
Hgure 8. Electron Microg---?^:
of honodisperse Gold Particle'
and Reference Polystyrene La"*
Particles (0.83 ;j'm Rulinq),
£ s tj mate of L_APS_S ainpj i no_ _T iroe__Ne_c esjs/j ry- to _0bta i n jj_s efuj _Nurrh e r ?_ o
Wanted:
Estimated run time utilizing two LAPS units to obtain
gold particles for one generator loading for a single s-;
, JLB for a Beagle doq - 1.0 ;.Ci .
! ivfla'.e ^erierat^r (0,')5 '•il/ri'ir output), 4 l/^in ui'ulir;
anu ^O'J iv'' /nr^n, i. Af'5 sriinolinq rate, 'ir.c* c '^enc.f a r.o-1 -'U-.,-!f^
concert ration of 1 my A.(/'n>! ^nd 5;' ,rf LnPS sample ner sp<.:p'
Then xh« sarrspHnq rate ~, ~ qivyn
-, -.r- -. • •
x 0.05 ;Tii/rci
H X 0 "i ! ''!"i >•'!/ ' O i 1
--- — "•>:- -r/-; ~
7
3.75 x 10"' g/nin/foi! segmtnt
•S'. ^>suming a Qer-erd1- or l^.aa for- exp^urf; of 10 ;." at a con-'encrac
6C ..Ci/nil, the LAPS sanpling time required per generator load is giv
60
,_.. -
10 m",
= -!3.3 min/uenerator l
•MM x 3 G94 > 1C' ..C'/y
A 10 nl generator s.-.-'.utiop wa:3 adequ?..e for 1ZG nun of aerosol qc TJ
-------
RESULTS
DISCUSSION
198
The yield of Au in monodisperst1 gold particles neutron irradiated in
the Omega West reactor at the Los Alamos Scientific Laboratory was com-
parable to the calculated values and the difference was due to lack of
precision in estimating the mass of gold. Numerous small animal inhala-
tion studies^ have been conducted using monodisperse ^Sftu-gold aerosols
to radioactivity levels between 0.1 yCi and 0.8 uCi per liter of air.
The most significant problem encountered was difficulty in breaking up
of agglomerates in the smaller particles.
CHARGE DISTRIBUTION
Both monodisperse aerosols, 67Ga-aluminosilicate and 198/\u_g0id, were
studied for charge effects using a miniature charge spectrometer by Yeh
et al.H After aerosolization of the purticle suspension from 0.001 M
NH40H, the aerosol was passed through a charge neutralizer containing
2 rnCi of 85j
-------
SUMMARY
The production of monodisperse spherical particles of 67Ga-labe1ed
aluminosilicate and 198Au-labeled gold has been described. Gallium-67
labeled aluminosilicate particles (p = 2,3) and a specific radioactivity
of = 0.5 Ci/gm can be produced to yield aerosols ranging in size from
3.0 to 0.60 urn aerodynamic diameter. These particles are suitable for
human inhalation studies involving initial deposition and early clear-
ance. The maximum time required to achieve a 0.5 yCi initial lung
deposit is 25 to 30 min for the smallest particle size. Similarily,
monodisperse radioactive gold aerosols can be used for similar studies
in Beagle dogs, rodents and man. Gold as a basic laboratory aerosol is
desirable because it is stable under the electron beam of an electron
microscope, insoluble, chemically inert, non-hygroscopic, non-toxic nnd
Available in high purity Radioactive gold aerosols could also find
applications where a unipolar charge distribution is desirable, Goid
aerosols also have a high density (o - 19,3} which might be important in
evaluating inertial sampling devices designed to sample atmospheres con-
taminated with compounds of actinides.
ACKNOWLEDGEMENTS
The authors are indebted to Mr. Emerson Goff and Mr, Michael Rios for
preparation of illustrations, to Mr. Fred Rupprecht for editoris" as-
sistance and to Mrs. -Judith Miller ror r./oing the manuscript. This re-
port was prepared under research supported by the National Institute of
Environmental Health Sciences via ERUA Contract E(29-2)»1013.
REFERENCES
1. Kotrappe, P. and ?1. £, Liqht. Dei., »:; Study of the Exchange fv.r-
acteristics of Monx^tori T'oni^p L'l.i;, "':>r F1r,sion Product Caticfv- (-.:•?
Use in the Generation uf Iri^oluble Aerosols USAEC Research :->f
-------
5. Raabe, 0. G., H. A- Boyd, G. ?1. Kanapilly, C. J. Wilkinson and G.
J. Newton. Development and Use of a System for Routine Production
of Monodisperse Particles of 23SpUQ2 and Evaluation of Gamma
Emitting Labels. Health Phys. (in press)
6. Sneed, M. C., J. L. Maynard and R, C. Brasted (eds.). Comprehensive
Inorganic Chemistry, Vol. II, Cooper, Silver and Gold, pp. 222, Van
Norstrand Co., Inc., New York, 1954.
7. Raabe, 0. G. Aerosol Generation and Characterization, In: Inhala-
tion Carcinogenesis, Hanna, M. G., P. Nettesheiiri and J. R. Gilbert.
(eds.). Oak Ridge, Tennessee, U. S. Atomic Energy Commission
Division of Technical Information. 1970, pp. 123-172.
8. Raabe, 0. G. The Dilution of Monodisperse Suspensions for Aerosol-
ization. Amer Ind Hyg Ai-cc .). 29:439-443, 196C.
9. Boecker, B. B., F. L. Aguilar snd T. T. fiercer. A Canine Exposure
Apparatus Utilizing a Whole Body Plethysmograph. Health Phys.
10':1077, 1964.
10. Raabe, 0. G., Hsu-Chi Yeh, G. J. Newton, k. F. Phalen and D. J.
Velasquez. Deposition of Inhaled Monodisperse Aerosols in Small
Rodents, To appear in Proceedings of Fourth International Sympo-
sium on Inhaled Particles and Vapours, Edinburgh, Scotland, Sep-
tember 22-26, 1975.
11. Yeh, Hsu-Chi. G. J. Newton, 0. G. Raabe and D. R. Boor. Self
Charging of l^Au-Labeled Monodisperse Gold Aerosols Studied With
a Miniature Electrical Mobility Spectrometer, Aerosol Science
(submitted for publication).
143
-------
AEROSOL OS fit RATION FOR
INDUSTRIAL RESEARCH ,"ND PRODUCT ITS
by
EUGENE, ?, . GRASSEL
SR. RESEARCH CrlEMS'I, DONALDSON COMPAQ/, INC,
ABSIRACY
Aerosol generation plays an important role if dr"'»=? ^ping reliable air
cleaning systems. Aerosols needed for product testing, where the em-
phasis is on life performance, are different and consequently require
a different method of generation thar. those normally used in research
and theoretical studies, A number of aerosol generators and dust feed-
ers have been used during the past 20 years to supply the broad range
of aerosol needs for research, qualitv rontrol and performance testing.
The aerosol needs range in type troti lerosols containing uniform parti-
cle size to aerosols having br^«d distribution? in a spr.etfric size
range. They range in capacity fren> small volumes sr low concentrations
to large volumes at high concentrations. The systems reviewed include
commercially available units and specially adapted equipment. They are
described in terms of aerosol needs, method of generation, kind of
aerosol produced, methods of particle sizing and method of measuring
performance. Experience with these general methods is discussed in
terms of equipment, limitations and problems in producing fire particle
aerosols for life testing air cleaning systems. Also discussed are
methods of fine particle aerosol generation that have been adapted for
specific applications in lieu of roors general sysueir.s.
Preceding pap blank
-------
AEROSOL GENERATION FOR
INDUSTRIAL RESEARCH AND PRODUCT TESTING
by
EUGENE E. GRASSEL
SR. RESEARCH CHEMIST, DONALDSON COMPANY, INC.
INTRODUCTION
Aercsji general ion is an important. element in developing reliable &i <
cleaning systems. It is industry's tuner, ion to translate understanding,
material?, ideas and manpower into reliable products with a competitive
edge to a profitable market. This competitive edge is defined in terms
of 3ise, price or performance or, in some cases, as sinpjy a solution
co an unsolved problem. This goal it .; -womp] ished by developing uk^
materials and concepts, evaluating prototypes and providing a met hot' of
measuring quality. Each of these steps rray require aerosol testing,
The purpose of this paper is to re i ..ew
aerosols ar.J aeiosol generators used r.u
needs for research and product testing
3s life efficient:;,, dust: capacity' <.n,J c
product design sa p;* -ti-v le :;ize c- ? i ': c ic-
• ien-ry tests, IJ .,- ftMamete-s art arc;'
ili u~ lined b} i!'e ,e r is "i «f f Ijt-ni.- s'!
h^cji of tiie generfcitots cesciibeu r.ere i
as ",.*f 1 S as thei1- present appiicaLion •
Liioas. An effor'; is trade fo ciefr'ne i\
<-:L3 f'ii'ti needed t L- JT.IJKSUS'.; ;j«>r
pa r i. i c lr remove 1 .
twe
sa
nee
nty years experience witn
tisfy an air cleaner company ':-
ds. The life parameters .'nub
r^'~ i lity ^ro as ipip»>r:a;>< ; •
ts determined by initial i-t> „-
:••£;_> sf.em interactions not .-.r.'r1
i..j' 1,0 It M Ja L Lons . Thfr> ,?.rt
i e.c.ed to ^x.-unir.p how we'' ^
i:y ot t!iese test.s have l:Tit'
in acaJemie research.
• J SJHT^ io/i of 3 cclle-
-------
pie or a standard dust, Such gr-ner- -f >'•'?, however, should not be
on cbe acaceraic standard of aerugo i *• ^e. uniformity, but on bow well trie
l meets the industrial r!;-cteti Lo Isolate and quantitavely measure n single mech.inism, phenomen-
on or property a& a function of vat Labi*-": . The emphasis is in quality
not quantity. The quantity .:an be !-ept small by keeping the test sample
small b i< the test particle shculc be round with a veiy small geometric
stan^ar-' deviation to simplify rnaLnerrit ical manipulation of data. Ef-
f?c':s r, £ s;.^s;. em-aei osoj interactions, that mi^ht change the experimental
cor.Jjtion's wi'h i. ime . a>e minir-:; L\:>.\ by using low aerosol concentration
and kfivpin.:; expos t?re tines a-~ slv-s; a? is conrjisuent with the sensitivity
of the n2asa"'.4ng metnod. A» fent^",: •!' that proviti-as 3 few liters per
tnifit'e 01 mon-jii j sper sed ae^osc-1 v /-, '• -, ro.'at i "e Lj ] '-v (Concentration is
usi!'>l L\ adeq;ia*"D.
Most C-L *• he aerosol generators as^-d Lr, indust'cy are for Improving pro-
ducts -ind customer qualification te<.;:inp, T\f eT.rhasis is on quantity
and repr.tduci bil ity , The aerosol must provide an ostiinate &s to how
,v,-i,?lsj prototype-; or nrodection sv&tems
will perform in che field. This requires measuring the at-roaol -system
Li'.terac c Lot.s as life parai!e('ei.s Lhit make up a oif-aiilug system's total
The co'npiexitv of life performance parameters make comparative testing as
useful ;,.s absolute measurements in evaluating th-? merits o£ a system
modiiication or change. Tor such letting, idesli/t-d aerosols ace neither
practical nor desirable, as tuey usually shed little light on overall
per (romance under lield -:onrl j r ] ons - For such studies, broad distribu-
tions are desirable; and, in maiw cases, atterrpts are Fiade to appiroy-
imaLe application conditions in c, t laboratory by dispersing a typical
sample of the mace-rial to be encountered. Because l,f.; producible aerosol cloud with long
*!':'!!*;, .'..'.Ability.
Dcnaldso't Company furnishes f iteration products to tour market area.> as
sho-.)n in Taite 1 Products developed for each of these markets require
a different qualification tesc to evaluate performance in their ajplica-
tiori aro.a. The need to retain a method of testing for the market life of
a pto-rjt.ct and th? range of airfl"v,s, ^.--ncentrations , and kinds rf aero-
147
-------
to
0!
o;
o
a
aj
tt!
e
en
P5
Ex3
tH
,J
M
U.
P*
S
£?
n
Z
a
o
w
<
&'
c/;
i
r'
••,
i E-.-
I
£*-»
£
t *~ >
)~i
>X"
f-
^ ^ '
? ;r'
,< '
> .
' „ „ , _J
I 1
f ...
;' J; !
i <
! >!
1 ^ i
I 5 '
1 5 1
,
•"•""•"——"I
t ',
I :
i
' !
,
! s
i !
•o
e
CO 4-J
C
0.' Cl)
r^ 3
4J K-l
C! M-i
H W
O
"O l~i
a ro
^
| *i- C^
; "£ ^>
*J 0
a o
L
r™* **™~~™m~™"
I
! K
.-<
^t
4»j
^5
i
^
; r"1
' C*
Lw
!
f
!
- .-^
, iif
! *
1 c
[
,
't
r5
'4.
C
f *
' l-<
^
,.
c
-*••*.
5. *,
"C
, ,
*^J -—
'cO
2s
"O
w «
-a i"
. ^ '•— •
01
o
CO
I-l
a.
41
Pi
O
CJ
'"3
r-l
a.
a.
cri
c:
( ,
^~"
"" "" — ™
'ij
j
•X
1 ^
JH'-fltWWW IM«I.>
,,
j^
**"* 2
-.
C
'U CS
:J n
"M °
o --3
3 CO
C
V ^
*4«f -,j
.J
**« ""-
^ r
L'
^.^
"!
^
* ^
g
I,- •"?
•*-* >-
.i
O
v -i J*.
r
'H *! *<
i-^ r s
c^
*- * i
r ~,
0
f— i
cc
i~-t
•* v
™* -, .-- *•- ^ ~ *~- «
?•<
A
ij.
0
r', r''
C' -*-:
" ." S-4
£-< x W
M -f-
j . ^
, -' -~"
'-, ^
' t~;
in
sr
y**4
L'
'^
C i
lO
r—t
^>
*
'^T
'w.
,—
C^t
-
V*
-
V,
O
i-
O^
*
i' >rf
tj ,0
i" >•*-*
•>. d
tT-tj _ ^e
•^ O -r- 'is
y> — r -•; »„
*. Sv -w
'" > S
O
CO
o
c.
1
*&<
r"i|
tft
1
4-
; M
1
i
^
S
en
* >
j
p
o *
e^.
cn
•
c^
L „ j
r — - 1
!
i
••*•> '<•< 1
o o
!^
•^ 0 ,
*j '
^ 5
t? '** 1
a/ - s
p£5 Ci 1
•c-i j
•^ ^^ i
--. -j
•*\l f -\ ^
j
r._ - ^
1
^ j
*
'• '
,
r1'
^
.^ ;
--" .
-
b
*
>•
J ,.
--^«-,w«- ««™--^
L- ' ".
O -;!,
""'
v- V- —
AT -- T
•p^ Ci
^) W *
-^ r •"
"+-« w v
H«< C
**"" *"' * *""
-------
••;.'•' -* ~.: iro:! has result eu ' >'• .:< .v-.i • r '£ aerosol lit-nv:: colors to tree*. Ci
•la\ tc J,3y testing needs. f:.!ieri-c \ ;;->-jtors havs- h'1*.*: gtuupcc- accord inr
i ."iVket area. Th"1 r.ane'rai- r.*. _!,,>,';.•, si ;>f two typ, •; : 1) Those th&-
,wo*?i;,-r ;,r-rosol by dry disp*? /si on of a standard ci'ist or a collected
'•if., .30-' 2) Chose that geneva?-" ••*«:. ngol s either by c •..•nr'C-usa hioa or atom-
iza1 ion of a li-jinc. Most ^.oncr-M - ,'s used for i. , '> resting, are large
py ;. ', !.-• Le generators «ad fall into (he f irpt - -jat-igory , r» he research and
'. i'.it -.article r,«.''n'p Caters usual i > fa ! I into t*.c- spconc- category. The fine-
~.i.t;.c't perr^r^f.or ?rouj) -.vi 1 i ."' "; ' J J ni," rt ' rdiisfr J a'i ar ' -;-ntion as t!.-u
m0. frt...- 1, s FT.' .j»ir cleaning s /s •:.".». ;,'fe .•-'>:, increabed P f f .;.o ieT'c v on fine
par Lit 1 :£. =,..:.n/f .
INDUSTHL- I GENERATORS
Ex:(.INI AIR Cl£^T.Nr MARKET
," rates •;;>,! n i r t"' •.- "•> cov^rad h) ''oarL'<;- pui't-rle genera -
L'o\ ? i;:-*jd in i-v,-1 engine raarkei, art: ; ,r,i!:L«riK.e'J n\ Tabi-i 2, To better
,.n.'' „•; - raaci , Tidus t .•"-,,? I aer^scS ^e: •"; -,i . ion, it is nfc.c-s:;ary ro exairire tnr-
;•>•. Jt ••'> t>t i>ir cl'/fiaig a«_ t; t rir.-"- ;: prc>'''!C'. "as xntrudiicc-J to a market
an.! -• tr!1'. rn^.^ho'- ,/aa -"><-:•'» «t LFO :',•- •.; , s.-'ucu •.:>i^ ' : f ^ ,\i t ion
indv.&t 'ial pr^riices ma'bf- it. >i;ffii ..]' to cK • un.c-'vt Jaces ' f which some
trst pr-">cedurps were first ad^lrd. Even roday. -. quipmeni. or materials
w;. M d:'f-'VS-"> in and '>e well -:-3 •' ,ib • ,• shed befov:.> f.l.t:y arc rf-jc ordf' . The
eoi I . ':isftivy .;-i .'•••• 3 i r-f Is --. ', f ^ Jt:,-t. feeder r*c..d tie evoiticton of stand1-
«rc!'/. d ttsr, dusc are good e»u,pJ eb .
i'.-'H nocu 1C. ran.ove airborne silica .-just in the intake air to tninitnize
vert \as rccogni zod as an ;iir: a^'r cle.iaing designs wore tested on etrsipsnent at an A"i;:ona
i^iock. There is no record as t .. w'len field testing was; fir&c
s.ip; i t-i .. aled ;-. hand dribbling dust from a scoop Into an engine air
c".i »'t r ; -.let ar. an <-'.xped ierit :n design. The du.^i us^ti xvas coll^cr.es''
'•••ii -,- .'-r;:>,cru engine prcvi.,^ ,. /•: >,nd and shLpcod L, t.u> iaburatory ui
3'. •• -:o 1 1 ;,i dr^ms. At first. It war. i r.' pared bv ccrerni.ne throuyii a ,'!00
•'ues • ;; Lc\ 3 a:iu an?lv7cd b-- a .-sieve e;ial}Si3 to d».v3de the mass into two
'.-/-. r-;.ipes. The t?st cooe adopivd r>^ the SAE in 1^41 recomiuendtc eithe
,\ Sfed !. i .'lie?: t. von anal -sis or r. .•• 1 let analysis tor evaiuating the fines.
'•it ' . i. 'r- -, 1.2.1* di s cri bcir i v~t "r , ; L^m co tf.e .. ir c .t^r-.-.r maMcfacturer *-.
549
-------
Table 2, Aerosol Generators used for rhe Fngine Air Cltr.u'. a
Market
| Ff»ed Re,.p Ties: Ve
~sr.gf i (c,,-: ct .f";. •.
',•" r i K h t i ;u r-1 11
-------
It was necessary to increase the p"-r»,entage of fines in order to measure
improvements oil bath nir cleaner uesigns in the test dust and an air
t;i,t; i id'-or method was adopteJ to Jo uiis. During the war years, the air
elufriator method was replaced by a ball milling operation^, Shortly
thereafter, coarse and fine roller analyzed dust became commercially
available. The commercial availability of standardized Air Cleaner
Coarse and Fine dust was not recognised by the SAE until the I960 change
in tho SAE code3.
With the introduction of dry air c -.fanirig systems to thf engine air
cleaning market, it has been necessary to supplement standardized coarse
and fine testing with other dusts to provide adequate design information.
Staticized AC Coarse dust was found to be unsuited for loading studies used
for improving the dust capacity of paper. Total cake depth relative to
dust, paper intei action depth aiad^ t'-e test insensitive to the paper
structuto. Use of fine test dus'_ improves the definition of the paper
strict 'iro but provides little inforraati ->n abc-nt fine particle loading
tha, causes high pressure drops with light loadings,
llciing dust-carbon mixt .res to simulate high pressure drops at light loads
vac unsuecesr.lul , although the dust-carbon ir,ixt;res were of similar com-
position to chose determined from problem situations in the field. This
is undoubtedly due to the inability to deaggiomc rat" the carbon-dust '.sis
and fefd each particle as its discrete size. Standardized coarse or fine
dust does not provide adequate Information abo-st cake support. Du,st
cak*::; formed by most dusts on 1 tber beds will become unstable and migrate
if e-.r.her the pressure drop and/or velocity become too high. The values
at. wiiich this takes place are dependent on distribution of the dust and
the structure ot the fiber bed supporting the cake. For papers with
90 micron maximum pore and dust with a mixture containing a large portion
if -;-l'i micron particles, the values are 20 ft/min 1'ace velocity and lf;ss
than 20 in. wp, pressure drop. A (is.: it containing sect; a mixture is ur.ed
to develop filter papers w'th proper fiber spacin^a to support these
pool" cake rormevs.
A_i r - F lea te_-J Dust Feeder
Tl c air floated feeder shown in Fipure 1 is a fabricated item and con-
sists of a 2.5 in. o.d. tube 21-1/2 in, long with a rounded bottom con-
';aLniru a 1/8 in. dent. A 5/1. in. o.d. copper tube brings air into the
oystym. ThP operation of the air-floated dust feeder depends on the
range of sizes in the distribution of the dust to he fed. It feed:; dust
fv classifying the floating out the fines first. The rate is maintained
by increasing the air pressure to float out larger si^es. The range of
sizes in the distribution of both coarse and fine Air Cleaner dust per-
m1' t.v a reasJOtiable uniform feed. The feeder is operated manually ^r depam'finl on the particle size distribution of the dust and
151
-------
-------
operator judgement. It is unmiiteri for dusts having a distribution with
a Sfnal 1 O~~ g.
The air-floated dust feeder was adopted to test the oil bath air clean-
ers (s type of oil wetted scrubber) sometime prior to 1940. The life
efficiency of oil bath air cleaners, like many steady state air clean-
ing systems,, is a function of the size distribution of the particle? or
agglomerates fed and is only slightly influenced by particle classifi-
cation during feed. This feeder represented an advance and, in the hands
of only a few experienced operators, the disadvantages of the air-floated
feeder were not apparent until it was well established as a standard
method.
4
In 1940 this feeder was proposed to the Society of Automotive Engineers.
The proposed feeder and test procedure were adopted in 1941 . 1953
changes in the Code recommended improvements for increasing test repro-,.
ducibility along i-iith a new type dust feeder called the Ordnance feeder .
Although the new feeder waF m^tor driven and did not classify dust, it
was more difficult to use. Action of the vibrators and piston compacted
the loose dust in tho feeder, sometimes forming a cake that did not move
readily in relation to the cylinder wall causing the feeder to jam.
Until its recent displacement by the automatic feeder, the air-floated
feeders simplicity, portability, ease of set-up and use, and our exten-
sive experience of interpreting results made it the preferred feeder for
most testing. It was used as a single unit for airflows less than 400
cfm and in multiples necessary t;o meet flow requirements on higher air-
flows. It is still specified for qualification Letting of military
air cleaners.
Wright Dust Feeder
The smallest dust feeder and the standard for paper research in our
laboratories is the Wright dust feeder. It is a purchased item normally
employed by others for toxicological studies of animals. Its operation
depends on air picking up the small fraction of a dust cake scraped
into n channel as the compacted dust cake rotates past a fixed scraper
blade. The usable range with silica dust is not as great as the gear
combinations furnished v.-ich the instrument. At zero visibility, 0.02i>
gm/ft^, the lowest feed rate is liirited by the amount of air pressure
and resulting airflow (0.5 cfm) needed to pick up the separated dust.
The highest feed rate is limited by the available motor power and
difficulty of cutting the compacted dust cake. Uniformity of feed is
dependent upon the way the dust is packed in the dust cup. When the
dust cup is properly loaded with silica dust, the average dust feed over
both reasonably long and short incervals is constant, although the in-
stantaneous feed may vary greatly. To obtain good feeds, the cake
153
-------
must be built very slowly in extremely thin layers packed under urr.2-
sure. The feeder cannot be used for every kind of dusu> especial!-
soft materials with high cohesive forces (for example, coal dust';, fhr
size distribution of the airborne coal fed from the Wright fet^-rr r-.v
ve no relationship to the distribution packed in the dusc cup,
Tar Wright feeder has the ad-. ant-age of being light and port -Vole
c?asy to set-up and use,. Except for rb. care necessary in pack >':.--
dust C'i'ps, results are opciator independent. The degree of ag«.". •".
icion is cub jeer to the conctnlration of vaLer v-ipor in tV ;rrr" • r ;•
air^ Loath n'4 r* sv.lt.', ovt filter nrt ^rial •? show vir-teTr-suimrier •/,'-;: ' ; . ' "
char CcT< be n>od ;. r tec* <\Y huouJity cor'tiox. To iir.prove reps oc'jcif" ;
br-'wpen t<'St;s_ che fc.Crr should :v "-vied in a chamber t.> !t«? n-. " •• ^ •
it'ok-t j'-'pi. dc-"«'iop i.n^ ?i*'.rj.f ten;." iy ..u'ii'V-ci threads:' arf' m£cM"Hi "I -i , '- •
Thf -1 ov flow rate,- ef t':u- feeder li,r'i.« -ts use to fi^er ^«';c •' • ••
,RI i,t .TH.C;*---?, anc a v -:», l;uw lev flow c^lt, Its '/ri! ;.=•»: v ,-,. •
I."!/nr'i*'." r > i. o-.i 'vji-tl flow >>nchps s7,;;r life te.'?' J nr il.;", ; -; : ,
p i }',v r a -;:HI r<_ us. It h;>j be^n useti in ;..• 1 ;•,(.> r mefli.^ KISI..'-, f-;. , .- .
2.;.;h £ic',ors as: i) Effect of fiber r.(=d structure on loadiag c:ur---
• -,-r 5 st'-.cs ef .v.^dia, 2} s-ftc-cts of fio-r shape, length, <*>--:>,- tc- :; ' -
^ofi;iiiy_ oa --"-hoct pr-tp^-rt i cs rel.itivr tr( efficiency and ic-M1! -v>;,
C;i.'v x ':iec i' or avornge fihtc sr»?!Ci.^«; cat dv^sr caVe Sir,jno'.;?:,
;o^--- (\u't- r^r'der, H' cf- >• i> i i^ure' 2 and 1>. v^s • ;-
^d. Tr was D.i^c^ '.'-n I -.•>? '•• <<:<.: piinciplc- as s s--s -:'_/-•:
t!,:' 'Tiv>.-rs: ' ; ;,f Mi v,-t .%-:,- ;r Dr. W'u.by's gro>,, '
;" •>.-:• :?• • :n' ^s?1..!'! ;:-' "' •.:„. • -r cru t - • > .1;.; n '.->• * t •.
.' <•* ..
-. 1 •:,' •' -.' ,i .,•<>. - ' . < - .> p, ;
''r.p .. . 1 ';!:• 7 .<" £o*"'i v. i--;; v i.tl t'-f •>... .--.-,;.
••"•.' '' . ., • >k. -.- •? ' o•"•','.•.:'. f k" : .' *.'\ • . : i ;
t .*«•' ' • 7-; - •.
<"...- ?>••>•,.
'"".<•• ' ''. i :. i oi' r- .... ,-3^ ve v^^-.oa fro.n ?ini,:F.
: -. '-.-d(.-r . •• of!f«-' , a? j,i a.j1. •":;. . . >; "> i ")C Of-.Ir.yp " ^ fe» n d in i963
-------
-------
,-?:
i^/,._..^_J_._,.w^ ^
: I ' 3! fe* '•; <"
i''«S*?fi15jB*JS5,»5tr- -.^ «-j>tft'^''^,'*JHK-itafc'^jJuWn'lJgS .''VMC&KW fc.^4 '
I 1 fr '
i: If
?5<
-1 g
r- u y
• 5 n
*ij Jw5
.: •« rv
fr; ,
K-, 1
M*
*-* 5 sii ;; . _
'0:; i
-------
by using different size dust cups and > •..1
The sonic feeder is heavy and difficult to Lraiu.po.c~ - r, cue rxnnr in
the lab to another. Ease of over-pressurizing the cn.r.i.^r to sick up
too much dust or under-pressurizing causing compact-ou oi tne dust by
the pick-up foot, alters the bulk density -and makes tKo letd rate
difficult to estimate for critical testing.
This feeder found its major use in selecting candidate materials for
self-cleaning air cleaners. Its presen*' use is product qualification
testing.
Barrel Feeder
The barrel feeder is a scaled-up version of the sonic d.,st feeder. It
is limited to a single dust container -
-------
Automatic Feeder
The automatic feeder, shown in Figures 4 and 5, is the preferred feeder
for most of our present engine air cleaner test purposes. It was de-
signed and fabricated to supplement a flow controller which automati-
cally compensates for pressure drop increases across an air cleaner as
it. loads with dust. The feeder consists of a straight-sided hopper with
a screen false bottom separated from and mounted above the true bottom
containing a spring auger. A low velocity scraping device is mounted
above the screen and a second identical unit scrapes the- material fall-
ing between the screen and bottom into the offset auger channel. Such
an arrangement minimizes bridging when feeding very fine dust and keeps
the auger channel filled with dust which has a reasonably constant bulk
density. The auger speed is set by digital controls. The feeder is
mounted on a 20 Kg balance so the total dust fed is easily obtained at
the end of tests or at intermediate points. The dust feeder, balance,
&-id all its controls are mounted on a cart for ease of t.vsu3port3tion to
any test site.
This feeder is compact, mechanically reliable, and easy to operate,. It
delivers a consistent particle size over reasonably long and short inter-
val.;. The feed rate can be adjusted for feeds between !J and 230 gm/min
to accommodate testing at zero visibility for airflows between 200 and
V.200 cfm. The major disadvantages are tight packing and auger clogging
••'hen feeding soft or very fine dusts.
Methods of Evaluating Efficiency and Par^icla Size
Life efficiencies are determined from .ueasuring the weight of <"otal dust
fed and material collected on an absolute downstream of the test device^
Almost all of the information about particle size effects, from both thr
laboratory and the field, are genera tea from collected samples,, Size
distributions are determined on a mass basis utLig either the Waitby MSA
Sedimentation method or a Coulter Cornier,, Boih methods provide the
most information about the larf-d.' £i_zef and the Coulter has a lower
rv.-.c -\ .„ al limit of: 007 micron j, With these met iions, however, particle
>:~c- efficiency measurement;? ruude Iw feeding a poiy.'iisperaed sample -uvJ
.-, .-.'ly'ij ng collected samples may tend to show a better fir.cS efllciency
'~'T*n should be expected because there is uo vay to estimate th
1 <;,' uest oust to be dispersed moves toward the fitu.-: sites, ic becomes
,.f
-------
-------
Figure 5 Exploded Drawing c£ the Automatic Dust Feeder
Dual Mecering Section
-------
INDUSTRIAL EFFLUENT MARKET
Duet collection systems such as cyclones, electrostatic precipitators,
scrubbers, and baghouse filters for the industrial effluent market also
pre-date the extensive interest in submicron particles, refined air
cleaning theory, or legal needs to clean air0 At the time of their
introduction, they were sold on the basz.3 of either economics or public
relations. In all cases, the emphasis was on the removal of coarse
particles and mass efficiency.
The portion of the market covered here is baghouses and cabinet cleaners
with airflows from 200 to 30,000 cfm. Conventionally, these cleaners
are physically larger for the same airflows and require a different kind
of filter material than is used in engine air cleaners. This is because
they are intended to operate in a steady state condition, i. e., a con-
stant pressure drop. They require constant cleaning by reverse air jets
in order to operate in high dust concentrations for extended periods of
time. Materials used in filter bags are purchased to specifications.
The filtration characteristics for selecting materials are determined
by testing on a small flow bench with a back flushing capability. Dusts
are applied to the surface using a spring type auger transport system
and a strong air jet to aspirate and deagglomerate the sample to form
the aerosol cloud. After a number of loading-unloading cycles have been
run and a steady state has been^reached, penetration of the materials,
normally expressed in grains/ft air, is determined by sampling down-
stream of the test sample using a millipore filter. If size efficiency
is determined, it is from Coulter results run on samples of feed dust
and the dust on the millipore downstream of the test sample.
Tests are run on full size units using either a straight through or a
looped system. There are no standard test procedures, dusts, or feed
rates in this industry as provided by the SAE code in the automotive
industry. The method of testing has been left to the discretion of the
manufacturer. Duramite (a calcium carbonate dust), fly ash and iron
oxide are representative of the test particles used. Test criterion
requires a 24-hour continuous test without a change in pressure drop,
called steady state conditions. After a steady state condition is
reached, the penetration should not exceed 0.0046 gm/m using a straight
through test.
Two types of feeders, SYNTRON^and VIBRA SCREAK, are used interchange-
ably. The preferred feeder for life testing is the VIBRA SCREW feeder.
SYNTRON is a registered tradename of Syntron Co,, Hornet City, PA.
VIBRA SCREW is a registered tradena-ne of Vibra Screw, Inc., Totowa,NJ
-------
This is a purchased item capable of feeding 3 to 32,000 gms/min. The
feed rate selected for testing may be varied from 1 to 40 grains/ft
(2.29 to 91.5 gm/m3). Assuming a feed of 10 grains/ft3 (22.9 gm/m ) as
normal, this feeder has a capacity of testing systems with airflows of
5 to 50,000 cfm. Under a 24-hour life test, a rotary valve is installed
in the hopper outlet. The dust collected in the hopper ifi returned to
the feeder which recycles it back to the cleaner inlet. Penetration is
determined after steady state conditions are reached by sampling down-
stream of the bag, using either an absolute or an EPA probe.
Sl'NTRON feeders can also be purchased in a number of sizes and feed
rates., They are best used in a straight through mode, although they
can be looped for recycle. For longer runs under recycle operation,
the SYNTRON tends to give ever increasing feed rates with respect to
time as the viscosity decreases with time as powder is recycled., This
increase is accentuated with certain powders if a method of dispersing,
such as a materials handling fan, is used to break up the agglomerates.
In such cases, Coulter measurements indicate a loss of the measurable
fines. This fines loss appears to stabilize the flow characteristics
C'jf the bulk sample. The size distribution in the fead and the method
of feeding dispersed dust cannot be expected to present either the total
'•'inss cr fines at the same rate as can be expected in some Industrial
applications even though the total concentration of airborne duat may be
the sarre. In applications where a large percentage of dust is fines,
such a tfist method may not be a good indicator of either penetration or
rate of blinding.
>i.;'.iag baghouses to an application is usually dependent on the experience
and judgement of an applications engineer and, ac test, suoplementt-d by
a Coulter analysis of a collected sample. If there is doubt about the
percentage of fines, it is the practice to recommend -. larger uui t i.hat
provides more media area per cfm of airflow. Within the limits of
available filter media, design improvement of bnghcuses is a packaging
problem. Applying engineering prdevice normally reserved for the engine
air cleaning market to the bagbouse market has reduced the unit size and
I ucrtv.'sed the media area withoi. _ cHanging the throughput.
SPACE CLEANING MARKET
rht ASHRAE Standard
rhe apace cleaning market presents an entirely different set of para-
:u ters for filtration than those in the engine or industrial effluents
r,;irket.sc The application requires moving large quantities of air with
liuwers having low static pressures. 'cV>r this application, filter media
fac,- velocities may reach 300 ft/mln, but pressure drops, even in the
condition, cannot exceed 1 in. of water and may be United to
162
-------
),5 inches of water.
In the home comfort market,, the: f.io .-./ Kurpose of deep bed filters con-
taining 50 microns and larger fibers ^,s to reduce lint collection on
heat exchangers and in hot air duct;:. Higher performance filters are
specified by efficiencies measured K t .^rding to the American Society
of Heating, Refrigeration and ;^r !,.;,ii<*: i: j oning Engineers (ASHRAE)
standard . Filters with efficiencies less than 20 percent or greater
than 98 percent fall outside the scope of this standard.
All ASHRAE standard testing is performed in a large duct with direct
access to the outside atmosphere. The. test involves a specific pro-
cedure and depends on the unlvf-rsal aaiure of aerosols found in the
atmosphere. A test requires an initial estimate of the filter's per-
formance. After its performance is escimated, the filter is placed in
the test duct and the flow rates of the system adjusted. Sampling tips
i*re selected from one of two sizes that best matches the duct velocity
to provide semi-isokinetic sampling of the duct. Flows through the
sampling tips are controlled by critical orifices and measured with dry
gas meters. The downstream sampler is run continuously and the upstream
sampler is set for intermittent operation, so the difference in darken-
ing between the two sampling filters is no greater than 20 percent. The
test length must be adjusted so the staining opacity of the samples
shall not be greater than 40 percent or less than 10 percent. The
opacity of staining is determined by an opacity meter calibrated to
read relative light transmission.
Although the atmosphere provides -3 convenient source of fine aerosol
for filter testing, it probably is far from a standard. To be a valid
standard, it must be assumed that atmospheric aerosol is of universal
composition, although experiment,o Lly if ic is found that the length of
the test varies from day to day an: day to night. In three consecutive
tests on the same filter, one tes" ma^ be higher or lower than the
other two. Information published by others indicates the particulate
matter in the atmosphere is not uni,-ei:?ally uniformly distributed . The
second assumption is that the darkening power in the aerosol is equally
distributed among all particle sizus in proportion to their mass. With-
out a knowledge of the relationship between staining power and particle
size at the time of the test, it is Difficult to correlate this stand-
ard to any other initial effieiv.uc\ Lest, Products manufactured to
this specification are best tested a'. Che customer's site.
It. takes one to four hours or longer to run an initial efficiency test
on atmospheric aerosol. In tests to determine life-loading character-
istics using atmospheric aerosols, the filter of interest is compared
to a standard by running the two in parallel in a common test duct.
Such life tests usually average far^ t.o six months, which is too long
-------
to be of much value in product improvement research, furthermore, the
competitive nature of many segments of this market limits the type of
improvements that can be justified.
AFI Feeder
The AFI feeder provides a method of feeding materials which are difficult
to feed uniformly by other means. These include materials that contain
fibers, those that might deform and fuse on packing, and those tnat are
too free flowing because of their particle size distribution. The AFI
feeder was until recently commercially available. The feeder consists
of a support frame, a tray to hold the dust, a gear train to move the
tray and metering gear, a motor to drive the gear train, a gear that
meters the dust from the channel to the pick-up area, and a pick-up tube
and aspirator for dispersing the dust. The original feeders have been
racdified to provide a wide range of fet-d rates. A full tray holJs about
150 giris AC Fine or 50 gms coal. Feed time can be varied from 2-2/3
minutes to 146 minutes. It is the feeder specified in the ASHRAE code
for feeding ASHRAE dust consisting of 72 percent standardized air clean-
er fine dust, 23 percent Molocco Black and 5 percent #7 cotton linters.
Experience with this feeder has been in feeding AC Fine for military
qualification testing and coal dusts to study performance of air clean-
ing devices for the mining industry.
Turntable Feeder
The turntable feeder, shown in Figure b, is similar in operation but not
design to the AFI feeder. It was developed at Stanford Research Insti-
tute to feed small quantities of ground materials at a precise rate to a
classifier system.
The feeder consists of a turntable rotated by a Variable speed motor.
The table contains a groove to accept dast from a hopper which holds the
supply. The hopper contains agita :ing bars to ktep the dust moving
toward the groove. A scraper oar icmcves excess dMSt from the table to
: ne groove and keeps the dusr in the channel level with t'ie tat-ls cop
The svstem can use eithe. of two types of pick-up:,. The dus.'_ can be
aspirated out of the channel as in the AFI feeder or the system can be
r.ressurized to force the dust out the feed tube, as in the son-!.c feeder,
i'rie turntable feeder has been built to provide a number of feed rate
rauges. The practical limit on both the upper and lower ends <>t' feed
La",e depends on the ability of the balk powder to flow. On the upper
end, the maximum depth and width channel are limited to R size ;:'iat
permits the powder to flow so freely it floods tne table. The icwer crci
Is limited by the groove size which ; s so small that a^i ioir«?raf:ec fror"
•he hopper will bridge and prevent the channel from fillj.n,t evenly. For
164
-------
Figure 6 Tarr IVoie Feeder
-------
AC Fine, the lower practical feeding limit is between 0.1 to 0.5 gm/min,
The turntable feeder supplements the automatic feeder and is the pre-
ferred feeder for low feed rates. It provides a consistent feed and can
be equipped with large hoppers for unattended runs for extended periods.
Its compact size make it a convenient feeder for many tests.
HE PA MARKET AND MONODISPERSED AEROSOL GENERATORS
Interest in fine particles, filtration theory, and HEPA filters grew
out of a World War 11 military need9. Pre-World War H gas mask filters
did not provide the needed protection against toxic war gas when these
gasses were presented as finely divided particles. The application re-
quired a filter having a small size, a low pressure drop at flow rates
below 85 liters/min, and an efficiency of 99.97 percent minimum for the
removal of all airborne particles. HEPA filters, however, have low dust
capacities and a fragility that prevents cleaning,, This prevents their
use in solving most industrial effluent problems, limiting their use to
space cleaning for contamination sensitive applications and industrial
effluents containing highly toxic or radioactive materials.
Academic interest in filtration theory has provided a number of monodis-
persed aerosol generators which we have purchased or copied for research,
calibration, or quality control,, These can be split into drop gener-
ators and thermal dispersion methods. As the design, operation and
application have been adequately described by others in the literature,
only applications of industrial interest will be covered.
Spianing_Disc Generators
Experience with drop generators includes both a high speed and a lov
speed spinning disc generator. The high speed generator is a purchased
item and is used mainly to provide particles to calibrate s
The low speed spinning disc generator'.' was fabricated and used to provide
8 micron and larger particles to study velocities in fiber beds at which
;. v-llec tlon by inertial deposition decreased with increasing velocity.
Thermal Dispersion Generators for Research
Several generators based on the thermal dispersion of a liquid have been
:".'iit and used. The earliest was a generator built un'Jer a government
v-cr.'i-ract in 1955 to study inertial deposition in fiber pads of very fine
t-articles (0C3 micron) using very high velocities (up to 100 ft/sec) .
This generator was capable of producing 50 cfm of aerosol and delivering
it under pressure to a test manifold. Particle size was determined using
a tour position jet impactor with the jets operating La p^taS, lei. The
166
-------
iets covered the 0.1 to 1 micro- rv" . .
aiass for both this and filter jijdies '<•.:• '
Dust loading studies were made >:8Mv .••..• :xi/. o-:.".»).,*
there is some doubt as to the pr;sctl--.-.-, ••;-£> oi th: >
loading for comparing performnnc ' par-i-.rrters of •> • . , < . -
loading, as the particle is depositr; •"•-•• :n oil -'r. :
It is doubtful if thermally generator. '••-•rf.i-.'Ii-^ -. •:, :• • - * "U.uir
solid particles in industrial effluents for lo-dirt,? • i < ;.;.•- in? ay .-vet?,
in baghouse fine particle studies„
The oil orange generator was followed by a Letter tv;..<. i • i^i^i ,>r, tcr
laboratory studies of filter materi-'--. It is tlv- t:nj]!. -. ••••. rhe
available thermal dispersion generator-, piodiu -•<.>., ',,>> • " '.>t 1 ^v
concentration aerosol. 1C is tne oiuv i/a. t-i-i' h •_•:."'•.." ; r:T. IP • pc:t'-:
and the only one that can be ea-ilv •<: lusted to ;>.'. .-:'.; ;• ..-r ic le'. in r
wide range of sizes from DOP, steric acid, or o? >.• -• - /cdios-sners
in measuring both the particle size and ccncenv.T. r'.-' ; ;.r- .1 tru; ex-
tensive use of these earlier generators.
Thermal Dispersion Generators for Qualij..jy ton. .-»; 1_
Early use of DOP as a standard test for HEFA filTr-' %••=? !:s.1; : .1 mili-
tary contract using a standard DOP machine of Vot V ''•:-" j • '-f&g'-.
Particle size adjustment of such generators i . t:-ii> cn::?^' o«nd
of sizes near 0^3 microns,, Partic.lt- size cc.nlroi •;• - r,^,-, ti/o co
temperature changes in the laboratory., A s^milj'r tnr >•' icdein '/err - ott
of equal capacity has replaced this earlier DOf ••:. -,'1 i • f nn-J !' used
routinely for quality control and custome^ r.c";-^1 • ••• '.i;v ,
The largest of the available thermal Jisprtsio;; ;iie:;;' -?";• . • 3 I'-OO '-f:
DOP machine c It is larger and se IL-U;IC ioat1 ,'.r ''.••;- ..-us in ?.V
same manner as the smaller mach.lm.t-1
POTENTIAL NEW MARKET
Recent scrutiny of fine particles in the atn>oiM/a r,- • :-, •••;..?
marketo The market requires cleaning et C. r i e'sr ,. •• ..- •-i
filters with capacities of baghout>e filters. TM <•> ,-.'" u'i.k. h-.r
not a new problem. Until recent ye^rs indu-rtT- -". •• > n,.,
cleaning devices could be, and were, exhpusif-c . • : , -^ , •'<: Tnr;
relatively small mass made them unecouom,'a'l • • • ?;v rfiec
difficult to trace and measure. This is fat-t i • , .-Vfs " c5'
market area with a problem to be solved, "h*
tion needed to measure achievemert toward scK.i ;- , . .-•
quire one capable of producing aercfols in -v/l
a 1000 cfm system. The aerosol sho 1
penetration and cleanability obsefv-.c ..a *~rf •
-------
methods normally used in the engine, baghouse and space cleaning markets
are not capable of deagglomerating the fines to provide a measure of the
problems of field efficiency and cleanability. Thermally dispersed
aerosols provide a measure of efficiency but are either a liquid or a
solidifying liquid which may not duplicate most industrial effluents for
filter loading and cleanability. Preliminary studies to meet our needs
in this area have involved the following generators,
A Carbon Generator
Premature shortening of filter life in air cleaners exposed to diesel
pxhaust has generated an interest in filtration problems associated with
fine particles.
Trie laboratory carbon generator shown in Figure 7 is based on an acety-
lene burner. It is built in an enclosure so the flow of both the air
and gas released to the system can be controlled. The test syste:n uses
pressure regulators, rotometers and metering valves to measure, adjust
and control the flow of both gas and air. Two types of burner heads
have been tested. The first head was fabricated from a pipe cap and
capillary tubing. This head showed decreasing aerosol output as a
spiralling rim of carbon would build around the point at which the gas
emerged, causing short term changes in aerosol output. The second head
design was simply an orifice drilled in a pipe cup. This appeared to
•.ave a better long-term stability, but day-to-day reproducibility of
filter media loadings could not be held closer than a factor of 10 by
'ssirig repetitive settings on the controls. Measurements of the flame
heigl't at which smoking became visible was found tc vary from day to
dc',y, These variances in flame height, were related to atmospheric con-
dition?, primarily changes in atmospheric pressure. The variability of
outputs was determined by flat sheet loading tests i.o 3 constant pressure
ckop at a constant velocity. The generator requvr-..5 monitoring instru-
mentation if it is to provide the reproducibility n/edeJ for industrial
.studies.
A particle even finer than that produced by the acetylene gen.era.tor can
be produced by burning commercial mHthane or propane. TH« particles may
not be carbon, since they come from neither the propane nor methane.
When either gas was filtered through a charcoal bed, neither formed
particles, even when the gas to air ratio was* var?'ed over a wide range.
1i-.e investigations were conducted using a Nuclei Counter.
-!£! -able Air Operated Generators
Atomization of solutions appears to hold tne most promise for solving
•r.-JuGlry's needs of large quantities of fine particles for fiber-bee
.oading studies. One such generation is the Model II NRL air operated
168
-------
ti
O
a
-------
generator. It consists of six modified Laskin jets suspended in a lug
top five-gallon can. This is connected to a high pressure air nozzle
through a pressure regulator. Its design and operation was reported by
NRL in 1963 . A modified version is available commercially.
The system is reasonably portable and easily set up and adjusted. When
set at 30 psi upstream pressure, it provides a reasonably stable, repro-
ducible output as determined with a light scattering photometer. The
six jets are reported to produce 512 L/min with a mass concentration of
4.95 mg/L . Because of contract restrictions, no tests have been made
on liquids other than DOP. Until an instrument free of restrictions is
obtained, we can only speculate as to its ability to provide an aerosol
for industrial testing. If it is assumed soluable salts can be substi-
tuted for DOP, then the methylene blue-fluorescein^->or sodium chloride
techniques ®> •*•'> could be applied. Echols and Young measured the
mass medium diameter for DOP as 0.8 micron with a geometric standard of
1.37. Assuming a saturated solution of sodium chloride, the generator
should produce 0.9 gms of smoke with a HMD of 0,43. Each nozzle oper-
ates independently of the others, which should permit scaline from a
single nozzle to multiples of six to cover a range of outputs to pro-
vide the aerosol needed for most industrial testing.
The primary contract use of this generator has been to examine 400 cfm
HEPA elements for leaks. It provides a field test means of examining
the integrity of installed banks of HEPA filters.
CONCLUSION
T^e performance of a number of generators providing aerosol for indus-
trial testing have been explored. For most industrial purposes, com-
parative testing on a mass efficiency basis is adequate for both product
improvement and sales. It has been shown that the largest number of
aerosol generators to be found in the air cleaning industry are for
generating coarse aerosols from standard dusts or collected samples.
These generators produce the aerosols that support product improvement
end customer acceptance testing. Because the largest dollar volume is
i •- markets where mass recovery is the problem, only a few fine particle
generators will be found. Most of the available fine particle generators
do not provide a solid particle adequate for life testing products. The
use of fine particle generators in industry has been slow to develop be-
cause of the small size of the market for fine particlp air cleaning
devices.
A basis exists for generators and techniques of generating, fine particles
to meet industrial needs for evaluating products for the fine particle
market. For such generators to receive extensive industrial acceptance,
they should provide a reproducible aerosol cloud that can relate to mar-
ket problems and be easily ported, setup, and operated.
170
-------
REFOESCE;'
1. 1941 SAE Handbook. Society of Automotive Engineers, New York,
No Yo, 19410
2. W. W« Lowther, "Dust and its Effects on Air Cleaner Design",
SAE Preprint 365, SAE National Tractor Meeting, Milwaukee,
Wise., Sept. 13-15, 1949,
3. 1960 SAE Handbook , Society of Automotive Engineers, New York,
N. Y., 19600
4. W. H. Worthington, "Proposed Air Cleaner Test Code", SAE Journal
(Transactions). 47. No. 1, pp. 294-99c
5. 1953 SAE Handbook, Society of Automotive Engineers, New York.,
N» Yo, 1953o
6. 1963 SAE Handbook, Society of Automotive Engineers, New York,
N,, Y., 1963.
7. "Method of Testing Air Cleaning Devices used in General Ventila-
tion for Removing Particulate Matter", ASHRAE Standard 52-680
8. Robert E. Lee, "The Size of Suspended Particulate Matter in Air",,
Science. Vol. 178, No0 4061, pp, 567-75, 10 Nov. 1972.
9o Hc Fo Johnstone, Preface, AEG Handbook on Aerosols, Wash» D0 C.,
1950.
10c K. T. Whitby and R0 C. Jordan, Progress Report Research on Air
Cleaning, Aerosol Generation and lonization, U, of Minn,,
Jan. 1962.,
11., To Wright, Rp J. Stasny and C. £„ Lappel, "High Velocity Air
Filters", WADC Technical Report, pp. 55-457, 1957.
12o E» E. Grassel, "Construction of a LaMer Type Thermal Aerosol
Generator for Radioactive Compounds", ORNL 54-3-46, April 19540
130 Edgewood Arsenal Penetrometer, filter testing, OOP, Q107,
B76-2-639 (Chemical Corps Drawing Number).
14. We H. Echols and J. A. Young, "Studies of Portable Air Operated
Aerosol Generation", NRL Report 5929, July 1963.
171
-------
REFERENCES (continued)
15. R. T. Whitby, D. A. Lundgren and R. C. Jordan, "Homogeneous
Aerosol Generators", Technical Report 13, U. of Minn., Jan. 1961.
16. B. I. Ferber, F. J. Brenenborg and A. Rhode, "Respirator Filter
Penetration Using Sodium Chloride Aerosol", Report of Investi-
gations 7403, Bureau of Mines, June 1970.
17. R. G. Dorman and L. E. J. Yeates, "A Comparison of the Methylene
Blue and Sodium Flame Methods of Measuring Particulate Filter
Penetration", Filtration & Separation, September/October 1966.
18. R. G. Dorman, L, E. J. Yeates and P. F. Sergison, "An Apparatus
for Testing High-Efficiency Particulate Filters", Porton Technical
Paper No. 873, September, 1963.
19, A. B. Aigren and K. T. Whitby, "Tentative Recommendation for a
Method of Evaluating Air Cleaning Devices", Tech. Rpt. No. 7,
U. of Minn., June 1957.
172
-------
AEROSOL GENERATION H.-ING FLUIDIZEU BEDS
J.C GUICHARD
IRCHA BP N°l 9171O VERT-le-PETIT
FRANCE
Abstract
A well-know method to obtain airborne solid particles for in-
dustrial or laboratory purposes is to disperse pneumatically
the corresponding powder. The classical way which uses an air
ejector is not entirely satisfactory neither from the point of
view of the break-up and separation of particles and nor from
that of the constancy of concentration. Due to these shortco-
mings, these last years, we tried to build new powder dispersal
device which makes use of fluidizfd beds. These are of two
kinds.
1) Some of them uses the flow properties of the fluidi-
zed powder which is then fed to a separate dispersion system
(which may be an air ejector or else) at a controlled rate.
A system called "pulverotron" and a fluidized bed nebulizer
are described.
2) Other devices use the elutriation properties of
fluidized beds. In this case the fluidized bed by itself cons-
titutes the dispersion device. The most simple case is that of
a bed of monodisperse fluidized particles operating at a flow
rate which is greater than the eiutriation velocity of the
particles. An alternate system is a normal fluidized bed where
elutriation is locally forced. But the most useful system is
that we called "puldoulit". In this case the powder to be dis-
persed is a added to a fluidized bed of larger particles (for
example glass beads). When the system is in operation, the
fine particles are elutriated from the bed.
-------
AEROSOL GENERATION USING FLUIDIZED BEDS
J.C. GUICHARD
IRCHA -France
During the past ten years, we developed new aerosol generators
in the field of fundamental research arid for semi-industrial
applications. They use some properties of fluidized beds and are
able to disperse fine powders. This text explains some of these
new aerosol generators without describing all possible systems
but only a choice among those which were most pratical in use™
For each of them we shall give the basic principles, some numeri-
cal examples of their results and the limits of their pratical
applications but without details as those will be published
elsewhere.
In a first part we shall consider some special properties of the
fluidized bed that we studied because they are of current use in
our generators. Then we shall describe devices which take advan-
tage of the well-know liquid-like properties of the fluidized
bed. Finally we shall speak about more original devices where the
fluidized bed itself is the dispersing system.
I. SOME PROPERTIES OF THE GAS FLUIDIZED BED,
1. 1-The fluidization of very fine powders.
It is not easy to fluidize fine particles because their natural
tendancy to agglomerate. We understand that it is difficult, if
riot impossible, to put them in thp quasi -browniam motion which
is typical of the fluidized bed. If we try to fluidize a fixed
oed of fine powder, we current]} observe the formation of chan-
nels through which the air flows. In view of this experiment, we
understand that it is necessary to fight this channeling and
also to promote some mobility of the particles or of the agglo-
merate egainst each other.
A first measure is to use, as a porous plate, some microporous
membrane (reinforced nylon) because it is advantageous to have
a high pressure drop associated with fine pores. But that is not
sufficient to promote fluidization. A well known although appro-
ximative solution is to use a tapered oed build with some electro
magnetic vibrator fastened to the walls of the fluidization ey-
lynder . That gives good result for powder which are not too
sticky (carbon black for example). This device although easy to
174
-------
operate is unsatisfactory, Ursa we developed another solution^.
We mounted under the microporous j-iombrane a Cagniard La tour Siren
{a device use by firemen) which is essentially a ro^atiny per-
forated disk facing a identical fixed disk. With an air flow
around fifty liter/minute, we obtain a powerful audible sound
(see figure 1). The vibrating energy generated promotes vibra-
tions in the microporous membrane and is gradually absorbed by
the fine powder in such a way that no sound is heard oi.'tsi de . The
effect of the sound is to avoid tne channeling phenomena and to
help the dispersion of fine particles. This system works very
well for fluidized layer not thicker than twenty centimeters.
1.2 - Powder flowing from a fluidized bed through a hole in a
thin wall
In some of the systems using liquid-like properties of fluidized
bed, there is a flow of matter out of the bed through some ori-
fice and this raises the question of the relationships governing
the phenomena. It is a problem similar to that of the flow of a
liquid through a hole in a thin wall, (for this latter case, the
solution is found by application of Bernouilli's theorem)^. We
studied the problem, only from an experimental point of view,
using a variety of different powders and we came to the following
formule
Q = K S
(1)
where
Q is the flow of matter out of the hole in g/s
S the surface of the hole in cm
p the pressure in the fluidized bed at the level of
the hole expressed in baryes
K a and (3 are parameters function of the nature oi
the powder (cohesion for example)
It is useful to notice that Q is independent of the air flow
through the fluidized bed (p is also independent).We give some
numerical values of the parameters in the following table;.
Reference of the powder
Lycopode powder
Glass beads smaller than 63V
with a median diameter 4Oli
Glass beads of 8001J. diameter
0.
0.
O.
K
0568
910
885
0.
1.
1.
a
7'0
13
1?
0.
0.
0.
3
614
548
'i?2
-------
We remark that the numerical values are little dependent on thr
particle size distribution, except if the powder is a very firie
one.
T.3 - Introduction of powder in a fluidized bed - fluidized bed
with a constant level
For most of our generators, it is necessary to supply the flui-
dized bed with powder. There is some solutions available but our
choice was limited by the necessity of getting simple and cheap
device and also because some physical restrictions coming from
the generators themselves (for example introduction at the bottom
of the fluidized bed). Finally wo selected arid archimedean screw
fed by a hopper containing a fixed bed (see figure 3)- Pratically
wv use two models of screws : for free-flowing powder a home-
made teflon screw and for the others (which are numerous) a spi-
ral metallic brush of that sort use for cleaning plumbing. This
last solution, in spite of his simplicity, gave good results.
Very often the output of the archimedean screw is at the bottom
of the fluidized bed for reasons that will become clear later
on. With free-flowing powder, we saw that the teflon screw is
able to push the powder into the fluidized bed only until the
pressure at the end is not too high. For the spiral brush this
pression is zero. This last properties give us the possibility,
if necessary, to build a constant level fluidized bed where the
height is fixed by the position of the screw. If we want (with
one or the other type of screw) introducing powder at the bottom
of the fluidized bed, it is necessary to compensate the opposing
pressure. A simple solution is to make tight the hopper and to
connect its top to the bottom of the bed via ~_ tube, the output
of which is protected by a fine mesh (bv this way the pressure is
sent, to the hopper).
If we use fine or sticky p^w.^et ,• only the metallic brush is able
to hold a constant and regular- flow of matter It ir- further
necessary to vibrate the hopper, using some electromagnetic
vibrator, to ensure a constant flow of the fixed bed into the
screw. On the other hand, we observed th^t the screw is able to
push such cohesive powder against higher pressure than in the
case of free-flowing powders. In any case, the flow of matter
can ba adjust using different rotation speed of the screw.
When it is necessary to maintain a constant level of the. flui-
djr'ed bed (for example because we want a constant pressure at a
point inside the bed) we can solder an overflow where it is
needed .
-------
IT. SYSTEMS USING LIQUID-LIRE PH< J-PTIES OF FLUIKIZED BEDS
The best known way to disperse pooler into aerosol is to wend ii
through an air ejector. Generally the liesagglomerntion is good
but the emission is not constant because of the cohesive pro-
perty of the powder to be dispersed. A lot of work was done to
improve this system and it was natural to try a solution taking
into account the liquid-like properties of fluidized beds,
Nevertheless it is not enough to introduce the sucking tube of
the air ejector into the fluidized bed and some device has to be
conceived to solve the problem. We shall describe two of them.
II.1 - The "pulverotron" (see figure 1)
The fluidized state is made possible by a Cagniard-Latour Siren
which turns at 15OO rpm 2. The sucking tube crosses the fluidized
bed near its bottom. The idea is to try to deliver the matter
across a fine hole drilled in it. The formula (1) srhows as that it
is necessary to have a constant pressure at the level of this hole
however the depth of the fluidized bed should be. For this reason
the fluidized bed is put under pressure by a calibrated capillary
which restricts the output of the airflow. When the level of the
bed sinks, the airflow blown by an air compressor increases due
to the decrease of the total pressure and the pressure over the
bed increases tending to restore the initial value at the level
of the hole. A special chamber went to control and to regulate the
flow of matter surrounds the hole. This chamber, constituted by
a tube 4 mm diameter, is soldered outside the sucking tube. Its
axis is that of the hole, inclined about 15° below the horizontal
because this is the convenient position to mininize the perturba-
tions induced in the bed. This space is surrounded by a bigger
tube and its walls are perforated by a lot/fine holes. This /of
system allows to blow clean air inside the chamber. The space if,
closed (side of the bed) by a hole drilled in a thin wall. The
diameter of this hole gives a means of controlling the flow of
matter. The clean air which is blown into the chamber plays seve-
ral roles. Firstly it avoids blocking the hole by the powder,
secondly it increases the effective pressure in the chamber and
consequently reduces the pressure drop across the hole, thirdly
it disperses roughly the powder avoiding big agglomerates which
could interfere with the regularity of the emission. When the
diameter of the hole is chosen, any flow of powder can be obtained
by adjusting the clean air blown into the chamber. An example of
this is given by the figure 2; the air flow in the sucking tube
modifying the effective pressure drop across the chamber.
-------
FLUIDIZATION AIR
POWDER FLOW
DILUTION CHAMBER
WITH ITS HOLE
CLEAN AIM
CAGNIARO-l ATOUS SIREN
TO THE
AIR EJECTOR
Figure 1. General Structure of a "Pulverotron",
178
-------
FLOW OF MATTER
ISILICE K) IN G/MN
15
10
150L/MN
AIR FLOW THROUGH
THE SUCKING TUBE
k240L/MN
FLOWRATE OF THE CLEAN AIR
THROUGH CHAMBER IN L/MN
i
7
riot-* of nattt1" d3ainst flow rate of the clean a:" for a
"Pulvei">tron" eij'iin^ed with a hole 1.3 mrr. 41
"ofal pre?sur<" at the le^el of the hole 40 c'V'G
-------
This device called "pulverotrom" is well suited for dispersion
of fine powders from some g/mn to 100 g/mn.
II.2- The fluidized bed pulverizer (see figure 3)
With the fluidized bed pulverizer, the powder is pushed out of
the bed and the sucking tube of an air ejector collects it.
The two stages are completely independant. The complementary
air comes from the ambiance, when the air flow sucked by the air
ejector is larger than those coming from the tube which is im-
mersed in the fluidized bed. So modifications of the working
conditions of the air ejector a~e without effect on the perfor-
mances of the pulverizer.
The tube which is immersed in the fluidized bed has a 22 mm
inside diameter and is perforated (about 3O mm of it end) by
four holes of 10 mm diameter. Each hole is covered by a fine
mesh. A larger tube surrounds it arid allows clean air to be
blown into the fluidized bed across the meshes. For the parti-
cular model which is presented on figure 3, the fluidization is
obtained using an electromagnetic vibrator of 5OOg weight. The
powder is introduced at the bottom of the fluidized bed by the
metallic screw and the level is fixed by an overflow. The air
which is blown into the fluidized bed by the perforated tube
carries up the powder at a rate which increases with the air
flow. The air ejector collects the totality and disperses it
into airborne particles. Another way to control the powder flow
is to change the depth where the clean air is blown but is ap-
pears that 7 centimeters under the surface of the bed is a sui-
table disposition. The figure 3 shows how the air allows to con-
trol the flow of aerosol.
The maximum flow of solids which can be obtained is a function
of that which is introduced by the screw ; it is currently about
iiO g/minute. There is no limitation towards weak concentrations
except stability of emission.
I IT. SYSTEMS WHERE THE FLUIDIZED BED IS THE DEVICE WHICH DISPER-
SES THE FINE POWDER
The elutriation of fluidized beds is a well-known phenomena and
was the subject of numerous publications*. Neverthless these
works have not given, for the time being, a satisfactory knowledge
of the mechanisms involved or a full experimental description
valid for each of the different systems. In fact this is a com-
plicated phenomenon and we think that the failure of many a
180
-------
a
110 mm
22mm
120mrn
TO THE AIR EJECTOR
MICROPOROUS MEMBRANE
FLUIDIZATION
AIR FLOW
Figure 3. The fluidized bed pulverizer
181
-------
Research was the lack of simple and well defined system for
which the analysis could be successful. We will present below
two such systems which are useful to study this problem of
fluidized bed elutriation but which are also new systems to
disperse fine powders into airborne particles.
Notice - We did not make use of systems where a powder is flui-
dized at flowrate such that the surface of the bed is elutria-
ted, except for monodisperse particles. The reason is the lack
of laws discribing the phenomena and, consequently, difficulties
to operate satisfactory systems of this type.
III.l - Generators of monodisperse pollens
Pollens are equipped with special structures which facilitate
wind transportation and desagglomeration. For this reason
they give easily good fluidized beds even if they are fine. When
elutriation is active for such fluidized beds, the pollens are
carried away in form of monodisperse aerosols. One example is
lycopodium perlatum (2611) largely used by pharmaceutical indus-
try and which is commercially available. We made a detailed stu-
dy of the laws governing the elutriation of such a fluidized
bed and we came to the formula ':
20.7 - 4.74 LnD
n = 7.96 (q-150)e ~ h (2)
where n is the numerical concentration in the aerosol phase
measured at fifty centimers maximum above the bed
D is the column diameter
q is the flowrate of the fluidizing air for one cm
of free surface of the bed.
h is the height of the bed
In a more general form (2) can be writed
a - b LnD
n=k(q-q0)e~ h (3)
where qo is the airflow at the incipient fluidization
k, a and b are constants which characterize each
powder
Starting from this law it is easy to build any generator with
constant level fluidized bed. The concentration will be adjusted
only by the airflow because n varie linearly with it (see also
figure 7)
Notice - The generators with forced elutriation - A microscopic
examination of a boiling fluidized bed shows that the bubble
182
-------
15
10
FLOW OF MATTER
IN G/MN
10
20
CLEAN AIR FIQWRATE
IN L/MN
Figure 4. PowHer flow of a fluidized bed pulverizer
183
-------
plays an important role in the phenomenon of elutriation.
Firstly the bubbles take an aerosol during their ascension and
particles are liberated when they burst. Secondly when they burst
at the surface, they throw up pieces of the fluidized bed which
can be disintegrated and carried away if the mean air velocity
is sufficient. It is also why the concentration increases with
the air flow. An effective idea is to collect bubbles inside
the bed and to force them to burst through a smaller surface.
That is done by an inversed funnel (see figure 5). The concen-
tration of the emerging aerosol increases when the funnel is
sunk into the bed (that is equivalent to increase q in the
formula 3)•
This system allows weak flow of aerosol from a larger fluidized
bed. It can work with any fluidized bed but the fine particles
are desagglomerated only in the case of pollens.
III.2 - Systems using fluidized beds with two differents consti-
tuants - devices of the "puldoulit line".
Let us considerer a fluidized beds of glass beads diameters
between 100 and 200 11. A good fluidization is obtained for air
flow giving a mean velocity above the bed around 5.3 cm/s. Let
us suppose that fine particles are clung on to their surface.
From an experimental point of view we observed that such par-
ticle are progressively loosened and evacuated in the airborne
state if their elutriation velocity is less than 5.3 cm/s (cor-
responding to particle of 42 "p. for a density of unity). This
experiment gives the principle of the aerosol generation by a
fluidized bed with two constituants. It is possible to make
use of this principle by different devices. We shall give the
description of three of them which are able of producing weak
medium and high aerosol concentration. They have common fea-
tures which are :
good desagglomerat in'j efficiency which was also noticed
by Liu 6~7
good stability of the concentration without any instan-
taneous fluctuations
particle size distribution of the aerosol identical
with that of the fine powder
The "puldoulit" model A 5- The general arrangement of the device
is identical with that shown by figure 3« In place of fine powder,
there is a mixture of glass beads and powder in the hopper. When
the screw pushes this mixture in the fluidized bed, an equal
quantity of glass beads falls down into the overflow and the
184
-------
AEROSOL
FLUIDIZED BED
WITH BUBBLES
1
rFi u-e 5. Pr'nci le o1. the g^r'p-r •*" '" »'ith lorccd <••! utr i ..f'
185
-------
fine powder is dispersed as airborne particle. It is possible
to get some idea concerning the emission mechanisms at work
by considering the simple situation where the fine particles
are totally adhering on the glass beads.
The inside of the fluidized bed is a very turbulent uni-
verse and the air velocity fluctuates : it can reach locally
the values needed for loosen the fine particles. Further, the
bubbles bursting at the surface throw up pieces of fluidized bed
and liberates inner aerosols as explained before .
The glass beads are in quasi-brownian motion and collide.
This mechanical effect can also loosen fine particles as it is
currently observed at macroscopic scale. Besides, there is
another phenomenon which plays a major role. The fluidized bed
is a very efficient filter for aerosols and the particles libe-
rated near the bottom have little chance to reach the surface.
In fact they are captured by some bead and re-emitted later.
This mechanism has probably a favorable effect on the desinte-
gration of agglomerates and it shows us that different parts
of the fluidized bed play different roles in the emission of
aerosols. The main consequence is the build of a gradient
concentration in a complex manner. A good approximation, useful
in practice,is that of the aerosol emission kept under control
by a top layer of fluidized beads (for example this layer would
be 15 mm thick for a bed 120 mm height fluidized at 301/mn)
We also understand that the best result are obtained when the
introduction of mixing is made at the bottom of the fluidized
bed.
The preceding description shows the importance of adhesive
forces and consequently we are waiting efficient parameters
correlated with this forces. In fact, for a stable and repro-
ducible emission, it is necessary :
- to work with constant humidity of the air and to
avoid large temperature fluctuations. The best is often to ope-
rate with dry air.
- to use a mixture where the humidity is kept under con-
trol - the most simple being to dry the mixture before use.
- to ground the fluidization column in view of electros-
tatics phenomena existing in the fluidized bed8.
186
-------
- to use the same fine powder because adhesive forces
depend on the physico-chemical properties of the particles but
also on their previous history (grinding for example).
- on the contrary, for the conditions prevailing in a
"puldoulit",although the fact that adhesives forces are size
dependent, it seems that the probability of reemission for a
particle is independent of its size and it is why the particle
size distribution of the aerosol is just that of the powder
(if the desagglomeration has been good)
Always starting from the preceding description, we understand
how differents parameters control the emission. Let us consider
a fluidized bed where the fine particle concentration is kept
constant.
The aerosol concentration increases with the concentra-
tion of the powder in the bed (pratically we can increase the
flow of mixture or increase the partial concentration in the mix-
ture) .
The aerosol concentration increases linearly with the
air flow as shown below (the law is similar to 3)
But this static description does not apply exactly to the device
working continuously because of an equilibrium between the feed
and the output. Taking into account the hypothesis of a top
layer controlling the aerosol emission. It is easy to write
the fundamental equation :
(4)
where
Q is the volumetric air flow
M the weight flow of the mixture
Q the weight concentration of fine powder in the
mixture
a the weight concentration of fine powder in the
fluidized bed falling into the overflow
C the weight concentration of aerosol
This equation allows to calculate C if the other quantities are
know or to calculate 6 and a to obtain an emission of a defi-
nitive C. In this latter case, it is necessary to know the re-
lationship C = F (a), which is fundamental for a good knowledge
of the device. This law depends on the physical nature of the
fine powder and is only experimentally known. An example is shown
by the figure 6, for an aloxite powder with particles between 1
and 2OVI and a median diameter, by weight, of 5.5"U.
187
-------
10
M AEROSOL CONCENTRATION
IN G/M3
^
/
0,1
/
0.01
0,0001
0,001
0,01
Figure 6. Relationship Of (r<,) for a "puldoulit" working with glass beads
(100 to 200u) and aloxite powder (1 to 20y with 5.5y median
diameter by weight)
188
-------
For the model currently used in our laboratory (see figure 3)
we have obtained good results with aerosol concentration from
some mg/m3 to 1 g/m3 for an air flow between 2O and 60 1/mn.
The emission is very stable but there is some details to observe
if we went reproductible emission. The main problem is a good
preparation of the mixture. It is sufficient to mix the two
constituants in any bottle by shaking, if we want only a stable
emission. But for any mixture a part of the fine powder is adhe-
ring onto the beads and the other is free in the insterstices
of the bed. If the proportion between the two states of particles
is not the same from one to another experiment it could result
a different aerosol concentration. For this reason it is ad-
visable to fluidise independantly the mixture during a few
minutes before use. The shortcoming is the necessity to deter-
mine directly the real value of Q in the mixture.
Notice - If we want to obtain a known value of C when starting
with a fine powder that has not been used previously, it appears
the problem of choosing an initial partial concentration for
the fluidized bed in view of a minimum delay before reaching
equilibrium.
Let us consider a fluidized bed without fine particles and at.
t = o let us introduce the mixture. Let us suppose that the
concentration a of fine particles is instantaneously the same
inside the layer controlling the emission. If we borrow the
simple hypothesis that the relationship between C and a is
approximated by C = ka, the law governing the aerosol concen-
tration is
-( M ^ —) t
C = Cm
where
1 -
P
Cm is the equilibrium concentration
p is the weight of the layer
t is the time
We see that the delay before equilibrium is independent from
6 and is reduced if M increases. This delay is around ten
minutes with the conditions specified by the figure 6, and a
motor running at 1O rpm (M in the order of 0.9 9/s). But it is
more lenghty with a motor 1 rpm and it is advantageous to start
with a predusted fluidized bed.
If all the quantities of equation (4) are know, it is not pos-
sible to calculate the partial concentration of fines inside
the fluidized bed because a is valid only for a top layer. From
189
-------
a practical point of view, taking into account that a is a
maximum for the unknow concentration, it is convenient to try ~?
The "puldoulit" model D - The device is identical with model A
but without overflow and with an electromagnetic vibrator
(Approximatively 10O g) fastened on the hopper. The induced
vibrations help to have a good flow of the fixed bed here which
is of fine powder alone. The archimedean screw is of metallic
brush type and is currently able to send between 1 and 10 g/fnn
of the fine powder at the bottom of the bed. Another useful
detail is the possibility to separate immediately the hopper
and the screw from the device and consequently to know by weigh-
ting the quantity of powder used up during an emission.
With this model where the powder is not premixed with glass
beads, the mechanism of capture and reemission plays a major
role. But the lack of predispersion as for the model A explains
that the desagglomeration efficiency is lower for sticky powder.
Consequently it is necessary to introduce the powder near the
bottom, to use a thicker fluidized beds/to work with dry / and
powders. Nevertheless it can happen that somme agglomerates
are not broken. In this case they rise and float on the sur-
face giving a layer which has to be eliminated from time to
time. Finally it appears that the desagglomeration capacity
related to the flow of solids introduced is limited but this
phenomenon was not studied quantitatively.
With the present model of "puldoulit D" it is possible to
generate aerosols with concentrations from some 1 g/m3 to
1OO g/m3, the air flow being between 'ir> and 6O 1/mn.
Generator giving weak concentrations - Starting from the same
idea it possible to have a generator giving weak aerosol concen-
trations which can be measured with an optical counter. That
device is very usefuL for experiments in aerosol physics. Let
us consider a fluidization column in which we poured a mixture
of glass beads and fine powder prepared as explained earlier,let
us follow the concentration function of time. At the begin-
ning of fluidization we observe a high concentration that de-
creases rapidly and appears stable after some hours. At this
point the consumption of powder is small compared to the quan-
tity available inside the bed and which is totally sticked on
the surface of the beads (the powder that was free has been
eliminated). The stability of concentration and particle size
distribution is maintained during a long time giving a calibra-
ted generator. Nevertheless three points have to be noticed.
190
-------
- At the beginning of each experiment, there is a peak
Concentration. A progressive increase in place of starting with
full airflow, smoothens this phenomenon.
- Starting from two mixtures with the same value of Q ,
we do not generally end with the same numerical concentration
of aerosol.
- It is necessary to operate with air flow of controlled
humidity and to keep the fluidized bed at this level of humi-
dity between two successives operations.
This generator gives us a useful way to study the fundamental
problem of fluidized bed elutriation in the particular case
of two constituants. Such studies have not been systematically
undertaken. Nevertheless the figure 7 shown the relationship
between numerical concentration and air flow for a fluidized bed
where the initial value of Q was 1% and the equilibrium concen-
tration in the top layer O.Ol4%. The concentration increases
linearly with the total air flow and it is possible from the
results of the figure 7 to show that the particle size dis-
tribution is independent of that air flow.
CONCLUSION
The details given in the text are sufficient to build with
success the generator well adapted to the particular problem
in view. The dispersion of fine powders using fluidized beds
is the best solution for a lot of pratical problems.
191
-------
AEROSOL CONCENTRATION
IN NUMBER/CM3
PER SIZE RANGE
0,3- 0,5 \i
50
10
0,5-0,8^
4 (q-qj IN cM3/s
Figure 7. Aerosol concentration versus flowrate foi a fluidized bed
110 nun diameter 40 mm height usinc a mixture of glass beads
and aloxite powder
192
-------
REFERPMt Ei:
1.- F.A ZENZ arid D.F OTHMER - Flu idi zation and iluid-particle
systems - Reinhold publishing Company - New York I960
2.- J.C GUICHARD - Approvisionnerct-nt constant en poussiores a.
1'aide d'un lit fluidise (Presented at. the 8eme Colloque
International sur les poussieres - IRCHA - STAUBFORSCHUNG-
INSTITUT - Paris 17 - 18 Novembre 1966 )
J.- L. MASSIMILA, V. BETTA and C. BELLA ROCCA - A study of
streams of solids flowing from solid-gas fluidized beds
A I Ch E Journal 5O2-508 September 1961
4.- J.F DAVIDSON and D. HARRISON - Fluidization
Academic Press London and New York 1971
5.- J.C GUICHARD - Anwendung der wirbelbett-Ausschlammung
bi der Herstellung von Aerosolen
Dechema Monographien 59 N°1O45 : 247-269 - 1968
6.--:B.Y.H. LIU, V.A MARPLE, K.T WHITBY and N.J BARSIC - Size
distribution measurements of Airborne coal dust by Optical
Particle Counters
American Industrial Hygiene Association Journal 443-431
August 1974
7.- K. WILLEKE, C.S.K LO and K.T WHITBY - Dispersion charac-
teristics of a fluidized bed. Aerosol Science 5 : 449-4^ri
1974
8.- J.C GUICHARD - The triboelectrification of metallic parti-
culate aerosols generated hy a two-component fluidized bed
Staub Reinhaltung der Luft (English) 33 : 174-179 April 1973
L93
-------
LARGE FLOW RATE REDISPFFSJON AEROSOL GE^ifRATOR
Fred Moreno, Dale Blann
Aerotherm Divislon/Acurex Corporation
485 Clyde Avenue
Mountain View CA 9404.ect currently under construction.
195
Preceding page blank
-------
LARGE FLOW RATE REDISPERSION AEROSOL GENERATOR
Fred Moreno, Dale Blann
Aerotherm Division/Acurex Corporation
485 Clyde Avenue
Mountain View CA 94042
BACKGROUND
In Spring of 1974 Aerotherm began a program for the EPA Control
Systems Laboratory to develop and construct a large flow rate redisper-
sion aerosol generator for use with their Particulate Aerodynamic Test
Facility. The test facility is a specially designed and constructed
wind tunnel dedicated to aerosol research. The Facility specifications
are shown in Table 1.
Table 1
Particulate Aerodynamic Test Facility Specifications
Test section size 2 ft diameter by 40 ft long
Test section velocity 5 to 90 ft/sec
Flow rate Up to 17,000 cfm
Temperature 70°F to 450°F
Dew point 50°F to 130°F
Test gases Air or combustion products
Configuration Closed loop gas flow
Open loop dust flow
Dust is injected upstream of the test section and removed from the gases
downstream of the test section in a baghouse (see Figure 1).
When first constructed, the facility had a small aerosol injection
system capable of generating very light loadings typical of stack outlet
conditions in the tunnel test section. It was decided that it would be
desirable to create dust loading conditions within the tunnel which are
more typical of the conditions at the inlet of particulate control
196
-------
(S,
oj
T3
O
O>
3
O
i-
ro
en
u_
-------
devices so that instruments designed to operate in this environment
could be evaluated and prototype-sized particulate control devices could
be tested in a laboratory environment.
The general requirements of the aerosol system are as follows:
t Three grains per cubic foot dust loading in the test section
at 17,000 cubic feet per minute flow rate implying a feed
rate of 480+ pounds of dust per hour
• Capability for around-the-clock operation
t A large fine particulate fraction in the dust injected into
the test section.
In addition, the following specific requirements peculiar to this instal-
lation had to be met:
• No dilution of the tunnel gases (which may consist of combus-
tion products doped with SOX and other gaseous constituents)
t Temperature matched flow streams such that the temperature of
the gases injecting the particulate into the tunnel has the
same temperature as the gases flowing through the mainstream
of the tunnel
t Tight space and layout constraints dictated by the physical
facility
APPROACHES CONSIDERED
A number of approaches were considered in the initial phases of
the program and each was judged on its merit relative to this installa-
tion. The first considered was a single component fluidized bed utili-
zing particle dispersion by means of aspiration (see Figure 2). This
system fluidizes dust directly and then extracts it in the fluidized
form by means of an aspirator, as illustrated. The feed rate of dust
into the aspirator is controlled by the flow of dilution air injected
at the extraction point within the fluidized bed. Our experience with
a prototype of this system indicated very good flow control capability
and moderately good redispersion of the dust. However, this approach
requires high pressure air which must be carefully dried to prevent
icing within the aspirator. Reliability was questionable because erosion
frequently takes place within the aspirator. There is an additional
problem of dilution of the tunnel gases by means of the compressed air
used to drive the aspirator and convey the dust into the tunnel test
section. Finally we noted high levels of static charge generated at the
system outlet.
198
-------
cuvr
t
\ \
ME-
A-KC.
Figure 2. One Component Fluidized Bed with Aspirator
199
-------
The second approach considered is the two component fluidized bed
illustrated in Figure 3. This approach is based upon a design originated
by Dr. J. C. Guichard of I.R.C.H.A., France1. The system utilizes a
fluidized bed of glass beads of approximately 0.010 to 0.030 inch diam-
eter fluidized by the passage of the appropriate gases. The dust to be
dispersed is injected into the side of the fluidized bed and subsequently
coats the glass beads. Dust particles are removed by the aerodynamic
action of the fluidizing gases and by the mechanical action of the glass
beads as they impinge upon one another. This system utilizes low pres-
sure gases and therefore has low power consumption. Further, it has
demonstrated good dispersion efficiency. The bed can be fluidized by
means of the gas available within the tunnel, and the flow rate of dust
is controllable by means of a variable speed screwfeeder as illustrated
in Figure 3. By calibrating the screwfeeder for each dust of interest,
the operator can inject a known and controllable rate of dust into the
tunnel test section. We experimented with this approach previously in
work done for the EPA Chemistry and Physics Laboratory wherein we de-
veloped a redispersion aerosol generator having a maximum feed rate in
the range of 60 to 80 pounds per hour. Table 2 summarizes the advantages
and the disadvantages of the two approaches considered as the most via-
ble for this particular application.
Table 2
Advantages and Disadvantages of Various Approaches
Fluidized dust
with aspiration
Two component
fluidized bed
Compact, small feed
1ines, large turn-
down ratio.
design.
_ turn-
simple
Can utilize tunnel
gases, good disper-
sion, low power con-
sumption, low static
charge, flow rate
controlled by screw-
feeder.
Requires dust that
can be fluidized;
high pressure air.
Air dilutes tunnel
gases. Possible
erosion of aspirator.
Generates high sta-
tic charge.
Turndown limited by
screwfeeder. More
complex system, pos-
sible attrition of
beads.
200
-------
Figure 3. Two Component Fluidized Bed
201
-------
The approach selected was to utilize a large version of the two
component fluidized bed examined in the earlier program. The approach
was simply to increase the bed diameter until an adequate area was ob-
tained for conveying dust into the airstream at the desired flow rate.
It was quickly observed, however, that there were problems with dust
injection and mixing in the bed. The dust was fed into the side of the
bed by means of the screwfeeder as before, but failed adequately to mix
throughout the bed before being elutriated into the fluidizing gas stream.
As a result there was a stratification of the exit gas stream with a
higher dust loading in the vicinity of the injection location, and a
lower dust loading on the opposite side of the fluidized bed. In addi-
tion, the excessive concentration of dust in the region of the screw-
feeder inlet resulted in poor dust dispersion because of saturation of
the glass bead media.
The fluidized bed aerosol generator requires a fairly good quality
of fluidization with no violent bubbling, slugging or fissuring within
the glass media of the bed. If any of these conditions occur, it per-
mits the passage of dust out of the bed without being first acted upon
by the glass beads. In an attempt to increase the mixing in the bed
the bed depth was increased. This succeeded only in increasing the size
of the bubbles which were generated in the bed, and also increased the
tendency of the bed toward a slugging mode of operation. The result of
the increase in bed depth was thus a poor quality of fluidization and
thus poor redispersion of the dust into the gas stream.
A variety of other dust injection schemes were considered in order
to circumvent the problems experienced with the large diameter fluidized
bed. The first and most obvious is multiple injection ports at various
locations around the circumference of the circular bed as illustrated in
Figure 4. The problem with this type of approach is that the larger
fluidized bed exhibits recirculation cells in the glass media, as illus-
trated in Figure 5. The circulation cells are fairly stable in their
behavior and are quite effective in preventing dust injected at the
perimeter of the bed from mixing into the center of the bed. Thus the
problem of saturation of the glass beads media in the vicinity of the
injection ports would remain.
Injection through the gas distributor plate at the bottom of the
fluidized bed was examined but rejected because of the expectation that
there would be poor quality fluidization in the region of the injection
location due to the lack of fluidizing air in this area. In addition,
this approach would require the operation of the screwfeeder in a ver-
tical mode while in fact the screwfeeder operates satisfactorily only in
a horizontal orientation.
*
*
202
-------
Figure 4.
Circular Fluidized Bed with Multiple
Injection Parts (Plan View)
Figure 5.
Recirculation Cells in Larger
Circular Fluidized Beds
203
-------
The final approach considered was a rectangular fluidized bed
With multiple injection locations as illustrated in Figure 6. The rec-
tangular fluidized bed removes the problem of circulation cells if the
bed width is carefully selected. However this approach requires numer-
ous screwfeeders in order to effect multiple-injection locations and
this would be attended by a complex materials handling system to feed
dust to the several screwfeeders as would be required for round the
clock operation.
Because of the problems and complexities associated with large
scale two component fluidized beds, a new approach, the three component
fluidized bed (based upon an idea of Dr. J. C. Guichard) was considered.
The three component system shown in Figure 7 utilizes a screen support-
ing a layer of steel or lead beads which in turn is covered by another
layer of glass beads of comparable size. The fluidizing gas passes
through the screen and is distributed by the layers of heavy beads. The
gas velocity is selected such that the heavy beads remain stationary
while the glass beads are fluidized. The heavy beads are, however, near
incipient fluidization. The advantage of this approach is that dust
laden gas can be fed into the bottom of the apparatus passing through
the screen and the stationary bed of heavy beads and subsequently into
the fluidized bed of glass beads where the normal redispersion action
takes place. As the heavy beads become plugged with dust the pressure
drop across the layer of heavy beads achieves the point where fluidi-
zation occurs. At this point the layer fluidizes thereby cleaning itself,
and then quickly reestablishes stationary behavior when the pressure
drop falls. The result is a fluidized bed capable of being operated with
dust laden air without plugging of the gas distributor.
The primary problem in the initial design was determining the best
technique to introduce dust into the bottom of the system. The selected
approach was to utilize a smaller diameter two component fluidized bed
using steel or lead beads as the fluidizing media. Dust is fed into the
side of the two component bed by means of a screwfeeder as illustrated in
Figure 8. The lower bed operated in a violent bubbling mode which serves
to break up the agglomerates. The gas stream coming through the lower
bed carries the dust to the bottom of the second fluidized bed. Redis-
persion is completed by passage through the second bed. The advantage
of the two stage systems created by the vertical stacking of two fluidized
beds is that (in theory) the system can be scaled to virtually any size
because the dust stream conveyed into the bottom of the fluidized bed
can assume virtually any diameter.
It was anticipated that a "puffing" type of behavior might occur
when the stationary heavy bead layer fluidized during the cleaning cycle.
It was decided that if this problem were severe it could be solved by
204
-------
8fD
Figure 6. Rectangular Fluidized Bed with
Multiple Injection Ports
20!
-------
PUVT
Figure 7. Three Component Fluidized Bed
206
-------
OUST
Figure 8. Two Stage Aerosol Generator
20"
-------
stirring the layer of heavy beads with moving fingers reaching through
the fluidized bed. The stirring action would then keep the layer com-
paratively clean. An experimental program was designed to examine this
and other potential problems.
EXPERIMENTS
The experimental apparatus utilized an Acrison Model 105Z variable
speed screwfeeder feeding into an 8-inch diameter fluidized bed utiliz-
ing 0.020 to 0.030-inch diameter steel beads as the bed media. The
gases passing through this fluidized bed subsequently passed through a
low angle diffuser and thence into the bottom of a 16-inch diameter
three component fluidized bed.
The initial experiments were conducted utilizing flyash derived
from a coal fired power plant. The flyash was fed through the screw-
feeder and the particulate laden gases emanating from the top of the
system were collected in a small baghouse. The first problems noted
when operating at the design point (approximately 500 pounds per hour
feed rate) was that the bottom screen assembly of the top fluidized bed
plugged with dust over a period of several minutes. The solution to
this problem was to vibrate the screen assembly from below by means of
a rod which passed from the bottom of the screen assembly through the
center of the lower fluidized bed into the air plenum where it was con-
nected to a variable amplitude vibrator. (See Figure 9.)
The immediate plugging problem was solved by vibration, but it was
observed that as time progressed the quality of fluidization began to
degrade with channeling and fissuring occurring within the upper fluid-
ized bed. It was hypothesized that the poor quality fluidization was
probably due to the plugging of the heavy bead layer by dust because
flow rates were substantially below those required to permit fluidization
of the heavy layer. The solution to this problem was to increase the
depth of both the glass and steel bead layers, increase the flow rate of
the air, and increase the vibration amplitude of the supporting screen
assembly. It is postulated that the increased vibration gives the heavy
beads sufficient relative motion with respect to one another that they
can continually clean themselves without having to be fluidized.
The above modifications permitted operation for a greater period
of time before air channeling and fissuring were again observed in „ ,
localized regions of the upper bed. Unlike the previous case, the re-
gions of poor quality fluidization appeared to be fixed in location '
rather than random in occurrence as was the case previously. There
appeared to be two reasons for this. First, there was a selective flow
of air through one side of the bed which was traced to a failure to ade- ••
quately level of assembly thereby causing a greater depth of beads to k
208
-------
Figure 9. Vibrator Arrangement for Upper Bed Screen
Assembly
209
-------
migrate to one side of the bed than the other. Air then passed prefer-
entially through the shallower portion of the bed. This gross maldis-
tribution of air flow was fixed by careful leveling of the screen and
bed assembly. Second, it was necessary to reconstruct the screen assem-
bly in order to reduce the size and number of vibratory nodes. These
regions were regions of inadequate agitation and thus plugging of heavy
beads occurred within these locations.
Correction of these minor mechanical problems permitted stable
operation for a still greater period of time. However it was observed
that under some conditions an unusual instability would occur wherein
the upper fluidized bed exhibited an unusual "hopping" or "slugging"
behavior. The behavior of the bed was such that the entire mass of
glass beads contained in the upper fluidized bed would jump up and down
in an oscillatory behavior. This unusual phenomena was traced to a
pneumatic coupling of the upper and lower bed through the air stream.
Instabilities in the upper bed would change the pressure drop of the sys-
tem. The reduced pressure drop in the upper bed would then cause an
increased air flow in the lower bed which in turn caused another change
in flow rate. The result was a feedback system with a resonance occur-
ring at a frequency of one to two cycles per second. During the hopping
behavior both beds would jump up and down in concert with one another.
The problem was ultimately corrected by adjusting the air flow rate and
the depths of the upper and lower beds in order to get an appropriate
pressure drop regime such that instabilities did not occur.
The solution of the above problems led to satisfactory operation
with flyash. The system operated successfully at feed rates of up to
600 pounds an hour in a consistent and reliable fashion.
Once operation with flyash was established the dust was switched
to black iron oxide particulate in order to see if the system could
handle a dust with a much finer size distribution. Initial tests again
showed a plugging problem associated with the support screen of the
upper fluidized bed. The plugging was avoided by increasing the screen
size. This largely eliminated any problems with plugging except for
occasional erratic behavior of the upper bed. Several times the system
exhibited poor quality fluidization and ejection of glass beads from the
bed indicating severe channeling and constriction of flow. The source
of this final problem was not ultimately explained because it happened
so irregularly. It is theorized that it was due to the moisture content
in the dust or the air stream which causes this particular dust to
assume a caking behavior which makes it difficult to handle.
SAMPLING RESULTS
A brief set of particle sampling experiments were conducted
towards the end of the experimental program. The objectives of these
210
-------
;.ests was to give a quick look to evaluate the redispt-» s :;;r. >-i-.';:. ;f."r
01 the system. It was felt that an in-depth measurement prr,yr.^, «a< ;'ot
required because the fluidized bed technique has been well examined ir,
'he past.
The sampling apparatus employed utilized a series of sUged
..-/clones calibrated and loaned by Joe McCain of Southern Re?f.au,n ]nsti~
••jte. The apparatus is shown schematically in Figure 10. 'he cyclone-.
have a fairly well defined cutoff point with the larger cvJ^ne na>,n?,y
j 50 percent cut point of 3.8 microns while the smaller cyclone had a
;-<(j percent cut point of 0.63 microns. Normally the cyclones are foil civ-;
:.v a filter, but a variety of problems dictated that only V'>e approxi-
4ce percentage of dust falling ir, the smaller cyclone c.;i te -;cterm'red.
The results of the brief sampling experiments are shewn in f-'i'j-
vires 11 and 12 for flyash and iron oxide respectively. It ibn^:! bo
nted that in some conditions with iron oxide in excess >.>f if; pe; cent o*7
:.he total dust emitted from the bed was smaller than 3.8 rr-c.-ois. Tre*-e
;"i an unexplained trend wherein the dispersion efficiercy o~ re cyst em
appears to increase with higher feed rates of dust when one would expect
d decrease. If one were to continue increasing the feed rate one would
expect saturation to occur, but apparently the system was operating in a
-egime sufficiently far removed from saturation that dispersion efFi~
ciency was maintained over the range of the feed rates examined. We
have no current explanation for this anomolous behavior,
OVERALL SYSTEM DESIGN
Based upon the excellent experimental results derived froni the
crude laboratory apparatus, an overall system has been designed. It -'s
currently being fabricated for installation at the Paniculate Aero-
dynamic Test Facility. The system block diagram is illustrated in Fiq-
ure 13. The gas utilized to operate the fluidized bed is extracted Vorn
the wind tunnel and under conditions when high moisture content s beiny
•ised in the tunnel gases, the gas is passed through a chiPtr to remote
excessive amounts of water. The gases then pass through a biower, and
then through an electric heater which heats the gas stream such thit the
fixture of gas plus dust has a temperature matching that in the wind
tunnel test section. Gas is then fed into the lower flu^'d-ized bed wnere
.lust is injected by a variable speed screwfeeder with dust supplied f^o.':
din integrated materials handling system drawing upon bin storage located
in an adjoining room. The dust is dispersed in the first fluidized bed
and then passes through the second fljidized bed for final rechspersion.
'"he dust exiting the top fluidized bed enters eight 2-inch diamf.er
s-ansport tubes which carry the aerosol mixture into the wind tunnel for
•njection upstream of the tunnel test section. The aerosol is ir.jectea
through nozzles with tip diameters seVcted to optimize f:he irixi/io
211
-------
PUMP
Figure 10. Sampling Apparatus
212
-------
too
10
O.I
0 O
- ° O
O
&
A
O >=»
100 loo
DUST
4oO
Figure 11. Sampling Results for Flyash
213
-------
D
0
0
s
s
100 loo 5oo
(oOO
Figure 12. Sampling Results for Iron Oxide
214
-------
OUVT
Figure 13. Block Diagram of Aerosol Generation System
215
-------
^n
*
§
h
0
•>
0
ll
fO
tl
0^
m
-ml ,1
h
d
^X
»-
\r
1-
— — —
________
-,
-1
J
— UJ—
©!
i.
o
"7
w/
e
•4
^•z
o
s
U-
e
(O
o
U
O)
c
Ol
O
(U
O)
o
OJ
3
0>
216
-------
15. Facility Layout Showing Wind Tunnel, Aerosol
Generator, and Dust Conveying System
-------
diameter of the jet issuing from each of the eight tubes. Dust mixes
with the tunnel gases within the stilling chamber upstream of the test
section and then passes through the contraction into the test section
where instruments can be tested or prototype particulate collection de-
vices can be evaluated. Figure 14 shows a detailed side view of the
aerosol generator itself and illustrates the sizes of the equipment that
will be installed. The system is currently under construction and will
be operational in the fall of 1975.
CONCLUSIONS
• The two component fluidized bed has been demonstrated to be
incapable of operating at high flow rates of dust because of a
variety of scale-up problems.
• The new three component fluidized bed combined with a two com-
ponent fluidized bed in a two stage system appears to be
scaleable to any reasonable size.
• The two stage system has demonstrated capability to redisperse
collected dusts with substantial amounts of particulate in the
fine particle range, and thus the system is appropriate for
fine particulate research involving either instrumentation or
particulate control device technology.
REFERENCES
Guichard, J.C., and Magne, J.L., Applications of Liquified Bed in Gaseous
Atmosphere to the Production of Aerosols. Part IV: Study of Elutria-
tion of Certain Simple Liquified Bed - Applications. July 28, 1967.
Translated from the French for the National Air Pollution Control Ad-
ministration. APTIC-Tr-0169.
218
-------
GENERATION OF INORGANIC AEROSOLS
FOR WEATHER MODIFICATION EXPERIMENTATION
William G. Finnegai; and John W. Carro;
Naval Weapons Center
Michelson Laboratory
China Lake, California
ABSTRACT
Since 1947 submicron-size aerosols of materials which can initiate ice
nucleation activity have been used in weather modification experiments
and in operational programs. Many aerosol generating methods have been
tried; the two methods most used at this time are combustion of certain
flammable solutions and the burning of specially formulated pyrotech-
nics. Other aerosols have also been generated for atmospheric tr-icer
and warm cloud modification experiments. These aerosol generation
techniques are quite flexible, in that many different mat-rials can be
generated in a wide range of particle size distributions. Aerosol
technology has been and is continuing to be advanced by the weather
modification effort.
-------
GENERATION OF INORGANIC AEROSOLS
FOR WEATHER MODIFICATION EXPERIMENTATION
William G. Finnegan and John W. Carroz
Naval Weapons Center
Michelson Laboratory
China Lake, California
INTRODUCTION
This paper describes methods used to generate ice nucleating aerosols
for weather modification experimentation. The earliest methods used
were sparks, hot filaments, burning charcoal, and burning solutions.
Some of the propellant and pyrotechnic formulations used are discussed
in some detail. Other aerosols used to release tracer elements and
hygroscopic salts into the atmosphere, particle sizes, and a computer
program for calculating the chemical species in the hot gases of com-
bustion are also briefly described.
After Schaefer (1946) demonstrated that supercooled clouds could be
greatly changed by seeding them with many small ice crystals produced
by falling "dry ice" and after Vonnegut (1947) observed that silver
iodide (Agl) particles serve as nuclei for the formation of snow
crystals, there was considerable interest in generating ice nucleating
aerosols to test the possibility that the weather might be influenced
over large areas by dispensing large numbers of ice nucleating particles
into the atmosphere.
The efforts to generate aerosols for weather modification started in
1947 and have continued to the present. Many different methods have
been tested. The two methods most used are burning solutions and
burning pyrotechnics; however, several of the other methods may be as
useful to environmental protection technology. This paper will attempt
to cover most of the methods used.
EARLY EFFORTS
The production of submicron-size Agl smokes is easily accomplished by
first evaporating Agl at a high temperature and then rapidly cooling
the vapor so that it condenses into many small particles. Early methods
used to produce Agl nuclei for laboratory studies were to pass a spark
220
-------
between silver electrodes in the presence of iodine vapor, to beat Apl
on a hot filament or to disperse it in 3 flame (Vonnegnt, 1947)^.
Another method of doing this was to burn Agl-impregnated charcoal in a
stream of air (Vonnegut, 1951)3. The heat of the burning charcoal
'evaporates the Agl on the surface of the charcoal and the air stream
quenches the Agl vapor, as well as promoting the combust Lou of the
4 charcoal.
1 Although Agl is highly insoluble in both water and organic solvents,
it is, nevertheless, quite soluble in acetone solutions of a soluble
iodide, such as Nal, KI, and NIfyT. Act-tone solutions can be diluted
'with acetone to any concentration without precipitation of the Agl.
,i Accordingly, acetone solutions were and still are widely used in
'wejther modification.
Using solutions in the spray nozzle type of generator is one of the
simplest and most efficient methods for producing large numbers of
i\gl nuclei. The early method used a two-phase conventional nozzle; the
Agl solution was atomized by the actiuiv of a jet of compressed gas
(Vonnegut, 1949)^. Because of the small size of the solution droplets,
they were very rapidly vaporized in the flame. The Agl and Nal vapor
in the flame rapidly condensed when they mixed with the cool air of the
atmosphere to form a smoke of very small particles. By varying the
concentration and the rate of flow of the solution to the spray nozzle,
'the size of the resulting particles could be varied over a wide range.
In general, the greater the flow rate to the nozzle, the larger were
the resulting particles, Vonnegut, 1949 . Solutions containing up t.o
10% of Agl can be used in spray nozzle type burners.
Additional methods were developed after 1951. They are buining of a
celluloid film, toilet tissue, or confetti impregnated with Agl
(Vonnegut, 1957) , and vaporization of a solution of Agl in anhydrous
ammonia, or isopropylamine (Davis and Steele, 1968)".
PYROTECHNICS AND PROPELLANTS FOR WEATHER MODIFICATION
3 "Pyrotechnics, the Fire Art - from the Greek words py_r (fire) jt^
; (an art) - is one of three closely related technologies, those of
, » ' explosives, propellants, and pyrotechnics proper .... Explosives
: perform at the highest speed of reaction, leaving gaseous products;
propellants are gas formers of brisk reactivity, but slower than
explosives; and pyrotechnic mixtures react mostly at visibly observable
rates with formation of solid residues- Numerous exceptions to these
* definitions may be cited" (Ellern, 1968) . A fourth class of materials
can be defined as gas-generating substances which are not intended to
22J
-------
produce heat or light as such but are sources of gas used to disperse
other materials (e.g., colored organic dyes).
Propellants usually are burned under pressure; pyrotechnics usually
are burned at ambient pressure up to altitudes of 40,000 feet.
PRECAUTIONS
A word of warning may be unnecessary to most readers but may not be
amiss for some who may wish to make their own pyrotechnics. The
mixtures described are all dangerous if not properly handled. Mixing
and firing pyrotechnics is safe only when done under proper conditions,
and these should be rigorously observed. Advice should be sought from
someone skilled in the field before undertaking such work.
PROPELLANTS FOR FREEZING NUCLEANTS
Q
Vetter et al. (1970) summarized the production of cloud nucleants
from propellant formulations.
While working on the production of colored smokes, Drs. Burkardt and
Finnegan of the Naval Weapons Center (NWC) generated pure Agl aerosols
by burning mixtures of organic fuel binders and silver iodate
the A.gI03 is the oxidizer. These first studies (1957) at NWC were
made using the exotic fuel poly(2-methyl-5-vinyl)tetrazole, C^H^N^.
The general reaction, balanced to produce both CO and C02> is
3AgI03 + C4H&N4 ——> 3AgI + 2CO + 2C02 + 3H20 + 2N2
This composition produced copious smoke, the solid particles of which
were identified by X-ray and chemical analysis as Agl and a trace of
metallic silver. This composition, an ideal choice from a chemical
point of view, is not widely used, however, because of the unavailabil-
ity of the fuel-binder.
Polybutadiene is a less expensive binder. Several satisfactory poly-
butadiene compositions for the dissemination of Agl have been developed.
A large number of polybutadiene formulations were investigated using a
computer to find the best possible mixture. The basic chemistry is
shown in the unbalanced equation:
AgI03 + tCH2^]n + A1 >
Ag + Agl + H20 + nCO + nCO + HI + H + A120
222
-------
Folybutadiene has a high oxygen requirement for complete combustion,
and AgIC>3 is a poor oxldizer. It is impossible to use enough AglO-j to
completely oxidize the polybutadiene and still have enough polybutadiene
to mix and cast the propellant. The heat of combustion of polybutadiene
is not great enough to vaporize all Agl, so aluminum is added to the
formulations to raise the heat of combustion and the flame temperature,
, The presence of aluminum presents an additional oxygen demand which can
tonly be met by adding another oxidizer or a high-energy explosive
plasticizer.
1
Double base propellant systems are interesting because they produce
*an aerosol of only small particles. When burning they do not produce
i, a visible smoke. A maximum of 15% by weight unrecrystallize AgI03 can
rbe added to this propellant. A 5% formulation is shown below:
Nitrocellulose ....... 47.3
Nitroglycerin . . ..... 38.0
Di-N-Propyl adipate .... 2.9
2-Nitrodiphenylamine .... 1.9
Lead beta resorcylate ... 2.4
Cupric salicylate ..... 2.4
Candelilla wax ....... .1
AglO powder ....... • . 5.0
The products of combustion are Agl, a trace of copper oxide and lead
oxy iodide, and large amounts of water vapor, N£> CC>2, and CO. The high
ratio of gas to solids minimizes agglomeration and the yield of nuclei
per gram of Agl is similar to the high values from NH^I-Agl-acetone
solutions which when burned also have very high gas to solid particu-
late ratios.
PYROTECHNICS FOR FREEZING NUCLEANTS
Since several hundred different pyrotechnics have been formulated for
atmospheric modification, this paper will cover only a small portion
of what has been done. St.-Amand et al. (1970)9 summarized the
production of nucleants from pyrotechnic compounds.
The simplest composition is an explosive mixture of pyrotechnic-grade
magnesium and AgI03 powder. The mixture is sensitive, quite dangerous,
and should be kept dry.
AgI03 + 3Mg ----- ;> Agl + 3MgO
223
-------
A rapid burning composition suitable for a pressed flare or fusee for
use at high altitudes is
Wt. %
AgI03 ....... 84
Mg ........ 8
Binder ...... 8
Slower burning mixtures may be made with aluminum.
The reaction with aluminum is
+ 2A1 - > Agl + A10
Mixtures containing only aluminum, AgIC>3 and binder are hard to ignite,
so enough magnesium is usually added to ensure an easy ignition and a
quiet burn. An example of a practical formula is
Wt. %
AelOo . . .
Al
MP
Binder . . .
... 78
... 12
... 4
... 6
When burned, the solid products are Agl, and a mixture of magnesium
and aluminum oxides; the binder is oxidized to H^O, CC^, and CO.
It is frequently desirable to decrease the amount of AglOo in a mix-
ture in order to reduce the particle size of the resultant Agl or to
produce a hygroscopic wetting agent to go along with the Agl. For this
purpose another oxidizer, such as KNO-, or KIO-j, may be added to the
formulation. One such mix is
Wt. %
The smoke from this
of a mixture of Agl
AelOi . .
KNOT . .
Me ...
Al . . .
Binder
formulation
, K2C03, MgO,
28
44
11
11
6
is yellowish white
and A120 .
in color and consists
Drew (1966) has shown that aluminum oxides produced from the combus-
tion of aluminum powder consist of spheres of a variety of sizes
224
-------
dependent primarily upon the sizr ,: f -r , u,;rir,: ir,;,
nature of the environmental flarcc t-^sf -. ":Tc alunu:
occur in two particle size distributifua. .'"he ]ar£.,.-
ing of spheres 1 to 100 y in diametc-i ••: b 1 i(jyi>.! r
from accumulations of oxide on L:;-. p.:.: ' c. i-.j :,urfjcc
The smaller material, comprising perray,: .•(.•% cf cro
spherical powder usually less than 1 L :i th r.h' r^.i >
less than 50 A in diameter. Undt./ sore c oad i t io--"--,
hollow spheres are produced which reach n v.xis-i'jni
the diameter of the original particle c~ c lur.iTi>nn.
'i 100 y diameter starting material, 300 j c'air*. er :
measured.
[f an alio}/ of aluminum and magnr..-. i.u •; ;- L;
is produced, its yield being depender.' on
magnesium in the combustion process.
In general, the hottest reactions occ -.r .--'-
are used. With Agl, however, any ttv%C; ar :rc "l^> rc '";"'.''•'' . h.. ur:-e
for dissemination. The temperature can bt control].
action is still properly oxidized by Ir •• '.-v
value as desired, compatible with satisfactory hura\:ig
Most of the AglO-j pyrotechnic formulatiunc contain iiif/'^ d i enl, wK
react with or shift the final chemical equilibrium ;jw.', i .-~~i ,'.1-:
particulates. Thus, each formulation has to b^ ,--:rc «. '•, -.-.;' : :H-;
studied before its cloud seeding potential is unders ion-! ; in-
materials when added to a formulation greatly lower n ^ : ',--. ,. c
irjclei output.
An important consideration for envi re". ,i,-nta.l pr ".--.'rr j' .1 .: •'•1iri; ^r.
well as weather modification is the Dai ''.i c-i i.-iLt: u> .-;,; ..-o->n •• ; --'> ; •
the pyrotechnic flame, the size of" t '-•« Claiuo, a-^1 ' ' ; • r T < •:>,;
They all affect the size of the res;..' tin,.- par1-*ru! .• ..>s . • ~ ;,j
SOLUTIONS AND SOLUTION tflJRM;;1,."
4
Bernard Vonnegut (19^9) published a j^i|..er c\, in i.. •-=, • •• , - -i ,n
burner for seeding clouds. Because Ag f v/ilJ nrn . j.-,'-> ;. : r> /• ••••.
without excess iodide ions, he added V, i I? t's- .-!••":' .
tion. Then, in a later paper, he discrssed u;;e f'f " • , ; ; L
additive tc replace the NH/.T (Vonneg* .- 9!.QV! ; , i . : ,
[x;inted out that the Agl nuclei ;:i<'si1, •_'•<' u eh j^,.-" .-. -"
olexes. Although these complex nir .•">', .-*•/« .-. ^'j-.;-. ; ..
•ibiJity than the Agl nuclei T>T/>,-:,•,..-., •' ,-.
-------
1950 paper had a great deal of influence, and for 20 years the scien-
tists who were engaged in weather modification used the TSIal to dis-
solve Agl in acetone. Consequently, from 1950 to 1970 the results of
weather control efforts were poor and indeterminate.
12
Thompkins et al. (1963) carefully analyzed the burner situation, with
particular regard to the use of Kl as a solubilizing agent. They
(and Vonnegut) pointed out the importance of the chemistry involved.
Finnegan et al. (1971)1^ again emphasized the importance of differences
in the chemistry of the solutions, and most weather modifiers became
convinced that NH^I is a better solubilizing agent. Currently, solu-
tions combining Agl with NH^ are used almost universally in acetone
burners. The NH^I is destroyed in the flame and pure Agl is the only
solid emitted from the burner.
Several burners were developed to burn Agl-acetone solutions. Some are
ground based, others are operated from aircraft. Most of the ground-
based burners were designed to burn propane (or a similar fuel) along
with the acetone solution. The airborne burners often were designed
to operate without an auxiliary fuel. In Patent 2,527,231, Vonnegut
(1950)1^ describes a ground-based solution burner with and without
auxiliary fuel. Vonnegut and Maynard (1952)15 described the design
and operation of an airborne burner which burned butane and acetone
solution. Fuquay (1960)16 described the development and calibration
of an airborne solution burner without auxiliary fuel. Carroz and Noles
(1972)17 described a ground-based solution burner which operates with-
out an auxiliary fuel (see Figure I). Carroz (1973)18 described an
airborne burner which generates more measured nuclei at -5°C per grain
of Agl than any other system (see Figures II and III).
OTHER AEROSOLS
Aerosols of the tracer elements lithium and indium have been used to
aid in the identification and understanding of atmospheric processes.
Efforts to formulate both solutions and pyrotechnics for rapid release
of kilogram quantities of a water soluble lithium tracer aerosol were
unsucessful because of the low atomic weight of lithium. Large kilo-
gram quantities of water soluble cesium tracer, however, can be pro-
duced by burning a pyrotechnic containing 70% by weight CSN03 (Finnegan
and Carroz, 1974)19. The weight % of elemental cesium in this pyro-
technic formulation is 47.7%. A water soluble rubidium tracer can also
be formulated by using RbN03 in place of CsN03. A high purity aerosol
of In903, a water insoluble tracer, was produced by burning a pyro-
technic made by mixing indium metal with nitroplasticized smokeless gun
powder. The formulation contains 16.7% by weight elemental indium.
226
-------
Figure I. Ground-based solution burner; 6-galion capacity
-------
NOSE
SECTION
RELIEF VALVl
CHECK VALVE
NOSE COVER
CENTER
SECTION
FILLINl, VALVL
SIHONGBACK
I I I I I
ill I I I /
TAIL
SECTION
IGNITER POWI.H SUPfl V
INLET AIR , IGNITER
SCOOP //FLAME HOLDER
SOLUTION TANK
WITH BAFFLES-
GAS REGULATOR
SOLUTION OUTLET
STAINLESS STEEL PIPE
NITROGEN PRESSURE TANK AND SOLUTION TUBING'
NOZZLE
FLAME TUBE
SECONDARY AIR PASSAGE
Figure II. Airborne burner with solution supply system
INLET AIR SCOOP -
SECONDARY AIR PASSAGE
INLET AIR SCOOP
Figure III. Airborne burner
228
-------
I »
The smoke from this pyrotechnic is free of a contamination from other
water soluble particulates. Except for tht-; indium metal and trace
amounts of lead and copper, all of the ingredients decompose to gaseous
products.
Large 250-pound pyrotechnic units were manufactured to dispense NaCl-
KCl-LiCl aerosols. These and smaller cnits were used in warm cloud
modification experiments.
AEROSOL PARTICLE SIZES
The size range for Agl-Nal particJcs j/c-nerated by the S'>cyfire generator
was found by Kissinger and Mitchell (date unknown) to be 10~2 to 10"-'- \i
(Fuquay, I960)16. Mossop and Tuck-Lee (1968)20 reported the size
distribution of Agl-Nal particles generated by the Warren-Ncsbitc
generator. The particle distribution followed a log probability law
with a median diameter of 0.085 \i and a standard deviation factor of
1.47. Combustion of an atomized ethyl alcohol (95%) solution with
18.9% LiClO^ yielded particles ranging between 0.27 to 0,63 \i with a
mean of 0.45 y.
In many cases the nuclei measurements of active nuclei exceed the
measurements of particles. It is difficult to capture and to measure
all of the smallest particles in an aerosol.
MEASUREMENTS OF NUCLEI
In an effort to compare one generator with another and with pyrotechnic
formulations the National Science Foundation has for several years
supported the Cloud Simulation and Aerosol Laboratory at Colorado State
University. Here generators and pyrotechnics are tested for measured
nuclei output. Even though the measured nuclei may differ from the
actual nuclei, this facility does provide some comparison of generator
output. The many complications of nuclei measurements will not be
covered in this paper because of their limited application to pollution
technology. A sample of the measured nuclei output from a solution
burner is shown in Figure IV.
229
-------
10
15
<
<
10
13
8"
UJ
Z
o
UJ
£ 10
12
10
11
5% Agl
2% Agl
2 GPH NOZZLE
1111
-5 -10
TEMPERATURE, °C
-15
Figure IV. A sample of the measured nuclei output
COMPUTER CALCULATIONS
The NWC Propellant Evaluation Program (PEP) can be used to calculate
flame temperatures. The calculations are more realistic for chemical
reactions at equilibrium (e.g., rocket motors, power plants, etc.) than
for small pyrotechnics burning in air.
The PEP is based on a straightforward thermodynamic model consisting
of two processes: (1) constant pressure, or adiabatic combustion, and
(2) isentropic pressure, or adiabatic expansion.
The assumptions behind the combustion process include:
1. Reaction kinetics are fast enough so that chemical equilibrium is
attained before the products leave the combustion chamber.
230
-------
2. No heat is exchanged between the sy.Jtom and the surroundings.
(In ramjets, the stagnation energy of the incoming air becomes
part of the system. This may simply be added to the heat of the
formation of air.)
3. Gaseous species individually obey the perfect gas law and collec-
tively obey Dalton's law of partial pressures.
When such assumptions are made, the system enthalpy and system pressure
completely determine the final state and chemical composition of the
system after combustion. The solution to this state and composition
is found by a computer technique called "snthalpy balance."
An example of a computer calculation ••;, shown below for an Agl pyro-
technic formulation burning at one atmosphere of pressure.
INGREDIENTS
WEIGHT
SILVER IODATE 78300
ALUMINUM (PURE CRYSTALLINE) 10300
MAGNESIUM (PURE CRYSTALLINE) 5200
100DER321/43DEH14 (EPOXY BINDER) 5.700
GRAM ATOM AMOUNTS FOR PROPELLANT WEiGHl OF 100.000
(H) (C) (N) (0)
.461183 .339340 012526 .892143
Reaction Pressure 1 atmosphere
Flame Temperature: 2346 K
Chemical
AL
AL20
CO2
H
H2
"MGO
,0
12
AGH
AL203S
AGIS
(MG) (AL) (AG) !U
.213816 .400297 276884 .276884
Species m Flame- Number of Moles
.0002/
.00001
.0002/
.0065/
.2129;
.OOOO/
.OOOO/
.0003'
.00087
.00007
.00007
2.18-04
3.51-05
2.27-04
6.4903
2.13-01
1.41-05
8.72-08
2.57-04
8.14-04
1.00-25
1 00-25
ALH
C
C2H2
MGH
H20
N
O2
ALI
AOO
AL203"
ACS
00007
.OOOO/
.00007
.0015/
.00087
OOOO/
.0000,'
.0322,
00017
18397
.00007
1.
6.
7
1
8.
5.
1.
3.
5
1
1
08 05
16- !l
9203
4903
07-04
81 09
,39 10
2202
.4505
.84-01
.00- 2 "7.
ALHO2
Ch4
CNH
NH
i\H3
NO
hi
ALI3
AG
CS
OOOO/
.00007
.00007
OOOO/
.00007
OOOO/
0249'
00007
27607
.OOOO/
2.
1.
5,
1
5
4
2
7
?
1
59 1i
.44-03
,/306
.7008
9008
.16-08
.49-02
.14-06
.7601
00-25
ALC
CO
C3
MQ
MG
N2
J
NHO
ALNS
MGOS
U0007
339 I /
.00007
.OOOO/
.2123,'
.0063/
2192'
0000;
OOOO/
OOOO/
8 12 O/
3.39-01
5 14-16
4.37 06
2.1301
6.26-03
2 1901
7 U9 12
1 00-25
1.00-25
231
-------
CONCLUSION
The preceding brief description of the methods used to generate
aerosols for atmospheric experimentation is necessarily limited.
Weather modification experiments during the past 25 years have provided
much experience in aerosol generation. Aerosol generation technology
has been advanced by the weather modification effort and we look for-
ward to applying our knowledge of these advances to meet the require-
ments of environmental protection technology.
REFERENCES
1. Schaefer, V. J., 1946: The production of ice crystals in a cloud
of supercooled water droplets. Science, pp. 457-459.
2. Vonnegut, Bernard, 1947: The nucleation of ice formation by silver
iodide. J. Appl. Phys., Vol. 18, pp. 593-595.
3. Vonnegut, Bernard, 1951: Final Report Project Cirrus. General
Electric Research Laboratory, Report No. RL-566, pp. 27-56.
4. Vonnegut, Bernard, 1949: Nucleation of supercooled water clouds
by silver iodide smokes. Chem. Rev., Vol. 44, pp. 277-289.
5. Vonnegut, Bernard, 1957: Early work on silver iodide smokes for
cloud seeding. Final Report to the Advisary Committee on Weather
Control, Vol. II, pp. 283-285.
6. Davis, C. L. and R. L. Steele, 1968: Performance characteristics
of various artificial ice nuclei sources. J. Appl. Meteor., Vol. 7,
No. 4, pp. 667-673.
7. Ellern, Herbert, 1968: Military and Civilian Pyrotechnics. Chem.
Publishing Co. N. Y., p. 3.
8. Vetter, R. F., et al., 1970: Pyrotechnic production of nucleants
for cloud modification - Part III: Propellant compositions for
generation of silver iodide. J. Weather Modification, Vol. 2,
pp. 53-64.
9. St.-Amand, P., 1970: Pyrotechnic production of nucleants for cloud
modification - Part II: Pyrotechnic compounds and delivery systems
for freezing nucleants. J. Weather Modification, Vol. 2, pp. 33-52.
232
-------
10. Drew, C. M., 1966: Aluminum particle combustion in gas burner
flames. Naval Ordnance Test Station, China Lake, TPR 415, NOTS
3916, pp. 8-56.
11. Vonnegut, Bernard, 1950: Techniques for generating silver iodide
smoke. J. Colloid Sci., Vol. >, Nc. 1, pp. 37-48.
12. Thompkins, L. M., et al., 1963: Water adsorption in the system
! AgI-KI-H20. J. Geophys. Res., Vol. 68, pp. 3537-39.
1*3. Finnegan, W. G., P. St.-Amand, and L. A. Burkardt, 1971: An
evaluation of ice nuclei generator systems. Nature, Vol. 232,
pp. 113-114.
14. Vonnegut, Bernard, 1950: Method of generating silver iodide
smoke. U.S. Patent 2,527,231.
15. Vonnegut, Bernard and Kiah Maynard, 1952: Spray nozzle type
silver-iodide smoke generator for airplane use. BullA Am. Meteor.
Soc., Vol. 33, No. 10, pp. 420-428.
16. Fuquay, D. M., I960: Generator technology for cloud seeding.
ASCE Proceedings, J. Irrigation Drainage Division, Vol. 86,
pp. 79-91.
17. Carroz, J. W. and R. C. Noles, 1972: A new ground based solution
burner for cloud seeding and production of atmospheric tracer
materials. Proceedings Third Conf. Weather Modification, Am.
Meteor. Soc. pp. 37-40.
18. Carroz, J. W., 1973: Airborne jet seeder solution burner. JL^
Weather Modification, Vol. 5, pp. 166-177.
19. Finnegan, W. G. and J. W. Carroz, 1974: Tracer element from
pyrotechnics and solution burners. Proceedings Fourth Conf.
Weather Modification, Am. Meteor. Soc., pp. 214-217.
20. Mossop, S, C. and C. Tuck-Lee, 1968: The composition and size
distribution of aerosols produced by burning solutions of Agl and
Nal in acetone. J. Appl. Meteor., Vol. 7, No. 2, pp. 234-240.
233
-------
GENERATION OF AEROSOLS BY BURSTING OF SINGLE BUBBLES
Milos Tomaides*, and K. T. Whitby**
Interpoll Inc., f-t. Paul, Minnesota, and
University of Minnesota, Minneapolis
ABSTRACT
The mechanism of aerosol generation by bursting of single bubbles from
aqueous solution of sodium chloride has been studied. A single bubble
aerosol generator has been designed to evaluate the size ana concen-
tration of droplets formed by bursting of a bubble film and also by the
liquid jet action subsequent to the bubble burst. A solution of 0.1
percent sodium chloride in distilled water was used in these experiments
to generate bubbles of 1.1+ and 5-5 mm in diameter. While bubbles of
1.1+ mm size generated both the film and also jet droplets the 5-5 mm
bubbles result in generation of the film droplets only. The burst of
a 1.1+ mm bubble resulted in a generation rate of twenty film droplets
and four jet droplets on the average. The 5-5 mm bubbles formed two-
hundred film droplets. Semi-empirical equations have been developed GO
describe the total mass of droplets generated by bursting of bubbles of
various sizes. A superior monodispersity of jet droplets over film
droplets has been found. The rnonodispersity of film droplets improves
rapidly with increasing bubble size. Practical aspects of the ex-
perimental results on the design of a bubble aerosol generator are pre-
sented.
This paper is based on work performed at the University of Minnesota
1 * '• and in the laboratories of Interpoll Inc., St. Paul, Minnesota.
* Present Address - Interpcll Inc., 1996 W. Co. Rd. C., St. Paul,
Minnesota 55113
** Present Address - University of Minnesota Particle Technology
Laboratory, Minneapolis, Minnesota 55^55
235
Preceding page blank
-------
NOMENCLATURE
B - Constant
(.' - Constant
p - Solute density, g cm » ,
t - Time
236
-------
GENERATION OF AEROSOLS BY BURSTING OF SINGLE BUBBLES
Milos Tomaides*, ana K. T. tfhitby**
Interpoll Inc., St. Pau1 , Minnesota, a.nd
University of Minnesota, Minneapolis
INTRODUCTION
The behavior of single bubbles and also of a cloud, of bubbles rising
in liquids has been studied by many in/estigators both theoretically
and experimentally. Studies reported, by K. Koide, et a].-, E. Kcrn-i,
et al.2, S. Krishnamurthy, et al.-^, V,. G. Glejm, et al. ', V. C. G.lc.im
and V. M. Vilenskij5} and R. L. Datta -ind R. Kumar0 are on:;/ some of
many others. The majority of these studies have primarily been in-
itiated by needs of the chemical industry. Variety of mass and energy
transfer operations and processes in which a gas is dispersed in x
liquid are usually accompanied by more or less vigorous entre.inment
of liquid in the form of droplets *>s for example studied by F. M.
Garner, et al.T. An original work by L>. C. Blarichard*- ha* been devoted
to the generation of droplets by bursvJnp, of bubbles on the sea ^s a
part of a meteorological and oceanographical research.
The generation of small droplets by bursting of bubbles started to De
considered a possible, simple source of artificial aerosols only re-
cently. For example, an aerosol generator in which a flow of air
bubbles through a layer of aqueous solution of methylene blue has been
studied by J. C. Guichard and J. L. Magne9. These authors produced
successfully an artificial aerosol of a rnonodispersity which is corn-
parable to the quality of artificial aerosols generated ny the Col ti t,">r
atomizer aerosol generator which h&o be^n accepted in a variety of
aerosol studies. Because the "fod,nun(j;" aerosol generator has an
advantage of being easy to construct n question arises how versatile
device it is and up to what extent thf quality of the aerosol gener-
237
-------
a I,I'd by Mili; ;>ppro;u:h can he maintained, varied or improved comparing
I" Ciller aerosol, generators. Trying to analyze these questions a
La,ek oT technical information on generation of droplets by bursting
of bubbles beca.me obvious.
An attempt was made in this work to develop a technique which can be
utilized to study the physical properties of aerosols generated during
the burst of the individual bubbles floating on a liquid surface.
238
-------
MECHANISM OF BUBBLE BURSTING
Two independent physical processes are involved in generation of
aerosols by bursting of bubbles. The first one is the formation of gas
bubbles and their motion through a liquid. This process Is followed
by a decay of bubbles after they penetrate the liquid surface.
The formation of bubbles and their behavior in liquid is an important
factor which influences the final size arid number of bubbles pene-
trating the liquid surface film. It has also direct impact on the mass
of droplets generated during the bubble decay. The size of bubbles i:;
also a basic parameter which can be directly related to the size of
minute droplets generated during the bubble burst. The formation and
dynamics of bubbles has been subjected to many investigations. The
results of these,studies are well described and summarized in the
literature1'2'3' .
Of primary interest in this study was the process of the generation
of minute droplets during the bubble decay and the evaluation of their
physical properties. The minute droplets are the result of two inde-
pendent physical mechanisms. A bubble .-ising through a liquid pene-
trates eventually the liquid surface forming a thin, dome-shaped f i*:..
This film is stretched above a more or less spherical, sha.tp.ow de-
pression in the liquid surface. Formation of the bubble film is
accompanied by the liquid drainage from the dome apex causing the fi.!m
thinning in this area. The drainage is very rapid with pure liquids.
Any impurities, monomolecular films or dissolved mate-rials tend to
stabilize the bubble and the liquid drainage is slower. If these
stabilizing agents are present in large quantities a very stable foam
is usually formed. Thinning of the bubble film results eventually in
its rupture forming an opening on the top of the dome. The p;-.j trapped
inside the bubble under positive pressure leaves the bubble rarddly,
The high velocity gas jet helps to break up the bubble film surrounding
the initial opening. A cloud of so called film droplets is formed. The
rest of the bubble film is rapidly attracted to the liquid surface and
disappears.
239
-------
In the process of the bubble film rupture the decreasing pressure
within the bubble is followed by a motion of the surrounding liquid in-
to the bubble crater. For small bubbles the potential energy at the
bottom of the bubble crater is high enough to eject a jet of liquid
from that area. The jet rises above the liquid surface and disin-
tegrates partially into droplets before it disappears back into the
Liquid. The droplets resulting from this liquid jet action are called
jet droplets. They are generally of much larger size comparing to film
droplets.
DROPLETS
The process of the generation of film droplets is very complex and very
sporadical experimental results have been published so far on this
subject. Some of the recent studies suggest that the air jet escaping
from the bubble has only limited influence on the generation of film
Droplets. Blanchard" came to the conclusion that hydrodynamical dis-
turbances in the bubble film after its puncture are the primary
mechanism of film droplets generation. This opinion seems to be real-
istic and can be supported by a simple calculation. It is known that
depending on surface tension and viscosity of liquid the ruptured
bubble film moves outwards with the speed of several hundred to several
thousand cm per second. At this speed a 0.25 mm water bubble collapses
Jn about 3 x 10"7 seconds. Even assuming that the gas leaving the
ruptured bubble attains speed of the sound instantly, it travels only a
distance of 0.1 mm during the above calculated rupture time. It is less
than a half of the bubble diameter. This supports the opinion that the
film droplets probably originate during the decay of a disturbed liquid
toroid whose circumference is proportional to the bubble film diameter.
The published experimental data on number of film droplets generated by
rupture of bubbles of various sizes are scarce. Only qualitative con-
clusions can be drawn from the published data. The number of film
droplets can be considered approximately directly proportional to the
diameter of the film cap. No film droplets can be expected for water
bubbles smaller than 0.2 mm, but this limit depends strongly on even
very small amount of organic impurities in the liquid. The impurities
may prevent generation of film droplets even from 1.6 mm bubbles. To
our knowledge no reliable systematic data on the size distribution of
the film droplets are available.
JET DROPLETS
More attention has been paid in various research activities to the jet
droplets. As described above the jet droplets originate from a breakup
240
-------
of a bubble ,)et. Thene droplets urc ejected upward to the height- nn
r.coat as ?0 cm and with speeds that, rom-h Govern 1 t,houL'.nergy of a bubble
with the energy of jet droplets Glejrrr' has expressed total mass of
droplets generated by a single bubble .jet as
M = B x d
J -D
Based upon the experiments with sea vate.- bubbles by Hay ami ana
the mean diameter of jet droplets generated from bubbles J arger than
0.3 mm can be described by the following empirical equation
!2)
Number of jet droplets liberated from one bubble can also be calculated
from
MJ = \* dDJ X nDJ X ?D
Plugging equation (2) into (3) and equating equation (3) and (l) re-
sults in
Because term E is a constant the equatJon (U) predicts a quite rapid de-
crease of the jet droplets number concentration with increasing bubble
size. On the contrary a large number of jet droplets must be expected
from the burst of small bubbles. Experimental results have shown the
equation (h) be valid only up to the bubble diameter of approximately
5 mm for pure water. For 6 mm diameter bubble of pure water the jet
ejection velocity is already so small that no jet droplets are forrr.co.
any more.
241
-------
EXPERIMENTAL PROCEDURES
As a part of a program to evaluate the feasibility of bubbling aerosol
generators an attempt was made first to collect more information on the
size and number of droplets generated by bursting of small bubbles.
For this purpose a single bubble aerosol generator has been constructed
to study the properties of both the jet and also the film droplets.
This device proved to be a useful tool for this purpose.
SINGLE BUBBLE AEROSOL GENERATOR
The single bubble aerosol generator is shown in Figure 1. Its design
allows only one bubble to burst at a time. It consists of the air
humidifier, the aerosol generating section and of the aerosol sampling
filter. The humidifier is made from a piece of 18 mm I.D. glass tube
being plugged from both sides and nearly filled up with distilled water.
Compressed air enters and rises through water in a form of small bubbles
formed on the tip of a glass capillary. The bubble-water contact time
is sufficient to nearly saturate bubbles with water vapor. This is
necessary for trouble-free operation of the aerosol generator which
follows the humidifier section. The jet droplets generated by bursting
of bubbles in the humidifier are removed by an impactiori disc which is
mounted very close to the water surface. Humid air continues to flow
into the aerosol generator through another capillary tube. The gen-
erator section is of a design similar to the humidifier. The diameter
of a capillary tube used to form bubbles is selected such to form a
bubble size desired. The generator is filled with an aqueous solution
as for example of sodium chloride. This helps to determine the size
of original droplets generated if the size of residue particles is
measured. All droplets generated by bursting of bubbles on the solution
surface are removed from the generator by a specially designed cone-
shaped flushing air deflector. They are transported in the stream of
clean, dry flushing air and dried completely before reaching the Milli-
pore membrane and collected on its surface. This generator was utilized
successfully to generate aerosol by bursting of single bubbles of l.U
and 5-5 nun diameter. The aqueous solution used in these experiments was
0.1 per cent of sodium chloride.
242
-------
DILUTION
AIR
ABSOLUTE
MLET FILTER
NOCI- WATER
SOLUTION
DISTILLED
WATER
VACUUM
PUMP
CRITICAL ORIFICE
MILLIPORE AEROSOL
SAMPLING MEMBRANE
DROPLETS
ENTRAPMENT
REGION
CAPILLARY TUBE
- IMPACTION DISK
- AIR HUMIDIFIER
- CAPILLARY TUBE
COMPRESSED
AIM
Figure 1. Single Bubble Cap' .lary Tube Aerosol Ge-ierato:
243
-------
The bubble size was measured using a direct and also an indirect
technique. In the direct sizing the bubble size was compared with a
scale engraved on a thin glass slide which was immersed in the solution
close to the capillary tube outlet. The indirect technique was based
on the measurement of the bubble rising velocity which was compared
with the experimental results of Harberman and Morton1 . To determine
the bubble rising velocity the total number, N, of bubbles seen station-
ary within the distance, H, when illuminated by the stroboscopic light
of frequency, f, were counted and rising velocity calculated from the
following equation
H f (5)
The preference was given to the direct technique because the bubble
rising velocity does not vary significantly within the bubble size range
of 1 to 10 mm and the indirect technique is thus not too accurate.
AEROSOL SIZE DISTRIBUTION AND CONCENTRATION
The residue aerosol particles generated by the single bubble aerosol
generator were collected on a O.U5 >um pore size Millipore membrane.
The membrane was divided into five equal annular areas and a replica
prepared of a small portion of each of these areas. The particles
were sized and counted under the transmission electron microscope.
The size of droplets generated by the bubble rupture was then calculated
knowing the size of sodium chloride particles collected on the membrane
with the use of the following equation
dP
(6)
Total number of droplets generated during each generation experiment was
calculated knowing the surface of the Millipore membrane and the area of
the membrane replica in which particles were counted.
244
-------
RESULTS
:i\s an example of the performance o? the single bubble pero;;-^ generate:
the results of the size distribution of droplets generated oy bursting
of l.U and 5-5 mm bubbles of 0.1 per ^ent sodium chloride aqueous eola-
tion are presented and discussed. The size of the droplets generated
by bursting of l.h mm bubbles is shown in Figure 2 and for 3o mm
bubbles in Figure 3-
The size distribution of droplets shown in Figure } has a clear ti-
modal characteristic. This means that both the film and also the jet
droplets were generated from l.U mm bubbles. Larger droplets resulted
from jet action while smaller droplets are the result of the film
rupture. Very few submicrone droplets were found in samples and the
size distribution curve falls off rapidly for sizes below 1 /ijm for both
bubble sizes tested. As it follows from Figure 3 the spread of the
sizes of the film droplet is about 1:50 while the majority cf jet
droplets is within 1:U limits. Superior n.onodispersity of jet oropleU
for l.H mm bubbles is obvious. On the average 20 film droplets and
h jet droplets were liberated by each bubble.
Examining the data plotted in Figure "' -pore carefully a roiid lire eon
be drawn through the data points having at least threo rnaxinas ror fiIre-
droplets and two for jet droplets. It Ls not difficult :: believe crat
this curve is much closer to reality -ban the dashed curve is. The fa:'
.that the jet droplet created on U:e tcp of the jet is alwayr, larger in
'comparison to the rest of jet dropler.- generated closer to the liquid
surface explains the presence of a I/', .-.'le peak in the J.et droplet si".e
distribution curve. For very small bubbles even another peak can be
probably expected located toward snaller droplet sizes that would
correspond to generation of small satellite droplets. These satellite-
are created during the separation of larger droplets along th- bubble
jet. This mechanism is well known when aerosols are generates by
spinning disk aerosol generators.
The 5.5 mm bubbles do not result in any jet droplets as seen in Figure
2. The ratio of the smallest and the largest film droplet size is
245
-------
about 1:20 because only insignificant number of droplets was generated
in the r.ubmicronc ran^o. Kxper i menti; wi t,ti larger bubble trizer. have
shown that the monodispersity of film droplet,i; is; improving with in-
creasing bubble size. Number ol' film droplets generated by 5-5 nut!
bubble is about 200.
Similarly to Figure 3 several maximas can be visualized on the size
distribution curve in Figure 2. This fact may support the idea of a
wave-form disturbance in the bubble film during the film rupture and
decay. Such a disturbance would break the bubble surface into several
concentric toroids. Each toroid breaking up into nearly monodisperse
droplets of the size which corresponds to the toroid diameter.
246
-------
"FILM" DROPLETS
REGION
I I I I 11
I I ! I t ill I
I I
BURSTING OF SINGLE BUBBLES
BUBBLE OIA 5.3 mm
0.1* Wda -WATER SOLUTION
-4
—I
0.001
10 100
DROPLET SIZE (>um)
i
Old
1000
Figure 2. Size Distribution of Droplets Generated by
Bursting of 5-5 mm Diameter Bubbles on a Surface
of 0.1 Per Cent Sodium Chloride Solution
247
-------
1.0
'E o.i
zo
a:
ui
0.001
0.0001
"FILM" DROPLETS
REGION
"JET" DROPLETS
REGION
I I T| Mill I I 1 | I 111 I I I III!
— \
BURSTING OF SINGLE BUBBLES
BUBBLE OIA 1.4 mm
O.I%NaCI WATER SOLUTION
I
10 100
DROPLET SIZE (xim)
IOOO
Figure 3. Size Distribution of Droplets Generated by Bursting
of l.U mm Diameter Bubbles on a Surface of 0.1 Per
Cent Sodium Chloride Solution
248
-------
DISCUSSION
The design of an aerosol generator which would employ bursting of
bubbles to generate artificial aerosols can be discussed based upon the
single bubble experiments. It is obvious that the best aerosol mono-
dispersity can be achieved by bursting of very small bubbles. The
bubbles would have to be below the critical size at which no film drop-
lets are already generated. Monodispersity of the aerosol of geometric
standard deviation of about 1.2 would be quite feasible. Unfortunately,
it is very difficult to generate larger number of very fine bubbles and
prevent their coalescence on the solution surface. For this reason,
the generation of artificial aerosols from very large bubbles is more
promising. The monodispersity of such an aerosol would not be much
better than 1.1*. This aerosol quality is already very close to the
quality of aerosols generated by the best atomization aerosol generators.
The aerosol generation from bubbles between 0.5 mm to 6 mm diameter
should be avoided because these bubble sizes result in generation of
both the jet and also the film droplets.
An estimate of the aerosol output from a bubbling aerosol generator
can be made based partially upon the results of this study. Let us
suppose the case of generation of droplets from bursting of single
monodispersed bubbles on a solution surface in the absence of any foam
that would otherwise change drastically the aerosol generation conditions.
If a porous plate with the air bubbling through it is used to generate
bubbles and the flow rate of the air is constant and only the plate pore
size changes, then the number of bubbles generated for these conditions
can be expressed by
n - Kl (7)
"B - d
The rate of droplet generation is described by
dn
dt
D - = n_ x n_ (8)
249
-------
The number concentration of film droplets can be calculated from
nDp = K2 x dBj (9)
as described earlier in the text.
Plugging (9) and (7) into (8) the rate of the film droplet generation
obeys the equation
dnDF Kl X K2 , (10)
dt
indicating a rapid decrease of number of particles generated with in-
creasing bubble size for constant air flow rate.
Similar tendency can be expected for jet droplets as calculated from
equation (8) with equations (H) arid (7) plugged in, which results in
dnnT E x K
DJ = r^-^ (11)
This shows an extremely fast decrease in number of jet droplets with
increasing bubble size for constant air flow rate.
The same procedure can be used to evaluate the mass rate of droplets
generated by bursting of bubbles. This can be done without problems
for jet droplets using equation (l) but lack of data on total mass of
film droplets makes this calculation impossible for film droplets. But
the experimental result published by Garner et al.T shows that the
total mass of both the jet and also the film droplets decreases with
increasing bubble size. The interesting phenomena found by these
authors is levelling off of the total particle mass for bubbles larger
than approximately 10 mm for water. This suggests limited validity
of equations (ll) and (10).
250
-------
REFERENCES
fi Koide, K. , S. Kato, Y. Tanaka, and H. Kubota. Bubbles Generated
from Porous Plate. J. of Chem. Eng. of Japan, 1968, Vol. 1,
p. 51-56.
2 Kojima, E. , T. Akehata, and T. Shirai. Rising Velocity and Shape
of Single Air Bubbles in Highly Viscous Liquids. J. of Chem.
Eng. of Japan, 1968, Vol. 1, p. U5-50.
3 Krishnamurthi, S. , R. Kumar, and N. R. Kuloor. Bubble Formation in
Viscous Liquids Under Constant Flow Conditions. I and EC Funda-
mentals, 1968, Vol. T, p-
h Glejm, V. G. , I. K. Shelomov and B. R. Shidlovski j . 0 processach
privodjascich k generacii kapel pri razryve puzyrej na poverchnosti
radela zidkost-gaz. Zhurnal Prikiadnoj Chimii, 1959, Vol. XXX11,
p. 218-222.
5 Glejm, V. G. , and V. M. Vilenski j . Vlijanie stepeni dispersnosti
parovych (gazovych) puzyrej na kapelnyj vynos pri kipenii i
barbotaze. Zhurnal Prikladnoj Chimii, Vol. XXX11 , 1966, p. 82-814.
6 Datta, R. L. , and R. Kumar. Gas Bubbles in Liquids (Part l).
Indian Chemical Engineer, 1961, Vol. 3, p. 7J-75-
T Garner, F. H. , S. R. M. Ellis, and J. A. Lacey. The Size Distri-
bution and Entrainment of Droplets. Trans. Inst. Chem. Engrs . ,
1951*, Vol. 32, p. 222-235-
8 Blanchard, D. C. The Electrification cf the Atmosphere by Particles
from Bubbles in the Sea. Progress in Oceanography, The Macmillan
Company, Vol. 1, 1963.
9 Guichard, J. C. , and J. L. Magne. Etude D'Un Pulverisateur A
Membrane. Institut National de Recherche Chimique Appliquee.
VERT-le-PETIT. 1966.
251
-------
10 Hayami, S. , and Y. To~ba. Drop Production by Bursting of Air
Bubbles on the Sea Surface. J. Ocean. Soc. Japan, 1958, Vol.
lU, p. 1^5-350.
11 Harberman, W. L., and R. K. Morton. An Experimental Study of
BubbJes Moving in Liquids. Trans. Am. Soc. Civil Engrs., 1956,
Vol. 121, p. 227-250.
* 4
252
-------
AN INVESTIGATION OF AN EXPLODING WIRE AEROSOL
by
James E. Wegrzyn
Cloud Physics Research Center
University of Missouri-Rolla
Rolla, Missouri 65401
ABSTRACT
The concentration, shape, mass, arid total charge of a gold
aerosol (produced by the exploding wire technique) is monitored
leading to discussion of the effect that a bipolar charge has on the
stability of a submicron aerosol. Values of K (coagulation kernel)
and 3 (wall loss coefficient) for the bipolar aerosol are compared
against the values of K and 6 determined from an exploding wire
aerosol that was neutralized to their bipolar charge equilibrium.
This aerosol is neutralized by use of a K-85 neutrali^.er. The
results of this investigation showed that the charges on the gold
aerosol enhance the coagulation in the first 600 seconds , and that
gas adsorption on the newly formed aerosol is considerable.
253
-------
AN INVESTIGATION OF AN EXPLODING WIRE AEROSOL
by
James E. Wegrzyn
Cloud Physics Research Center
University of Missouri-Rolla
Rolla, Missouri 65401
INTRODUCTION
The field of aerosol mechanics can be divided into the following
three groups: aerosol generation, aerosol sampling, and aerosol
analysis. This report deals mainly with the generation of a gold
aerosol by an exploding wire (E.W.) technique. Aerosol generation,
however, has to be complimented by correct sampling procedures and
analysis if it is to be understood. Therefore, this paper will also
discuss the use of the Electrostatic Aerosol Sampler-'- (produced by
Thermo-Systems Inc.) as a sampling instrument. Other instruments
used in this experiment are: the Particle Mass Monitor (produced
by Thermo-Systems Inc.), a Gardner counter, a diffusion battery , a
K-85 neutralizer (produced by Thermo-Systems Inc.), and a second
electrostatic precipitator which was designed in this lab. The
analysis will consist of intercomparison of these instruments for
measurements on the concentration, shape, size, total charge, and
reproducibility of an aerosol generated by exploding a wire. Values
of K (coagulation kernel) were determined from the solution of the
equation,
dn ..2 „
- dt = Kn + 3n
where 3 (wall loss coefficient) was evaluated experimentally. The
K-85 neutralizers effect on the coagulation kernel was separately
evaluated. The overriding purpose of this investigation is to
determine the feasibility of using an exploding wire aerosol as a
trace aerosol in a scavenging experiment.
254
-------
PROPERTIES OF AN EXPLODING WIRE
The physics of the E.W. phenomenon is well documented in the
literature . Likewise, the aerosols generated from 'rhis explosion
have been studied by several investigator s^'~ . The report by
R. F. Phalen in 1972 gives a thorough review on t:iis subject, and, in
general, the previous studies indicate that if the E.W. is completely
vaporized it results in the following characteristic":- Pictures 1, 2, 3.
t
1) The primary particles form a chain-like structure of smooth
spherical particles whose size distribution ran be cha -icterized by a
Jog normal or normal distribution.
i
2) The mean diameter of these primary parti oles are between .01
and .1 micron in size depending on the parameters of tneir generation
13 , 14
3) There exists a symmetric bipolar charge on the aerosols. '
4) The aerosols are reproducible if generated in the carne
manner.
Unless the joule heating is enough to completely vaporize the
wire in a manner of 300 nanoseconds or less, the "f W. aerosol will not
have all these characteristics. Karcdsis and FishJ i : lu-Trated the
decrease in the primary size of the particles wit'i injrearing electri-
cal energy. The reason being t.iat sorne of thir eie. ' r ical energy
goes into creating new surface area. Thereby, a ^vrger .input of
energy yields more surface area and results in smaller particles:,
The size of the primary particles has little deocaderice r>n ^--: tempera-
ture of the surrounding gas since the teir.perat-jre of *:L)9 plasma is
between 7000° and 8000°K6. Several exp lesion? were d-ne at the liquid
nitrogen temperature of 77°K with the primary r, iz« of these particler
generated •> remaining the same. However, the pre.-.,R-. < -Iocs '•irLaenoa
the size of the particle because at reduced pre.-: .-uros fne •••. ;rt5~a
spreads over a larger region. With a prescure 71 car n PI;,I Hg 'i'. -30 Lute,
an aerosol is formed that is too small *^o be detected bv *"'ne aero?,ol
3ampl;r and the mass monitor; and yet, the Gardner1 ^our.'ter ^ ; ;e
readings in excess of 100.000 particles r:,er ci;r rongfc. Vo ii.su^e that
this aerosol was due to the gold wire and not '\.\ ar-tifac.. c-f the high
temperature and radiation of the arc, a similar exi e^iTiem was
repeated without the gold wire. An arc was in* en t Ion?-.] ly produced
all cases the condensation nuclei was less than 1,000 j£irvicle3 oer
aA
255
-------
B~-™- «•
Picture 1 Gold Aerosol x 32,000 Magnifications.
256
-------
I MICRON
GOLD AEROSOL
32,000 X
Picture 2 Gold Aerosol x 32,000 Magnifications Neutralized by Kr-85.
-------
I MICRON
GOLD AEROSOL
32,000 X
Picture 3 Gold Aerosol x 32,000 Magnifications Collected Without the
Use of the Corona Field.
258
-------
In order to gauge the reproducibility of the E.W. aerosols, the
parameters of generation were kept the same throughout the experiment.
These parameters are: diameter of wire - .002 inch, length of wire -
2 cm, composition of wire - gold, pressure in chamber - 20mm Hg
absolute, electrical energy - 2000 volts at 3 micro-farads, and
carrier gas - argon. Argon was used to reduce any particulate
matter of nitrogen that might form from an arc in a nitrogen, oxygen,
and water environment. Another problem at reduced pressure in argon
gas is that some of the electrical energy can be discharged around
'the wire rather than through it. To overcome this problem, the
wire was kept hot, and a hydrogen thyratron (EGSG 7322/1802) was
!used to short the wire to ground.
To reduce the agglomeration of the E.W. particles, it was
necessary to use a two stage diluation system (Figure 1). This two
stage dilution system can control the agglomeration size and concen-
tration of the E.W. aerosol. The mean number of particles in the
agglomerates can be varied between two and several hundred while the
size distribution of the primary particles remain the same.
259
-------
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
x = o
UJ Q O
< CD 6
260
en
^
en
fi
O
•H
0)
60
CO
O
0)
M
M
•H
-------
GENERATION PROCEDURE
The procedure used for aerosol generation was as follows: A
gold wire was mounted between the electrodes, and a fine stainless
steel mesh (175 per inch) was placed between the exploding wire
chamber and the dilution chamber. To clean the system, both chambers
were pumped down to less than 1mm Fig absolute and immediately re-
filled with filter argon gas. This procedure was repeated at least
three times. The coagulation chamber was simultaneously purged with
filter N2 gas until the concentration of Aitken nuclei measured by
i:he Gardner counter was less than 100 particles per cm . Also, the
total mass detected by the T.S.I, mass monitor had to be less than
t3.0yyg/cm , and the background current of the Keithley 410 electro-
meter had to be less than ±.25x10"^ amps before the experiment was
run. The pressure in the coagulation chamber was monitored by a
• iignehelic pressure gauge (±2 inches of water), and was kept at a
slight overpressure to prevent any contamination from the room.
The experimental setup is shown in Figure 2. This figure
illustrates the overall goal of this investigation which is to
measure the scavenging efficiency of the E.W. aerosol with a larger
^article. The time of fall of these larger particles will be between
S and 30 minutes. This paper deals with the first stage of the
experiment which is to measure the characteristics of the E.W.
aerosol over this time span. In Figure 2, blocks D, E, and F were
not employed for this investigation. The electrostatic precipitator
is shown in Figure 3. The charge on the top plate was ±2200 V.D.C.,
and the bottom plate was connected tc the Keithley 'HO electrometer.
Responce frequency of this electrometer is 8.7 sec. for an eternal
capacitance of SOyy.f. The capacitance of this setup, which includes
f:he connecting leads, was only lUyyf. Picture U shows the output
from the electrometer over a five minute span.
261
-------
1.
)—
^^
)
v: .
£1
®»
J
„
1
±
^_
'e
o
s
cc
>-
X
K
o:
UJ
CD
OC
CD
UJ
5
cc
UJ
CD
QC
Ul
N
_J
QC
I-
Z)
O
o
Ul
I
<
o
o2
ZH o
o5 ?
00
ii
UJ
o
Ul
QC
QC Z)
O CO
I- CO
^ UJ
QC OC
UJ 0-
OuJ
oh:uj^zz
-------
SCALE
i 1
2 IN.
TOTAL CHARGE MEASUREMENT
KEITHLEY 410 ELECTROMETER
FULL SCALE - \0 X \0~a AMPS
FLOW RATE - 60 CM/SEC
CHART SPEED - 20 SEC/IN
CONCENTRATION - 125,000/CM
Picture 4 Output from Keithley 410 Electrometer.
263
-------
RESULTS
The diffusion battery (L) was designed by T. A. Rich, and is
used for the detection of particles too small to be precipitated
out by the aerosol sampler (less than .Oly). Table 1 gives the
results from the use of the diffusion battery.
33 33
cone/cm cone/cm cone/cm cone/cm
time at position at position at position at position
(sec) 21 21
charger current-10 microamps charges current=0.0
t<0 500 500 200 200
60 8,000 -- 55,000
90 -- 4,800 -- 45,000
120 6,000 -- 66,000
150 — 4,400 -- 47,000
180 5,000 -- 59,000
210 -- 3,500 — 42,000
240 4,800 -- 50,000
270 — 3,000 -- 36,000
300 3,500 — 45,000
Table 1 Diffusion Battery Data.
A diffusion battery along with the Gardner counter were placed
between the aerosol sampler and its pump. The first column on the
left hand side of Table 1 gives the concentration of particles that
passed through the aerosol sampler when it was operating in the con-
tinuous mode with a charger current of 10 microamps. The second
column gives the concentration of particles that passed through both
the aerosol sampler and diffusion battery. With a background count
of 120,000 particles per cm , it is seen that the aerosol sampler is
95% efficient in removing the E.W. aerosol, and the diffusion battery
has a 75% transmission rate. Using the table published in Rich's
paper, this 75% transmission rate corresponds to an effective diameter
for this aerosol of .032 microns. If the particles were less than
.01 microns in diameter, the transmission rate would be less than
15%. This experiment indicates that the number of particles too
small to be precipitated out, yet able to act as condensation nuclei,
are negligible. The right hand side of Table 1 was run with the
charger current of the aerosol sampler set to zero so that only the
previously charged E.W. aerosols would be collected. (See picture 3.)
264
-------
Again with a background count of 120,000 particles/cm , it is seen
that the percentage of charged particles is a little less than 50%.
This agrees well with the other electrostatic precipitator. The
transmission rate was also 75% which yielded the same effective
diameter of .032 micron for the aerosol.
The data taken from the E.W. aerosol over a thirty min. time
span is presented in Table 2 and Table 3. Table 2 is the data
taken without the use of the K-85 neutralizer and Table 3 is the
data taken with the use of the K-85 neutralizer. The 100,000 charges
per cm3 at a flow rate of 10 Lpm is well within the operating range
of the 2 milli-Curie K-85 neutralizer4'15. B. Y. H. Liu and
D. Y. H. Pui^-6 verified that for a monodisperse aerosol in the size
range of .02 to 1.17 microns, the magnitude of the charge on the
aerosol is reduced to the Boltzmann's equilibrium value which is:
N
~ = exp(-n e /2akT),
o
where N is the number of particles carrying n elementary units of
charge, N is the number of neutral particles, e (=4.8xlO~10esu) is
the elementary unit of charge, a(cm) is the particle radius, k is
the Boltzmann constant, and T(°K) is the absolute temperature. The
validity of the Boltzmann's distribution of charge for the smaller
sized particles has been questioned, however, by N. A. Fuchs-1-'. The
values of K was determined from the solution of the equation
dn 2
- -— = Kn + pn
dt
which is
R
K =
n n , pt .
o (e -1)
where 3 = .00009/sec.
Use of this equation has been criticized since it applies only
for a monodisperse aerosol; but it is also found that this equation
'fits emperically the case of the coagulation of a polydisperse
aerosol. From Table 2, the values of K range from 21 to 29 cm /sec,
.and there appears to be no dependence with time. These values
compare well with the experimental value of 20xlO~10 cm3/sec found
by Phalen for a silver E.W. aerosol; and also with Rosinski values
of 24.3 to 28.2 x 10~10 cm3/secl8 for a gOid E.W. aerosol. It also
compares favorably with Zebel's theoretical value (taking into account
265
-------
CDC— 'H O rH CO OJ H C7) LO
co i
ocNCOaoHcocMJ-a>
H H H -H H H
CO
O bO
CO 3-
co •*->
CNi-Hi—1r-|t--CMHLOLOOOrH
CNCNCNCNHCOCOJ-d-LOLO
ooooooooooo
ooooooooooo
ooooooooooo
H H
ooooooooooo
OOLOOLOOOOOOO
COOOCOCNOCna>c£>tQCOcO
o\e o\° o\B <^e o\° oSe o\e
LO en H co en LO co i
LO CO 3T J± CO LO d"
0\°
o
LO
ooooooooooo
OOOOOOOOOOO
ooooooooooo
OLOOOLOOLOOLOOLO
CO LO UD r|- t> CO .3- CM CO .H
I I I I I in I i i i i
LO LO LO LO LO I LO LO LO LO LO
o o o o
CO CO CO CO
o r- o
co ~\ co
o
CO
o
•rl
•U
cd
M
•U
(II
4-1
o
) J- CO CO J- 0 CO CD
3 CJ1 CT> CO CD (_O 00 i — I |
* t^- LO Cx) t£? (X) <-D C*^-
00 CO
co H
to to
h^
CO
4J
cO
Q
rH
o
en
o
cd
H
f-
CMCNCMOMCSICMCMCNCNCMCNCN
OOOOOOOOOOOO
oooooooooooo
OOOOOOOOOOOO
oooooooooooo
OOOOLDOOOOOOO
COCOLOLncOCMCNCDCJltDtDCO
H rH H H rH rH rH
o\° o\B o\° oN° o\° o\e o\e o\e o\°
CO IOJ- IrHHrHCM ICOCO
co CNCO cococoro coco
t~-uD^ cncococncMCOcDCOt^
cncM'HcncocotDcnocDcnco
oooooooooooo
oooooooooooo
oooooooooooo
UOOOCOOOOOOCOCOOOCM
CNCOHHCOCDOCOCSlHCMCN
H H H H H H HHHH
CncOHCOLOCMLOt^CO IDJ-
I I I I I I I I I J- I I
LOLOLOLOLOLOLOLOLO ILOLO
CM
O
o
0)
W
H i-3 co
< w e
CD
o
o
<
I—I
P-.
o
o
H
X
o
o
o
o
55
O
1 — !
E-*
1-4
H3
CD
«3^
O
O
H
S3
W
0
(V,
W
O,
CO
S
CJ
\^
CO
CO
s
<
h- 1
E-H
M
M
,_,
H o
S W
1— 1 CO
E-i ~— '
Q
H
O
PS
<
D^
O
s-*,
bO
3-
~i
^"^
CO
o
^•v.
o
0
o
m
c
o
•H
I 1
cd
N
iH
cd
J_l
4J
3
0)
rC
1)
•H
cd
4-1
CO
Q
rH
O
(A
O
M
(U
<3
CN
-------
—- ( ) ^
'he gas kinetic correction) of 21xLC cm /see for a .02 micron
T.onodisperse aerosol.
The data run, without the use of the K-85 neutralizer, gives
mother picture. The values of the coagulation kernel, in this
case, is time dependent with higher values of K being recorded in
the first 500 seconds than in the longer time runs. Since the aerosol
:.ad essentially an identical size distribution (Pictures 1 and 2), this
effect has to be assumed due to the higher bipolar aerosol. This was also
.-hser-ved by Jerry Burford19 of this lab for his Master's Thesis.
;'his higher initial coagulation rate can be explained several ways.
-f the charge distribution was not completely pymmetric, the excess
i.,irge, would be forced to the wal.lc. due TO th::ir mutual electro-
static dispersion. This would, in effect, lower the concentration
•^r.til a symmetric charge distribution could be established. Another
•-•"•jason could be that some of the E.W. aerosol Had multiple charges
-••nd this would increase their coagulation rate. It is also inter-
esting to note that after 900 seconds th^. aerosols in Table 2 appear
to be more stable than those in Table 3. All this speculation leads
Lo the conclusion that the charge distribution as a function of time
•r.ust be known before this phenomenon can be understood.
The electrostatic precipitator show:i in Figure 3 was used to
. 1-i'termine the total charge per crn^ for -t-ne aerosol. Making the
approximation of only one charge per parcicle, the aerosols which
.-rare subject to the K-85 radiation were, on the average, 32%
barged; and the aerosols coming directly from the dilution chamber
ere '17% charged. Theory does not predi -t that the difference cf
r'-i'i'ai charge of 15% would effective!'/ d^aole the coagulation rate.
'•^r-iln, the answer to this problem mxast .le in the charge distribution.
The concentration was measured by a Gardner counter, which was
recently recalibrated by Gardner Associates, Inc. of Scheneotad-y , New
•"ork. The calibration curve supplied l,v Gardner Associates was used
o convert the output signal froD th ; p". jiOiuultipller tube to a con-
• r.tration reading. In light of the ri. ':".lts of the Fort Collins
• mposiuir.20 where the Gardner insrruir.er: counted EomewhaT. less than
" -• other condensation nuclei counters, 3 7-eca] i :>ration is in order.
.'his calibration will be done by a comparison with the University of
, issouri-Aitken Nuclei Counter designed by Dr. Kassner21. If this
".ardner co'onter is found to count low, '.Vis will lower somewhat all
he values of the coagulation kernels li?V2d in Tables 2 and 3.
The size of the E.W. particle? c-:-;i also be. de lerrained by com-
taring the Gardner counter with the -,:-.v,-. -i-niter. This comparison
, lelded an effective diameter of Lc,i«"-;. r: .03 and .Clf microns f-rr
267
-------
LU
g
«-
>-
Ld
.J
X
H
LU
*
t_
LU
O
CD
S
CO
O
z
LCVJ
O
4-1
cd
p*
•H
O
QJ
M
P-i
O
•H
•U
td
4J
CO
O
3
to
•H
268
-------
::he particles, and this agrees well with the electron microscope
pictures. An interesting side light to this investigation, is the
ability of the exploding wire aerosol to adsorb gas vapors. Looking
it the total mass/cm^ data listed in T-able 2 and 3, it is seen that
in nearly all cases an increase with total mass is recorded with
•.irae. Since the coagulation chamber is a closed system, one would
•ixpect a small decrease in total mass due to particle wall losses.
A net increase in total mass can be explained by condensation of
;>'AS vapors in the gaps between the primary gold particles. On the
;iverage, the total mass listed in Table 3 is greater than the total
•"ass in Table 2 yet the concentrations remain about the same.
'.'his indicates that the K-85 neutrelizer or the connecting tubing
-,-"! been previously contaminated ,vith jome form of condensable vapor.
269
-------
CONCLUSION
This paper reinforces the previous findings that the exploding
wire aerosol is reproducible, submicron in primary size, highly
charged, and distributed normal or log normal in primary size. The
initial aggregation can be controlled by a two stage dilution system,
but a comparison of pictures 1 and 2 with picture 3 shows larger
agglomerates in the first two pictures. Since picture 3 was
collected with the corona field turnoff, it must be assumed that the
field or ions from the corona wire enhances coagulation before the
particles are precipitated. This has serious consequences when a
representative sample is needed for analysis. It must also be pointed
out that this phenomenon is peculiar to E.W. aerosols since a high
concentration of .02 micron NACL aerosols showed no agglomeration
when collected by the same aerosol sampler.
This question could be answered by a paper recently published
by W. H. Marlow and J. R. Brock22, who found that the charging rate
for a conducting particle depend importantly and in a complex
manner on the particle size and total particle charge.
The data indicates that the initial coagulations of the E.W.
aerosols is more pronounced than the theory predicts, but, as stated
before, the charge distribution rather than the total charge must
be monitored as a function of time if this reaction is to be under-
stood .
Perhaps the thing that is most important about the E.W. aerosol
is its shape factor. Vomela5 found the charge on the E.W. aerosol
t:o be between 50% and 70% higher than on an equivalent volume aerosol.
The influence of the shape factor on the diffusion coefficient of the
oarticles still needs to be determined experimentally for this size
particle. Finally, the large surface area to volume characteristic
of the combustion aerosol provides many sites for the adsorption of
gas vapor. Corn and Reitz^ discovered that the specific surface
of an urban aerosol taken in Pittsburgh would increase by a factor
of about two after degassing at 200°C.
This large adsorption of gas by this type of Aitken nuclei
could be an effective mechanism for cleansing the condensable vapors
from the atmosphere.
270
-------
ACKNOWLEDGEMENTS
I wish first to thank Austin W, Hogan and Dr. Volker Mohnen
'.>f the State University of New York at Albany for their generous
loan of the diffusion battery. Further I express my regards to
Tr. Kassner for his initial guidance and suggestion of the problem.
I also must thank Dr. Podzimek for his steadfast advice throughout this
investigation, and finally, I mur.t thank Mrs. Sandy Shults for typing
this manuscript. This work is being done as a partial fulfillment
"f mv Ph.D. Dissertation.
271
-------
REFERENCES
I. Liu, B. Y. H., K. T. Whitby, and H. H. S. Yu. Electrostatic
Aerosol Sampler for Light and Electron Microscopy. Rev. Sci.
Inst. 38: No. 1, 100-102, Jan., 1967.
2. Olin, J. G., and J. S. Gilmore. Piezoelectric Microbalance
for Monitoring the Mass Concentration of Suspended Particles.
Atmospheric Environment, Vol. 5: 653-668, 1971.
"5. Rich, T. A. Apparatus and Method for Measuring the Size of
Aerosols. J. Rech. Atmos. Vol. TI, 2e Annee: 79-86, 1966.
4. Liu, B. Y. H., and D. Y. H. Pui. Electrical Neutralization of
Aerosols. Aerosol Science. Vol. 5: 465-472, 1974.
5. Chace, W. G. and H. R. Moore, eds. Exploding Wires, Vols. 1-4.
Plenum Press, N.Y. 1958-1968.
6. Phalen, R. F., Evaluation of an Exploded Wire Aerosol Generator
for use in Inhalation Studies. Aerosol Science. 3: 395-406,
1972.
7. Karoisis, F. G. and B. R. Fish. An Exploding Wire Aerosol
Generator. Jour. Colloid Sci., 17: 155-161, 1962.
8. Karoisis, F. G., B. R. Fish, and G. W. Royster, Jr. Exploding
Wires, Vol. II (eds) Chace, W. G. and H. K. Moore. Plenum
Press, 1962, 299-311.
9. Vomela, R. A. and K. T. Whitby. The Charging and Mobility of
Chain Aggregate Smoke Particles. Jour. Colloid Sci., 25:
568-576, 1967.
10. Rosinski, J., D. Werle and C. T. Nagamoto. Coagulation and
Scavenging of Radioactive Aerosols. Jour. Colloid Sci. 17:
703-716, 1962.
11. Sherman, P. M. Generation of Submicron Metal Particles. Jour.
Colloid Sci. 51: 87-93, April, 1975.
12. Harvey, J., H. I. Matthews and H. Wilson. Crystal Structure
and Growth of Metallic or Metallic-Oxide Smoke Particles Produced
by Electric Arcs. Discussion of the Faraday Society, 30: 113-124,
1960
272
-------
13. Whytlav/-Cray, R. and K. Patterson, Smoke, !.'. Arnold,
London (193?).
...4. DallaValle, J. M. , C. Orr, Jr., B. L. Hinkle. The Aggregation
of Aerosols. Brit. Journ. App. Physics, 5: 5198-5208, 1954.
Ib. Cooper, D. W., and P. C. Reist. Neutralizing Charged Aerosols
with Radioactive Sources. Jour. Colloid Sci. 45, Vol. 1:
17-26, Oct., 1973.
£F>. Liu, B. Y. H. , and D. Y. H. Piu. Equilibrium Bipolar Charge
Distribution of Aerosols. Jour. Coijoid Sci., Vol. 49, No. 2:
305-312, Nov., 1974.
17. Fuchs, N. A. and A. G. Sutugin. Topics in Current Aerosol
Research. G. M. Hidy and J. R. Brock (eds.). Fergamon Press
Ltd. , p. 42-47, 1971.
18. Zebel, G. Aerosol Science. (ed.) C. W. Davies. Academic
Press Inc. Chapter II, 31-57, 1966.
L?. Burford, J. N. The Effects of Washout in Polydisperse Metalic
Aerosols. Master's Thesis, University of Missouri-Holla, 1974-.
'.„'(:, The Second International Workshop on Condensation and Ice
Nuclei (held at Colorado State f.'n Lvsr-si try) National Science
Foundation, Aug., 1970.
<-'i, Kassner, J. L, J. C. Carstens, M. A. Vietti, A. H. Biermann,
P'aul C. P. Yue, L. B. Allen, M. K. fastburn, D. D. Hoffman,
H. A. Noble and D. L. Packwood. Expansion Cloud Chamber
Technique for Absolute Aitken Nuclei Counting, Jour. Rech.
Atmos. 1968.
''!'. Marlow, W. K. and J. P. Brock. Unipolar Charging of Small
Particles. Jour. Colloid Sci., W . f.O, No. 1, 32-37.
Jan., 1975.
: 1. Corn, H. and R. Reitz. Atmospheric Par •.iculates: Specific
i Surface Areas and Densities, Scier.co 159: 1350 (1968).
273
-------
AEROSOL PARTICLE FORMATION FROM PHOTO-OXIDATION
OF SULFUR DIOXIDE VAPOR IN AIR
Kanji Takahashi and Mikio Kasahara
Institute of Atomic Energy, Kyoto University
Uji, Kyoto 611, Japan
: ABSTRACT
6 Aerosol formation from photo-oxidation of SO^ in air is studied
both experimentally and theoretically. Experiments on the effects of
environmental factors, such as S02 concentration, relative humidity, UV
.light irradiation intensity, irradiation time, and the presence of
foreign aerosol particles, on the formation and the evolution of aerosols
;4te conducted. A kinetic model for aerosol formation is also presented,
and calculated examples are shown.
Aerosol formation and evolution are considerably influenced by SOo
vapor concentration and relative humidity, and it is noticed that the
effects of those environmental factors are markedly varied with the
irradiation time.
275
Preceding page blank
-------
AEROSOL PARTICLE FORMATION FROM PHOTO-OXIDATION
OF SULFUR DIOXIDE VAPOR IN AIR
Kanji Takahashi and Mikio Kasahara
Institute of Atomic Energy, Kyoto University
Uji, Kyoto 611, Japan
INTRODUCTION
Sulfur dioxide vapor in air is photo-oxidized by ultraviolet(UV)
irradiation in sunlight to form sulfuric acid aerosols, and this process
plays an important role in air pollution.
The process of aerosol formation from photo-oxidation of S02 vapor
is considered to consist of three stages,i.e.: 1)photochemical oxidation
of S0« to SO-,, and followed production of l^SO/ vapor through the combi-
nation with H20 molecule, 2)heteromolecular nucleation of l^SO^ vapor to
a critical sized cluster or an embryo through the combination of several
number of molecules of t^SO^ and H^O, and growth of the nucleated embryo
to larger aerosol particle through the condensation of I^SO^, 1^0 and SO-j
molecules, and through the coagulation with other particles.
Number of experimental works have been reported on photochemical
aerosol formation from S02 in air by Gerhard and Johnstone^, Renzetti and
Doyle^, Cox and Penkett^, Quon et al.^, and Clark^. We have also shown
some experimental results and a kinetic model' on the photochemical
aerosol formation. In this work, those results of our previous studies
on the effects of various envirnmental factors, such as S02 vapor concen-
tration, relative humidity, UV light intensity and irradiation time, on
the formation and the evolution of aerosol particles are summarized, and
an additional study on the influence of the presence of foreign aerosol
particles on sulfuric acid aerosol formation are conducted.
EXPERIMENT ON FORMATION AND EVOLUTION OF SULFURIC ACID AEROSOLS
FROM S02 - H20 - AIR SYSTEM
Sulfur dioxide vapor was mixed with carefully purified and condition-
ed air, and passed through the reaction chamber surrounded by blacklight
source. S02 vapor concentration ranged from 0.05 to 10 ppm. Relative
humidity was kept at about 80 %, 60 % or less than 10 %. The maximum
irradiation intensity of UV light was about 0.15 mW/cm^ sterad, and this
value is equivalent to 0.05 hr~~l for the specific absorption rate const-
ant of S02 molecule(relative light intensity to the maximum is denoted
by Ir, hereafter). The maximum irradiation time was 750 sec. Particle
number concentration was measured with a condensation nuclei counter, and
the smallest size of detectable particles with this type of counter is
about 0.0025 )Jm. Size distribution of formed particles was determined by
276
-------
diffusion tube method.
Generally, as shown in Fig.l, the number concentration of formed
aerosol particles increases rapidly after some period from the onset of
Irradiation, and decreases gradually after reaching a maximum value.
The volumetric concentration of particles increases almost proportion-
ally to the increase of irradiation time, and the surface area concent-
ration approaches an equilibrium value. Aerosol particle number concen-
tration is strongly dependent on SC>2 vapor concentration and the rela-
tive humidity as shown in Fig.2. The particle number formation rate
is in proportion to the product of SC>2 concentration and the irradiation
ilntensity, particularly in the case of lower formation rate. The maximum
particle number concentration and the irradiation time required to reach
£he maximum value are expressed as a function of formation rate.
^article size just after formed is found to be smaller than 0.0025 ym
and almost homogeneous. Fig.3 shows variations of average particle size
which are increasing with the irradiation time.
As a whole, particle number concentration is strongly dependent on
both S02 vapor concentration and relative humidity, and the particle
size is mainly influenced by S02 concentration. Volumetric formation
rate of particles is also dependent on both SC>2 vapor concentration and
relative humidity, and the value ranges 0.15 to 18.7 ym3/cm3/hr within
this experimental condition. The oxidation rate and the over-all quantum
yield of photochemical reaction of 862 are 0.04 %/hr and 8 x 10~3,
respectively. And the former value is equivalent to be 0.7 %/hr for
noonday sunlight in summer of the middle Japan.
t, Sec
Fig.l Variation of particle number concentration(Np), volume
concentration(V) and surface concentration(SA) with
irradiation time: S02=l ppm, r.h.=60%, Ir=l.O
277
-------
0
600
800
400
t, sec
Fig.2 Variation of particle number concentration of formed
aerosols with irradiation time: Ir=1.0
x 10
,-6
1.5
1.0
0.5
- so
0.125
0
0
200
AOO
t.sec
600
800
Fig.3 Variation of average particle radius(r) of formed aerosol
particles with irradiation time: Ir=1.0
278
-------
EXPERIMENT ON SULFURIC ACID AEROSOL FORMATION
FROM S02 - H20 - PARTICLE - AIR SYSTEM
Metallic fume particles were generated using an electrical furnace
i'r, N2 atmosphere. A modified Rapaport type generater was also utilized
or production of searic acid aerosols. Particle size distribution was
>.<-asured by Electrical Aerosol Size Analyser(Thermo-System Inc., USA).
•erosol particles were neutralized electrically before mixed with an
irradiation system of S02 - ^0 - Air by passing over a radioactive
: .nirce(10 mCi of Kr-85). The irradiation and the measurement procedures
'-••....re almost the same as those used for the experiments of foreign particle
" ree.
Fig.4 shows the time change of number concentration of formed aerosol
-..Ticles in both cases of foreign particle free and added. At the early
::t.age of irradiation, aerosol formation is markedly hindered by the
ir.rHr.ioti of foreign particles. However, as the irradiation proceeds this
M:\iderance effect becomes smaller, and at a period of longer irradiation
ime, number concentration of formed part.icJes appears sometimes larger
than in the case of foreign particle free.
E
o
106
105
Id*
i i i i i i i
-
-
—
^=£===fe^
\
material j Np,cnrf3 1 r^m
*l • none
1 j o Lst«iric acsij 8 x 1 04 j Q. 1 3
^ Zn i 1 x105 iO.018
i '
,
i
T i 1 i 1 i 1 i
0 200 AGO 600
t, sec
Fig.4 Variation of particle number ,."-'.\centration with irradiation
time: S02= Ippm, r.h.=60%, 1 =0.2: Np, r; number corxentra
tion, average radius of foreign particles, respectively.
-------
In Fig.5, the number concentration of formed particles is shown as
a function of surface concentration of added foreign particles at the
fixed irradiation time of 50 sec. Even though the material and the size
of added particles are varied, aerosol formation is markedly affected
when the surface area concentration is ranged from 10 to 10 ym^/cm^
and almost disappears in the case of larger surface area concentration
than 10^ ym2/cm3.
10
n
i
o
ID
O
Zn, r = 0.12
Pb, r = 0.09
Stearic acid, r = 0.13 JJPD
102 103
SAF,
0
Fig.5 Particle number concentration of formed aerosol particles as
a function of surface area concentration(SAF) of added
particles: S02= Ippm, r.h.=60%, Ir=0.2, t=50sec.
KINETIC MODEL OF SULFURIC ACID AEROSOL FORMATION
The nucleation rate for a system of f^SO^ - H^O vapor was calculated^
at various relative humidities according to the heteromolecular nuclea-
tion theory. Nucleation rate is remarkably dependent on both S02 vapor
concentration and relative humidity. The embryo usually consists of
several molecules of ^SO^ and ten or twenty molecules of H^O, and the
size is about 10 to 15 A.
Taking into account the concentration of l^SO/ vapor and the formed
particles, kinetic model for each component can be expressed as,
dNr
dt
dNT
dt
= R - ( a
a
Np -
NP ) NG
280
-------
where, NQ is the number concentration of t^SO^ molecule, Np is the
number concentration of aerosol particles formed including embryo, t is
the time of aerosol formation and irradiation continues throughout this
time, R is the conversion rate of SC>2 to I^SO^ by photo-oxidation and
is given as a function of SC>2 vapor concentration and irradiation light
intensity, a is the conversion rate of
vapor to embryo, io is
the number of H2S04 molecules contained in a nucleated embryo, 8 (or
is the deposition rate of I^SO^ vapor (or particle) to boundary surfaces,
•y is the deposition rate of ^2^04 vapor to already formed aerosol
particles, and K is the coagulation constant.
^ Calculations of aerosol free conditions are performed for various
values of relative humidity and of photo-oxidation rate of SC^. Fig. 6
shows one of the calculated example? cf aerosol formation, by Eqs.(l)
•_\nd (2). Generally, vapor concentration oE ^SO^ molecule increases
rapidly at the start of irradiation and decreases slowly after reaching
3 maximum value. The maximum particle number concentration appears a
little later than for vapor concentration, but this time lag is not so
large. Surface area concentration approaches an equilibrium value.
This implies that at the early stage of the process nucleation is pre-
dominant, however, condensation of vapor on already formed particles
becomes significant in later stage.
109
1000
150C
t , sec
Fig. 6
Time change of some characteristic values ia aerosol
formation: Ras10^cm~2f,fo~'> , r.h.=50%, w is the weight fraction
of ^SO^ in a particle, other notations are the same as in
Figs. 1 through 3.
281
-------
When foreign particles exist, Eqs.(l) and (2) are also applicable
by assuming that foreign particles act as adsorbing matter for vapor and
that all other effects such as catalytic reaction and coagulation are
negligible. Adsorption effect of foreign particle can be taken into
account by letting 3c = YF Np, where NF is the number concentration of
foreign particles, and YF is the adsorption rate of vapor molecule on
a foreign particle surface and is given by,
TT r2 v
YF
1 , rz v 4 TT
6 4( r + I )D
where, r is the radius of foreign particles, v is the mean thermal
velocity of vapor molecule, <5 is the condensation probability of vapor
molecule onto a particle, 1 is a length nearly equal to mean free path,
D is the diffusivity of vapor molecule in air, and S is the covered area
of particle surface by adsorbed vapor molecule and is regarded zero for
perfect sink surface. Eq.(3) is reduced to YF= ^ r v f°r very small
particle with perfect sink surface.
Judeikis and Siegel^ estimated that the value of 6 is 1.0 to 20
x 10~" for the mixture of S02 and atmospheric dust and it is raised more
than 0.5 to 10 x 10" ^ in the presence of trace quantities of metal salts.
According to our previous experiments^, adsoption of SC^ vapor onto
lead fume particles in humid air is much greater than monolayer, and the
adsorption rate is so slow that the order of magnitude of 6 is 10"^.
Adsorption of 862 vapor by other particles such as iron fume is found to
be very slight. The values of 6 of ^SO^ vapor for various kinds of
aerosol particles are unknown, however, they are considered much larger
than the value of 862 vapor. Hence, in this calculation, adsorption of
S02 vapor is neglected, and it is assumed that foreign particle has
perfect sink surface for H^SO^ vapor. Various values of 6 are tried to
use for calculations, and it is found that the magnitude in the order of
10 is most suitable to the experimental results.
Fig. 7 shows a calculated example of the time change of the number
concentration of formed aerosol particles, where the value of R is
consistent with that obtained from the experiments. In Fig. 8, the
number concentration of formed aerosol particles at various irradiation
time is given as a function of surface area concentration of foreign
particles. Within the calculations for such a short irradiation time,
the value of S/4TTr2 is so small that the assumption of perfect sink
surface has insignificant meaning.
DISCUSSION AND CONCLUSION
Due to the different conditions between the experimental and the
theoretical studies, such as the discrepancies in the smallest size of
282
-------
10
10*-
0 500 1000 1500
t, sec
Fig.7 Variation of number(—) and volume(--) concentration of
formed aerosol particles with irradiation time: B=5 x 10
1, r.h.=50%, 6 =0.01
Fig.8 Particle number concentration of formed aerosols as a function
of surface area concentration of foreign particles at various
irradiation times: conditions used for calculations are sente
as in Fig.7
283
-------
particles taken into account and in the boundary conditions of wall, the
agreement and the validity of the kinetic model are unable to be dis-
cussed in detail. However, general tendencies and the order of magnitude
of various factors obtained from the experiments are well elucidated by
the present kinetic model in all cases.
As to the influence of foreign particles on sulfuric acid aerosol
formation, some other workersl»10 have shown enhancement or neutral
effect. However, it is noticed that, even in the presence of adsorbing
foreign particles, number concentration of formed aerosol particles can
be enhanced sometimes in the course of long irradiation as shown both
experimentally and theoretically. In such a case volumetric concentra-
tion of formed aerosol particles is always lowered by the presence of
adsorbing foreign particles.
REFERENCES
1. Gerhard, E. R., and H. F. Johnstone. Photochemical Oxidation of Sulfur
Dioxide in Air. Ind. Eng. Chem. 47:972-976, 1955
2. Renzetti, N. A., and G. J. Doyle. Photochemical Aerosol Formation in
Sulfur Dioxide-Hydrocarbon System. Int. J. Air Poll. 2:327-345, 1960
3. Cox, R. A., and S. A. Penkett. The Photo-oxidation of Sulfur Dioxide
in Sunlight. Atmos. Environ. 4:425-433, 1970
4. Quon, J. E., R. P. Siegel, and H. M. Hulburt. Particle Formation
from Photolysis of Sulfur Dioxide in Air. 2nd Int. Clean Air Congress,
Wash. D. C., CPID. p.330-335, 1970
5. Clark, W. E. Measurements of Aerosol Produced by the Photochemical
Oxidation of S02 in air. Thesis, Univ. of Minnesota, 1972
6. Kasahara, M., and K. Takahashi. Experimental Studies on Aerosol
Particle Formation from Photochemical Oxidation of Sulfur Dioxide.
submitted to Atmos. Environ. 1974
7. Takahashi, K. , M. Kasahara, and M. Itoh. A Kinetic Model of Sulfuric
Acid Aerosol Formation from Photochemical Oxidation of Sulfur Dioxide
Vapor. J. Aerosol Sci. 6:45-55, 1975
8. Judeikis, H. S., and S. Siegel. Particle-Catalyzed Oxidation of
Atmospheric Pollutants. Atmos. Environ. 7:619-631, 1973
9. Takahashi, K., and T. Tamachi. Some Experiments on Adsorption of S02
Gas on Metallic Fume Particles. Taikiosen-Kenkyu(J. Japan Soc. Air
Poll.). 7:1-6, 1972, in Japanese
10. Urone, P., H. Lutsep, C. M. Noyes, and J. F. Parcher. Static Studies
of Sulfur Dioxide Reactions in Air. Environ. Sci. Techn. 2:611-618,
1968
284
-------
PART III
AEROSOL SAMPLING
-------
SIZE-SELECTIVE SAMPLING FOR INHALATION HAZAKD EVALUATIONS
Morton Llppsann, Ph.D.
Institute of Environmental Medicine
New York University Medical Center-
New York, New York 100J6
The basis for the increasing application of size-selective aerosol
samplers is reviewed in terms of the factors affecting and parameters
1 ascribing particle deposition and retention in the hutian respiratory
tract. The available experimental data describing total and regional
particle deposition in man are discussed, as are the sampler acceptance
criteria adopted by BMRC, AEC, and ACGIH, and the ICRP Task Group
'vposition model. Techniques and equipment used for size-selective
•i^rosol sampling are evaluated in terms of the design principles applied,
o correspondence between design and performance of specific samplers,
.eir applicability to field conditions, and their ability to satisfy
npler acceptance criteria.
287
Preceding page blank
-------
INTRODUCTION
A given aerosol can be described by at least as many characteristic
parameters as there are measurement techniques. Several papers in this
session describe techniques for determining aerodynamic diameter distri-
butions using cascade impactors and aerosol centrifuges. Papers in the
following sessions describe techniques for chara cterizing aerosols by
their optical and electrical properties, by total particle counts, and
by total mass concentrations. However, none of these parameters
provide any direct measure of inhalation hazard.
The human respiratory tract is a size selective particle trap with
a continuously variable collection efficiency at all particle sizes
below its effective upper cut-size. Furthermore, inhalation hazards
are usually more dependent on the concentration of deposits at specific
sites or regions within the respiratory tract, than on the total amount
deposited. The deposition pattern is particle size dependent, and the
problem is further compounded by a very large intersubject variability
in deposition pattern even under identical exposure conditions. The
deposition characteristics of the human airways must be known in order
to realistically evaluate the inhalation hazard potential associated
with a given aerosol.
INFLUENCE OF DEPOSITION PATTERN ON INHALATION HAZARD
The toxic dose delivered to respiratory epithelium by inhaled
particles depends on their surface density and residence time. In the
conducting airways, where the air velocity is relatively high and there
are sudden changes in air path directions, the particles which deposit
tend to be concentrated on a small fraction of the interior surface.
Such areas, i.e., the dividing spurs of large bronchial airway bifur-
cations, the larynx, and the nasal septum, also happen to be the loca-
tions most frequently identified as sites of cancer origin among
populations exposed to occupational and/or environmental carcinogens.
Although the residence times on the surfaces of these airways are
relatively short, i_^e_._, generally less than about 2 hours, the exposures
may be frequent or continuous, and a limited number of cells may receive
very high dosages.
Another region where the dosage may be high is much deeper in the
lungs, i.e. , in the alveolar region where the gas exchange takes place.
Here, most of the deposition takes place by sedimentation and/or
diffusion, and therefore the particles deposit relatively uniformly
over a huge surface area, estimated to be 70-80 m2 JL/. Insoluble
particles which deposit in this region have very long retention times,
which vary with the physical and chemical properties of the particles.
288
-------
in some cases, these can be characterized by half-times measured in
months or years. Furthermore, such particles tend to be gradually
concentrated in semi-permanent storage depots as peribronchiolar dust
foci. Such accumulations are associated with chronic lung diseases,
e-j^, silicosis, coal-workers pneumocon:iosis (black lung)and emphysema.
EFFECT OF PARTICLE SIZE ON REGIONAL DEPOSITION
It follows from the preceding discussion that the hazards associated
with airborne particles are dependent on where they deposit within the
respiratory tract, or if in fact, they deposit at all. A summary of
the best available human in vivo regional deposition data was recently
completed 2, 3 /. For very large particles, i.e. -' 50 nn, the suction
created by the nose or mouth is Insufficient for them to be inhaled.
Particles which are drawn into the nose will all deposit there unless
they are smaller than ~ 9 urn. For 3 urn particles, nasal deposition
falls of to ~ 50%, and tracheobronchial deposition will remove about
20% of the remainder, i.e. , ~ 10% of the amount inhaled. Of the - 40%
which reaches the alveolar zone, about i will deposit and the balance
will remain airborne and be exhaled. Head and bronchial deposition of
i pm particles are very low, and almost all of these particles penetrate
to the alveolar region with the inspired air. However, only about 20%
will deposit there and the balance will be exhaled. For still smaller
narticles, deposition by diffusion increases, and the exhaled fraction
decreases. For molecular sized particles, i.e. , unattached radon
daughters, the rms diffusional displacement becomes so large that these
/articles deposit efficiently in the upper airways.
The particle sizes of interest in inhalation hazard evaluations
obviously depend on the toxic material involved. For materials associ-
ated with an elevated incidence of bronchial cancer, the particles
depositing on the bronchial tree are of primary concern. On the other
hand, for silicosis hazard evaluations, the only particles of interest
are those which deposit in the alveolar region.
AEROSOL SAMPLING FOR INHALATION HAZARD EVALUATIONS
Two different sampling approaches can ba followed in evaluating
.inhalation hazards. One approach, which has the virtue of general
applicability, and the possibility of retrospective reinterpretatiou,
Is to determine the aerodynamic size distribution of the whole aerosol.
With this information, the particle fraction in every size range of
interest is known. The disadvantage of this approach is in the
complexity and cost of the sampling and/or analytical effort- The
alternate approach is selective sampling, i.e *, to collect only that
fraction of the aerosol which is capable of reaching the target sites.
This approach has been widely used for sampling pneumoconiosis produc-
ing dusts, where the target sites are in the alveolar spaces. Selective
289
-------
sampling greatly simplifies the analytic task. The major problem and
limitation of this approach is the requirement for a definition of the
aerosol fraction which is potentially hazardous. Two similar, but
somewhat different definitions of pneumoconiosis producing dusts are in
widespread use, as will be discussed in the next section. For some of
the carcinogenic aerosols, where the large airways may represent the
critical deposition region, there have been no recommendations on
sampler acceptance criteria proposed to date.
Since the particle size parameter which has the greatest influence
on respirable mass deposition is the aerodynamic diameter, the most
reliable measurements can be made with instruments which classify
airborne particles by this parameter during the sampling process. It
is seldom possible to perform an accurate size distribution analysis on
collected samples, since the particles can no longer be restored to
their original state of dispersion. Thus, particles which were unitary
in the air may be analyzed as aggregates and vice versa. Furthermore,
particles analyzed by microscopy will be graded by a linear dimension
or by projected area diameter, and these are normally larger than the
true average diameter.
HISTORICAL BACKGROUND FOR RESPIRABLE MASS SAMPLING
The first standard sampler for pneumoconiosis producing dusts, the
Greenburg-Smith impinger was developed in 1922-25 through the cooper-
ative efforts of the U.S. Bureau of Mines, U.S. Public Health Service,
and the American Society of Heating and Ventilating Engineers 47. This
impinger arid the midget impinger, developed in 1928 by the Bureau of
Mines _5_/ , collect most particles larger than about 0.75 micron in a
liquid medium. Samples collected in impingers are analyzed by counting
the particles which settle to the bottom of a wet counting cell and are
visible when viewed through a 10 x objective lens. Particles larger
than 10 urn observed during the count are rejected by many industrial
hygienists as "non-respirable." The only alternative approach available
prior to 1952 was gravimetric analysis of the tota1 airborne particulate
sample. With total mass sampling there is no practical way to
discriminate against the oversized particles which tend to dominate
the sample mass.
STANDARDS AND CRITERIA FOR RESPIRABLE DUST SAMPLERS
British Medical Research Council (BMRC)
In 1952, The British Medical Research Council adopted a definition
of "respirable dust" applicable to pneumoconiosis producing dusts. It
defined respirable dust as that reaching the alveoli. The BMRC selected
the horizontal elutriator as a practical size selector, defined respir-
able dust as that passing an ideal horizontal elutriator, and selected
290
-------
the elutriator cut-off to provide the best match to experimental lung
deposition data. The same standard was adopted by the Johannesburg!!
International Conference on Pneumoconiosis in 1959 6/.
In order to implement these recommendations, it was specified that:
1. "For purposes of estimating airborne dust in its relation to
pneumoconiosis, samples for compositional analysis, or for assessment
of concentration by a bulk measurement such as that of mass or surface
area, should represent only the 'respirable1 fraction of the cloud.
2. "The 'respirable' sample should be separated from the cloud
while the particles are airborne and in their original state of
-dispersion.
3. "The 'respirable fraction' is to be defined in terms of the
free falling speed of the particles, by the equation C/CQ = l~f/fo,
where C and CQ are the concentrations of particle?; of falling speed /
in the 'respirable' fraction and in the whole cloud, respectively,
and fQ is a constant equal to twice the falling speed in air of a
sphere of unit density 5 urn in diameter."
U.S. Atomic Energy Commission (AEG)
A second standard, established in January 1961 at a meeting spon-
sored by the AEC Office of Health and Safety, defined "respirable dust"
as that portion of the inhaled dust which penetrates to the non-ciliated
portions of the lung Tj. This application of the concepts of respirable
dust and concomitant selective sampling was intended only for "insoluble"
particles which exhibit prolonged retention in the lung. It was not
intended to include dusts which have aa appreciable solubility in body
fluids and those which are primarily chemical intoxicants. Within
these restrictions, "respirable dust'" was defined as follows:
Size* 10. 5. 3.5 2.5 2
Respirable (%) 0 25 50 75 100
*Sizes referred to are equivalent to an aerodynamic diameter having
the properties of a unit density sphere,
For these "insoluble" internal emitters, the concentration of the
respirable fraction can be compared to a modified maximum permissible
concentration (MFC ) on a simple basis. In the procedure for calculating
MPCa values as estlblished by the ICRP, the equation is:
29]
-------
10
y/a(i-2
where
q = maximum permissible body burden,
/„ = fraction of isotope in critical organ, relative to whole body
burden
/ = fraction of inhaled aerosol that reaches critical organ
t = period of exposure (set at 50 years)
T = effective half-life, days.
The published MPCa standards for insoluble particles were calculated
from this equation using the assumption that fa = 0.12 in all classes
where the lung is the critical organ, i.e. , that 25% of the airborne is
"respirable" and one-half of the "respirable" dust is retained. Thus,
the MPCa for insoluble "respirable" dust should be 25% of the MPCa for
total airborne dust.
American Conference of Governmental Industrial Hygienists (ACGIH)
The application of respirable dust sampling concepts to other toxic
dusts and the relations between respirable dust concentrations and
accepted standards such as the ACGIH Threshold Limit Values (TLV's) are
more complicated. Unlike the MPCa's for radioisotopes, which are based
on calculation, most TLV's are based on animal and human exposure
experience. Thus, even if the data on which these standards were based
could be related on the particle size of the dust involved, which
unfortunately is unlikely, there would probably be a different correction
factor for each TLV, rather than a uniform factor such as 0.25.
The task of establishing alternative or revised TLV's based on
"respirable" dust concentrations, was begun by ACGIH at its annual
meeting in 1968, when ACGIH announced 8/ in their "Notice of Intended
Changes" alternate mass concentration TLV's for three forms of crystal-
line free silica to supplement the TLV's based on particle count
concentrations.
292
-------
The U.S. Department of Labor 9_/ adopted the ACGIH size-selector
criteria for respirable dust and extended its application to coal
dust and inert or nuisance dust.
Discussion of Standards for Respirability
Basically, the two sampler acceptance curves described in the
preceding discussion have similar, but not identical, characteristics.
This is illustrated in Figure 1. The shapes of the curves differ
because they are based on different types of collectors. The BMRC
curve was chosen to give the best fit between the calculated character-
istics of an ideal horizontal elutriator and lung deposition data,
while the AEC curve was patterned more directly after the Brown, e_t
a_L 10/ upper respiratory tract deposition data and is simulated by the
separation characteristics of cyclone type collectors. Some of the
practical considerations and problems in using elutriator and cyclone
samplers for these applications will be discussed in a subsequent
section. In most field situations, where the geometric standard
deviation (a ) of the particle size distribution is greater than two,
samples collected with instruments meeting either criterion will be
comparable. For example, Mercer calculated the predicted pulmonary
(alveolar) deposition according to the 1CRP Task Group deposition
model 11/ for a tidal volume of 1450 cm3 and aerosol with 1.5 < o < 4.
He found that a sampler meeting the BMRC acceptance curve would have
about 10% more penetration than-a sampler meeting the AEC curve 12/.
Calculations by Lynch 13/, Moss and Ettinger 14/, and Coenen 15_/
indicated cyclone/elutriator penetration ratios of 0.81 to 0.83.
<
o:
BMRC
LOS ALAMOS.
O Z 4 6 6 IO
DIAMETER UNIT DENSITY SPHERE, MICRONS
figure 1. BMRC, Los Alamos (AiIC), and ACGIH sampler acceptance curves,
(Reprinted courtesy American Industrial Hygiene Association).
293
-------
HISTORICAL BACKGROUND FOR SIZE-MASS DISTRIBUTION DETERMINATIONS
For practical purposes, aerodynamic size classification of "respir-
able" aerosols began with the development of the cascade impactor by
K.R. May 16/ during World War II. Up until this time, essentially all
particle size analyses were done by optical microscopy, which was
inherently tedious. Furthermore, optical measurements cannot provide
useful data on airborne density, or the aerodynamic drag of nonspherical
particles. Cascade impactor samples could be analyzed chemically, and
therefore could not only avoid these limitations, but could also be
used to determine the size-mass distributions of multiple aerosol
constituents. The real advantages of this approach led to a
proliferation of instrument designs 17-22/. Unfortunately, the cascade
impactor has some inherent limitations which are too often not recog-
nized by the users. There is a tendency to overload the collection
surface in the small particle stages in misguided attempts to collect
measurable sample masses. Attempts to avoid this problem by the use
of adhesive layers or fibrous filters on the impaction plates have
often created other unanticipated problems, which will not be discussed
here since another paper in this morning's session will be discussing
inertial impactors.
The size-mass distributions of industrial and ambient aerosols can
also be measured using aerosol centrifuges, cascade centripeters and
multicyclone samplers 23—27/. The aerosol centrifuges, which will be
discussed in two papers later in this session, provide better resolution
and lower size limits than the other instruments. The cascade centri-
peter and multicyclone have poorer resolving power, but are least
subject to problems associated with overloading, i.e. , they can be run
for relatively long intervals and collect relatively large sample masses.
The performance characteristics of small cyclones used in air sampling
will be discussed in a later section of this paper.
The criterion most often followed in applying size-mass distribution
data to hazard evaluations has been the 1966 Task Group Deposition
Model 11./. This model received the approval of ICRP in 1973, and may
be promulgated by ICRP during 1975.
The Task Group defined a model respiratory tract described as
follows:
"(1) The nasopharynx, (N-P)-This begins with the anterior nares and
extends through the anterior pharynx, back and down through the posterior
pharynx (oral) to the level of the larynx or epiglottis...
294
-------
"(2) Continuing caudally, the next component, (T-B), consists of
the trachea and the bronchial tree down to and including the terminal
bronchioles...
"(3) We recommend the third compartment be entitled pulmonary
(P). This region consists of several structures, viz, respiratory
bronchioles, alveolar ducts, atria, alveoli and alveolar sacs...
The region can be regarded as the functional area (exchange space)
of the lungs. Its surface consists of non-ciliated, moist epithelium
with none of the secretory elements found in the tracheobronchial
tree..."
Assuming all dusts to occur as log-normal distributions,
deposition calculations were made for three different ventilation
states typified by tidal volumes of 750, 1450 and 2150 cm3,
respectively, at a respiratory frequency of 15 cycles/min.
One of the more significant conclusions of the Task Group study
was that the regional deposition within the respiratory tract can be
estimated using a single aerosol parameter, i_-_e_., the mass median
diameter. For a tidal volume of 1450 cm-^, there were relatively
small differences in estimated deposition over a very wide range of
geometric standard deviations (1.2 < aq < 4.5). Also, there were
relatively small variations associated with the substitution of 750
and 2150 cm3 tidal volumes.
Using the mass median diameter, as determined using any of the
instruments discussed in this section, and the Task Groups estimates
of regional deposition for that size, one can produce predictions of
the fractional deposition in each component region. These can be
converted to predictions of inhaled dose, using the measured value
of the overall mass concentration of the aerosol.
RELIABILITY OF BMRC, ACGIH, AND ICRP TASK GROUP DEPOSITION CRITERIA
The human in vivo deposition data base used by BMRC, ACGIH, and
the Task Group was not very strong, as was freely acknowledged by all
concerned. Many previous reviews on deposition have called attention
to the very large differences in the reported results 11, 28-31/.
COMPARISON OF DATA USED IN CRITERIA DEVELOPMENT WITH RECENT
MEASUREMENTS
Much of the discrepancy can be attributed to uncontrolled
experimental variables and poor experimental technique. The major
295
-------
sources of error have been described by Davies 32/. Figure 2 shows
data from studies of total respiratory tract deposition which have
been performed with good techniques and precision. None of these
data were available to BMRC. The data of Altshuler et al. 33/ and
Landahl 34, 35/ were not used by ACGIH or the Task Group, and the
others were not available. All were done with mouth breathing at
respiration frequencies of from 12 to 16. Tidal volumes varied from ^
to 1.5 liter. All appear to show the same trend with a minimum of
deposition at ^ h ym diameter. The four studies using di-2-ethyl
hexyl sebacate (DBS) 36-40/ all appear to have somewhat lower
absolute values.
It is also apparent that in most studies involving more than one
subject, there was considerable individual variation among the
subjects. Davies et al. 37'/ showed that some of this variation could
be eliminated by standardizing the expiratory reserve volume (ERV)
and thereby the size of the air spaces. They found that deposition
decreases as ERV increases. This was confirmed by Heyder et al.
38, 39/ who reported that there was little intrasubject variation
among six subjects when their deposition tests were performed at their
normal ERV's.
The DES data of Heyder et al. appear to represent deposition
minima for normal men. Their test protocols were precisely controlled.
There were no electrical charge on particles. With more natural
aerosol and respiratory parameters, higher deposition efficiencies
would be expected. The Landahl, Altshuler, Giacomelli-Maltoni, and
Martens data provide the best available estimate of total deposition
in normal humans breathing through the mouth. For nose breathing
the rate of increase in deposition with particle size above 1 ym would
be greater.
DEPOSITION IN THE ALVEOLAR ZONE
Brown et al. 10/ studied regional deposition during nose breathing
using china clay aerosols in narrow size ranges (CU = 1.25), with the
count median diameters between 0.9 ym and 6.5 ym, and these data
formed the major basis for the BMRC and ACGIH criteria. They
collected the exhaled air in seven sequential components, and used
the C02 content of each fraction as a tracer to identify the region
from which the exhaled air originated. The validity of these data
depends upon the accuracy of the association between the various
exhaled air fractions and their presumed sources. The Brown data
have been criticized 41/ on the basis that their simple two filter
model was inadequate, and that the rapid diffusion of CC>2 caused the
measured CO2 values to differ significantly from the corresponding
CC>2 concentration in the alveolar spaces. The resulting error in the
volume partitioning caused an underestimation of the alveolar
deposition.
296
-------
0.2 as as o.r i.o 23 s r
Aerodynamic Diam«t«r - pm
10
Figure 2. Total respiratory tract deposition during mouthpiece
inhalations as a function of aerodynamic diameter, except below ^ ^a,
where deposition is plotted vs. linear diameter. Data of Lippmann 2; 3/
are plotted as individual tests, with eye-fit average line. Other data
on multiple subjects are shown with average and range of individual
tests. Monodisperse test aerosols used were "Fe203 (Lippmann), triphenyl
phosphate (Landahl 34, 35/ and Altshuler _33/)» carnuba wax (Giacomelli-
llaltoni 5^57), polystyrene latex (Martens 5&_/)> and di-2-ethyl~hexyl-
sebacate (liuir 36/, Davies _37/, Heyder 38,_.39./» and Lever 40/).
(Reprinted courtesy The American Physiology Society).
297
-------
Altshuler, Palmes and Nelson 41/ estimated regional deposition
from mouth breathing experiments on three subjects in which measure-
ments of both the concentration of a monodisperse triphenyl phosphate
aerosol and the respiratory flow were made continuously during *
individual breaths 33/. Using a tubular continuous filter bed model
as a theoretical analog for the respiratory tract, regional deposition
in the upper and lower tract components was calculated for various
values of anatomic dead space. The upper tract penetration during
inspiration, pause and expiration was derived from the expired aerosol
concentration corrected for aerosol mixing. The alveolar deposition
estimates varied from subject to subject, and for each subject varied
with the volume of anatomic dead space. The particle size for
maximum alveolar deposition was estimated to be greater than 2 ym, but
its value could not be estimated since only one particle size > 2 ym
was used in the experiments. The alveolar deposition estimates of
Altshuler et al. in the 0.1 to 3.6 ym range and the alveolar
deposition values for 2.5 to 12 ym particles measured by Lippmann 2, 3/
using y-tagged aerosols and external in-vivo retention measurements
are in good agreement in the region of particle size overlap. The
alveolar deposition curve in Figure 3 labelled "in-vivo via mouth" is
based on the Altshuler and Lippmann data.
Figure 3 also shows an estimate of the alveolar depos .tion which
could be expected when the aerosol is inhaled via the nose It is
based on the difference in experimental in vivo head reten :ion data
during both nose breathing and mouth breathing tests 2, 3/. It can be
seen that for mouth breathing the size for maximum deposition is
^ 3 ym, and that ^ h of the inhaled aerosol at this size deposits in
this region. For nose breathing, there is a much less pronounced
maximum of ^ 25% at 2.5 ym, with a nearly constant alveolar deposition
averaging about 20% for all sizes between 0.1 and 4 ym.
The sampler acceptance criteris of ACGIH, BMRC, and the alveolar
deposition according to the in vivo data of Figure 3 and the ICRP
Task Group Model are illustrated on Figure 4. It can be seen that
the Task Group's model, which is based on nose breathing, over-
estimates alveolar deposition at all particle sizes. It should be
noted that some of the difference is due to the fact that the Task
Group's model is based on the MMD of a polydisperse aerosol rather
than a monodisperse aerosol. The overestimates would tend to be *
somewhat lower if the Task Group calculations were repeated for
monodisperse aerosols.
Samplers meeting the BMRC and ACGIH criteria would collect the
larger airborne particles with very similar efficiencies as the
conductive airways of normal mouth breathing humans. However, they
would remove less of the large particles than most nose breathing
humans, and "respirable" dust samples as defined by these criteria
298
-------
1.0
.9
.8
.7
.6
.t .5
8
.3
.2
6
Alveolar
via -
mouth
0.1 0.2 0.3 0.5 0.7 1.0 13 5 7 10 10
Aerodynamic Diameter - pm
Figure 3. Deposition in the nonciliated alveolar region, in 7, of
aerosol entering the mouthpiece as a function of aerodynamic diameter,
except below £ urn, where linear diameter was used. Individual data
;points aad eye-fit solid line are for the same Fe203 aerosol tests
plotted in Figure 2. The dashed line is an eye-fit through the r.edian
best estimates of Altshuler 4l/ on 3 subjects who varied as shown by the
vertical lines. The lower curve is based on the additional head
* correction resulting from inhalation via the nose. (Reprinted courtesy
: The American Physiology Society).
299
-------
1.0
0.8
0.6
0.4
0.2
0
ACGIH
Task Group Model (750cc)
Sampler Acceptance
Criteria
-BMRC
01 02 0.5 1.0 2 5 10 20
Aerodynamic Diameter
Figure 4. Comparison of sampler acceptance curves of BMRC and ACGIH with
alveolar deposition according to ICRP Task Group Model and median human
in vivo data from Figure 3.
300
-------
would provide conservative estimates of the pneumoconiosis hazard.
It should also be noted that for nose breathing, the deposited dust
will most likely be about 20% of the "respirable" dust.
PERFORMANCE CHARACTERISTICS OF "RESPIRABLE" DUST SAMPLERS
This discussion will emphasize data developed since 1969, when
an earlier review of the literature was prepared as a background
document 30/ for the AIHA's Guide for Respirable Mass Sampling 42/.
At that time, the field performance was not fully satisfactory for
either the elutriator samplers designed to meet the BMRC criteria,
or for the cyclone samplers selected on the basis that they followed
the AEC-ACGIH criteria.
The chief advantage of the horizontal elutriator over the cyclone
as a pre-collector is that its performance can be predicted on the
basis of gravitational sedimentation theory and the physical
dimensions of the device. Thus, it was sometimes claimed that
laboratory calibrations with carefully characterized test aerosols
were not needed. This generalization was true, at least in a relative
sense, in comparison to cyclone collectors where no adequate
predictive relations exist for collection efficiency. However, in a
absolute sense, the prediction of performance of actual elutriator
samplers is not completely reliable, as discussed in the earlier
review 30/. Some of the discrepancies observed in elutriator
performance have been attributed 43/ to technical shortcomings in the
manufacture of the elutriators, especially in the non-uniformity ot
the plate spacings.
One of the major limitations of all elutriator pre-collectors is
that it is difficult if not impossible to recover the collected
material for analysis. In many cases, it is even difficult to
periodically clean out the collected dust to minimize contamination of
the second stage collection by reentrained dust. For research studies
and other situations where the concentrations of both fractions are to
be determined, cyclone pre-collectors are generally used.
Another limitation of the elutriatior type of sampler is that it
must be operated in a fixed horizontal position. Cyclones, on the
other hand, can be operated in any orientation without significant
change in their collection characteristics 26, 44/. This independence
of orientation, combined with their smaller physical size at
comparable flowrates, are among the reasons that most of the recent
two-stage personal sampler designs have been bin It around miniature
cyclones as the pre-collectors. These samplers, combined with filter
collectors as the second stage and miniature battery powered air
pumps are small and light enough to be worn throughout a work shift.
301
-------
Most of the miniature battery powered pumps are diaphragm or
piston type air movers and therefore produce a pulsating flow, which
is basically incompatible with their usage with pre-collectors whose
collection characteristics are flowrate dependent. Reports by
Anderson, Seta, and Vining 45/, Lamonica and Treaftis 46/, Caplan
et al. 47/ and Blachman and Lippmann 26/ have demonstrated that the
instantaneous flow can be as much as four times the average. Since
there is a greater increase in collection efficiency at flows above
the average than there is a decrease for flows below it, the net
effect is to produce reduced cyclone penetration in pulsating flow as
compared to a constant flow at the average rate. Earlier field
samplers with pulsating flows underestimated the respirable mass, and
more recent models have been equipped with pulsation dampers in order
to overcome this problem.
Small variations in steady-state flowrate in cyclone collectors
may not be a severe problem, at least for those applications when the
parameter of interest is the "respirable" mass measured on the second
stage. Knight and Lichti 48/ demonstrated that variations in airflow
are corrected to some extent by changes in cyclone collection
efficiency. For nonfibrous test aerosols, including mica and silica,
there was essentially no change in the mass collected on the filter
for flowrates between 1.3 and 2.65 1pm. As the flowrate increases,
the aerosol mass entering the cyclone increases proportionally, but so
apparently does the collection efficiency. In an elutriator, the
effect would be the opposite; an increase in sampling rate would
result in an increase in penetration to the filter.
Despite all of their practical advantages as pre-collectors,
cyclones have not met with universal favor. Their major limitation is
the continuing uncertainty about their efficiency calibrations. There
are no predictive equations capable of providing useful calibration
estimates for the small cyclones used as pre-collectors 49/. Further-
more, for most cyclones, there is insufficient and/or conflicting
empiric data. The greatest amount of attention has been directed
twoard the Dorr-Oliver 10 mm nylon cyclone used extensively in under-
ground coal mines, and most recent studies 26, 47, 50/ confirm the AIHA
Guide 42/ which indicated that it comes closest to matching the ACGIH
criteria when operated at a 1.7 1pm flowrate. However, the developing
consensus does not include MESA, and they continue to require the 2.0
1pm flowrate indicated by their own laboratory calibration 51/ for
samplers used in the mines.
The recent calibrations of the 10 mm cyclone have also clearly
established that this cyclone cannot really match the ACGIH criteria at
any flowrate, because it has a cut-off characteristic which is much
sharper than that of the criteria. In other words, when the cyclone is
operated at a flowrate which produces a 50% cut at 3.5 ym, it collects
302
-------
more than 75% of 5 pm particles, and less than 25% of 2.5 um particles.
This relatively sharp cut-off characteristic makes this cyclone useful
in a itmlticyclone sampler which is used to determine aerodynamic size
distributions 26/.
. One other problem with the nylon cyclone is that, being an
insulator, it can accumulate a static charge. Vfaeu sampling aerosols
with very high charge levels, this can significantly affect collection
efficiency. Blachman and Lippmann 26/ showed that highly charged
aerosols with aerodynamic diameters below ~ 4 ym were collected with
higher efficiencies than charge neutralized aerosols of the same
aerodynamic diameter. Almich and Carson 52/ reported that the average
collection efficiency for 4 to 5 um charged particles was not signifi-
cantly increased, but that the variability in collection efficiency in
replicate runs was increased. This variability was absent when using
10 mm cyclones of the same design which were constructed of stainless
steel.
While the 10 mm cyclones are injection molded and virtually
identical to each other, all of the larger cyclones used as pre-
collectors are hand assembled, and the quality control exercised by
their manufacturers has frequently been quite poor _53/. On the other
hand, the larger cyclones have the advantage, at least for their
application as respirable dust pre-collectors, of much less sharp cut-
off characteristics. Their cut-offs turn out, probably fortuitously, to
closely match the ACGIH criteria, as illustrated in Figure 5. The flow-
rates at which these cyclones most closely match the ACGIH sampler
acceptance criteria are 9, 25, 75 and 430 1pm for the i-inch HASL,
Aerotec 3/4, 1-inch HASL, and Aerotec 2 respectively. For still higher
flowrates, Battelle _54_/has developed what they are calling a massive
volume air sampler. It includes an impactor collector which was designed
to match the ACGIH criteria at 1.3 x 106 ft3/day (21.2 m3/inin). This
multistage series sampler also includes an entry section with a cut-off
at 20 vim, an impactor with a 1.7 ym cut-off, and an electrostatic
collector to remove the particles penetrating the 1.7 ym impactor. This
sampler is currently being evaluated by EPA, and its ability to avoid
the characteristic limitation of impactors, i.e., particle bounce and
reentrainment, remains to be demonstrated.
SUMMARY AND CONCLUSIONS
Inhalation hazards are dependent upon the aerodynamic size distri-
butions of the airborne particles, and the. size-selective deposition
characteristics of the human airways. Realistic hazard evaluations must
therefore be based either on the determination of the aerosol's
aerodynamic size-mass distribution, or on concentrations determined from
samples collected with devices which simulate the human airway's
deposition characteristics. The latter approach is more direct and
303
-------
10mm Nylon
H A S I 1/2
A If OllC 3/4
H »S I 1
AEROTEC 2
• ACGIH CRITERIA
J : I I I
0 5
NORMALIZED PARTlClE DIAMETER
Figure 5. Collection efficiency v^ normalised particle size for four
dual-inlet stainless steel cyclones and the 10 ram Dorr-Oliver nylon
cyclone. (Reprinted courtesy American Industrial Hygiene Association)
304
-------
economical, but requires an accurate quantitative knowledge of human
regional particle deposition.
Recent human in vivo deposition data have helped to resolve some
of the long-standing confusion concerning regional deposition in man,
and provide an improved basis for the utilization of size-selective
sampling for inhalation hazard evaluations. Samplers which satisfy
the BMRC or ACGIH sampler acceptance criteria simulate the collection
Characteristics of the conductive airways of normal mouth breathing
humans. For particles between ~ li ym and the samplers upper cut-off,
the alveolar zone deposition is about twice as high for mouth breathers
as compared to nose breathers. For particles between 0.1 and 1.5 ym,
the route of entry has little effect on alveolar deposition. For nose
breathing, the fraction deposited in the alveolar zone remains nearly
constant at ~ 20% for all sizes between 0.1 and 3.5 ym. The ICRP Task
Group Model overestimates alveolar zone deposition at all particle sizes,
with the greatest discrepancies for particles > 5 ym and < 0.5 ym.
A large number of elutriators which can satisfy the KMRC criteria
are available, as are a large number of cyclones which can satisfy the
ACGIH criteria. In most cases, elutriator performance can be predicted
from physical theory, but the samplers are large in comparison to their
sampling rate, and must operate in a horizontal position. Cyclones are
much more convenient and compact field instruments, but must be
calibrated empirically. Unfortunately, the quality control in small
cyclone manufacture has frequently been poor, and the calibration data
needed to verify adequate performance have frequently been lacking or
of questionable reliability.
Despite the somewhat unhappy past history of cyclone usage in
inhalation hazard evaluations, their practical advantages are considerable
and there are no fundamental limitations to their proper usage. With
better quality control and the use of reliable calibration procedures,
they should continue to be widely used-
ACKNOWLEDGEMENTS
This investigation was supported by Grant /fES-00881 and is part of
a center program supported by Grant ES-00260, both from the National
Institute of Environmental Health Sciences. It is also part of the
center program supported by Grant #CA-13343 from the National Cancel-
Institute.
305
-------
REFERENCES
1. Liebow, A. A. Recent Advances in Pulmonary Anatomy. In:
Ciba Foundation Symposium on Pulmonary Structure and Function,
De Reuck, A. V. S. and M. O'Connor (eds.). Boston, Little,
Brown and Co., 1962. p. 2-25.
2. Lippmann, M. Regional Deposition of Particles in the Human
Respiratory Tract. In: Handbook of Physiology
(Environmental Physiology), Lee, D. H. K. and S. D. Murphy
(eds.). The American Physiological Society, Bethesda, Md.,
1975.
3. Lippmann, M., and B. Altshuler. Regional Deposition of
Aerosols. In: Proceedings of 20th OHOLO Biological Conference
at Ma'alot, Israel, 16-19 March, 1975. (In Press. John Wiley
and Sons, Ltd.).
4. Katz, S. H., G. W. Smith, W. M. Myers, L. J. Trostal, M.
Ingels, and L. Greenburg. Comparative Tests of Instruments for
Determining Atmospheric Dust. Public Health Bulletin No. 144,
Public Health Service, DHEW, Washington, D. C., 1925.
5. Littlefield, J. B., and H. H. Schrenk. Bureau of Mines Midget
Impinger for Dust Sampling. Bureau of Mines RI 3360, U. S.
Department of Interior, Washington, D. C., Dec. 1937.
6. Orenstein, A. J. (ed.). Proceedings of the Pneumoconiosis
Conference, Johannesburg, 1959. London, J. and A. Churchill,
Ltd., 1960.
7. Lippmann, M., and W.B.Harris. Size-Selective Samplers for Estimating
"Respirable"Dust Concentrations, Health Physics 8:155-163, 1962.
8. Threshold Limits Committee. Threshold Limit Values of Air
Borne Contaminants for 1968. American Conference of Governmental
Industrial Hygienists, 1014 Broadway, Cincinnati, Ohio,
1968.
9. U. S. Dept. of Labor. Public Contracts and Property Management.
Federal Register 34 (No. 96)-.7946, May 20, 1969.
10. Brown, J. H., K. M. Cook, F. G. Ney, and T. Hatch. Influence of
Particle Size Upon the Retention of Particulate Matter in the
Human Lung. Amer. J. Public Health 40:450-457, 480, 1950.
306
-------
11. Task Group on Lung Dynamics Committee II - ICRP. Deposition and
Retention Models for Internal Dosimetry of the Human Respiratory
Tract. Health Physics 12:173-208, 1966.
12. Mercer, T. T. Air Sampling Problems Associated with the
Proposed Lung Model. Presented at the 12th Annual Bioassay and
Analytical Chemistry Meeting, Gatlinburg, Tenn., Oct. 13, 1966.
13. Lynch, J. R. Evaluation of Size-Selective Presamplers. I.
Theoretical Cyclone and Elutriator Relationships. Amer. Ind.
Hyg. Assoc. J. 31:548-551, 1970.
14. Moss, O. R., and H. J. Ettinger. Respirable Dust Characteristics
of Polydisperse Aerosols. Amer. Ind. Hyg. Assoc. J. 31:546-547,
1970.
15. Coenen, W. Berechung von Umrechnungsfaktorem fur Verscheidene
Feinstaubmessverfahren. In: Inhaled Particles III, Walton, W. H.
(ed.). London, Unwin Bros., 1971.
16. May, K. R. The Cascade Impactor, An Instrument for Sampling
Coarse Aerosols. J. Sci. Inst. 22:187-195, 1945.
17. Mitchell, R. I., and J. M. Pilcher. Improved Cascade Impactor
for Measuring Aerosol Particle Sizes. Ind. and Eng. Chem.
51:1039, 1959.
18. Brink, J. A. Cascade Impactor for Adiabatic Measurements.
Ind. and Eng. Chem. 50:645, 1958.
19. Andersen, A. A. A New Sampler for Collecting, Sizing and
Enumeration of Viable Airborne Bacteria. J. Bact. 76:471, 1959.
20. Lippmann, M. Compact Cascade Impactor for Field Survey Sampling.
Amer. Ind. Hyg. Assoc. 22:348-353, 1961.
21. Lundgren, D. A. An Aerosol Sampler for Determination of Particle
Concentrations as a Function of Size and Time. J. Air Poll.
Cont. Assoc. 17:4, 1967.
22. Mercer, T. T., M. I. Tillery and G. J. Newton. A Multi-Stage
Low Flow Rate Cascade Impactor. J. Aerosol Sci. 1:9-15, 1970.
23. Stober, W., and H. Flachsbart. size-Separating Precipitation
in a Spinning Spiral Duct. Env. Sci. & Tech. 3:1280-1296, 1969.
307
-------
24. Kotrappa, P., and M. E. Light. Design and Performance of the
Lovelace Aerosol Particle Separator. Rev. Sci. Inst. 43:1106-
1112, 1972.
25. Hounam, R. F., and R. J. Sherwood. The Cascade Centripeter: A
Device for Determining the Concentration and Size Distribution
of Aerosols. Amer. Ind. Hyg. Assoc. J. 26:122-131, 1965.
26. Blachman, M. W., and M. Lippmann. Performance Characteristics
of the Multicyclone Aerosol Sampler. Amer. Ind. Hyg. Assoc. J.
35:311-316, 1974.
27. Bernstein, D., M. T. Kleinman, T. J. Kneip, T. L. Chan, and M.
Lippmann. Development of a High-Volume Sampler for the
Determination of Particle Size Distributions in Ambient Air.
J. Air Pollut. Control. Assoc. (In Press)
28. Davies, C. N. Deposition and Retention of Dust in the Human
Respiratory Tract. Ann. Occup. Hyg. 7:169-183, 1964.
29. Hatch, T. F., and P. Gross. Pulmonary Deposition and Retention
of Inhaled Aerosols. New York, Academic Press, 1964.
30. Lippmann, M. "Respirable1 Dust Sampling. Amer. Ind. Hyg. Assoc.
J. 31:138-159, 1970.
31. Stuart, B. O. Deposition of Inhaled Aerosols. Arch. Intern.
Med. 131:60-73, 1973.
32. Davies, C. N. The Deposition of Aerosol in the Human Lung. In:
Aerosole in Physik, Medizin und Tecknik. Bad Soden, W. Germany,
Gesellschaft fur Aerosolforschung, 1973. p. 90-99.
33. Altshuler, B., L. Yarmus, E. D. Palmes, and N. Nelson. Aerosol
Deposition in the Human Respiratory Tract. AMA Arch. Ind.
Health 15:293-303, 1957.
34. Landahl, H. D., T. N. Tracewell, and W. H. Lassen. On the
Retention of Airborne Particulates in the Human Lung II.
AMA Arch. Ind. Hyg. Occup. Med. 3:359-366, 1951.
35. Landahl, H. D., T. N. Tracewell, and W. H. Lassen. Retention of
Airborne Particulates in the Human Lung III. AMA Arch Ind.
Hyg. Occup. Med. 6:508-511, 1952.
36. Muir, D. C. F., and C. N. Davies. The Deposition of 0.5 um
Diameter Aerosols in the Lungs of Man. Ann. Occup. Hyg. 10:161-
174, 1967.
308
-------
37. Davies, C. N., J. Heyder, and M. C. Subba Ramu. Breathing of
Half-Micron Aerosols I. Experimental. J. Appl. Physiol.
32:591-600, 1972.
38. Heyder, J., J. Beghart, G. Heigwer, C. Roth, and W. Stahlhofen.
Experimental Studies of the Total Deposition of Aerosol
Particles in the Human Respiratory Tract. Aerosol Sci. 4:191-
208, 1973.
39. Heyder, J., L. Armbruster, J. Gebhart, and W. Stahlhofen.
Deposition of Aerosol Particles in the Human Respiratory Tract.
In Aerosole in Physik, Medizin und Technik. Bad Soden, W.
Germany, Gesellschaft fur Aerosolforschung, 1973. p. 122-125.
40. Lever, J. In Press, Aerosol Sci. 1974, quoted by C. N. Davies
in Reference #32.
41. Altshuler, B., E. D. Palmes, and N. Nelson. In: Inhaled
Particles and Vapors II, Davies, C. N. (ed.). Oxford,
Pergamon Press, 1966. p. 323-335.
42. AIHA Aerosol Technology Committee. Guide for Respirable Mass
Sampling. Amer. Ind. Hyg. Assoc. J. 31:133-137, 1970.
43. Thurmer, H. Investigations with the Horizontal Plate
Precipitator. Staub (English Translation) 29:35-42, 1969.
44. Treaftis, H. N., and T. F. Tomb. Effect of Orientation on
Cyclone Penetration Characteristics. Amer. Ind. Hyg. Assoc.
J. 35:598-602, 1974.
45. Anderson, D. P., J. A. Seta, and J. F. Vining III. The
Effect of Pulsation Dampening on the Collection Efficiency of
Personal Sampler Pumps. Report #TR-70. Cincinnati, Ohio,
Nat'l Inst. for Occ. Safety and Health, Sept. 1971.
46. Lamonica, J. A., and H. N. Treaftis. The Effect of Pulsation
Dampening on Respirable Dust Collected by Coal Mine Personal
Samplers. (Presented at 1972 Amer. Ind. Hyg. Conference.
San Francisco, California. May 1972).
47. Caplan, K., S. Sorenson, and L. Doemeny. Evaluation of a
Coal Mine Personal Dust Sampler. (Presented at 1972 Amer. Ind.
Hyg. Conference. San Francisco, Calif. May 1972).
48. Knight, G., and K. Lichti. Comparison of Cyclone and
Horizontal Elutriator Size Selectors. Amer. Ind. Hyy. Assoc. J.
31:437-441, 1970.
309
-------
49. Chan, T. L., and M. Lippmann. Prediction of Cyclone Collection
Efficiency. (Presented at Amer. Ind. Hyg. Conference. Miami
Beach, Fla. May 1974).
50. Seltzer, D. F., W. J. Bernaski, and J. R. Lynch. Evaluation
of Size-Selective Presamplers II. Efficiency of the 10 mm Nylon
Cyclone. Amer. Ind. Hyg. Assoc. J. 32:441-446, 1971.
51. Tomb, T. F., and L. D. Raymond. Evaluation of Collecting
Characteristic of Horizontal Elutriator and 10 mm Nylon Cyclone
Gravimetric Dust Samplers. (Presented at the 1969 Annual
Meeting of the American Industrial Hygiene Assoc. Denver, Col.
May 1969).
52. Almich, B. P., and G. A. Carson. Some Effects of Changing on
10 mm Nylon Cyclone Performance. Amer. Ind. Hyg. Assoc. J.
35:603-612, 1974.
53. Lippmann, M., and T. L. Chan. Calibration of Dual-Inlet
Cyclones for "respirable" Mass Sampling. Amer. Ind. Hyg.
Assoc. J. 35:189-200, 1974.
54. Mitchell, R. I. Massive Volume Air Sampler for Large Quantity
Particulate Size Fraction Collector. Internal Report.
Battelle Columbus Laboratories, 505 King AVe., Columbus, Ohio
43201.
55. Giacomelli-Maltoni, G., C. Melandri, V- Prodi, and G. Tarroni.
Deposition Efficiency of Monodisperse Particles in Human
Respiratory Tract. Amer. Ind. Hyg. Assoc. 33:603-610, 1972.
56. Martens, A., W. Jacobi. Die in vivo Bestimung der
Aerosolteilchendeposition im Atemtrakt bei Mund-bzw.
Nasenatmung. In Aerosole in Physik, Medizin und Technik.
Bad Soden, W-Germany, Gesellschaft fur Aerosolforschung, 1973.
310
-------
DICHOTOMOUS VIRTUAL IMPACTORS FOR LARGE SCALE
MONITORING OF AIRBORNE PARTICULATE MATTER *
Billy W. Loo, Joseph M. Jaklevic, and Fred S. Goulding
Lawrence Berkeley Laboratory
University of California
Berkeley, California 94720
ABSTRACT
The bimodal particle size distribution in the urban aerosol sug-
gests the use of a dichotomous sampling method with a cut point near
two microns. The collection of the two size fractions separately
should aid the identification of pollutant sources and the evaluation
of fine particle-related health hazards. We have constructed such a
sampler in which size segregation is effected by inertial impaction
across a virtual surface into a volume of relatively stagnant air.
Following a two stage separation, the particle sjze fractions are
deposited on membrane filters. One filter collects 951 of uncontam-
inated fine particles while the other collects all of the coarse par-
ticles along with 5% of the fine fraction.
The sampling scheme is tailored to match the requirements o£ ele-
mental analysis by X-ray fluorescence and total mass measurements by
beta gauging. The design parameters of the virtual impactor have been
optimized for minimum loss and sharp cut characteristics. A servo flow
controller is utilized to maintain a constant sampling rate of 50 Vniin
with a particle size cut at 2.4 ]_tm Stokes' diameter. Fully automated
units have been developed for a computer controlled monitoring network.
The compatibility of the virtual impactor with other analytical
instrument and its adaptibility to automation suggest that it may be
suited for wider use.
This work was supported by the Environmental Protection Agency
under Interagency Agreement with the U. S. Energy Research and
Development Admin ist rat ion.
311
-------
DICHOTOMOUS VIRTUAL IMPACTORS FOR LARGE SCALE
MONITORING OF AIRBORNE PARTICULATE MATTER
Billy W. Loo, Joseph M. Jaklevic, and Fred S. Goulding
Lawrence Berkeley Laboratory
University of California
Berkeley, California 94720
INTRODUCTION
Airborne particulate matter consists primarily of natural aerosols
such as soil dust, pollens and sea spray as well as a multitude of man
made pollutants in the form of hydrocarbons, sulphates, nitrates and a
broad spectrum of trace elements, many of which are considered poten-
tially toxic. Any effective air quality control strategy must be based
on detailed knowledge of the generation, transformation, dispersion
and depletion of those undesirable substances. Since gas-particle
interaction is recognized to play a key role in forming secondary
aerosol from primary pollutants, the study of particulate pollutant
is an important facet of the total problem.
Unfortunately the problem is made unwieldy by the complex meteoro-
logical, geographical conditions and anthropogenic processes. The sys-
tem under study has no steady state. Any comprehensive air pollution
study will require the large scale deployment of instruments to pro-
vide both time and position data on the concentrations of the polluting
species so that practical models of the systems might be developed.
Energy dispersive X-ray fluorescence analysis has now emerged as
a potent and versatile technique for trace element analysis due to its
capability for simultaneous rapid multi-element analysis. Quantita-
tive non-destructive analysis on a large and economical scale is pos-
sible for most elements of interest1. A fully automated analysis sys-
tem can measure 35 elements ranging from Al to Pb collected on two-
hour air filter samples through a sequence of three five-minute runs
with a detection limit on the order of 10 ng/m3 2. Such a system can
comfortably analyze some 20,000 samples a year.
312
-------
In understanding the evolution of pollutants and in assessing
potential hazards, it is desirable to obtain particle size information
in addition to elemental concentrations. It has been observed that
urban aerosols tend to have a bimodal size distribution with a minimum
at about 2 ym, (Whitby, et al3). The fine particle fraction peaks at
about 0.3 urn and is considered primarily to be the accumulation of com-
bustion products by condensation and coagulation. The coarse particle
fraction consists mainly of mechanically produced aerosol with particle
size peaking around 10 ym, although this is subject to large fluctua-
tions caused by gravitational settling and impaction losses. There
appears to be little mass transfer between the two modes'*. The occur-
ance of a fairly distinct minimum may correspond to a natural physical
effect since the creation of larger surface areas by mechanical means
becomes energetically unfavorable at sub-micron sizes. Further inter-
est in separating particles whose sizes lie on either side of the 2 ym
point arises because the deposition and retention of aerosols in the
human respiratory system is such that particles smaller than 2 or 3
microns are not efficiently removed in the nasal-pharyngeal region and
penetrate to the tracheo-bronchial and pulmonary regions5. The large
surface area (some 60 m2) in the lungs renders the human body vulner-
able to sub-micron pollutants. The coincidence of the size transition
at 2 ym in both the particle production and the health effects is a
prime motivation for developing a dichotomous sampler to collect size-
segregated fractions from each size range for subsequent X-ray fluores-
cence analysis.
OBJECTIVE
As part of the St. Louis Regional Air Pollution Study (RAPS), a
network of 25 monitoring stations (RAMS) has been set up to collect
meteorological, gaseous and particulate pollutant data. The initial
plan called for the installation of automated dichotomous air samplers
(ADAS) in ten of these stations to be operated under the control of a
central computer. Figure 1 illustrates an integrated program of sampl-
ing and analyzing some 20,000 samples per year. To facilitate data
handling, each filter carries a computer readable digital label. Total
mass loading on filters will be determined by beta-gauge measurements
before and after exposure and elemental analysis will be performed by
X-ray fluorescence analysis. These results will then be merged with
meteorological and other relevent data for inclusion in the main data
bank.
313 :
-------
oo
to
LO
I
f>
«*•
r^
10
•H
•a
T3
H
=>
Q.
H
O
(/I
rt
in
SP
•H
•XJ
O
O
bfl
•H
314
-------
Our objective was to develop a dichotomous sampler that matdies
the requirements of analysis systems. It should collect uniform
depositions on a thin substrate of low atomic number and should exhi-
bit a sharp particle size cut with very small or zero particle losses
in the apparatus. It should provide a high sampling rate to permit
adequate sample collection in a reasonably short time. Ease of sample
handling and reliable operation are also very important considerations
for large scale monitoring. The desired sampling rate and particle
size cut point have been chosen to be 50 H/min and 2 ym respectively.
APPROACH
One of the prime considerations in particle sizing is the selec-
tion of a physical mechanism by which size separation can be accom-
plished. Since the shape, density and dielectric properties of the
atmospheric aerosols cannot be predicted, the interpretation of the
term "particle size" depends largely on the method of measurement.
Since effects of particulates depend on their transport processes in
the atmosphere and on their uptake and retention through respiration,
inertial methods of separation should provide more relevant data than
electrical, thermal or optical methods. We shall therefore regard the
size of a particle as being the diameter of an aerodynamically equiva-
lent unit density sphere (i.e., the Stokes' diameter).
Conventional methods of inertial separation in the 2 ym range
often uses an impactor \\rhich is an air jet impinging on a collection
plate as shown in Fig. 2. The jet size, velocity and geometry are
controlled so that the inertia of particles above a certain size
causes them to overcome the drag force as they leave their deflected
stream lines and to impact onto the plate. A useful concept in impac-
tor studies is the Stokes' number Stk defined as
p C V D 2
Stk - IE 9L.JL. (1)
9 y D1
where p = particle density
C = Cunningham slip correction for the discontinuous nature
of fluid interaction when the pressure is low or when
the particle size is small compared with the molecular
mean free path in air
315
-------
V = mean fluid velocity of the jet
D = particle diameter
P
y = viscosity of air
D, = a characteristic size usually taken as the jet diameter.
At low Reynolds' numbers, the drag force on a particle is 3 iry VEL as
given by the Stokes' law, where V is the relative velocity between the
fluid and the particle. The Stokes' number is a measure of the ratio
of inertia! force (in this case centrifugal force) to the drag force.
It is also equal to the stopping distance of a particle in stagnant air
divided by the jet diameter. Under a given set of flow conditions,
particles with a Stokes' number higher than a critical value are impac-
ted onto the collection plate. It can therefore be regarded as useful
for scaling purposes.
\BI 7*1 I HI H
Fig. 2 Schematic of a conventional impactor
316
-------
Impactors with one or more size cuts have been widely used for
collecting size-segregated particulate samples. However, the inher-
ent difficulties with particle bounce, reentrainment, non-uniform
deposition and cumbersome sample handling have limited their potential
for large scale applications. A virtual impactor uses the principle
of inertia! separation, but the impaction plate is replaced by a
region of relatively stagnant air (receiving tube in Fig. 3 and 4).
The virtual surface formed by the deflected streamlines realizes
similar boundary conditions to those in real impactors. Large par-
ticles will pass into the forward low-flow region while small particles
will remain mostly in the high-flow air stream deflected radially
around the receiving tube. Both size fractions can subsequently be
deposited onto separate filters.
Fig. 3 Critical parameters of a virtual impactor
317
-------
FILTER A
:
\
Qo
U
A
,
FILTER B
FLOW
LIMITING
ORIFICE
CONTROL
VALVE
TO PUMP
Fig. 4 Schematic of a single-stage
dichotomous virtual impactor
A virtual impactor possesses several distinct advantages:
1) Particle depositions on the filters can be made quite uniform,
which is ideal for photon excited X-ray fluorescence analysis
and total mass measurement using a beta gauge.
2) The size and pattern of deposition can be controlled to match
the optimum requirement of the analysis system.
3) Since the object is to avoid collection within the apparatus,
particle bounce is a favorable phenomenon. In fact it is
desirable to shape streamlines as nearly tangential to the
surfaces as possible to encourage bounce and thereby reducing
losses.
4) Reentrainment of particles in the air stream is a second order
effect since only particles lost in the apparatus are sub-
ject to blow off.
318
-------
S) Collecting samples external to the apparatus not only facili-
tates automatic sample handling but also permits the design
of the impaction geometry to concentrate on maintaining pre-
cise and consistant separation characteristics.
6) The virtual irapactor can serve as an input stage for any down-
stream instrument.
VIRTUAL IMPACTOR DESIGN CONSIDERATIONS
An early study of a virtual impactor was done by William Conner6,
who utilized ideas from the cascade centripeter7. Although the
measurements and analysis were inaccurate, it did demonstrate the size
separating power of the virtual impaction principle. A more sophisca-
ted version using two cascade separation stages was developed by Carl
Peterson of the Environmental Research Corporation. We have made a
detailed evaluation on the ERC unit to assess the feasibility of the
virtual impaction scheme for large scale sampling8. Emphasis in our
work has been placed on improving the sharpness of the particle size
cut characteristics and reducing losses. The result of this and other
related studies revealed the importance of several factors which govern
the performance:
1) Jet Reynolds' number - The Reynolds' number is defined as
pVD/y, where p and y are the density and viscosity of air
and V is the mean air velocity and I) is the diameter of the
jet. Excessive turbulence which tends to occur at high
Reynolds' numbers sets an upper limit to the maximum flow
in a jet for a given size cut.
2) Flow symmetry and alignment - The sharpness of the cut char-
acteristic is degraded by the azimuthal flov; asymmetry about
the axis of each jet. In our design this effect is minimized.
It is clear that rectangular jets have inferior performance
in this regard and stable symmetrical flow configurations are
difficult to maintain. The coaxial alignment of jet and
receiving tube is obviously essential.
3) Flow control - A feedback system is needed to maintain a con-
stant operating point independent of variations in the fil-
ter impedance (including changes due to particle loading).
4) Particle bounce - Taking advantage of the non-sticking proba-
bility of solid particles, streamlines are designed to be as
nearly tangential to the physical surfaces of the impactor as
319
-------
is possible. Contours of parts are shaped to reduce impac-
tion losses.
5) Critical parameters - Referring to the notations in Fig. 3
on the relevant parameters of a simple jet and impaction tube,
it is necessary to determine an optimum set of parameters
QQ> DI> Ql/Qfl' D2/% an^L S/DI for minimum loss at a given size
cut. A fundamental study of conventional impactors has been
done by Marple9. However, the theoretical analysis is not
readily applicable to the case of the virtual impactor due to
the much more complex boundary conditions. We have adopted
an empirical approach, first measuring the range and sensitiv-
ity of each parameter then converging onto a region where the
cut point is relatively definitive and stable and where re-
maining parameters are optimized.
METHODS OF MEASUREMENT
Due to the close approximation of their geometric diameters to
unit density Stokes' diameters, dioctyl phthalate (DOP) droplets are
well suited as test particles. Particles between 1 to 10 ym in size
can very readily be produced in a Berglund-Liu monodisperse aerosol
generator with uranine (fluorescein sodium) used as a tracer for
quantitative measurements. The generator utilizes the uniform breakup
of a liquid jet into droplets as it passes through a vibrating orifice.
The final size of a liquid or solid particle is calculated from the
initial solvent concentration, jet flow rate and vibration frequency.
The monodisperse aerosol output is then used to test a particular
impactor configuration. Particle depositions can be dissolved in water
and the uranine content of the solution determined by UV fluorescence
techniques. Such procedures to determine deposits on parts have been
described in detail in an earlier report8.
The Stokes' diameters of solid particles are less well defined
because of uncertainties in the final density of the particles.
Figure 5 is a picture of an "8.8" ym NaCl particle as viewed by a
scanning electron microscope at 7000 magnification. The void fraction
after the solvent is evaporated is estimated at 50%. The sizes of
solid particles (uranine) used in our final evaluation were assigned
using the size cut characteristics of the virtual impactor as measured
with liquid DOP particles in order to compensate for such uncertainties.
320
-------
Fig. 5 A "8.8" \m NaCl particle as viewed
by a scanning electron microscope.
Flow measurements were made with pressure-corrected rotameters
and cross checked with a temperature-compensated integrating flow meter.
All calibrations were performed at 20°C and 735 mm Hg pressure.
We shall now digress to a problem of special interest which is
the measurement of the flow division within a two-stage vii^tual impac-
tor where the internal flows are not directly assessable. Referring to
the schematics of a two-stage virtual impactor (Fig. 15), the flow
conditions may be closely approximated by the equivalent circuit shown
in Fig. 6. RQ, R^ and R£ represent the non-linear flow impedances
associated with the inlet orifice, and the orifices in the coarse and
fine particle streams respectively. The flow division Qj/Qo may be
derived from the measurement of the total external flow as a function
of a single static differential pressure Po between the inlet and the
second stage of the impactor as shown in Fig. 7. Curve A is the char-
acteristic when RI and R2 are made zero by removing the inner section
of the virtual impactor. Curve B is the characteristic with all the
parts in place and Curve C is similar to B except that R^ is made
infinite by temporarily sealing the orifice such that R and R£ are in
321
-------
series. Under normal operation, the total flow is QQ and XY is the
pressure differential across R2. Hence, Q2 is determined graphically
by locating UV = XY. In this case Q-J/QQ is measured to be 24.9?.
Fig. 6 An equivalent circuit of fluid
flow in a virtual impactor.
322
-------
u
S
E 5
u
D,=.152
Q,/Q0=24.9%
25 30 35 40 45
FLOW (l/m)
50
55
XBI. 7411-8546
Fig. 7 A graphical solution to the internal flow
distribution problem in a two-stage
virtual impactor.
323
-------
RESULTS AND DISCUSSION ON DESIGN STUDIES
A series of measurements have been made to probe the behavior of
a single-jet virtual impactor. The coarse and fine particles were
collected on filters A and B corresponding to greater and less than
2 ym respectively. For each set of measurement, the collection effi-
ciency, defined as E = A/(A+B), and/or particle losses were observed
as a function of the parameter being varied. Starting with a set of
parameters near the 2 ym cut point, the objective was to develop an
optimum set of parameters with a minimum number of excursions in the
multiparameter space. The selection of 2 and 10 ym test particles
were used as fiducial sizes in this regard.
The results of such measurements are shown in Figs. 8 through 14.
Table 1 is a summary of the conditions under which each profile is
observed, using the symbols of Fig. 3.
0.8
0.6
0.4
Q0=16.7l/m
Q,/Q0=15%
o s
0.2
.100
.120
.140
D,(INCH|
.160
.180
Fig. 8 Behavior of E vs. D,
324
-------
0.6
0.5
0 4 -
0.3
0.2
0.1
Q0=16.7l/m
D, = D,=0.157"
I I
9 12 15 18 21 24
Q,/Q0(%)
Fig. 9 Behavior of E vs. Q-,/Q0
3U(
40
g
in
PURBULENCE LOS
rsa <*»
0 0
10
i i i • i • • r - i
\ lOp OOP
\ D,=D2=.157"
' \
\
\
V
\
\
\
\
\
1 1 1 1 I 1 ~ — <
579
Q,/Q0(%)
11
13 15
Fig. 10 Turbulent mixing as a function of Qj/Qn
325
-------
0.50
0.40
I r~
DP=2H
D2/D,=1.25
Q,/Q0=15%
0.4
0.6
0.8
S/D,
1.0
1.2
\BI. 741 I 8352
Fig. 11 Behavior of E vs.
326
-------
in
d
CO
d
03
X
N
O
(Nl
Q
in
in
(D
in
in
o
§
m
o
o
•H
•s
PQ
bo
•H
(i,
;%) ssoi
327
-------
10|i OOP
Q,/Q0 15%
1 4
D,/O,
1 6
1 8
Fig. 13 Losses of 10 \m DOP
particles as a function
of D2/D!
0°
§ 10
_J
1
10y DOI
Q,/Qo=
D,
A .157"
B .136"
C 128"
•%=_
I / ' '
' / '
15% • /
/ r
/ i
1 i
Dj/D, / o /
1 27 / / -
1.30 / / _
1.38 / /
/ ?
I /
! / '
/ ° /'
/ ^"''\___-^-;
10
20
30
40
50
Fig. 14 Losses of 10 \m DOP
particles as a function
ofQ0
328
-------
UJ
u
rH
2?
cx
1
p-
o
|
CO
rH
a
rvj
fZj
g
"
^ — '
. — 1
Q
0\=
O
Cr-
i — l
cx
!=
^j
S
CD
O'
Sa
Q cQ
^ y
p
.-H
U-.
t<
1— H
Pi
CNJ
M
to
o
rH
UJ
3
fjQ
*%
i — i
f|X
^
*•-*
t-T)
rH
Iv.
o
t-H
UJ
GO
S
h— *
•/,
rxj
O
"^J-
o
i-H
O
^rt"
UJ
^
3
^
^
^o
rH
UJ
CTi
a,
o
I-O
K">
*
r-H
O
.
r^-
UJ
^
g
^
f"~
O
rH
i_O
LO
3
0
i-H
t-'.
rH
'/.
a
^
i — \
o^
*T;
^
0
•
rH
o
-r
to
, — 1
t^
vO
rH
UJ
r-H
^.
r-H
B^
D
CM
C-J
I-O
a
23
t— i
CK
•-t.
***
m
to
LO
rH
f^-
\o
rH
UJ
rH
S
t—J
X
g
(VI
rg
hO
^
h-H
DJ
•
rH rH fH
CD to ro
. • •
Tj- Kl I--O
to
i — 1
a
S?
• — i
rv*
v-Ti
"^
(J~)
3
rH
rH T3
ni 0)
•H -P
•M O
•H 3
q 'O
•3 C
s
JS O
u
•H 11
& r-
r= •"
SH 12
(1)
13 w
2 w>
3 c
•H
t/1 -M
C m
o ,rH
5-. 03
n3 |3
1-H
to >
r-t
a
§
t-H
32'.'.
-------
Figure 8 shows the variation of E with Dj_. This strong dependence
on jet velocity is expected to be the dominant factor governing the cut
point. Figure 9 shows the variation of E with the flow division
The corrected values of E, after contamination of fine particles in
the coarse particle stream has been subtracted, indicate the need for
a significant amount of flow in the forward direction, particularly
for the particles near the cut point. At a given Q]/Qg there will al-
ways be the same fractional contamination of fine particles in the
large particle stream but more than one stage of separation can be
used to reduce this below the amount desired. The results in Fig. 10
were taken with a single-stage virtual impactor with three jets in
parallel. The cross-contamination of 10 ym particles in the fine par-
ticle stream is taken as a measure of turbulent mixing. It is obvi-
ously desirable to use a value of Q]/Q2 greater than 151 for each
stage of separation.
The dependence of E on the spacing S between the inlet jet and
the receiving tube is shown in Fig. 11. It is seen that E is about
constant for S/D^ greater than 0.5.
Figure 12 shows both E and the losses as a function of D^/Dj.
Note that E has a flat region for 1.1 < (D2/D^) < 1.5 and the impaction
loss on the lip of the receiving tube has a pronounced minimum at
r>2/D]_ = 1.3 and increased rapidly beyond 1.5. The same sharp increase
in losses at D2/Dj - 1.5 is also observed (Fig. 13) for 10 ym parti-
cles, suggesting that this extreme sensitivity is related to flow
geometry rather than to particle size. This may explain the poor per-
formance of virtual impactors with a rectangular slit geometry.
All these measurements were taken with the assumption that a total
flow rate of 50 £/min will be drawn through three inlet jets in paral-
lel to adiieve a cut point at 2 ym and the jet Reynolds' number will
typically be in excess of 5000. For best performance Marple has shown
that the optimum jet Reynolds' number is about 3000 for a conventional
impactor. Figure 14 illustrates the effect of particle losses as QQ
is increased. It appears that the wall losses are the result of the
increase in stopping distance of the very large particles as they leave
the diverging streamlines at the region of separation, although a small
component due to turbulence losses has not been ruled out. It might be
possible to overcome this impaction loss by using a more complex design
of the receiving tube.
The considerations of jet Reynolds' number, flow symmetries, tur-
bulent mixing, stability of the cut characteristic and flow geometry
for minimum loss have led to the conclusion that sampling 50 H/min with
a 2 ym cut point with a tolerable fine particle contamination in the
coarse particle stream requires two series stages of separation with
330
-------
three parallel jets for the inJet stage and a single jet for the
second stage., The appropriate parameters will be Qj/Qo ^ 15%,
D2/D1 = 1.3 and S « 1.
DESCRIPTION OF VIRTUAL IMPACTOR
The schematic of the complete virtual impactor design is shown
in Fig. 15. It consists of the following main components:
PARTICLE SIZE FRACTIONATOR
Details of this component are illustrated in Fig. 16 with the
key parts numbered. Air is drawn through three inlet jets (part 1)
in parallel. Their protrusion into the first stage cavity is neces-
sary to eliminate the "backwall" losses on part 2 due to the spatial
oscillation of streamlines as found in the ERG design, as well as in
some conventional impactors. Part 3 forms the first stage cavity.
The three small holes in this part are symmetrically located about the
central axis but are offset 60° azimuthally with respect to the coarse
particle receiving tubes (parts 4) to minimize flow interference.
These holes, in combination to the one in part 7, also govern the
internal flow distribution. The Q]/Qg for the first stage was adjust-
ed to be 25% to minimize wall losses in the cavity. The tapered lips
on the tubes have no significant effect on the cut point although they
do tend to defocus the streamlines and reduce cavity losses.
The three coarse particle jets are then converged by a 15° cone
(part 6) onto the second stage of separation after passing through the
drift tube (part 5). Parts 8 are three positioning rods which forms
an open cavity for the second stage jets. The ratio QJ/QQ here is
chosen to be 20%. Thus 2.5 £/min of air will pass through filter A
carrying all the coarse particles along with 5% of the fine parti-
cles. The fine particle stream of the second stage will merge with
that from the first stage and be deposited on filter B. In analysis,
a correction for the 5% contamination on filter A can be made based on
the amount of the uncontaminated 95% of the fine particles on filter B.
Table 2 summarizes the actual parameters used.
331
-------
IU
O
o
ui
HORIZONTAL SHUTTLE
SOLENOID
VALVE
XUI. 741 l-8r. K)
Fig. 15 Schematic of the automated dichotomous air sampler.
332
-------
INTAKE
(50 l/m)
20
XBL-749-1688
Fig. 16 Details of the vir
333
-------
PARAM-TliRS
First Stage
Each of 3 Jets
Second Stage
Single Jet
Qo
(£/min)
16.7
12.5
Qi/Qo
(*)
25
20
Hi
(mm)
3.86
2.87
DZ
(mm)
5.05
3.86
S
(mm)
3.81
3.18
Table 2 Final choice of virtual impactor parameters
The overall construction utilizes all stainless steel parts
(excepting part 6) for mechanical integrity and corrosion resistance,
compression 0-ring seals and tie rods (parts 13) with thumb nuts
(part 12) for easy disassembly. Strategic corners are shaped to
minimize losses. Figure 17 shows the actual components of a virtual
impactor.
FLOW CONTROLLER
Flow regulation is essential for precise measurement of the air
volume sampled and the maintenance of a fixed particle-size cut point.
This is accomplished as shown in Fig. 15 by sensing (through part 15)
the pressure differential P0 between the inlet and the second stage
of the impactor with a diaphram operated null switch (Dwyer Model
1640-5) which,in turn, causes the opening in a motor driven valve to
be increased or decreased to maintain the preset null condition. The
valve is simply a 5.1 mm diameter orifice pierced by a travelling
micrometer shaft with a 2° taper. A fixed orifice limits the flow
through filter A to 2.5 H/min.. The variable orifice and the null
switch thus form a feedback loop to compensate any impedance change
in filter B. The carbon-vane vacuum pump used (Cast Model 0522-103-
G18D) has adequate pumping power to overcome an increase of about 70%
in impedance from a typical initial value of 26.2 torr-cm2/£/min
(1.2 Mm cellulose membrane filter manufactured by Nuclepore Corporation).
334
-------
-------
AUTOMATED SAMPLING SYSTEM
Our objective has been to develop a fully automated sampling
system for extended continuous operation. It is designed to cycle
samples with the minimum amount of handling and preparation. A net-
work of such samplers is to be controlled by a remote computer which
also monitors the system status and possible failure modes. Key com-
ponents include:
FILTERS
The filters used are 37 mm discs of cellulose membrane filters
supplied and mounted in 5.1 cmx 5.1 cm plastic frames by the Nuclepore
Corporation. Up to 36 of these slides are carried in a linear array in
standard 35 mm projector cartridges (Argus Camera). Figure 18 shows
such a pair of cartridges containing the digitally labeled filter
holders.
SAMPLE CHANGER
The function of the slide changer is to extract a matched pair of
filters from side by side slide trays corresponding to the A and B fil-
ter stacks, (only one stack is illustrated in Fig. 15). A horizontal
shuttle manipulates the slides into their sampling positions where they
are clamped in the output tubes of the virtual impactor. Upon the com-
pletion of the sampling interval, they are undamped and withdrawn back
into the slide trays. The over-travel of the shuttle actuates a "Geneva
wheel" which advances the stack by one vertical increment to be ready
for the next insertion. A single motor drives the shuttle which per-
forms the function of transporting, clamping and unclamping slides
together with advancing the trays with a single forward and return
stroke.
336
-------
-------
FLOW MONITOR
Several types of out-of-range conditions in the flow circuit are
detected and indicated by the system. Excessive travel of the microm-
eter valve due to the presence of leaks or broken filters causes out-
of-range switches to be activated. ATI auxiliary pressure sensing snap
switch PI (Fig. 15) is used to detect improper clamping or a broken
filter. Since the vacuum needed for the fixed limiting orifice results
from the proper flow condition in the fine particle stream, P^ actually
monitors the conditions at filter B even before the micrometer valve
reaches its limits.
ELECTRONIC CONTROLLER
The selection of sampling intervals, execution of the sequential
steps, regulation of flow, detection of errors, monitoring and display
of the system status and communication to an optional remote computer
are performed via the control module shown in Fig. 19.
In order to maintain the synchronism of the samplers with the
clock, ten seconds are allowed for a sample insertion or withdrawal
cycle, which normally requires about seven seconds, to complete.
Figure 20 illustrates the time sequence of a typical sampling period.
While the vacuum pump is turned on continuously, actual sampling starts
at the twenty second mark when a so1enoid valve is opened. Another ten
seconds are allowed for steady flow conditions to be established before
the flow controller is enabled. The right hand column of the figure
indicates the sequence in which error conditions are checked. The maxi-
mum times allowed to complete a sample transport and flow adjustment are
10 seconds and 12 minutes respectively.
To ensure synchronization ir the event of short ac power failures
(< 10 min) the elapsed time clc;.*. and logic control circuits are auto-
matically switched to a rechargeable battery.
Figure 21 shows a complete sampler. It is contained in a portable
soundproofed electronic rack with a plastic dust cover and a special
inlet pipe designed to draw an isokinetic sample from a larger rooftop
sampling port. Figure 22 is a closer view of a sampler with the dust
cover removed. It shows the final modification of adding a 7.6 cm
section to the height of the virtual impactor to reduce losses in the
drift tube.
338
-------
o
sc
z
o
o
oc
UJ
en
o
o
CD
X 2 i
e
® 0
$ f
0)
8
u
•H
•p
u
-------
START
o
UJ
Q
UJ
O.
UJ
10
20
30
WITHDRAW
SAMPLES
—UNCLAMP
-CLAMP
RAISE
STACKS
"STANDBY
INSERT
SAMPLES
UNCI AMP
•Cf.AWP
— OPEN SOLENOID VALVE
FLOW
ADJUST
~ ENABLE
FLOW
CONTROLLER
— LAST SAMPLE
— WITHDRAWN
K10 SEC.)
L_
I FLOW ~"|
ADJUST.
I
END OF j CLOSE SOLENOID VALVE
PERIOD —-—,—. -~™™ —™™.,
INSERTED
«10 SEC.
- PROPER VAC SEAL
FLOW LIMITS
EXCEEDED
FLOW ADJUSTMENT
«12 MiN. EACH)
XBL 741 1-8538
Fig, 20 Tbr 4~i'Hctional time sequence
of a Soinplmg period.
340
-------
Fig. 21 Oblique view of a completed ADAS.
341
-------
Fig. 22 A close view of the ADAS with
dust cover removed.
342
-------
RESULTS AND DISCUSSION
Thirteen sampling units of the type described have been built. I
high degree of uniformity has been achieved particularly on the
critical virtual impactor parameters. For example, the spread of the
pressure differentials P^ required to draw 50.1 5,/min of air is less
than II among all the samplers. The difference in cut point between
two different units was measured to be less than 0.1 \m.
SIZE SEPARATION AND LOSSES
The results of an evaluation of a typical sampler are summarized
in Table 3 and 4 and the overall results are plotted in Fig. 23. Loss
measurements are tabulated for specific regions. Region 1 losses in-
clude those washed from parts 1 and 2 of Fig. 16, region 2 from parts
3 and 4, region 3 from parts 5 and 6, and regions 4 and 5 are from
parts 7 and 9 respectively. There are two obvious components of losses.
A loss peak near the size cut point reflects the intrinsic tendency for
particles of the cut point size to be intercepted by the physical sur-
face which deflects the streamlines (in this case the inner rim of the
receiving tubes). The observed fact that there is no deposition on the
top of the tubes suggests that the virtual impaction surface is some-
what below the opening to the tubes. The coarse particles loss compon-
ent is mostly gravitational settling on the cone (part 6) in the drift
tube. In the specific design shown in Fig. 16, impaction losses were
found for large particles on part 6. A later modification added
7.6 cm to the length of the drift tube to eliminate this component.
The tabulated measurements reflect this modification.
The high sticking probability of the liquid DOP particles to the
wall and to each other leads to the exaggerated loss condition. A pro-
nounced reduction in losses is observed for solid uranine particles.
• The E vs. Dp plot shows a very sharp cut characteristic. Its long term
stability (months) enables us to define the aerodynamic size of solid
particles whose final density is less well known. The slight down
shift of the solid particle loss peak compared with that of the liquid
is consistent with the expectation that large solid particles are more
likely to bounce.
343
-------
o
o
rH
CD
OO
LO
O
0
rH
, — I
^3-
o
rH
to
\o
CM
rH
.
Os]
rH
CO
.
rH
to
rH
PJ
1 — 1
co
••— J ^-
&
<^
0,
to
.
0
rH
O
o
0
o
00
o
LO
,
o
o
o
rH
|
rj
a
CD
0
0
o
o
.
rH
rH
cn
rH
rH
rH
rrj-
.
O
CD
(Nl
c3
G
g
o
•
rH
co
•
0
LO
•
CD
.
O
rH
,
0
o
0
0
CD
to
|
U
g
0
0
o
o
o
0
.
o
o
CD
g
u
g
o
o
vO
•
o
•
CD
o
•
co
^
.
o
rH
cn
.
*rj-
cn
•
rH
rH
O
LO
O
Cj
g
to
•
rH
•
O
LO
•
CD
CO
•
O
rH
•
cn
i— i
•
to
r-j
«
o
rH
0
•
fxj
rH
O
cn
§
o
CD rH
344
-------
«
*
o
LO
•
Kl
CD
•
CM
oo
rH
•
CN]
iH
iH
O
O
.
,H
w
h- 1
C/3
u 3-
IH < '
fe
345
-------
-r
CM
CO
X
•H
P •
V) Q>
•H t-J
JH .H
-------
FLOW CONTROL
The null switch for the pressure sensor is set to reduce hys-
teresis errors to less than 0.51. The repeatability in flow rate is
typically better than 0.2%. If constant pressure drop is maintained
across a set of fixed orifices, the mass flow, to a first approxima-
tion, is inversely proportional to the square root of the absolute
temperature. Aerosol concentrations are conventionally expressed in
units of mass per unit volume. Thus, the volume flow needed is direct-
ly proportional to the square root of the absolute temperature.
As the in-take air is cooled from 20°C to -35°C, the automatic
micrometer valve shows little adjustment as long as clean filters are
used in the device. When heavily-loaded filters are in place, a
slight closing of the micrometer vavle is observed as the temperature
is lowered, suggesting the larger viscosity contribution to the pres-
sure drop across the loaded filter.
STABILITY AND RELIABILITY
The flow calibration drift over a three month period was measured
to be under 0.5%. The almost continuous sampling over the same-period
produced no detectable change in the 2.5 &/fliin limiting orifice behind
filter A.
The samplers have been extensively tested under laboratory condi-
tions by continuously recycling filters over an extended period. An
equivalent of 15,000 samples have been run as part of this study.
After eliminating obvious problems in the initial debugging period,
the average failure probability has been reduced to less than 0.11
per sample.
LIMITATIONS
A real instrument often falls short of ideal performance due to
** compromises made to satisfy practical boundary conditions. The phys-
ical size of the apparatus sets an upper limit on the largest •"article
that may be efficiently sampled. For example, the 10 ym partite loss
is less than 1.51 for liquid particles, but there is a sharp rise to
701 at 20 ym caused by impaction on sidewalls and by gravitational
V settling.
347
-------
Some of the limitations arise from the properties of the filter
medium selected. An ideal filter should exhibit high filtration effi-
ciency, homogeneity, mass loading capability, and mechanical strength,
and should exhibit low trace impurity content, flow impedance, mass
thickness and moisture uptake, as required by beta gauge and X-ray
fluorescence measurements. Some of these requirements are obviously
mutually exclusive. The 1.2 ym cellulose membrane filter used is con-
sidered a good compromise. The power and weight considerations on the
vacuum pump lead to the choice of a pump that will maintain the desired
sampling rate of 50 £/min for up to a 701 increase in filter impedance
over its nominal value of 26.2 torr-on2/£/min. The corresponding load-
ing on the fine particle filter, which bears the main flow, is about
200 |jg/cm2. This leads to the limitation that in heavily polluted air
with 100 ug/m3 of fine particles, the maximum sampling time will be
limited to approximately 4.5 hours.
As presently packaged in the sound insulated chamber, the pump
requires cooling air at a temperature below 35°C to avoid accelerated
wear.
CONCLUSION
Of the thirteen samplers, ten are installed and operating in the
St. Louis RAMS network. It has been demonstrated that the virtual
impactor, with its distinct advantages over its conventional countei-
part, has fulfilled the need for an instrument to collect aerosols in
two strategic size ranges. The adaptability of the virtual impactor
to automation and its compatibility to other analytical instruments
suggests that it may be suited for wider use. Refinements on the
range of acceptable particle size, filter impedance and filter mass
loading should further improve its potential for large scale deployment.
ACKNOWLEDGEMENTS
The authors wish to express their gratitudes to T. Dzubay of
EPA and C. Peterson of ERC for making the ERC virtual impactor avail-
able for evaluation. We acknowledge the contributions of the follow-
ing LBL staff: R. Adachi, 0. Arrhenius, C. Cyder, B. Jarrett,
N. Madden, J. Meng, H. Riebe, A. Roberts, W. Searles, D. Vanacek and
S. Wright. The fabrication of thirteen samplers would not have been
348
-------
possible without the great support from the personnels of the Elec-
tronics and Mechanical Shops.
We also appreciate the cooperation of H. Schneider and R. Leiser
of the Nuclepore Corporation in developing cellulose membrane filters
of acceptable quality and mounted appropriately. We are indebted to
B. Liu of the University of Minnesota for evaluating filter efficiency,
and R. Giauque of LBL for analyzing filter impurity content.
REFERENCES
1. Goulding, F. S. and J. M. Jaklevic, Photon-Excited Energy Dispers-
ive X-ray Fluorescence Analysis for Trace Elements. Annual Review
of Nuclear Science, 23:45-74, 1973.
2. Goulding, F. S. and J. M. Jaklevic, X-ray Fluorescence Spectrom-
eter for Airborne Particulate Monitoring. Environmental Protec-
tion Agency Publication No. EPA-R2-73-182, April 1973.
3. Whitby, K. T., R. B. Husar and B. Y. H. Liu, The Aerosol Size
Distribution of Los Agneles Smog. In: Aerosols and Atmospheric
Chemistry, Hidy, G. M. (ed.). Academic Press, 1972. p. 237-264.
4. Whitby, K. T., On The Multimodal Nature of Atmospheric Aerosol
Size Distribution. Particle Tedinology Lab Publication No. 218.
University of Minnesota, 1973.
5. Air Quality Criteria for Particuiate Matter, Chapter 9. National
Air Pollution Control Administration Publication No. AP-49, 1969.
6. Conner, W. D., An Inertial-Type Particle Separator for Collecting
Large Samples. Journal of the Air Pollution Control Association.
Vol. 16, No. 1:35-38, January 1966.
7. Hounam, R. F. and R. J. Sherwood, The Cascade Centripeter: A
Device for Determining the Concentration and Size Distribution
of Aerosols. American Industrial Hygiene Association Journal.
Vol. 26, No. 2:122-131, March-April 1965.
8. Loo, B. W. and J. M. Jaklevic, An Evaluation of the ERG V-rtual
Impactor. Lawrence Berkeley Laboratory Report No. LBL-24..".
January 1974.
349
-------
9. Marple, V. A., A Fundamental Study of Inertia! Impactors. Doctoral
Thesis, Department of Mechanical Engineering, University of
Minnesota. December 1970.
350
-------
f
DESIGN, PERFORMANCE AND APPLICATIONS OF SPIRAL DUCT AEROSOL CENTRIFUGES
Werner Stober
f Fraunhofer-Gesellschaft zur Forderung der angewandten Forschung
Institut fur Aerobiologie
59^8 Schmallenberg, Germany
ABSTRACT
This paper gives a review of the design, the performance and the appli-
cations of spiral duct aerosol centrifuges as they are increasingly
used in aerosol particle size spectrometry and related aerosol research.
Background, advantages and limitations of this type of instrument and
its modifications will be discussed. Precision measurements of dynamic
shape factors on certain types of nonspherical particles as well as
applications to high-resolution size distribution analysis will be
reported. Comparative studies of different authors on the possibility
of using the spiral duct aerosol centrifuge as an absolute instrument
for aerodynamic size distribution measurements are compiled from litera-
ture. Data on aerosol particle densities obtained with spiral duot cen-
trifuges are reported and problems of measuring cigarette smoke and
dense aerosols are discussed. An adaptations of a short spiral duct
centrifuge for applications to ambient particulate air pollution is
described and a recent feasibility study of developing the instrument
into an aerosol mass distribution monitor by way of using piezo-electric
quartz crystals as size-selective mass sensors is reported.
351
-------
V • "~ . v
\
DESIGN, PERFORMANCE AND APPLICATIONS OF SPIRAL DUCT AEROSOL CENTRIFUGES
Werner Stober
Fraunhofer-Gesellschaft zur Forderung der angewandten Forschung
Institut fur Aerobiologie
59^8 Schmallenberg, Germany
INTRODUCTION
Experimental research projects on airborne particulate matter generally
necessitate the analysis of aerosol particle size distributions at some
point along the course of the investigations. To serve this purpose,
various analytical methods, procedures and instruments have been devised
and used in the past. Most of these techniques involve a straightforward
precipitation of samples of the aerosol and a subsequent statistical
analysis by microscopic evaluation. During the last 30 years, however,
beginning with the development of cascade impaction by May ^3 in 191+5,
a size-related dynamic separation of particles prior to their precipita-
tion became of increasing interest to aerosol scientists. An obvious
reason for this was, of course, the simplification of the distribution
analysis by the preceding experimental size fractionation, but there was
also the additional advantage that the forces effecting the dynamic size
separation were directly related to the stopping distance and the aero-
dynamic diameter of the particles, both of which parameters are of
fundamental importance in aerosol dynamics and in health hazard evalua-
tions of inhaled industrial aerosols. Thus, dynamic size separation for
distribution analyses was very desirable.
The first real aerosol particle size spectrometer actually providing a
continuous size spectrum in terms of aerodynamic diameters was built ir.
1950 by Sawyer and Walton^. Their centrifugal device, called a coni-
fuge, deposited the particles according to their aerodynamic diameter in
a size range between 0.5 and 30 ym on the outer wall of a rotating coni-
cal annular duct. Figure 1 shows a schematic diagram. The size separa-
tion was achieved by a laminar stream of clean air enveloping the aero-
sol entrance at the apex of the cone and flowing down toward the base.
The aerosol particles, when entrained and leaving the apex, were subject
to the centrifugal forces and, thus, traversed the clean air layer in a
radial direction at a velocity determined by their aerodynamic size. Due
to this effect, a continuous size separation was obtained and particles
of equal aerodynamic diameter were collected in concentric rings on the
outer wall of the duct. At 3000 rpm, the instrument would permit an
aerosol sampling rate of 25 cm3 min~1.
352
-------
AEROSOL
FLOW VALVf
AtkGSOl
(N'rfANCF
Figure 1: Schematic Diagram of the Conifuge Potor and Housing
vispotf i:
300 crri-^ miii"
p ar t i c I e _• l o
T! r If.ck of
In spite of the promising aspects of i^ir. decigr. concept, the ccnifage
did not receive much recognition for *-.;re than 1 ^ years , R If. hough Kei'..h
and Derrick "^ extended the lo^er ;-.• - r,e V-.rr.it for
0.05 Mm and increased the sampling rat- tc
specific interest in the conifure1 i'" •- ^-'cr
gravitational versions of aerosol oi.-.f- ~Y>'.
by Timbrell"2 and Boose4 could rot r.;., . "';. :
conifuge. The smallest sizes deposit- ! 'n
not below 1.5 ym and the sampling rate ~:-.:
ctrometers ab ir1, educed later
>v.> performar.ee data cf the
th'.- gravit"" >-.nal devices were
or^ly i c.^ " .-n; n" .
In reviewing the situation of certri fvfal
after assessing the limitaLioiiS of a ."•-.••:.
'duct aerosol centrifuge ''Gcetz., Stf;venr.on
gested in 1965 (Stober and "<=:3sack;" }
fuge type; size spectrometers could lift
aerosol entrances in modified design;"
drical annular ducts. It vac anticira':
would permit increased sampling rat^s
purposes. The cylindrical design woi 1 >
facilitating an exact theoretical f-r:
small-size i-istrument of the latter K:
Berner and Peichelt"' and shcvod 'hat t
did, iii fa-et, follow the theoretical ;
-
L. • - "hr .metry and
^ r: J v e •.,' <.' } e - • -1' ..p'.- d h e 1 i c al -
r-'-inirg ' -') , ;i i vas sug-
hat the pei for T.rTct cf t.rie ooui-
i.Tiprov^ij by enYp_eving ring slit
f'T.t i, r-~ (j. jl-vri'lf.1" "on^s or cy.VLn-
r.j th^t. ring -.V-'t aerosol inlets
*," de&ired for ,v:,6r./ rraotioal
tne addiL-i^ral t-- --"-'t of
f-. evaluation. A.TL '• aal
"ii"., 7,.^ ,1 <--!i~" •»• r, :" ~* *' '".v
O U '- . M 1.. r i 1 , I _ 1.J J .- .. ^ • 'Jl
<;.] d.-\,'C^i'; patterns
to;jr,- 7>trner and
i4
•e
-------
In addition to the suggested conifuge modifications, the review also
conceived an entirely different centrifuge design by combining elements
of the semi-dispersive Kast centrifuge''" with the winnowing air flow
arrangements of gravitational size spectrometers: On a disk-shaped
rotor, a duct of rectangular cross section was to be wound into a spiral
in such a way that the aerosol inlet could be located at the axis of
rotation.
During the following years, a variety of ring slit centrifuges of the
conifuge concept as well as the first spiral duct centrifuge were built
and tested (Stober^5; Berner and Reichelt^; Hochrainer and Brown^3;
Stober and Flachsbart^°>^9). A comparison of the performance tests of
these devices indicated that from almost all practical points of view
the concept of the spinning spiral duct was superior to the other
designs.
The most important drawback of the ring slit instruments was the diffi-
culty of introducing the aerosol into the rotating duct without risking
substantial particle losses or significant flow disturbances. In addi-
tion, a desirable increase of the aerosol sampling rate of a conifuge
required a design of big instrument dimensions. Maximum rates reported
in such a case (Stober and Flachsbart^°) were 1.2 liters min"'' with a
range of deposited sizes about one order of magnitude (0.3 to 3-0 um).
Miniature designs suffered from greatly reduced sampling rates down to
12 cup min (Hochrainer and Brown'3) and the gain in size resolution
was practically offset by increases in particle losses. In contrast, the
original spiral duct centrifuge (Stober and Flachsbart °) permitted
sampling rates of several liters per minute. Furthermore, the particles
were deposited over a size range of almost two orders of magnitude,
while losses could be kept very low with suitable aerosol inlets. In
reducing the sampling rate to some ^00 cm? min~', an excellent size
resolution not surpassed by any other dynamic spectrometer design could
be obtained.
LONG-SPIRAL-DUCT DESIGNS
The first spiral duct centrifuge was built at the University of Rochester
(Stober and Flachsbart^9) with a total duct length of about 180 cm. Al-
though it was not a perfect instrument from the engineering point of
view and comprised certain elements of overdesign, the new device could
immediately be applied to the research problems it was intended for.
Figure 2 shows a photograpn of trie instrument with the rotor lid removed.
The disk-shaped rotor has a diameter of 26.2 cm. The essential part of
it, the spiral duct, is of rectangular cross section and begins off-
center. Then, with the inner wall touching the axis of rotation, the
duct leads in a narrow semicircle from the center toward and parallel to
354
-------
Figure ?: Original Spiral Duct Centrif ige (Rotor Housing Opened and
Rotor Lid Removed)
the periphery of the rotor, from \otr. it com imp", i r. vr!rl. c ./vrtures
fcr two and a half more turns. F-.c^.r-' •. tr.o.'j fhe or'.g:nct.' patter* v.'hi?(i
.is composed of six semicircles, A late • design (Vcss, rttjr.ger a;.L
Coulter-'1) utilized an Ar chime do;-.:. -i : , -11 instead of the Vi.-t f^ve ceir.i -
circles. This modification may h'-.v •o.'C.iio-ig.^s in malh-:rc.t ' c:t, .-'.'iluf-f
tions and descriptions hut is i jrana' C-T ; ^1 '-."'th record J o -;. cxf ^rimental
J:;;'T edge of Ihe rote, r vis
-./-• subsequent n.rxlf. ±; vrr > Vui 1 1
^8 cii d?-cp. The si £n ifi ;-prre of
"/.e wi'3th of !,he spiraj duct
r <.,f tv.s rotor to !.00 C'.c at the
'f the cv,ct is of constgrit width,
;i of ihe duct h^s .-> ^r0"] pr~cve
•) chroi:i: ^m-plate-l br-^s ' il of
.-ovc^ \ he --nhir' '. \ *; -: v^.'l >. f +r.c
performance of the inc trumer:4: .
r.'he depth of the spiral duct fn..r. 1 i:e
3-30 cm ir the origina] design, but :;
wi:,h dijcts cut U.30 cm, U.^0 cm ar •! c,"
tnese changes will "be discussed l;l'-r
narrows down from 1.73 cr, a1 the * er.t
end of the first semicircle, T'i.'.' rt :t
Along the outer wall, the spire,.: i .*•
cut into the oute:*- "bottom ed^e, -:>-.•"-
C.OV;) ctn tnickness can, Of M-.^crtj: *.c
duct. The total foil l-=r.gt.,r ir ' , •
In operation, the closed ro' •: r '-r, . :r
a clockwise direction when s>-^r, fror'
-------
Figure 3: Original Spiral Duct Rotor (Viev of the Top of the Rotor Disk:
Rotor Lid Removed, Laminator Block and Aerosol Inlet Section
Inserted)
duct at the off-center inlet and passes first through an inserted lami-
nator which subdivides the flow ty five thin parallel foils extending in
the coaxial direction. In this way, the clean air is quickly statilized
and emerges as a laminar flow from the down stream end of the inserted
section. A new simplified laminator will be described later in this
paper. The air then approaches an exchangable aerosol inlet at the cen-
ter of the rotor, where the aerosol enters ccaxially through a. non-
rotating center bore of O.^Q cm. The aerosol is released into the duct
as a thin layer parallel to the inner wall. Subsequently, it is entrained
into the laminar air flow.
Different aerosol inlets may be used for different purposes. For high
size resolution, where reduced flow rates must be employed, a narrow
slit design adjacent to the inner wall is very useful. For minimum par-
ticle losses, an open cut-away design of the rotating section of the
356
-------
Figure k: "Cut-away" Type of
Aerosol Inlet Section
for Minimum Aerosol
Losses at Entrance
center bore extending into the duct
is the best solution. Such an inlet
section is shown in Figure ]-i. A syste-
matic investigation of the influence
of different inlet arrangements vas
made by Moss, Ettinger and Coulter^5.
The size separation in the spiral duct
is a simple process. On leaving the
center of rotation, the aerosol par-
ticles are subjected to centrifugal
forces and start moving in a radial
ILrect.cn across the air stream. Their
trajectories depend upon the operating
:onditior/s of the centrifuge and the
aerodjfucuiilc size of ".he particles-.
Thus, while the air is drawn down the
spirfeJ duct tovard the outlet, the
particles are deposited, according to
their size, ir. different locations on
the chromium-plated foil or some other
inst-rtt-a cell eating surface along the
outer wall. The deposit represents a
continuous si.^e speotrvj,i in terms of
decreasing aerodynamic diameters be-
ginning near the aerosol inlet and extending to the end of the -spiral.
To facilitate the analysis, the collecting foil can easily be removed
from the instrument.
The original arrangement for the fjcv cor,trols of the spiral centrifuge
was essentially an open system reJ/ir.t on a source of compressed clean
air and a suction line. In principle. the suction line valve control Lr-d
the total flow through the spiral duct and the clear, air prese^r*3 had
to be adjusted in such a way that only the desired small fraction of the
' total flow vas drawn in through the -v.eroso." inlet. Thif- ^-r^ngement re-
01 ires extremely stable pressures ano subpressures but it i r- net very
sensitive to small leakages which occur particularly wht- t-he scaled
rf bearings of the shaft housing are wtrt.,
In more recent spiral centrifuge designs, modem V-rings replaced or
lightened the sealed bearings (e.g. Oeseburg and Rcos3i; Stoeer, Franzes
and Steinhanses^T) so that a simple closed air circulation system coulc
be devised where the total flow is set by a valve while th" c , rolled
bleeding from the pressure line determines the aerosol sample" _., rate.
Dx;e to the complexity of the geometiy and the physifp.l conditions ir the
spinning spiral duct, no quantitative theoretical model for the deposit
pattern was devised. Instead, empiric-si calibrations vere Piaile for
different operating conditions. Typical res alts for the original design
-------
0 1500rpm
3000 rpm
03000 rpm
6000 rpm
0.06 0.08 0.1
2 3456
D(10cm)
Figure 5: Calibration Curves of the Original Spiral Duct Centrifuge for
Different Operating Conditions
as obtained with monodisperse latex test aerosols are shown in Figure 5.
The curves indicate a maximum range of deposited particle sizes from
below 0.09 um to above 5 ym, thus covering almost two orders of magni-
tude.
The graph also reveals another significant feature of the long-spiral-
duct centrifuge: For small particle sizes around 0.1 ym, the deposit
location of the particles is no longer strongly dependent upon the cen-
trifugal forces acting on the particles. The upper two curves of Figure 5
show this quite clearly. They indicate that near the end of the spiral
duct the motion of the particles toward the collecting surface is pre-
dominantly controlled by factors other than the centrifugal forces.
There is, indeed, experimental evidence (Stober and Flachsbart ) that
the deposition of particles at long distances down the duct is primarily
due to an entrainment and transport of the particles by a slow secondary
358
-------
Figure 6: Secondary Double
Vortex Flow as Imaged
by an Aerosol Deposit
on a Filter in a
Curved Duct Centri-
fuge (Tillery61)
double vortex flow in the cross sectio-
nal plane of the duct. In a spinning
concentrical duct, Tillery"1 succeeded
to obtain a visible track of the double
vortex flow by collecting a fluorescein
aerosol on a filter placed into the
cross section near the end of the duct.
Figure 6 shows a photograph of the fil-
ter deposit. This double vortex flow
can be expected to exist because of the
curvature of the duct (Lean1) and, on a
spinning rotor, even more so because of
the Ccriolis forces. These forces are
created by the motion of the air rela-
tive to the spinning rotor. Under the
standard direction of rotation of the spiral duct -entrifugp they act
perpendicularly toward the outer wall of the duct and, except for a
slight reduction at the center of the rotor where the duct is wider and
the flow velocity reduced, they remain at constant strength throughout
the entire duct. A different assessment made earlier (Stober and Flachs-
bart^9) was incorrect because it accounted only for the circular compo-
nent of the Coriolis force.
Theoretical attempts of computing the secondary flow by assuming steady-
state conditions (Stober, Hederer and Horvath5°l were made according to
a solution from boundary layer theory (Ludwieg3°) and by deriving a
numerical solution of the simplified IJp.vier-Stokes equations for c^oep-
ing flow. In both cases, however, the results were quantitatively net in
keeping with experimental evidence. Instead, the Ludwieg approach con-
fined the secondary flow to the boundary layer vhile the creeping flow
solution gave flow velocities for ths double vortex which were higher
than compatible with the undisputedly proper function of the spiral duct
centrifuge as a size spectrometer. This suggests strongly that the actual
secondary flow situation in the duct is a case of developing double vor-
tex flow. Figure 7 shows the streamline pattern of a fully developed
double vortex flow in the cross sectional plane of the duct as osculated
under creeping flow conditions.
Apparently, a full development of the secondary double vortex flow vill
4 be detrimental to the instrument, performance- and has to be avoided. As
known from experimental and theoretical evidence, the vertices can be
; confined to areas adjacent to ths small sides of the cross section if
the duct approaches the shape of a slot. In ether words, a hifh aspect-
ratio of height to width of the duct is desirable. Thus, trie :se/elopment
of the double vortex will become less influential by either increasing
the depth or reducing the width of the duct, The former was dene ir the
modified designs mentioned earlier in this paper. Figure 8 shows a^com-
parison of two calibration curves obtained under comparable operating
-------
330
300
250
200
180-
165
150
y
100-
050-
050 x
Figure 7 '•
Theoretical Pattern of
a Fully Developed Secon-
dary Double Vortex Flow-
under Creeping Flow Con-
ditions
conditions for the original design and
for the 5.08 cm deep duct modification
(Moss, Ettinger and Coulter35), respec-
tively. Evidently, the development of
the double vortex in the latter design
(aspect ratio 5-1 : 1) is less pro-
nounced so that the deposition of small
particles around 0.1 ym diameter re-
quires greater distances down the duct
than in the original instrument. The
improvement is gradual only. Since tech-
nical reasons practically inhibit sub-
stantial further deepening of the duct,
a reduction of the width would be the
method of choice for more effective
vortex suppression.
In all performance tests and applica-
tions of the original spiral duct cen-
trifuge at ambient temperature, it be-
came apparent that the thermostat ar-
rangement for the rotor housing was
actually not needed and the housing
itself was a source of additional air
friction and heat generation (Moss,
Ettinger and Coulter35). Thus, in sub-
sequent models intended for work at
ambient temperature, the rotor housing
was omitted and cooling arrangements
were confined to the shaft bearings
(Stober, Franzes and Steinhanses57).
Without the rotor housing, a modified
design of the non-rotating aerosol in-
let tube as introduced by Moss, Ettin-
ger and Coulter35 became necessary.
Figure 9 is a schematic drawing of a
more recent design.
360
-------
— 10
E
ca
LU
Q
O
0
o
ce
LU
10
01
0.01
3000 RPM , 50 cm/sec
i i i i i i_i i t i i
i i I I 1 I i—L_J I—1—L
18 36 54 72 90 108 126
LENGTH OF THE FOIL (cm)
162
Figure 8: Influence of the Different Depth of the Spiral Duct on the Ca-
libration Curves under Comparable Operating Conditions (Open
Circles: Duct Depth 3.3 cm; Closed Circles: Duct Depth 5.08 cm)
Figure 9: Design of a Non-Rotating Aerosol Intake Tube on the Botor Lid
of a Spiral Duct Centrifuge without Rotor Housing
361
-------
PERFORMANCE AND APPLICATIONS OF LONG-SPIRAL-DUCT CENTRIFUGES
In the paper describing the first spiral duct centrifuge (Stober and
Flachsbart^9)5 the instrument was applied to the determination of rela-
tive aerodynamic diameters of clusters and chains of latex spheres of
uniform size. By exploiting the unprecedentedly high size resolution of
the instrument at low sampling rates around 1% of the total flow, it was
possible to discriminate clusters of up to 23 primary latex spheres. Thus,
their aerodynamic size could be measured directly by the location of their
discrete deposits on the collection foil as shown in Figure 1o. Electron
microscopic work was merely required to confirm the identity of the clust-
ers making up the particular deposits. Other electron micrographs further
revealed that chain aggregates of more than three spheres, which were
less frequent and not found as discrete deposits, were interspersed
among the clusters and could be identified in specific locations by elec-
tron microscopic screening. Figure 11 shows a deposit of quintuplet
chains interspersed with other aggregates of various forms.
Figure 10: Photograph of Deposits of an Aerosol of Uniform Latex Spheres
of 0.71 pm Diameter and their Aggregates on the Foil Strip of
the Original Spiral Duct Centrifuge as Obtained under high
Size Resolution Conditions
Tables 1 and 2 give a summary of the data obtained (Stober1*6). The se-
cond columns show the relative aerodynamic diameter fn, which'is the ra
tio between the aerodynamic diameters of an aggregate and its n uniform
primary spheres. The third columns give the parameter IT, defined as the
square of the ratio between the equivalent volume diameter and the Sto-
kes diameter of the aggregate, thus being equivalent to the dynamic sha
pe factor < except for the influence of the slip correction.
The most interesting experimental result for the aggregate particles
was the fact that fn and K were practically independent of the absolute
size of the primary spheres. The mean error of the mean values of the
°f Sizes betveen 3'5 ^d 0.13 M
362
-------
Table 1
Dynamic Data of Cluster Aggregates
(0.13 ym <_ D-| <_ 3.5 ym)
n
2
3
1*
5
6
T
8
9
10
11
12
13
11*
15
16
17
18
19
20
21
22
23
*n
1.189
1.3U3
1 .1*71
1.568
1.676
1.71*8
1.812
1.887
1.936
1.996
2.ol*3
2.107
2.159
2.211
2.25U
2.297
2.328
2.370
2.1*28
2 . U87
2 . 537
2.577
K
1.123
1.153
1 .165
1 . 1 89
1.175
1.198
1.218
1 .215
1 . 238
1.2H1
1.256
1.2l46
1.21*6
1.2UU
1.25o
1.253
1 .267
1 ,2bT
1 .250
1 .230
1 .220
1 .217
"1 K
1 .0 ym
1.115
1.1UU
1,156
1.180
1.167
1. ;89
1 .209
1.206
1.228
1.231
1.21*6
1.237
1.237
1.235
1.235
l.2Uh
1.258
1.258
1 .21+2
1.223
1.213
1.210
Di
0.1 ym
1.080
1 .102
1.113
1 .13o
1.123
1.1 UO
1.155
1.155
1,172
1.175
1.187
1.181
1.182
1.182
1.187
1.190
1 .201
1 .201
1.190
1 .176
1 .169
1 .167
Table 2
Dynamic Data of Chain Aggregates
(D1 > 0.3 um)
K D1 Di
1.0 ym 0.1 ym
2
3
U
5
6
7
8
1.189
1.280
1.380
1.U2C
1.U50
1.U80
1 .520
1.123
1.270
1.323
1.U50
1 . 570
1.671
1.731
1.115
1 .25!*
L305
1 .1*25
1.538
1.634
1 .691
1.080
1.177
1 .211*
1 . 297
1 .^76
1.UU2
1.1*83
363
-------
\
1
/
" *)
Figure 11: Electron Micrograph of a Deposit of Quintuplet Chain Aggre-
gates Interspersed Among Other Aggregates of Uniform Latex
Spheres of 0.71 ym Diameter as Obtained under High Size Re-
solution Conditions on the Foil Strip of the Original Spiral
Duct Centrifuge
This finding has an important consequence. Since ic~ is practically con-
stant for an aggregate of given shape regardless of the absolute size
within the slip regime, it is acceptable to calculate the actual dynamic
shape factor by
K =
where
bD
C(D) = 1 + — (A + Qe 2X
is the Knudsen-Weber slip correction with empirical constants
(A = 1.2U6 , Q = 0.^2 , b = 0.8?) and the mean free molecular path
length of air molecules (X = 6.53X10~6 cm). Values of K at primary
sphere sizes DI of 1.0 and 0.1 ym are included in Tables 1 and 2.
364
-------
Investigations on several shape factors of irregular particles and
aggregates were made with the original long-spiral-duct centrifuge "by
Kotrappa20,2l. Particles of a low grade coal, uranium dioxide, thorium
dioxide and quartz in a range of respirable sizes between 0.2 and 5 vm
were deposited and analyzed. Table 3 summarizes the results, which were
discussed by Davies . It was argued that the shape factors ic~ and a,
which both depend upon the particle mass, were too high for coal and
quartz to be consistent with other data. This suggests that the mass
determinations by beta-activity measurements of the activated ash con-
tent of the coal and by colorimetric chemical analysis of the quartz
may have been in error. In contrast, the mass-independent ratio between
;the projected diameter as observed in the electron microscope and the
Stokes diameter as obtained with the spiral duct centrifuge showed re-
gular values.
Table 3
Shape Factors of Respirable Particles of Low
Grade Coal, Uranium Dioxide, Thorium Dioxide
and Quartz
Size range ic
D (ym)
G ^^st
Low Grade Coal 0.56 * k.21 1.80 ,0.38 }
1.0 1.95 l±0.02; 1.53
Uranium Dioxide 0.21 * 1.68 1.22 ,0.33 » 1.39
1.0 1.28 (±Q.03> 1 .1*2
Thorium Dioxide 0.23 - 3-38 1 .06 ,.0.22 . 1.67
1.0 0.99 x±0.02j 1.57
Quartz 0.7 +2.0 1.90 0.35 1-55
0.2 +2.0 - ~ 1.53
1.0 1.82 0.3^ 1.53
Dp : Projected diameter in the electron microscope
~ : Stokes' diameter
K : Dynamic shape factor , average of
a : Volume shape factor ^ size range stated
Another study on dynamic shape factors focussed on elongated particles
like asbestos fibers (Stober, Flachsbart and Hochrainer56), The results,
-. when compared to theoretically available dynamic shape factors of pro-
365
-------
late spheroids, favored the assumption that, in laminar duct flow, the
fibers are predominantly orienting their polar axis parallel to the
streamlines. The study also showed that the aerodynamic size of a fiber
is closely related to the actual diameter of the fiber while the length
is almost immaterial. Figure 12 is an electron micrograph of a spiral
duct deposit of amosite asbestos fibers, all of which have an aerodyna-
mic diameter of 1.675 Vim as indicated by the black sphere in the graph.
By extrapolating the empirical findings with chain aggregates of uni-
form latex spheres to long fibers, a semi-theoretical model permitted
the derivation of a relationship
D
ae
)l/2U/D)l/6 D
between the aerodynamic diameter Dae, the actual diameter D, the^length
H and the density p of the fiber. With p0 representing unit density, the
empirical factor K can be calculated from experimental data obtained
with the spiral duct centrifuge. For amosite and crocidolite, a value
of K = 1.08 was found assuming p = 3 grams cm~3. This compares reason-
Figure 12: Electron Micrograph of a Deposit of Asbestos Fibers (Amosite)
of 1.675 ym Aerodynamic Diameter as Obtained under High Size
Resolution Conditions on the Foil Strip of the Original Spi-
ral Duct Centrifuge
366
-------
ably withk =0.86 for chain aggregates, although the asbestos data
showed considerable scatter around the regression line as indicated
by Figure 13.
1 ft
Recently, a study by Kops, Dibbets et al.'° indicated that the empi-
rical relations obtained for the aerodynamic diameters of chain and
fcluster aggregates of uniform spheres can be extended to aggregates of
a large number of small spheres as produced by the exploding wire tech-
nique. The authors measured the aerodynamic diameters of aggregates of
six labeled iron oxides (59pe) and two gold aerosols (19°Au) of diffe-
rent primary particle size distribution in a long-spiral-duct centri-
fuge. Subsequently, they evaluated the electron micrographs taken of
the deposits. For the primary particle sizes, log-normal distributions
were obtained from the micrographs while the pattern of the logarithms
of the equivalent volume diameters, as represented by the logarithms of
the primary particle numbers n of the aggregates, versus the logarithms
of the aerodynamic diameters had two apparent regimes divided at, a va-
lue of n < 1Cn. Within these regimes, the log-]og size pattern followed
distinct relationships correspondig to those which were respectively
derived for chain and cluster aggregates of a few uniform spheres (Sto-
ber, Flachsbart and Hochrainer5°). Figure 1H gives an example for iron
UJ
f~
UJ
UJ
0)
o:
UJ
t—
UJ
5 -05
o
o
oc.
UJ
<
o
o
-1L
0,5
15
Z3
LOG
FIBRE LENGTH
FIBRE DIAMETER
Figure 13: Regression Line of Experimental Data of Asbestos Flair
(Crocidolite) Arranged According to a Semi-Theoretical
Aerodynamic Diameter Model
367
-------
10'
10'
10'
104
10
I T I I I I I I I I I
ae
I I I I I I I I I
Figure ih: Relationship
between die Number n
of Primary Particles
and the Aerodynamic
Diameter Dae (in ym)
of Iron Oxide Aggre-
gates Generated by an
Exploding-Wire Techni
que (Kops et al
0.2
0.5
1.0
oxide aggregates of primary particles of a median size of D-i = 0.027 Mm
and a geometric standard deviation of og = 1.8. In the lower regime, the
aggregates, shown in the left half of Figure 15, resemble chain aggre-
gates. The regression line corresponds here to
1/2
p0k
In contrast, the aggregates in the upper regime as shown in the right
half of Figure 15 are closer to a cluster shape and follow a relation-
ship
n1/3
3(lnOg)2
With these designations, the values of k and K, the latter representing
the dynamic shape factors of the aggregates except for the influence
of particle slip, can be calculated in both regimes and, by comparison,
a value of n for the transitional range can be found. Table k gives the
data.
368
-------
Figure 15: Electron Micrographs of Iron Oxide Aggregates Generated by
an Exploding-Wire Technique and Deposited in the Spiral Duct
Centrifuge; left side: Strand Shapes (Low Number of Primary
Particles, Dae = 0.259 ym); right side: Cluster Shapes (High
Number of Primary Particles, Dae = 0.925 urn) ; (Kops et al.'8)
Table h
Dynamic Shape Factors K for Aggregates of Iron
Oxide spheres from an Exploded Wire
primary size chain shape
(lower regime)
D1 k K
ym g
cluster shape transition
(upper regime) range
K n
0
0
0
0
0
0
.020
.02U
.02?
.0^1
.0^3
.0^7
1.8 0
1 .8 0
1.8 1
1.8 1
1.8 1
1.8 1
.81*2
.939
.050
.29^
.3^9
.^59
0
0
0
0
0
1
• 596
.665
.7^3
.916
.955
.033
n
n
n
n
ri
n
1
1
1
1
1
1
/3
/3
/3
/3
/3
/3
7
13
1U
15
15
-15
.63
.13
.U6
• 7U
• 50
.02
2100
7700
7^00
5100
U300
3100
369
-------
The high size resolution capability of the long-spiral-duct, centr i I'ugv,
which was successfully exploited for dynamic shape factor studies and
similar investigations, can also be utilized for dynamical measurements
of the particle size distribution of nearly monodisperse aerosols. This
was done in a study of the parameters which influence and limit the si-
ze resolution of the long-spiral-duct centrifuge (Stober and Flachs-
bart^O). The study showed that the size resolution is directly proport-
ional to the absolute value of the slope of the calibration curves as
plotted in Figure 5. Thus, for the favorable range between 18 and 70 cm
along the sampling foil, it could be shown that the size resolution pri-
marily depends upon the ratio of the aerosol sampling rate Fae to the
total flow F, and a relationship
AE>
D
was derived. This was confirmed experimentally by the fact that the
area covered by the deposits of uniform latex spheres could be reduced
with decreasing aerosol sampling rates. However, no further reduction
was obtained for aerosol sampling rates below 2 %. At this rate, the si-
ze resolution seemed to be better than 0.5 %•> which can be concluded
from the fact that very small relative standard deviations of 1.U to
2.k % as obtained for latex spheres in the electron microscope (Heard,
Wells and Wiffen^ ') were redetermined with the long-spiral-duct centri-
fuge within 0 .h % or better. Figure 16 gives the number distribution of
quasi-monodisperse latex spheres of 0.357 ym diameter as obtained by
electron micrographic count evaluation of a spiral duct deposit.
Similarly, in an investigation by Oeseburg, Benschop and Roos-^ , the
narrow size distributions of dioctyl phthalate (DOP) aerosols from a
condensation generator (Lassen^9) were studied with a long-spiral-duct
centrifuge. The authors found relative standard deviations of about 10 %
in the micron size range and felt that small experimental deviations
from a theoretically expected log-normal distribution could be attribu-
ted to coagulation processes in the generating system. Figure 17 shows
some of their normalized data and the curves of the log-normal distri-
butions they approximate.
When utilizing the long-spiral-duct centrifuge as an absolute instrument
for sampling and analyzing polydisperse aerosols, two adverse influences
have to be considered: With increasing size, the particle losses in the
aerosol inlet will become increasingly significant, and, for small sizes
precipitated way down the duct, the deposit concentrations may be distor-
ted by the secondary double vortex flow discussed earlier. To assess the-
se effects, a comparative study was made (Stober, Flachsbart and Boose55)
to detect systematic deviations between the spiral duct centrifuge, an
electrostatic and two thermal precipitators as well as a cascade impac-
370
-------
Figure 16: Size Distribu-
tion of Quasi-Monodis-
perse Latex Spheres of
0.357 ym Nominal Diame-
ter as Obtained by Eva-
luating the Deposit
Concentrations of an
Aerosol Sampled under
High Size Resolution
Conditions in the Spi-
ral Duct Centrifuge
(Data Points and
Approximated Normal
Size Distribution
Curve)
D(10"5cm)
tor by analyzing model aerosols of fluorescein deposited simultaneous-
ly in these instruments. For three different test aerosols, it appeared
that practically no corrections were necessary because of the random na-
ture of the deviations between the instruments. The data of the long-
spiral-duct centrifuge were quite consistent but, possibly, somewhat
insensitive for sizes near 0.1 ym. Figure 18 gives a typical result for
the size distribution analyses to be compared. Furthermore, a^compari-
son of the cumulative data of the centrifuge and the cascade impactor
revealed that, with the particular aerosol inlet system employed in
t,he centrifuge for this study, there was an upper size limit around^
P, ym for particle sizes actually sampled on the centrifuge foil. This
cut-off size was lower than theoretically expected, a result which was
also observed by Ferron and Bierhuizen°. Figure 19 gives the probabili-
ty plots of the cumulative mass distributions to be compared.
In a very careful absolute calibration study which was made by Kops,
Hermans and van de Vate19 with regard to the deposition along the cen-
ter line of the sampling foil of a long-spiral-duct centrifuge with a
duct of k.5 cm depth, the/, authors did find systematic deviations be-
371
-------
800 -
1.00 -
Figure 17: Size Distribu-
tions of Dioctyl Phtha-
late Aerosols and Their
Approximations by Loga-
rithmic Normal Distribu-
tion Curves (Median Dia-
meters of 0.6U vim, 1.07
urn and 1.33 ym, respec-
tively; relative stan-
dard deviations 10 to
12 %} ; Oeseburg et al.33
040 060 080 100 120 UO 160 180
diameter Imicronl
dn
d°ae
7-
5-
3-
2-
Spiral Centrifuge
• Thermal Precipilator S
" Electrostatic Precipitator
0,1 0.2 0.3 0.4 0.5 0.6 0,7 0.6 0.9 1.0
Aerodynamic Diameter DQ^ I 10' cm)
Figure 18: Particle Size
Distribution of a
Fluorescein Aerosol
Measured with the Spi-
ral Duct Centrifuge
(Open Circles and Cur-
ve ), a Thermal Preci-
pitator (Closed Cir-
cles) and an Electro-
static Precipitator
(Triangles)
372
-------
99.9-
99-
95-
90-
50-
10-
5-
1-"
0.1-
0.03
0.05
~o!T
To"
0.5 1.0 2.0 5.0
Aerodynamic Diameter DQe (10"*cm)
Figure 19: Logarithmic Probability Plots of the Cumulative Mass Distri-
bution Data of a Fluorescein Aerosol Miasurcd with the Spi-
ral Duct Centrifuge (Circles) and a Cascade Impactor (Tri-
angles )
tween the size distribution analyses of the centrifuge and an electro-
static point-to-plane precipitator. Instead of using the derivative of
the calibration curve 1 = f(Dae) for the conversion of deposit concen-
trations into size frequencies as required by a simple mathematical mo-
del (Stober and Flachsbart50)5 they determined a special experimental
correction function for the actual operating conditions employed in the
study. Figure 20 shows the corresponding graphs. The experimental data
show good reproducibility and although the validity of the correction
function hinges somewhat on the assumption that sampling with ohe elec-
trostatic sampler was not size-selective (van de Vate^^), the relative
pattern of the correction factor P applicable to 7,he derivative of the
calibration curve is physically meaningful. Figure 21 is rescaled from
the data by Kops and coworkers to bring P close to unity in the almost
size-independent range between 0.2 and O.I ym. Tnen, for smaller sizes
the increase of P > 1 indicates the compounding of particles a I- <,. the
center line of the collecting foil due to secondary double vortj- flow,
while the decrease P < 1 for larger sizes may be due to particle losses
in the aerosol inlet system used in the study.
373
-------
5OO-
200
100
50
20
10
Figure 20: An Experimental^
Correction Function <•
fc(Dae) by Kops et al.19
Replacing the Derivative
of the Size Calibration
Curve (<3.£(Dae)/aDae> UP~
per curve) in Computing
Size Distribution Data
from Deposit Concentra-
tions in the Spiral Duct
Centrifuge
0-2
05
Dae(1(T4cm)
Kops and coworkers19 also investigated the influence of temperature im-
balances of the rotor housing on the deposition patterns of aerosols of
uniform latex spheres and their aggregates. They found that the deposit
location remained unchanged but that the regular shape and the symmetry
of the deposits could be distorted by thermal forces. Figure 22 shows
typical results.
The influence of the curved contours of the deposits obtained under re-
gular operating conditions and the changes of the deposit concentration
across the foil at given foil lengths have been investigated by Ferron
and Bierhuizen^. These authors used radioactive NaCl aerosols in their
study and found that, for given strip widths of 2.U and 0.8 cm around
the center line of the foil, the deposit concentration was representa-
tive for mass distribution measurements within the first half of the
374
-------
0.1 0.2
0.5 1.0
?igure 21: Relative Correction
Fa.ctor P for Transforming
the Experimental Correction
Functior fc(Dae) into the
beriv-iti vc of the Size Cali-
v'rat-'.on ."•,;r^fa (rescaled
aft.er ivops et al.19)
length of the loil. Further down the duct, there
sits along the edges of the downstream half or" tn,
apparently due to the double vortex secondly .."lo/
"been found for other polydisperse aerosols TOO (T
result obtained by Ferron a,nd Bierhuizen'-1 at, 15M.1
of 10 liters min""^ is shown in Figure 2k r~i t • '-
foil. The authors gave no correction function, al
worthwhile to establish such function by disrep ru
posits within about 5 mm of the edges of the foiJ
center line area as done for number dictrir^ticn/
(Fig. 21).
•red ,-rrat ic depo-
.." . 1 -us effect i s
; i", o r t i o r. a nd h a r.
•• '-3 j . A typical
-I.L-Q a total flow
•it, r>;is along the
;h i1 vouJ.d be
che parasitic de--
it ili? Ir^j -"'-i1 y the
'ov;> and -. -. orkers19
An interesting study of the size distribution of ciJv, >d. cigarette srncke,
375
-------
;-, &•
r
4.
Figure 22: Deposit Patterns of a Latex Aerosol (Primary Particle Size
0.357 ym) at Different Thermal Gradients Maintained at the
Centrifuge Rotor; (Temperatures T-| at the Rotor Top, T2 a^
the Rotor Housing, Tg at the Bearings) from Top to Bottom:
a) T-,= 35°C, T2 = 25°C, To = 15°C;
b) T-]= 15°C, T2 = T3 = 25°C;
c) T1= 15°C, T2 = 25°C, T3 = 35°C;
d) T-| = T2 = T^ = 25°C
(Kops et alJ9)
undiluted particulate car exhaust and atmospheric aerosols by means of
the long-spiral-duct centrifuge was made by Porstendorfer^^. The author
labeled the particles by attaching 220Rn decay products to the aerosol
particles. Then he used a multi-channel analyzer for the number distri-
bution analysis of the deposit. Depending upon the age and the dilution
of the cigarette smoke, he found mean particle sizes from 0.21 to O.Ulj-ym
diameter. For the particulate matter in Diesel and gas engine exhaust,
mean diameters of 0.1-4 and 0.18 ym, respectively, vere determined. The
atmospheric aerosol had a distribution similar to the Dae~^-distribution
postulated by Junge15.
In his measurements with cigarette smoke, Porstendorfer used a maximum
aerosol flow of about 2 % of the total flow through the centrifuge. The
aerosol sample consisted of at least 10-fold diluted cigarette smoke so
that the smoke flov in the centrifuge was only 0.2 % or less. This ar-
rangement gave apparently reasonable results for the mean diameters of
the smoke particles. In contrast, long-spiral-duct centrifuge experi-
ments with undiluted cigarette smoke at a high relative aerosol flow ra-
376
-------
Figure 23: "Parasitic" Deposits of Polydisperse Aerosols along the Edges
of the Foil Strip of the Spiral Duct Centrifuge;
Top: Cigarette Smoke with Heavy Edge Deposits beyond 30 to
UO cm Foil Length;
Bottom: Fluorescein Aerosol with Significant Sdge Deposits
beyond 80 cm Foil Length;
(unpublished University of Rochester Photographs, L, Schwartz
and H. Flachsbart, 1970)
te of 10 % (Stober and Osborne52) gave rather irregular results. A rou-
tine evaluation of the surprisingly short deposit obtained under this
condition indicated an improbably high minimum size of 0.88 ym for the
diameter of the smoke particles. On the other hand, running the instru-
ment as a semi-dispersive device by using no clean air and, at a compa-
rable total flow rate, filling the whole cross section of the duct with
cigarette smoke through the a.erosol inlet, a minimum diameter of 0.26 ym
or less could be observed. This suggested that either the deposition pat-
tern in the spectrometric test involved a cloud settling effect of the
dense cigarette smoke or a different density of the gas phase caused a
distortion, if not both effects contributed.
The possibility of density influences was systematically investigated
(Martonen and Stober-^1-) by testing a variety of clean winnowing gases
against latex aerosols with gas phases other than air. The ?.~"-er pho-
tograph in Figure 25 shows an impressive case of deposit di-'Vrtion cau-
sed by a relatively heavy aerosol gas phase convicting of carbon dioxide
entrained by air as the clean winnowing gas. This result indicates that
precise measurements with the spiral duct centrifuge should not involve
377
-------
I Am
I Ah
4C-i
30-
20 -J
,0-
0— '
20-
— 1
1
1
j
i?0cm
20-|
104cm
82 cm
57cm
24 cm
13 cm
9cm
Figure 2U: Mass Distribution
across the Foil Strip of a
Deposit of a Polydisperse
NaCl Aerosol at Different
Foil Lengths of the Spiral
Duct Centrifuge (Ferron
and Bierhuizen")
the confluence of gas phases of significantly different density. The
gas phase of cigarette smoke is about 6 % denser than air (Martonen^ )
which certainly accounts for some of the distortions of an undiluted re-
lative smoke flow rate of 10 % in the spiral duct centrifuge. The magni-
tude of the cloud settling effect of undiluted cigarette smoke is still
under investigation.
For all aerosol particles of known dynamic shape factor, the spiral
duct centrifuge can be used to determine the density of the particles.
To achieve this, a suitable electron microscopic inspection of the de-
posited particles permitting the determination of the particle volumes
must be performed. Such measurements can actually be made on single, se-
lected particles, but there is the disadvantage that only the square
378
-------
iiiimmimmmmmmmmmmmmtm
Figure 25: Foil Deposits of a Latex Aerosol; Top and Bottom at Same
.T-'-d.le in cm (Primary Particle Size O.H8 ym)
,.''Top: Regular Deposit (Aerosol Gas Phase: Air; Winnowing Gas:
Air)
Bottom: Irregular Deposit (Aerosol Gas Phase: Carbon Dioxide;
Winnowing Gas: Air)
root of the density is obtained experimentally and the precision is li-
mited to the accuracy of the determination of the aerodynamic diameter
of the particle. For this diameter, the absolute accuracy of the spiral
duct aerosol centrifuge is estimated as 5 % so that an error of some
10 % for the density value of a single particle is not unusual. However,
when a large number of chemically identical particles of different size
is investigated, the data may become quite consistent. In the simple case
of spheres, such investigations are reported. Moss, Ettinger and Coul-
ter35 used a long-spiral-duct centrifuge for measuring the density of
submicron iron oxide spheres (0.12 to 0,8 urn in diameter) and obtained
a mean value of 2.53 grams cm~3, which is within one per cent of the va-
lue found for larger iron oxide spheres with independent methods by Sper-
tell and Lippmann „ This indicated that there was no significant change
in the density of iron oxide particles over a wide range of sizes. Si-
milar conclusions could be drawn fron other measurements cy Moss and co-
workers concerning spherical fly ash particles in the O.U to 2.9 PK dia-
meter range (p = 3-83 grams cm ) and a laboratory aerosol (h : 1 methy-
lene blue: uranine) between diameters of 0,6 and 2.2 ym (p = 1.37 grams
cm~3).
Aerodynamic particle density determinations of this kind can reveal syste-
matic errors of the calibration of the spiral duct centrifuge if it is
safe to exclude density variations. This is shown in a study 01 ',ne den-
sity of laboratory aerosols nebulized from diluted ammonium fluorescein
solutions (Stober and Flachsbart-' ) . When these polydisperse aerosols of
spherical particles were sampled under high size resolution conditions,
a correction of the calibration curve became necessary to obtain consist-
ent density data for the different particle sizes.
379
-------
0.15
010-
P
005-
—ft—__
-9-
-6
0.03 0.04 005 0.07 01
02
03 0.4 05
07
—r-
1.5
Figure 26 : Relative Standard Deviations of Local Deposits on Electron
Microscopic Grids along the Foil Strip of a Spiral Duct Cen-
trifuge after Drying and Sampling Polydisperse Fluorescein
Aerosols Generated from Ammonium Fluorescein Solutions of
1 % (open circles) and 2.5 % (crossmarks)
In the same study, the evaluation of the electron micrographs further re-
vealed that in all locations along the foil strip of the centrifuge, the
local deposits had narrow Gaiissian normal distributions whose standard
deviations reflected the size resolubion in that location. Figure 2.6 pre-
sents the results in terms of relative standard deviations for the lower
end of the'size range of a long-spiral-duct centrifuge where the size re-
solution deteriorates for reasons discussed earlier. Nevertheless, even
at the small aerodynamic diameter of 0.08 urn, the standard deviation of
the sizes found in that location is .still less than 12 %. This is a value
which, "by several definitions, still permits the term "monodisperse" to
be used for the size dispersity of such local deposits (Fuchs and Sutu-
gin9, VDI Guidelines65).
The occurrence of highly uniform sizes at any given location along the
foil strip of the long-spiral-duct centrifuge was exploited by Kotrappa
and Moss25 for preparing monodisperse samples from polydisperse aerosols
of insoluble, almost spherical fused clay particles. After deposition,
the size-selected samples were resuupended in air, thus forming the de-
sired monodisperse aerosols of the clay material as needed for inhalation
studies.
380
-------
In many other applications of the long-spiral-duct centrifuge, the in-
strument has been used as a convenient laboratory tool to characterize
the aerodynamic size distribution of test aerosols. An example of this
kind is the study by Blachman and Lippmann3, who investigated the per-
formance of multicyclone aerosol samplers. Similarly, van Buitenen and
Oeseburg63 used a long-spiral-duct centrifuge for comparing "light scat-
tering diameters" of aerosol particles with aerodynamic data. Heyder and
Porstendorfer12 also reported optical and aerodynamic aerosol measure-
ments involving a long-spiral-duct centrifuge. The literature on appli-
cations of the long-spiral-duct centrifuge in the laboratory seems to be
expanding. However, no field applications have been reported so far, al-
though there is a potential for this kind of work in view of the feasi-
bility of sampling rates of some 3 liters min"1 at size resolutions
around 15%.
SHORT SPIRAL DUCT DESIGNS
For many applied research problems, particularly in dust control or
health hazard evaluations of aerosols in industrial hygiene, it would be
sufficient to obtain simply a gross value of the total mass concentration
of all particles of less than about 0.3 um aerodynamic diameter rather
than the actual mass distribution below this size. Thus, it would be
worthwhile to replace the corresponding far-end section of the long spi-
ral duct by a suitable aerosol filter and to confine the aerodynamic di-
stribution analyses to the larger sizes. For gravitational spectrometers,
Walkenhorst66 introduced such filter arrangement in 1965 at the end of
his horizontal duct and actually obtained filter deposits with a conti-
nuous size separation.
The first short-spiral-duct centrifuge with an exit filter was built by
Kotrappa and Light23. By leaving the range of smaller sizes to a back-up
filter, they reduced the length of the collection foil to U6.2 cm and
coiled it into a rotor of less than 18 cm diameter. Unfortunately, the
authors misconceived their design called the Lovelace Aerosol Particle
Separator (LAPS) by employing a duct of expanding radial width which can-
not possibly "bring an improvement of the instrument characteristics (Sto-
ber47) because of reasons of secondary flow discussed earlier ir this pa-
per. The expanding spiral was supposed to deposit a given particle size
at a shorter distance from the aerosol inlet than in a duct of constant
width. According to Kotrappa and Light21*, an increase of the ratio of ex-
pansion of the duct would increase the range of deposited sizes. It ap-
pears obvious and could be shown though117, that these expectations are
clearly unfounded.
38.
-------
Figure 27 : Expanding Spiral Duct of the Lovelace Aerosol Particle Sepa-
rator without Aerosol Inlet System and Laminator (Photograph
by Lovelace Foundation, Albuquerque, N.M.)
Figure 27 shows a photograph of the expanding spiral duct of the LAPS.
The increasingly unfavorable aspect ratio of the cross section of the
duct (2.1 : 1 to 0.8 : 1) must necessarily increase the detrimental influ-
ence of the secondary double vortex flow so that a successful operation
may be expected to be more stringently limited for the LAPS than for
ducts with more favorable aspect ratios (3-3 : 1 to 5•1 : 1 for long-spiral
duct centrifuges). From all this, it would appear that the LAPS is a
lapse. However, Kotrappa and Light23 claimed an excellent size resolution
over the entire length of deposition at rotor speeds of 6 000 rpm for the
LAPS and they mentioned operating characteristics at h 500 rpm and total
flow rates up to 15 liters min"1,
These surprisingly favorable performance data triggered the construction
of two experimental rotors with short spiral ducts of constant width (Sto-
ber, Hochrainer and Flachsbart59). The rotor diameters (17-5 cm) were al-
most the same as for the LAPS rotor (17-78 cm) and the dimensions of the
ducts were chosen so that one rotor would have a duct of the same cross
section and aspect ratio as the upstream end of the expanding spiral of
the LAPS, while the other rotor would have the same constant duct width
but a better aspect ratio and a cross section close to that at the down-
382
-------
Figure 28 :
Short-Spiral-Duct De-
sign with a Wide Duct
of Constant Width for
an Experimental Aero-
sol Centrifuge (Aero-
sol Inlet System Re-
moved, Laminator in
Place)
stream end of the duct of the LAPS. Thus, with a width of 1 .67 cm for the
spiral duct , the rotor top geometry was the same for both versions (Fi-
gure 28} but the duct depths of 3-3 and 6.0 cm provided different aspect
ratios of 2.0 : 1 and 3-6 : 1 , respectively. It was anticipated that these
rotors would outperform the LAPS.
A comparison between the calibration curves of the LAPS and the two expe-
rimental short -spiral-duct centrifuges at 3 000 rpm and a total flow rate
of
liters min
14 7
is presented in Figure 29. As shown for the long-spiral
version14 7, this comparison reveals again that the constant -width spirals
do deposit smaller sizes at shorter distances from the aerosol inlet than
the. LAPS. This result is contrary to the stated purpose of the expanding
spiral duct design^"4, notwithstanding an author's incorrect denial^ Vr.at
such erroneous assumption was made.
The calibration tests of the constant -width spiral duct rotors at increas-
ed total flow rates were disappointing. Apparently, the relatively wide
radial dimensions of these ducts, although generally smaller than in the
LAPS design (1.5 to 3-9 crn) impaired a proper size spectrometer operation
of the centrifuges at total flow rates as low as 10 liters mi,."1 . In view
of this, 3 1 is very difficult to accept the statement by Koti v:pa and
Light i-J thai they were able to establish operating characteristics of the
LAPS at U 000 rpm and 10 or 15 liters mirr1 .
',83
-------
u>
s
o
GC
LU
t—
UJ
3.0
2.0-
Q 1.0
O 0.8
< 0.6
Q
UJ
< 0.3
DUCT DEPTH
LOVELACE
CENTRIFUGE
SIMPLIFIED
SPIRAL CENTRIFUGE
33mmo
60mm A
3 4 6 8 10 20 30 50
DISTANCE DOWN THE DEPOSIT (cm)
Figure 29 :
Comparison of the Ca-
libration Curves of
the Lovelace Aerosol
Particle Separator
and two Experimental
Short-Spiral-Duct
Centrifuges of Diffe-
rent Duct Depth under
Equal Operating Con-
ditions (3000 rpm,
total flow rate of
5 liters min"1)
There was also a limit to the rotor speeds applicable to the experimen-
tal constant width short-spiral-duct centrifuges. Only the version with
the favorable aspect ratio of 3.6 : 1 could be operated at 6 000 rpm with-
out showing excessive turbulent deposition (Figure 30). Thus, for a de-
sirable increase of the total flow rate as well as for an extension of
the range of deposited small sizes, a widening of the spiral duct beyond
the width of the original long-duct design does not seem advisable. The
contradicting results with the LAPS appear to be questionable.
lllllllllllllllllllllllllllllllllllllllllll
Figure 30 : Slightly Distorted Deposit of a Latex Aggregate Aerosol of
Uniform Primary Particles of 0.23^ ym Diameter in an Experi-
mental Short-Spiral Deep-Duct Rotor Operated at 6 000 rpm
and a Total Flow Rate of 5 liters min"1
384
-------
Figure 31 :
Short-Spiral-Duct-Cen-
trifuge Rotor with a
Narrow Duct (Aerosol
Inlet System, Lamina-
tor and Back-up Filter
Removed)
As concluded earlier by considering tiic- re
the spiral duct, a practical improvemenI o
short-spiral-duct rotor can only b" expect
than an expansion of the width of Uie spir
of corresponding design was built recently
rotor diajp.et
17 o en., 11 L
1.6 cm at th
along t lu
The duct
Figure 32 :
New Aerosol Inlet Section
and Laminator for Spiral
Duct Centrifuges
condary double vortex flow in
f the characteristics of a
"i1 from a reduction rather
;J duct. Thus, an instrument
(Stober et al.60). While the
p.el. -r of this device remained at
11 L spiral duct narrows down from
the center of rotation to 0.8 cm
concentric part of the spiral.
is "].h cm de.-p and, thus, uhe
aspect T-a!ios range from
-------
2
1.5
1
08
05
0.4
0.3
0.2
10 liters I mm
o 2000 rpm
rpm
5 liters I mm
°3000 rpm
&5000 rpm
33 :
Calibration Curves for
Various Operating Conditions
of a Narrow-Duct Short-Spi-
ral Centrifuge (Established
by the Leading Edges of
Latex Test Aerosol Deposits)
56 8 W
15 20
LEM1H (cm)
30 40 50 60
Calibration curves for a few selected operating conditions of the new
short-spiral-duct centrifuge are given in Figure 33. The graph shows an
obvious improvement over the experimental rotor designs. The narrow-width
spiral duct permits now total flow rates of 10 liters min"1 at 3000 rpm.
Under favorable conditions (5000 rpm and 5 liters min"1) the size separa-
tion on the sampling foil can be extended to 0.23 pm .
PERFORMANCE AND APPLICATIONS OF SHORT-SPIRAL-DUCT CENTRIFUGES
Reports on the utilization of short-spiral-duct centrifuges are not abun-
dant in the literature. Since the narrow duct design is a rather recent
development which may still be considered as being in the experimental
stage with regard to applications, all published studies applying short-
386
-------
spiral-duct designs to practical problems refer to the Lovelace Aerosol
Particle Separator and all of these reports come from the same labora-
tory.
It appears from published data that the actual performance of the LAPS
has not been investigated very systematically so far. In their evalua-
tion of the instrument, Kotrappa and Light23 me.de some performance state-
ments which to date were not yet followed up by experimental data. For
instance, no quantitative evidence is available for the allegedly excel-
lent size resolution over the entire deposit length at 6 000 rpm . The
same lack of data exists for the capability of the LAPS to separate and
deposit sizes down to 0.2 ym diameters. No calibration curves are offer-
ed for quoted operating conditions at- relatively high total flow rates.
One exception will be discussed below. Similarly, fiaabe^11 gave no evi-
dence for his statement that the back-up filter of the LAPS collects the
smaller particles so that they are also separated vith respect to aero-
dynamic properties. Kotrappa alludes repeatedly22>23 to the same effect,
but if it is real, no quantitative investigation or analytic use of it
has been published to date. There is little doubt that some of the per-
formance claims for the LAPS need to be better documented or else, in
view of the recent experience with short-spiraj-duct designs not suffer-
ing from the misconception of an expanding du^t, may be considered in-
flated.
In an applied study, similar to the work with the long-spiral-duct cen-
trifuge on fused clay aerosols25, Kotrappa, Wilkinson and Boyd29 prepar-
ed monodisperse aerosols of plutonium cxi-le foi inhalation studies "i:y
precipitating a polydisperse aerosol on the ubmyling foil of the LAPS
and resuspending the local deposits of relatively uniform c,i-/e. This stu-
dy reports operating conditions of the LAPfo of i-'^OO rpni arid a total flow
rate of 10 liters min"1. In view of i,he failure to obtain regular depo-
sits under comparable operating conditions with the experimental ~,hort-
spiral-duct centrifuges described earlier, the quert!.r.n .-arises whether
the quoted operating data were not in error. Thi?; conj eclure is support-
ed by the fact that a recent paper by Enabe et al.1^ deali>if" with a rout-
ine system for the production of monodispcrfoft plutnnium oxiclL- aerosols
calls the questioned paper a pilot procedure and uj:es four Lovelace Aero-
sol Particle Separators at 3600 rpm and U.8 liters min"1, ajbhoagh, in
view of the use of four separators, higher flov rates would have been
very desirable. However, mention is made of fJcw rates as being measured
erroneously by Kotrappa et al.2^. Thus, it may be concluded that the same
is probably true for the earlier work with the LAPS.
Other potential applications of short-spiral-eentrifuges were t^umerated
by Kotrappa and Light23, Their listing does nor. inriude any prac+.ical
aspects which the long-spiral-duct centrifuges have not already been ap-
plie to. However, the authors stress a number of practical advantages,
387
-------
of which the more cc'nv.i nrj ng ^ne.:.; :ire the smaller rotor find the short.or
sampling foil. Both i'> vi.turer, f'ucH i I,ate an easier handling :i n routine
operations. Surprisingly, the 3iteraturo on LAPS apllications does not
indicate any significant role or specific use of the 3 x 3 cm exit filter
arrangement.
Kotrappa and Wilkinson^ used the LAPS to investigate the mass-specific
activity of polydisperse fused clay aerosols labeled with a variety of
radioactive isotopes (91Y, 90Sr, lltLfCe, 137Cs and 106Ru). By comparing
the calculated mass of the size-specific local deposits on the sampling
foil with the radioactivity measured for such samples, they were able to
prove that the mass-specific activity of the labeled aerosols did not
change with particle size, except in one case involving a more volatile
isotope (106Ru).
In another study , the same authors applied the LAPS to particle density
measurements of the kind described earlier in this paper. The investigat-
ed size range extended from 0.5 to 3-5 yrn of aerodynamic diameter. In
this range, the density of fused clay aerosol particles was found relati-
vely constant and comparable to independent measurements with a Millikan
chamber (2.1 grams cm~3). Other aerosols of constant density were obtain-
ed by nebulizing and drying uranine (1.5 grams cm~3 ) and iron oxide (2.2
grams cm~3). However, oxide aerosols of uranium, cobalt and zirconium
which were prepared by heat treatment of degradable salt aerosols showed
density variations depending upon the parent material. For a plutonium
oxide aerosol prepared by a two-stage heat degradation, a density of
10.5 grams cm was measured which is within experimental error of the
actual bulk density (11.5 grams cm ) . However, the same value found by
Kotrappa, Wilkinson and Boyd^^ was questioned by Raabe et al. ^ because
of low humidity conditions and erroneously measured flow rates. The lat-
ter authors used the LAPS at safe operating conditions and obtained den-
sities of the plutonium oxide particles between 7 and 8 grams cm~3 for
aerodynamic sizes of 0.65 to 1.33 ym.
Besides a new presentation by Newton et al.36, there is a total of fif-
teen other applied studies contained in two annual reports which make
use of the LAPS. These contributions deal primarily with the production
of monodisperse resuspensions of deposits of polydisperse aerosols for
inhalation toxicology research as described before. It appears that this
is now the main field of application of the LAPS.
No efforts have been published so far to employ spiral duct centrifuges
for air pollution field studies or atmospheric aerosol research. This is
obviously due to the relatively low aerosol sampling rates which, in most
laboratory studies, seldom exceed values of 1 liter min"1 because of the
associated loss of size resolution with increasing sampling rate. However,
field studies on particulate air pollutants do not require size resolu-
388
-------
Figure 3^ : Special Aerosol Inlet Section for Increased Aerosol Sampling
Rates with a Narrow-Duct Short-Spiral Centrifuge (The Inlet
Section is Mounted to the Rotor Lid)
tions to a few percent and a size resolution of some ^0 % would still "be
considered an improvement over cascade impactor data, where the deposits
on the various stages are of emulative nature with size variations of
more than 100 % . Thus, by accepting reduced size resolutions, thr sampl-
ing rate of spiral duct centrifuges could hf increased to values feasible
to suit air pollution measuring requirements for part i ?ulat,e mutter.
Such an attempt was made with a narrow-duct short -spi ral centrifuge (Sto-
ber et al.^O) although the long-spiral-duct centrifuge which permits to-
tal flow rates up to ?0 liters min~' would have oeen preferable because
of a better size resolution at hign sampling rates. By way of a special
aerosol inlet section of wide coaxial bore with a laminator insert as
shown in Figure 3^, the aerosol samp]ing rate of the short-spiral-duct
centrifuge could be increased to ^ liters mi';"1. A subsequent feasibility
study at operating conditions of ?000 rpm and a total flow rate of 10 li-
ters min"1 produced the upper calibration curve in Figure ^3 arid confirm-
ed that the size resolution did not deteriorate beyond Uo ;*> fc" ;izes be-
low U ym. For aerodynamic diameters of
-------
vimet.ri.cal mass distribution analysis required the use of a microbalance
and special thin plastic collection films which were weighed before being
placed on the campling foi.l strip. With a detection limit of 1.8 micro-
grains <:m~^ for t.he foil deposit at a <)0 % confidence level, particulate
air pollution mass d i str J but i ons can be analysed if the average airborne
concentration within the sampling period was hO micrograms m~3 or more
between 0.8 and 8.7 pm of aerodynamic diameter. Modest as these results
may be for air pollution surveillances, they certainly indicate a good
potential for air pollution source emirsion measurements.
FUTURE MODIFICATIONS AND DEVELOPMENTS
From the past experience with various designs of spiral duct centrifuges,
it appears that the 1ong-spiral-duct versions permit more favorable oper-
ating conditions, i.e., in particular, higher total flow rates and, thus,
higher sampling rates, than i, ie short-spiral-duct designs. While a flow
rate of some 19 liters min~-' is cri-- oT the standard operating conditions
at 3000 rpm for the original spiral duct centrifuge1*9, none of the small
rotor designs cat b" operate at, 'his flow rate. This came as a surprise
for the, narrow-duct sborr-r>j ira! < <-a'ri t\ige design wliose duct had a. more
favorable aspect ratio than the original instrument and, in proportion,
should permit a total flow rate of 15 liters min . The only explanation
for the failure is that, contrary to theoretical estimates^, the influ-
encp" of the iir.br lanced centrifugal forces due to the duct curvature is
and, thus, narrower curvatures on the
"'-;! of the secondary flow in the cross
the relatively wide curvatures of the
original design. Therefore, future modifications of the spiral duct may
have to go back to bigrc-r rotors and ducts with larger radii of curvature
and high aspect ratios.
In connection wi/; :i r.-'"]ern elemental mi croprobe analysis or similar me-
thods, it seems yLsc p^ssib]° to reduce the dimensions of the spiral
duct. Although a decrease of the depth of the spiral duct requires a
Tiroportional decrease of t\\r flow rate inorder to maintain the same flow
veljcitie: and depositjo;> characteristics in the duct, such modification
does not de^-, a.: - ' ,• .- "i. ;-osi t concentration but only the area of deposi-
tion and, thus, Ih<=- total, amount of the sample due to a narrower deposi-
tion foil. Fj.rthermor>., a reductio, of thw width of the duct will cause
an inversely proportional increase ;;f the flow velocities in the duct if
the flow rates are kept constant. However, this may be permissible since
the Reynolds number of the duct flow does not change. Hence, the deposi-
tion characteristics will not change significantly because the increase
in flow velocities is compensated by a reduction of the distance the
390
-------
0
20
40
60
80
100
120
140
160
180
200
220
240
260
280
300
f ' ' \ l>
iStOp, ,\
• i i i run
1 j v i *•
18h __ 6h
*** ~~^ p^^
^\
.
j
'- 'J
Af(Hz)
'', -0.,
run
18
!75h , 65h___|8^1]
_50jjgr
C =
t
XN '
Figure 35 : Recorded Frequency Shifts during Sampling and Rotor Stops of
h Quarts Oscillators Mounted along the Outer WaJ 1 of the Spi-
ral Duct Centrifuge and Used for Monitoring t.t1*1 Deposit, ion
of a Fluoroseein Test Aero:;o] of '^0 mi programs m~ Airborne1
(JonccntraL ion ;
From Top Lo Bottom : Reference Crystal without Deposit, fo ;
Crystals at Aerodynamic Sizes of 1.1 ym (f|), 0.8 ym (fgK
and 0.5 ym (f)
aerosol particles have to travel across the duct. Thus, if depos.it con-
centrations as presently achievable within reasonable sampling time arc-
acceptable, they could probably easily be ma'rJained in new designs with
reduced cross sectional dimensions of the spiral duct. In addition,
there is the chance that the smaller duct cross section may ^trmit high-
er rotor speeds and, thus, an increase of the deposit consent ,. vtions .
It has been shown in a feasibility study (Monig, Schwarzer and Stober34)
that the build-up of the foil deposit in the original spiral duct cen-
trifuge is sufficiently fast to permit the use of piezo-electric quartz
crystals as gravimetric sensors for measuring the deposit concentrations
391
-------
of moderate partieulate air pollution without going to impractically
Long sampling Limes. This sensing technique, first utilized "by Olin and
Sem3^ in an aerosol mass monitor instrument, does not suffer seriously
from the drawbacks described by Daley and Lundgren when applied in a
spiraJ centrifuge. The relative humidity identified as being the most
inl'luentin! disturb! rig factor can always be kept low or almost constant
at the sensor locations when the winnowing gas is dry or kept at a con-
stant humidity level. A rotor insert with four quartz crystals of a no-
minal resonance frequency of 10 megacycles was used in the feasibility
study. A telemetric circuit permitted to measure the frequency while the
rotor was spinning and sampling. With random frequency shifts of about
± 3 cycles due to variations of temperature and other surrounding physi-
cal conditions, the detection limit of the sensors was below a level of
20 nanograms cm~2 . Figure 35 shows a sequence of sampling runs and rotor
stops of different duration in the range of hours. The corresponding
frequency shift:; are given on an arbitrary ordinate scale for the refe-
rence crystal receiving no deposit (fo) and three sampling crystals at
locations of 1 . I yrn, 0.8 urn and 0.5 pm of aerodynamic size (f-| , f^ •>
fo), respectively. The test aerosol was generated from ammonium fluores-
cein and had an airborne concentration of 50 micrograms m~ , thus simul-
ating a moderate particulate air pollution level. The graph shows that
measurements during sampling as well as when the rotor was stopped were
consistent, although a frequency shift between the two situations was
involved. This shift was reproducible but individual for each crystal.
The graph clearly indicates that significant readings were obtained in
little more than an hour or two. This would make the spiral duct centri-
fuge an aerosol mass distribution monitot with a delay time of the order
of an hour.
1. Berner, A., and H. Keirh^lt, The Kotating-Slit Aerosol Spectrometer
(ROSL-Speetrometer): Prototype, in : Aerosol Research at the First
Physics Institute, Univ. of Vienna, Status Report, 1 -18, January
1968
2. Berner, A., and H. Reichelt, fiber EinlalS-Spaltsysteme in Konifugen,
Teil I : Das ROSL-System, Staub 29 : 92-95, 1969
3. Blachman, M.W., and M. Lippmann, Performance Characteristics of the
Multicyclone Aerosol Sampler, American Industr. Hyg. Assoc. Journ.
35 : 311 -326, 19TH
H. Boose, C., Ein Gerat zur fraktionierten Abscheidung von Staub fur
KorngroBenanalysen, Staub 22 : 109-112, 1962
392
-------
5- Daley, P.S., arid D.A. Lundgren, The Performance of Piezoelectric
Crystal Sensors Used to Determine Aerosol Mass Concentrations,
American Industr. Hygiene Conference, Paper No. 3, Miami Beach,
Fla. , May 12 - 17, 197^4
6. Davies, C.N., Particle-Fluid Interaction, Proc. of Harold Heywood
Memorial Symposium, Loughborough, England, September 17 - 18, 1973
7- Dean, W.R., The Stream-line Motion of Fluid in a Curved Pipe,
Phil. Mag., Ser. 7, 5 : 673 - 695, 1928
8. Ferron, G.A., and H.W.J. Bierhuizen, The Measurement of Polydisperse
Aerosols with the Spiral Centrifuge, J. Aerosol Sci. (in press)
9. Puchs, N.A., and A.G.Sutugin, Generation and Use of Monodisperse
Aerosols, in C.N. Davies (ed.): Aerosol Science, Academic Press,
London and New York, 1-30, 1966
10. Goet/, A., H.J.R. Stevenson and 0. Preining, The Design and Per-
formance of the Aerosol Spectrometer, J. Air Poll. Contr. Assoc.
10 : 378 - 383, 1960
11. Heard, M.J., A.C. Wells and R.D. Wiffen, A Re-determination of the
Diameters of Dow Polystyrene Latex Spheres, Atmosph. Environm. k :
1^9 - 156, 1970
12. Heyder, J., and J. Porstendorfer, Comparison of Optical and Centri-
fugal Aerosol Spectrometry: Liquid and Non-sphericai Particles,
J. Aerosol Sci. 5 : 387 - ^00, 1971*
13- Hochrainer, D., and P.M. Brown, Sizing of Aerosol Particles hy
Centrifugation, Environ. Sci. Technol. 3 : 830 - 835, 1969
1U. Inhalation Toxicology Research Institute Annual Reports, Lovelace
Foundation, National Technical Information Service, Springfield,
Va., 22151, LF-U6, 1972 - 1973 and LF-^9, 1973 - 197^
15- Junge, C., Die Rolle cler Aerosole und der gasformigen Beimengungen
der Luft im Spurenstoffhaushalt der Troposphare, Tel]us 5 : 1 -26,
1953
16. Kast, W. , Neuos Staubmessgcrat y.ur SohneJ Lbestimmurig dor Staubkorizen
tration und der Kornverteilung, Staub 21 : 215 - 223, 1961
17- Keith, C.H., and J.C. Derrick, Measurement- of the Particle Size
Distribution and Concentration of Cigarette Smoke by the "Conifuge",
J. Colloid Sci. 15 : 3^0 - 356, 1960
18. Kops, J., G. Dibbets, L. Hermans and J.F. van de Vate, The aerodyna-
mic Diameter of Branched Chain-like Aggregates, J. Aerosol Sci. 6:
XX - XX, 1975 (in press)
19. Kops, J., L. Hermans and J.F. van de Vate, Calibration of a Stober
Centrifugal Aerosol Spectrometer, J. Aerosol Sci. 5 : 379 - 386,
393
-------
20. Kotrappa, P., Shape Factors for Quartz Aerosol in Respirable Size
Range, J. Aerosol Sci. 2 : 353 - 359, 1971
21. Kolrappa, P., Shape Factors for Aerosols of Coal, UO and Th00 in
Respirable Size Range, in: T.T. Mercer, P.E. Morrow ,ind W. Stober
(ed's.), Assessment of Airborne Particles, C.C. Thomas, Springfield,
U L. , 331 - 356, 1972
22. Kotrappa, P., Reply to "Comments on 'Design and Performance of the
Lovelace Aerosol Particle Separator'", Rev. Sci. Instrum. hh : 1^38,
1973
23. Kotrappa, P., and M.E. Light, Design and Performance of the Lovelace
Aerosol Particle Separator, Rev. Sci. Instrum. h3 '• 1106 - 1112,
1972
,:•'<. Kotrappa, P., and M.K. Light, Centrifuge Separator, U.S. Patent No.
3, 698, 626; October 17, 1972
25. Kotrappa, P., and O.R. Moss, Production of Relatively Monodisperse
Aerosols for Inhalation Experiments by Aerosol Centrifugation,
Health Physics 21 : 531 - 535, 1971
26. Kotrappa, P., and C.J. Wilkinson, Measurement of the Specific Radio-
activity with Respect to Particle Size for Labeled Aerosols, J.
Aerosol Sci. 3 : 167 - 171, 1972
27- Kotrappa, P., and C.J. Wilkinson, Densities in Relation to Size of
Spherical Aerosols Produced by Nebulization and Heat Degradation,
American Industr. Hyg. Assoc. Journ. 33 : ^9 - ^+53, 1972
28. Kotrappa, P., C.J. Wilkinson and H.A. Boyd, Technology for the Pro-
duction of Monodisperse Aerosols of Oxides of Transuranic Elements
for Inhalation Experiments, Health Physics 22 : 837 - 8^3, 1972
29- Lassen, L., Ein einfacher Generator zur Erzeugung monodisperser
Aerosole im GroBenbereich 0.15 bis 0.70 p (Teilchenradius), Z. angew.
Physik 12 : 157 - 159, 1960
30. Ludwieg, H., Die ausgebildete Kanalstromung in einem rotierenden
System, Ingenieur-Archiv 19 : 296 - 308, 1951
31. Martonen, T.B., to be published at the University of Rochester
32. Martonen, T.B., and W. Stober, The Influence of the Gas Phase Density
of an Aerosol on the Aerodynamic Size Separation of the Aerosol
Particles in Spiral Duct Centrifuges, to be published
33. May, K.R., The Cascade Impactor: An Instrument for Sampling Coarse
Aerosols, J. Sci. Instr. 22 : 187 - 195, 19^5
3^. Monig, F.J., N. Schwarzer and W. Stober, Bestimmung der Aerosol-
Massenverteilung in einer Aerosol-Zentrifuge mit Hilfe von Schwing-
quarzen, in: V. Bohlau and H. Straubel (eds); Aerosole in Physik,
394
-------
Medizin und Technik, Jahreskongress 1973 der GeyoJ 1 schai't, fur
Aerosolforschung, Bad Soden/Ts., pp. 58 - 61, 1973
35- Moss, O.K., H.J, Ettinger and J.R. Coulter, Aerosol Density Measure-
ments Using a Modified Spiral Centrifuge Aerosol Spectrometer,
Environ. Sci. Technol. 6 : 6lU - 617, 1972
36. Newton, G.J., O.G. Raabe, R.L. Yarwood and G.M. Kanapilly, Genera-
tion of Monodisperse Aerosols of "7Qa-Labeled Aluminosilicate an
^9oAu-Labeled Gold Spheres, present at the Symposium on Fine Par-
ticles, Minneapolis, Minnesota, May 28 - 30, 1975
37- Oeseburg, F. , and R. Roos, Technical Note: Improvement of the Bear-
ing System of the Stober Aerosol Spectrometer, Atmosph. Environm.
9 : XX - XX, 1975
38. Oeseburg, F., F.M. Benschop and R. Roos, Particle Size Analysis of
Dioctyl Phthalate Aerosols Using the Stober Aerosol Spectrometer,
J. Aerosol Sci. 6 : XX - XX, 1975 (in press)
39- Olin, J.G., arid G.J. Sem, Piezoelectric Microbalance for Monitoring
the Mass Concentration of Suspended Particles, Atmosph. Environm.
5 : 653 - 668, 1971
UO. Porstendorfer , J., Die Bestimmung der GroBenverteilung von Aerosolen
mit Hilfe der radioaktiven Markierung und der Spiral centrifuge,
J. Aerosol Sci. U : 3^5 - 35^, 1973
'41. Raabe, O.G., Instruments and Methods for Characterizing Radioactive
Aerosols, IEEE Transact. Nucl. Sci. NS 19 : 6h - 75, 1972
h2. Raabe, O.G., H.A. Boyd, G.M. Kanapilly, C.J. Wilkinson, and G.J.
Newton, Development and Use of a System for Routine Production of
Monodisperse Particles of -^ PuOg and Evaluation of Gamma Emitting
Labels, Health Physics 28 : 655 - 667, 1975
1|3. Sawyer, K.F., and W.H. Walton, The "Conifuge" - A Size-separating
Sampling Device for Airborne ParticJes, J. Sci. Tnstr. 27 : 272 -
276, 1950
Ml. Spertell , R.B., and M. Lippmann, Airborne Density of Ferric Oxide
Aggregate Microspheres , American Industr. Hyg. Asrcc. Jo-urn. 3^ :
73]4 - 7^0, 1971
1*5- Stober, W., Design and Performance of a Size-Separating Aerosol
Centrifuge Facilitating Particle Size Spectrometry in the Submicrori
Range, in: Assessment of Airborne Radioactivity, Interns1'. Atom.
Energy Agency, Vienna, 393 - HoU, 1967
h6. Stober, W. , Dynamic Shape Factors of Nonspherical Aerosol Particles,
in: T.T. Mercer, P.E. Morrow and W. Stober (ed's.), Assessment of
Airborne Particles, C.C. Thomas, Springfield, 111., 2^9 - 288, 1972
395
-------
kf . Stober, W. , Comment on Design and Performance of the Lovelace
Aeroso] Particle Separator, Rev. Sci. Instrum. hk • 11^9 -1150
1973 ' '
H8. Stober, W. , and H. Flachsbart, Aerosol Size Spectrometry with a
Ring Slit Conifuge, Environ. Sci. Technol . 3 : 6U1 - 651, 1969
H9. Stober, W. , and H. Flachsbart, Size-Separating Precipitation in a
Spinning Spiral Duct, Environ. Sci. Technol. 3 : 1280 - 1296, 1969
50. Stober, W. , and H. Flachsbart, High Resolution Aerodynamic Size
Spectrometry of Quasi-Monodi sparse Latex Spheres with a Spiral
Centrifuge, J. Aerosol Sci. 2 : 103 - 116, 1971
51. Stober, W., and H. Flachsbart, An Evaluation of Nebulized Ammonium
Fluorescein as a Laboratory Aerosol, Atmosph. Environm. 7* 737 -
1973
52. Stober, W. , and J.G. Osborne , On the Limitation of Aerodynamic Size
Spectrometry of Dense Aerosols, to be published
53- Stober, W. , and U. Zessack, Zur Messung von Aerosol-TeilchengroBen-
spektren mit Hilfe von Zentrifugalabscheidern, Zentralbl . biol.
Aerosolforsch. 13 : 263-281, 1966
5H. Stober, W. , A. Berner and R. Blaschke, The Aerodynamic Diameter of
Aggregates of Uniform Spheres, J. Colloid Interf. Sci. 29 : 710 -
719, 1968
55- Stober, W. , H. Flachsbart and C. Boose, Distribution Analyses of the
Aerodynamic Size and the Mass of Aerosol Particles by Means of the
Spiral Centrifuge in Comparison to Other Aerosol Precipitators ,
J. Colloid Interf. Sci. 39 : 109 - 120, 1972
56. Stober, W. , H. Flachsbart and D. Hochrainer, Der aerodynamische
Durchmesser von Latexaggregaten und Asbestfasern, Staub 30 : 277 -
285, 1970
57- Stober, W. , N. Franzes and W. Steinhanses , Discussion: Improvement
of the Bearing System of the Stober Aerosol Spectrometer, Atmosph.
Environm., in press
58. Stober, W. , E. Hederer and H. Horvath, unpublished data, Rochester,
N.Y. and Vienna, Austria, 1971
59- Stober, W. , D. Hochrainer and H. Flachsbart, A Simplified Spiral
Centrifuge for Aerosol Mass Distribution Analysis, American Industr.
Hyg. Conference, Paper No. 96, Miami Beach, Fla., May 12 - 17, 197!;
60. Stober, W. , D. Hochrainer, H. Flachsbart, N. Franzes and ¥. Stein-
hanses, Fraktionierung und gravimetrische Bestimmung atmospharischer
Partikel mit einer kleinen Spiralkanal-Zentrifuge, to be published
396
-------
61. Tillery, M.I., A Concentric Aerosol Spectrometer , American Tridur.tr.
Hyp;. A;;soe. Journ. 3'; : 6P - r(h , ]<)'{h
(>',\ Timbre.il, V., The Terminal Velocity and Si /.a of Airborne Dust,
Particle:;, Brit. J. AppJ . Phyr, . [>, Suppl . 3, 86 - 90, 195^
63. van Buitenen, C.J.P., and F. Oeseburg, Comparison of "Light Scatter
ing Diameter" Based on Forward Scattering Measurements and Aero-
dynamic Diameter of Aerosol Particles, Atmosph. Environm. 8 :
885 - 896,
van de Vate, J.F., The Safety of SNR-300 and the Aerosol Model,
A Summary Report of the RCN Aerosol Research 196? - 1971, Reactor
Center Nederland Report Ho. 17)1, 23 - 2^, 1972
VDI-Richtlinie 3^91, Entwurf, Blatt 1: Messen von Partikeln, Pruf-
kriterien und Prufmethoden fiir Verfahren und Gerate zum Bestimmen
partikelformiger Beimengungen in Gasen, Begriffe und Definitionen;
Verein Deutscher Ingenieure, Diisseldorf, pp. 1 - 8, July 1975
Walkenhorst, W. , Untersuchungen an einem riach TeilchengroBen geord-
n' ^on Mischstaub im atembaren KorngroBenbereich, in: C.N. Davies
(ed.), Inhaled Particles and Vapours II, Pergamon Press, Oxford,
563 -571, 1967.
397
-------
PROBLEMS IN STACK SAMPLING AND MEASUREMENT
D. B. Harris and W. B. Kuykendal
Process Measurements Section
Control Systems Laboratory
Environmental Protection Agency
ABSTRACT
Though many techniques for monitoring particulates have been
studied in great detail under laboratory conditions, the application
to field situations of these techniques has often been difficult.
This paper describes the attempts to use cascade impactors for
manual fractional efficiency and optical and beta gauging for instru-
mental particle sizing. Though each technique seemed to be ideally
suited to particle measurement, the many different problems presented
by industrial gas streams and the solutions used to overcome them are
described. Particle bounce and reentrainment was the first problem
encountered with the impactors and was quickly followed by substrate
material selection due to reaction with gaseous components. Instru-
mental problems have centered on getting sensors to perform in the
hostile stack environment and on making reasonable compromises
necessary to apply these techniques to real stack environments.
399
Preceding page blank
-------
PROBLEMS IN STACK SAMPLING AND MEASUREMENT
D. B. Harris and W. B. Kuykendal
Process Measurements Section
Control Systems Laboratory
Environmental Protection Agency
As the concern about the effects of air pollution became focused
on meeting more stringent emission standards, it became apparent that
some information on the degree of control of particulate emissions
as a function of size was necessary. The Process Measurements Section
of the Control Systems Laboratory quickly became convinced that exter-
nal size measurements of collected masses of material did not yield
useful data and that some means of sizing the particles as they exist
in the ducting around the control device was needed. Thus, develop-
ment efforts were initiated to provide the needed techniques.
MANUAL METHODS
The first effort was to examine those devices already available
to determine if one could be used with little modification so that
useful information could be gathered without embarking on a long and
costly development program. Laboratory tests indicated that some
problems existed from particle bounce off the hard collection sur-
faces of the commercial impactor units identified as the best exist-
ing method. Problems were also noted during weighing caused by the
enormous ratio of the mass of the collector plates to the mass of
material caught. Experimental low tare weight substrates were made
and preliminarily tested in the lab with encouraging results.
Because of the need for a field method, we moved to an extensive
field test of the available commercial units and a few experimental
devices.
A three-week test, including 142 separate sizing device tests,
was conducted at the Marquette Station of the Upper Pennisula
Generating Company. An example of the results obtained is given in
Figure 1 and shows good agreement among the sizing devices tested.
A full description of this program is presented in Reference 2.
From this study, the first operational guides were set forth. To
ensure that inlet and outlet measurements covered the same time span,
low flow rates were suggested for the inlet measurements and high
flow rates for the outlet measurements. Substrates were found not
only to enhance the weighing, but also to reduce the bounce problem
if greased metallic foils or fiber glass was used. Thus, we felt
that we could recommend a method for determining fractional efficiency
of particulate control devices based on commercially available impac-
tors.
400
-------
10
a:
UJ
i-
LU
o:
UJ
a.
o
o
.01
.001
.0001
I I
I ' ' "I
II II I I I I-
ERC TAG
• BRINK
x ANDERSEN
o MARK III
I I I I I I I
I I I
III
.2 .512 5 10
PARTICLE DIAMETER , MICROMETERS
Figure 1 Impactor Pf-f >. rrance Comparison.
401
-------
Our testing proceeded following the new impactor method. Within
a few months, reports of net weight losses were received from a num-
ber of our testing contractors when using fiber glass substrates. A
number of quick tests were performed by several groups including EPA
and it was concluded that the "type A" glass was susceptible to hand-
ling losses if anything other than extreme care was used.3 The filter
manufacturers suggested using a new "type E" glass with better mechani-
cal properties. Laboratory tests indicated that it did improve the
weighing accuracy, so one manufacturer began to supply its substrates
in "type E" fiber glass.
A weight loss problem was also reported when using greased metallic
foils. Most of the people reporting the problem were using silicone
stopcock grease which had been used earlier with no abnormal weight
loss. Tests run by EPA confirmed the degradation of some grease sam-
ples as a function of increasing temperature in the range found in
process flow streams. We concluded that the problem was one of quality
control of a mass produced item. As a result of our tests with a num-
ber of greases, we recommended that gas chromatographic support greases
be used because of the high level of quality control applied to the
product. The temperature stability could be quickly ascertained from
the specification sheet.
The new grade of fiber glass performed adequately until our tests
were conducted on high temperature or high SOX gas streams. Excessive
weight gains were noticed by many investigators working in these gas
flows. It became apparent that the problem was serious enough to
cast great doubt on the total results of a number of tests. The
search for an answer to this gas reaction problem has been one of the
longest and most intensive of the particle sizing program. A detailed
description of the investigation will soon be available. Briefly, a
number of filter materials were tested within these problem environ-
ments by using the impactor with a prefilter to remove the particu-
late. Any weight gained by a substrate should then be only by gas/
solid reaction. The results are summarized in Tables 1 and 2. The
tables show that Teflon is unreactive and a probable choice for
temperature regimes under 260° C but we have not investigated this
material for reentrainment caused by the slippery surface. Some
fiber glasses were found to be significantly less reactive than those
we had been using. We believe that the pH of the glass is the driving
force behind the reactivity or lack thereof. Figure 2 shows the
effect of stack temperature on the reactivity of these materials.
As a result of this investigation we believe that a substrate material
can be found in the group which shows low level reactivity; however,
further testing is needed to find the right one. Thus, an approach
to the development of an acceptable method has been the testing of
a successful laboratory technique under various field applications.
Othec particulate measurement techniques are also undergoing a similar
development process.
402
-------
Table 1. ANOMALOUS FILTER WEIGHT GAINS4
Cement Plant
mm-Hg
°C
"C
Std. liters
°C
Mtn.
mm
ml
Actual
liter/min.
mm-Hg
mm-Mg
CC-1 CC-2
Date 2-12-75 2-18-75
Arab. Pres. 754.4 749.3
Arab. Temp. 16.6 20.0
Stack Temp. 266 249
Gas Vol. 348 674
Avg. Gas 14.4 23.3
Meter Temp.
Run Time 60 i20
Orl. ID 1.42 1.42
Cond. H20 74 12J
% H20 (29.6) 22% used (26.6) 22.0
for flowrate
Flow Rate:
Orl. 11.72, 10.84,
Gas Meter 12.40 ' 11.04
Avg. Probe 60.96 76.2
AP
Orl. AP 20.7 19.4
Filter No.
la GA* 3.22 mg GA 10.58
Ib - - Silicone GA 0.44
o-rings
stuck
2 SA* 0.28 to SA 0.92 =118ht;
,., brown
filters
ring
3 GA -0.48 RA* 1.16
900 AF
4 MSA 1.02 SU8ht
1106 BH °Town
ring
5 _
CC-3 CC-4
2-19-75 2-21-75
754.4 759.5
8.88 14.4
249 262
680 936
11.6 26.1
120 240
1.42 1.42
131 97
(27.2) 22.0 (28.0) 22.0
11.07, 4.16, ,
11.61 4.34J
76.2 381
19.6 4.9
GA 9.48 GA 4.38
GA 0.84 MSA 2.14
1106 BH
MSA 1.42 Teflon -0.02
1106 BH
RA 1.70 GA 0.90
900 AF
SA 1.66 Stuck RA 2.40
'° , 900 AF
metal
support
MSA 1.92
1106 BH
Teflon o-rings
Tef'.u 0.00
Teflon o-rings 'efi.on o-rlngs
*GA - Gelman type A
RA - Reeves Angel
SA - Gelman Spectrograde
403
-------
mm- HB
"C
"C
'C
Mln
mm
ml
Actual
lltera/min.
mm-Hg
mm-Hg
Date
Arab. Prea.
Amb Temp
Stack Temp.
Avg. Gas
Meter Temp.
Run Time
Ori ID
Cond. H20
Z "20
Flow Rate
Ori.
Gas Meter
Avg. Probe
AP
(+AP across
orifice)
Ori. AP
la
Ib
BRSP-1
2-25-75
738 6
12.0
135
27.2
60
3148- 059
6.0
(3.8) 7.5 used
in flow-
rate cal-
L;; 5 *
223
12 1
GA 12 29 tng
MSA 0 43
1106 BH
BRSP-2
2-25-75
738 6
16.0
129
2i 7
60
3348- 059
U 4
(4 9)
7 05
7.3
321
27.8
GA 26 50
GA 0.44
Coal-Fired Boiler
BRSF-3
2-26-75
747.7
11.6
135
27 2
240
3348- 059
57.5
7.5 (5 6) 7
7 10 8'09 «
'•19 8.07 8
320
13.6
GA
RA 0.77
900 AF
BRSP-4
2-26-75
747.2
20.5
135
27.7
120
3348-. 059
23.8
5 (5.7)
6.54
6.76
322
23.9
GA 37.99
M3A 0.88
1106 BH
BRSP-5
2-27-75
745.7
17.7
149
26.0
480
3348- 059
90.8
7.5 (5 7)
354
23.7
GA
GA 3.27
Moderat
BRSP-6
2-28-75
739.1
5.0
179
23.3
30
3348-. 059
4.8
7 5 (4.6)
317.5
24 0
GA 23 41
USA 0.69
ely 1106 BH
BRSP 7
2-28-75
739.1
10.0
179
-
0
7 5
7.58 "
-
GA 0.34
GA 0.38
brown on
edge
MSA 0 16 RA 0.83 MSA 3.65 SA -0 12 MSA 0.77
900 AF 1106 BH Moderately 1106 BH
RA 0 37
900 AF
Stuck
MSA -0 14 to Teflon 0 81 CA 0.24
1106 BH o-rlng
severely
Teflon 0.01 Teflon 0 00 SA* 0.12 SA 0.18 GA 0 53 SA 0.6? GA -0.10 Torn
MSA 0 27
1106 BH
GA -0.04 GA 0.41 GA 0 53 SA 0.66 GA 0.38
Severely No
' GA - C,elman type ft
RA - Reeves Angel
SA - Gelman Spectrograde
Stuck SA 0.31 Teflon 0.07 Teflon 0.00 Teflon-0 01 MSA 0 06
co 1106 BH
o-rlng
6 cut
by it
All filters stuck This (and all
slightly to support, that follow)
As much as possible, are Sp*ctro
recoverec Grade A from
Batch 8192.
MSA ll.OQ MSA 0.71 Teflon 0.0
1106 BH Slight 1106 BH ^ ^
rown
8
placed In stack and
taken out Inmedtatfel
404
-------
4-J
f.'
•H
T3
«
cn
en
c
o
•H
4-J
CJ
<£>
tf)
*• ro cvj
NIV9 1H9I3M
en
01
C
d/
cn
M-l
C
cn
CU
in
!-) Cfi
O QJ
C M
60 a)
•H Cu
3 tt,
o ns
r-H 00
C
•§ I
CNJ
cu
d
to
405
-------
INSTRUMENTAL TECHNIQUES
Although manual techniques have proven their usefulness and have
generated the preponderance of particulate data, most will agree that
they are cumbersome to implement. A more desirable approach would
utilize an instrumental technique to overcome the drawbacks of the
manual methods.
Before proceeding further it will be useful to define an ideal
particle sizing instrument. To do this the purpose and application of
the instrument must first be defined. For the purposes of this dis-
cussion the intended use of the data will be in the development and
evaluation of control devices for stationary sources; more specifically
for coal-fired utility boilers.
The ideal instrument must first measure the particle size in-situ
without altering the flow field or size distribution. Although the
instrument should be capable of sizing the particles in a nondisruptive
fashion, it would also be desirable to collect a sufficient size frac-
tionated sample for chemical and toxicological analysis. The data out-
put from this instrument should be in real time so that control devices
could be optimized on-line. In addition, the instrument should have
two or more sensors so that both inlet and outlet measurements could be
made simultaneously. The requirement for sampling before and after
control devices is also an implicit requirement for a large dynamic
range in particle mass and number concentration. Values for this con-
centration would typically be from 0.01 to 25 g/m^ or 10 to 10^® par-
ticles/cc.
The question of the size range of the ideal instrument could well
be the subject of a separate discussion in itself. First, the type of
particle size to be measured must be determined. The linear dimension
as measured by a microscope could be useful in determining filtration
characteristics of particles. The optical equivalent diameter, measured
by single particle light scattering instruments, is useful in determining
the effect of particles on visibility. The aerodynamic diameter, mea-
sured in a variety of ways, is an indication of how particles behave in
gas flow fields and appears to be the size parameter of most importance
in air pollution research.6 other size parameters could also be
selected and doubtless have specific applications, but these will not
be considered here. Aerosols have been defined by Fuchs as gasborne
solids or droplets whose size range is approximately 0.001 to 1000
micrometers.7 Although this may serve as a classical definition, a more
practical range would be 0.01 to 10 micrometers.
406
-------
In addition to the above ideal characteristics, the instrument
must be capable of performing the particle size measurements in the
hostile stack environment. Temperatures can range as high as 450° C
before a hot electrostatic precipitator, but a more typical value
for industrial gas streams is 200° C. The components of the instru-
ment should be portable and should require minimum support facilities.
Clearly, no existing or foreseeable instrument can meet all of
these criteria. However, several instruments are in various stages of
development which meet some of these requirements and at the same time
illustrate some of the problems inherent in the development of particle
sizing instruments.
One of the most ambitious efforts to date has been the development
of an in-stack cascade impactor with real time mass detection for each
individual stage via beta attenuation. The instrument, though in-stack,
is still not truly in-situ. Because it uses an inertial impactor for
aerodynamic size discrimination in the range of 0.2 to 5.0 micrometers,
it shares some of the problems discussed above for manual impactors. A
major problem in the development of the instrument has been the diffi-
culty in obtaining beta detectors that will operate at 200° C. By the
very nature of the detection principle employed, the instrument is com-
plex and expensive.8
Another instrument is under development which uses the beta
attenuation principle to detect the mass of particulate for each stage
of an inertial impactor. In this instrument a virtual impaction scheme
is used outside of the stack and the particulate sample is transported
to the instrument in a probe. The aerodynamic size range is 0.07 to
10 micrometers. Although this approach sacrifices in-situ sizing, the
operational problems of the sizing/detection system are significantly
reduced by removing the instrument from the stack environment. Never-
theless, this instrument is still quite complex and expensive.9
Work is also underway to develop an in-stack cascade impactor with
real time readout based on a detection system sensitive to the amount
of particulate removed by each stage. A virtual impactor is used for
the size discrimination over the range of 0.2 to 5.0 micrometers and a
simplified sensor is used to measure the size fraction for each stage.
Three candidate sensors are being evaluated: (1) light attenuation,
(2) electric charge transfer, and (3) pressure drop across a filter.
As with the in-stack beta sensing instrument, this instrument is still
not truly in-situ. Because the sensor does not actually measure the
mass of particulate collected in each size interval the instrument does
not truly measure aerodynamic particle size. However, the detection
principle is simpler and the cost of the instrument would be less than
that of either of the beta sensing instruments.10
407
-------
An instrument has also been developed which uses a laser light
scattering technique to size particulate in-stack. Because the instru-
ment optically defines the sensing volume it can be considered in-situ.
Since it is an optical instrument it senses the optical equivalent dia-
meter instead of the preferred aerodynamic diameter. This instrument,
like all present optical sizing instruments, is a single particle
counter and its maximum number concentration is limited by its sensing
volume to 10 particles per cm^. The size range for the instrument is
0.2 to 3.0 micrometers. Since a pulsed laser system is used and only
one particle is sized per pulse, a finite time period (approximately
five minutes) is required to measure the size distribution and the
instrument is not actually real time. Unlike the impactor based sys-
tems discussed above, it is not possible to collect a particulate
sample for subsequent analysis. The instrument is relatively simple
and cost is less than for the other techniques.H
Having now defined an ideal instrument and discussed several
instruments currently under development it would appear to be desirable
to define a practical instrument in light of the problems discussed
above. Unfortunately, this cannot be done au priori. Many factors
must be taken into consideration. In previously defining an ideal
instrument, cost was not considered whereas in any real system it will
have a significant impact. The trade-off of whether or not to place
the sensor in the stack must be made. In-situ measurements are always
desirable, but if the stack environment becomes too hostile, a remote
location would be the choice. The major decision point in the design
process is the selection of the sizing/sensing mechanism. Each user
must carefully trade off the available techniques and his requirements
against the problems to be encountered in applying these techniques.
In short, the selection of any particulate sizing technique, be
it manual or instrumental, must consider the practical limitations of
that technique.
408
-------
REFERENCES
1. McCain, J. D., K. M. Gushing, and A. N. Bird, Jr.. Field
Measurements of Particle Size Distribution with Inertial
Sizing Devices. U.S. Environmental Protection Agency,
Washington, D. C. EPA-650/2-73-035, October 1973, p. 45.
2. Ibid.
3. Smith, W. B., K. M. Gushing, and G. E. Lacey. Andersen Filter
Substrate Weight Loss. U.S. Environmental Protection Agency,
Washington, D. C. EPA-650/2-75-022, February 1975.
4. Smith, W. B. , K. M. Gushing, and G. E. Lacey. Particulate
Sizing Techniques for Control Device Evaluation. Southern
Research Institute, Birmingham, Alabama. EPA Contract
68-02-0273, Report No. 29, March 15, 1975, pp. 4-5.
5. Smith, W. B., K. M. Gushing, and G. E. Lacey. Particulate
Sizing Techniques for Control Device Evaluation. Southern
Research Institute, Birmingham, Alabama. EPA Contract 68-02-
0273, Report No. 30, April 15, 1975, p. 14.
6. Lilienfield, P. and D. Cooper. Literature Review and Selection
of Design Principle for the Development of a Fine Particulate
Source Testing Instrument. GCA Technology Division, Bedford,
Mass. Publication Number GCA-TR-73-22-G. October 1973. p. 30-
31.
7. Fuchs, N. A. The Mechanics of Aerosols. New York, The >farMillan
Co., 1964.
8. Lilienfield, P. Personal communication. May 1975.
9. Wagman, J. and C. M. Peterson. In: Proceedings of the 3rd
International Clean Air Congress. Dusseldorf, Ger., October 8-
12, 1973. p. C6-8.
10. Burckle, J. Personal communication. May 1975.
11. Shofner, F. M., G. Kreikebaum, H. W. Schmitt, and B. E. Barnhart.
Environmental Systems Corp. In-Situ, Continuous Measurement of
Particulate Size Distribution and Mass Concentration Using Electro-
Optical Instrumentation. (Presented at the Fifth Annual Industrial
Air Pollution Control Conference. Knoxville, Tenn. April 3-4,
1975).
409
-------
INERTIAL IMPACTORS: THEORY, DESIGN AND USE
by
Virgil A. Marple and Klaus Willeke
Particle Technology Laboratory
Mechanical Engineering Department
University of Minnesota
Minneapolis, Minnesota 55455
ABSTRACT
Inertial impactors are devices used to classify particles with re-
spect to their aerodynamic size. The theoretical analyzing techniques
currently available are sufficiently accurate to predict the impaction
characteristics of impactors, provided the impactor design conforms rea-
sonably well with the boundary condition used in the theoretical analysis.
Using results of a theoretical study employing these techniques, design
criteria are given for such parameters as jet-to-plate distance, jet
Reynolds number, and jet throat length to obtain sharp cut-off character-
istics for both round and rectangular impactors. Precautions which must
be taken in the use of impactors are discussed, and limitations of im-
pactors due to interstage losses and particle bounce from various surfaces
are presented. Various designs of impactor jets and impaction surfaces
are examined with special attention given to the use of multiple jet im-
pactors for the purpose of controlling the jet Reynolds number. Also,
the principle operating characteristics of commercial cascade impactors
are presented.
411
Preceding page blank
-------
INTRODUCTION
Inertial impactors have been used extensively for many years to col-
lect airborne particles for size and frequently also for composition anal-
ysis. Initially, they came into widespread use because of their simpli-
city of construction and operation. As modern instruments, they are be-
coming increasingly popular because recent studies-*->^ have shown that
they will classify aerosol particles into very distinct size ranges, If
they are properly designed and operated.
All impactors operate under the principle that if a stream of
particle-laden air is directed at a surface, particles of sufficient iner-
tia will impact upon the surface and smaller particles will follow the air
streamlines and not be collected, as shown in Figure l(a). Thus, an im-
pactor consists simply of a nozzle, either round or rectangular in shape,
and an impaction plate.
The configuration shown in Figure l(a) is a single stage impactor
which separates the particles into two size groups. The large particles
are collected on the plate and the small ones remain airborne. One may
define a cut-off size of a stage to be the particle size for which 50% of
the particles are removed from the air stream and collected on the plate.
By operating several impactor stages at different flow conditions,
the aerosol particles are classified into several size ranges from which
the size distribution is determined. These single stages can be operated
in a parallel or in a series (cascade) arrangement. In the parallel ar-
rangement, each of the stages classifies the airborne particles at dif-
ferent cut-off sizes, so that the difference in the amount of the deposit
of any two stages gives the quantity of particles in the particular size
interval defined by the respective cut-off sizes of the two stages. In
the series arrangement, also known as the cascade impactor, the aerosol
stream is passed from stage to stage with continually increasing veloci-
ties and decreasing particle cut-off sizes. Thus, each stage acts as a
differential size classifier. Of the two flow systems, the cascade ar-
rangement is by far the most popular, as is evident by the many commercial
cascade impactors currently available.
412
-------
NOZZLE
STREAMLINES
• JET EXIT
T-
IMPACTION
PLATE
-TRAJECTORY OF ^TRAJECTORY OF
PARTICLE TOO
SMALL TO
IMPACT
IMPACTED PARTICLE
(a) IMPACTOR STAGE
EFFICIENCY, E
3 C
1
IDEAL
i
/"STK"
(b) EFFICIENCY CURVE
Figure 1 Streamlines, particle trajectories and efficiency
curve for a typical impactor
413
-------
In the conventional impactor, the jet is formed in a nozzle (internal
flow) and then impacts onto a plate. It is also possible to pass the
impaction plate through the particle laden air (external flow). The ef-
fectiveness of particle collection in the latter arrangement is compar-
able to that of conventional impactors-^. In operation, these impactors
normally consist of impaction plates (or cylinders) mounted at the ends
of rotating arms. >5,6 ^s ^\ie arms are rotated through the air, par-
ticles are impacted onto the impaction surface. The size of the particles
collected depends upon the speed and width (or diameter) of the impaction
surface as well as the size and density of the particles. These devices
may be used to collect particles larger than 10 to 20 pm in diameter.
Thus, for the collection of large particles, which may be difficult to
sample efficiently in a conventional impactor, this type of impactor is
a suitable alternative.
The single most important characteristic of an impactor stage is the
collection efficiency curve, which gives the fraction of particles of a
given size collected from the incident stream as a function of particle
size. Ideally, an impactor should collect all particles larger than a
certain size and none of the smaller ones, which would correspond to the
sharp ideal cut-off illustrated in Figure l(b). The efficiency curve of
a typical real impactor stage, also shown in Figure l(b), spans over a
range of particle sizes, but still has a good "sharpness of cut".
It has been shown^ that ideal impactor behavior can be obtained if
two requirements are met: 1) in the region between the jet exit plane
and the impaction plate, the y component of fluid velocity is a function
of y only (y is parallel to the centerline of the nozzle); 2) the y com-
ponent of the velocity of the particles at the jet exit plane is uniform
across the jet. These two requirements are reduced to only the first
requirement, if the velocity of the particles is equal to the fluid veloc-
ity at the exit plane of the jet. Such requirements are nearly met> in a
real impactor as can be seen by the velocity profiles in Figure 2(a) and
(b) for rectangular and round impactors, respectively. In both cases,
Vy(or Vz) is quite independent of X(or r) over a major portion of the
jet. However, this criteria is not met in the boundary layer near the
wall. Particles passing through this region are the cause for the non-
ideal cut-off characteristics at the larger efficiencies of the real im-
pactor shown in Figure l(b).
The determination of the impaction efficiency curves for real impac-
tors has been the object of many studies reported in the literature.
However, only recently, with the aid of modern computers, has the aero-
sol flow in impactors become thoroughly understood. This paper reviews
these recent developments, describes how these developments can be used
to design impactors with predictable operating characteristics, presents
practical limitations of impactor use, and examines the operating
414
-------
characteristics of several commercial impactors which are currently avail-
able.
Exit Plane of Jet
(a) Rectangular Impactor (S/W= I,T/W= I, Re = 3000)
Exit Plane of Jef
///S/////S/////////,
(b IRound Impactor (S/W= I/2,T/W=I, Re=3000)
Figure 2 Flow fields in rectangular and round impactors
415
-------
THEORETICAL ANALYSIS
In performing a theoretical analysis of inertial impactors it is
necessary to first calculate the flow field within the impactor and then
determine the particle behavior in this flow field. Thus, an accurate
theoretical analysis of particle impaction can only be made if the flow
field is precisely known.
A review of the theoretical studies prior to 1970 shows that most
of the studies ' ' assumed simple approximations to the flow field.
One study!! macje a rigorous analysis based on the method of conformal
mapping and on the solution of Euler's equation for the flow of a fric-
tionless fluid through an impactor. However, to fully describe the flow
field, the analysis must include the viscous effects of the fluid.
1 7
An extensive theoretical study has been made of the flow fields
and of the impaction characteristics of rectangular and round impactors.
First the flow fields were determined by using numerical methods to
solve the Navier-Stokes Equations for viscous flow-^ following closely
the technique described by Gossman, et al. Then the particle trajec-
tories were traced through these flow fields by numerically integrating
the particles equation of motion from which the impaction character-
istics were subsequently found. In the study, the impactors were char-
acterized by the specification of three dimensionless parameters: S/W,
T/W and Re, where S = jet-to-plate distance, W = jet width or diameter,
T = jet throat length, and Re = jet Reynolds Number based on the hydrau-
lic diameter of the jet, D, (D^ = W for round jets, D^ = 2W for rectan-
gular jets). The resulting characteristic impaction curves are present-
ed in Figure 3 showing the effects of the parameters S/W, Re, and T/W.
Note that the abscissa of these curves, expressed in units of the square
root of the Stokes number, /Stk, is a dimensionless particle size. The
Stokes number is defined as the ratio of the particle stopping distance
to the halfwidth or the radius of the impactor throat
Stk = ,opCV0
416
-------
Roun
0'-
01 02 03 04 OS 06 07 08 09
/STK
(a) EFFECT OF JET TO PLATE DISTANCE (Re = 3,000)
100
80
Round
01
Re(S/W
10
100
500
3.0OO
25,000
02 03 04 05 06 07 08 09
(b) EFFECT OF JET REYNOLDS NUMBER (T/W = I)
100
^40 -
1 20
TVW(S/W=l/2)
0
i a 1/4
2
Round
_. 1 1 . 1.. L
If I ' '
III 11 T/W(S/W'li
/ '' ' '
1 // '°
1 ' '
' i'
i i
i i
i i;_.lj. i i___
0 01 02 03 0.4 05 06 0.7 08 09
_J ....1
10 II
(c) .EFFECT OF THROAT LENGTH (Re = 3,000)
Figure 3 Impactor efficiency curves for rectangular and round im-
pactors showing the effect of jet-to-plate distiince,
Reynolds number and throat length.
417
-------
where pp is the particle density, C is the Cunningham slip correction
factor, V0 is the mean velocity at the throat, Dp is the particle dia-
meter, and M is the fluid viscosity.
Comparisons of the theoretical efficiency curves of Figure 3 with
those of experimental investigationsl>16,17,18,19 have shown that the
agreement is very good for both round and rectangular impactors if the im-
pactor inlet conditions,shape, and Reynolds number are similar. It has
been shown, that as experimental techniques have been improved, the ex-
perimental efficiency curves have approached the predicted theoretical
curves.
Several interesting conclusions can be found by examining the curves
of Figure 3. First, all of the efficiency curves are similar in shape
except for the cases of low Reynolds numbers (Re < 500) or extremely high
Reynolds numbers (Re = 25,000). For low Reynolds numbers the poor cut-
off characteristics are caused by the thick viscous boundary layer in the
jet of the impactor. For high Reynolds numbers, the knee in the efficien-
cy curve at the low values of efficiency appears to be caused by a very
thin boundary layer over portions of the impaction plate adjacent to the
stagnation point-*- . This thin boundary layer, having a thickness about
equal to the particle diameter, allows smaller particles to impact than
in areas where the boundary layer is thicker.
Second, the efficiency curve's position on the /Stk axis is relative-
ly independent of these parameters, except for small numbers of S/W or
Reynolds numbers. This can be seen more clearly in Figures 4 and 5,where
the value of /StTkJg is cross-plotted as a function of S/W and Re respect-
ively. In Figure 4, it is important to note that /Stk drops sharply with
decreasing jet-to-plate distance for S/W < 1.0 for rectangular impactors
and S/W < 0.5 for round impactors.
Third, the round impactor will collect smaller particles than the rec-
tangular impactor for similar values of S/W and Re (i.e., the efficiency
curves for round impactors are at smaller values of /Stk than for rec-
tangular impactors). However, it nas been shownl? that if a modified
Stokes Number, Stk1, is defined as the ratio of the particle stopping dis-
tance to half of the hydraulic diameter, D^,
Stk' - PP
Dh/2
[2]
the efficiency curves for both round (D, = W) and rectangular (Di = 2W)
impactors nearly coincide on the /Stk1 axis for the same values of Re,
S/Dh and T/Dh .
418
-------
0.8
0.7
o 0.6
to
: 0.5
" 0.4
0.3
0.2
O.I
n
1 1 1 1
Rectangular (T/W=I)
/^
~ 1
^ 1 Round (T/W=2) _
V
/
— —
Re =3000
till
0
2 3
S/W
Figure 4 Impactor 50% cut-off size as a function of the jet-to-plate
distance.
419
-------
X.
t—
^
0.65
0.60
0.55
050 -
0.45-
RECTANGULAR (S/W =
' -
ROUND (S/W=l/3)
0.85
-0.80
-0.75
-0.70
-065
0401 i i i I i i ill | 1 | I i i i il 1 i—i I I I 111 1 IOSQ
10 10* 10s 10*
REYNOLDS NUMBER
Figure 5 Impactor 50% cut-off size as a function of the Reynolds
number.
DESIGN ARRANGEMENTS
Nearly all impactors can be classified into either one of two basic
types: round or rectangular. The main variations within these basic
types are given by the number of nozzles, by the arrangement of the noz-
zles, and the shape of the impaction plate.
Nozzle Types and Design Charts
Figure 6 shows the basic nozzle arrangements used in impactor designs.
Nearly all impactors can be classified as having a single round nozzle, a
single rectangular nozzle, multiple round nozzles, or multiple rectangular
nozzles.
42Q
-------
SINGLE NOZZLES
o
ROUND
RECTANGULAR
(INFINITE ASPECT RATIO)
RECTANGULAR
(FINITE ASPECT RATIO)
MULTIPLE NOZZLES
O O
o o
o o
o o
ROUND
RECTANGULAR
Figure 6 Basic impactor nozzle designs
1 *?
Theory can be used to predict the impaction characteristics for
impactors with round nozzles and rectangular nozzles of infinite aspect
ratio (Figure 6). This theory can also be applied to the flow in the
central portion of an impactor with a rectangular nozzle of finite aspect
ratio. However, because the flow at the ends of a rectangular nozzle is
radial, *.s in the round impactor, smaller particles are impacted out in
this region. In order to minimize this end affect, the aspect ratio
should be as large as possible and the impactor should be designed so
that the air at the ends of the jet is restrained from flowing radially
outward. The impaction efficiency curve of a rectangular impactor with
small aspect ratio matches that of a round impactor at efficiencies near
zero, and that of a rectangular impactor with an infinite aspect ratio at
higher efficiencies.
421
-------
Tlie theoretical predictions may also be applied to multiple nozzle ar-
rangements, Lf the aerosol flow through each nozzle of a stage is identi-
cal. Multiple nozzle impactors of both round and rectangular design are
commercially available. One reason for the use of multiple nozzles is
the compactness of the design. This is especially true of the rectangular
impactor where a long nozzle is divided into a number of small segments.
In round impactors, the use of a large number of small diame.ter nozzles
will allow a smaller jet-to-plate distance than for a large single nozzle.
Also, for a given flow rate through a multiple round nozzle impactor,
variation in the number of nozzles allows the setting of th'3 Reynolds
number to a desired value (Figure 3). The relationship between the num-
ber of round jets, n, and Reynolds number, Re, is found by expressing the
average velocity within the round jets, Vo ^as
= .
TinW2 [3]
where Q = total volumetric flow rate through the stage. This velocity
expression is now substituted into the expressions for Reynolds number
and Stokes number:
Stk50 = 4ppQCD502
where 059 and Stk^Q are, respectively, particle diameter and Stokes num-
ber at 50% efficiency, and p is the fluid density. By elimination of
W from Equations 4 and 5, we obtain the expression
I1/2 fRel3/2
By assuming unit density particles (pp = 1 gm/cm^) and air flow at nor-
mal temperatures and pressures (p = 1.205 x 10"-* gm/cm^, y = 1.81 x 10"^
poise)and noting that Stk5Q is a function of Re and a specific design,
equation 6 can be expressed graphically as shown in Figure 7. This fig-
ure is for S/W = 1, T/W = 1, and Re = 500, 3000 and 10,000. With these
values of S and T, small variations in S and T will have a negligible ef
fect on the impaction efficiency (Figures 3 and 4) .
Also note in Figure 7 that the parameter t/c" DrQ corresponds to a
specific value of W. This relationship is found by eliminating Q/n
from equations 4 and 5.
422
-------
a>
en
o
CO
a>
Q.
in
-------
[7]
9pStk50
As an example of the use of these curves, assume it is desired to
have a cutoff size of /C D5Q = 2 pm, a total flow rate of 40 1pm, and
Re = 500. From Figure 7 it is found that this stage must have 120 holes
of 0.0894 cm diameter. The number of holes can be reduced to 8 when each
bole is 0.226 cm in diameter which results in Re = 3,000. The number may
be further reduced to approximately one hole with a diameter of 0.426 cm
and Re = 10,000.
A similar analysis can be made of rectangular impactors. However,
instead of defining the number of jets, it is more conventient to define
the total length, L, of the jets, irrespective of how this total length
is divided into individual jets. Now the mean fluid velocity at the
throat, V0, is defined as
and the Reynolds number (defined in terms of the hydraulic diameter) and
Stokes number become
Re
= 2pVQW = 2 p Q [9]
M ML
Q C D3
9 u L W2 [10]
Since W does not appear in the expression for Re, we cannot eliminate
W from equations [9] and [10] as we did for the round impactor. However,
if the volumetric flow rate in a rectangular impactor of width W is ex-
pressed per unit length of the jet, the parameter Q/L versus W can be
plotted from equation [10] by allowing /C 050 to be a parameter as shown
in Figure 8. The value of Stk^Q in this equation is a function of Re
(Figure 3) and, thus, a function of Q/L as defined by equation [9]. The
corresponding value of Re is also shown in Figure 8. In this case S/W =
1.5 and T/W = 1.
424
-------
Rectangular Impxictor Design Chart
I- 1000
I0
>
£
CJ
II
0)
tr
-------
Impaction Surfaces
Besides variations in jet design, there can also be variations in the
design of the impaction plate, as shown in Figure 9. Although the most
common design is a flat surface, the variations shown in Figure 9 have
been made for good reasons and serve to expand the versatility of im-
pactors.
I I
I I
I I
I I
I I
I I
Flat
(round or
rectangular)
I I
I I
I I
Rotating Cylinder
(rectangular)
I I
Right Angle
(rectangular)
Cone
(round)
Figure 9 Types of impaction plates
Impaction from a rectangular jet onto a rotating drum is employed in
the Lundgren impactor^O for time-resolved studies of the deposit. This
is accomplished by rotating the cylinder at rates from one revolution
per 24 hours to one revolution per minute.
1 0
The cone was used in a round impactor by Schott . The purpose was
to develop an impactor which would allow particles which were blown or
426
-------
bounced off the cone, to be captured in a volume at the base of the cone,
thus having the ability to collect large deposits.
One might expect that impaction onto a surface which is not perpen-
dicular to the unimpeded jet flow, could be quite detrimental to the
sharpness of cut of the efficiency curve. However, the experiments >^
have shown that the efficiency curves are very similar to those for a
flat plate. In some of these experiments^-" the theoretical efficiency
curves were also found for impaction onto a cone by employing the same
technique used to obtain the efficiency curves for impaction onto a flat
plate (Figure 3). The theoretical calculations agreed very well with
the experimental data.
The final variation in impaction surfaces considered here is the
"right-angle" arrangement of the rectangular impactor shown in Figure 9.
This type of impaction surface must be used if the total flow is to be
directed to one side or the other. For analysis purposes, it can be
thought of as half of a rectangular impactor with a solid surface at the
center plane. Since the boundary layer along the center plane hinders
the development of a flat velocity profile at the exit plaiie of the jet,
the efficiency curve will not have as sharp a cut-off as the conventional
rectangular impactor.
OPERATIONAL PRECAUTIONS
As has been shown, an impactor has the ability to sharply classiiy
particles into distinct ranges of aerodynamic size. However, to obtain
reliable data from an impactor, one must be aware of their limitations.
For example, precautions must be taken in the use of the design charts
and of the calculated efficiency curves ior conditions other than those
specified. This is especially true if the inlet to an impactor stage is
obstructed or if the collection of very small particles is desired. Also.
an impactor cannot size particles which are not sampled. Thus, the inlet
collection efficiency must be known. Once the particles are sampled, thc-
particles must be deposited only upon the impaction plates and, once
deposited, must remain on the plates. Partic]e loss and particle reen-
trainment may result in indicated size distributions which are significant-
ly different from the actual ones.
Restricted Entrance Effect_s
Experiments-*-*3 > *' »•*•" > •'-" » 21 have shown that the theoretical efficiency
curves presented in Figure 3 agree well with experiments, if the experi-
mental design corresponds to the condition of the numerical model. How-
ever, if the flow conditions are different from those assumed in the theo-
retical treatment, the characteristic efficiency curve can be expected to
be somewhat different. For example, some cascade impactors are designed
42.7
-------
for compactness, and the flow approaching the inlet will be horizontal
with no tapered inlet, which is substantially different from the inlet
shown in Figure 1 and assumed in the theory.
The influence of horizontal flow at the entrance on the efficiency
curves has been shown 17,19 in an experimental evaluation of two commer-
cial cascade impactors: The Sierra Instruments High-Volume Sampler,
which has several parallel slots per stage, and the Andersen 2000 High-
Volume Sampler, which has a multitude of circular holes per stage. The
second stages of these impactors, which have comparable cut-off sizes,
were tested with and without the presence of the first stage. The re-
sults, shown in Figure 10, reveal that, when the inlet flow is not ob-
structed (stage 2 alone), there is good agreement between the experi-
mental efficiency curve and the theory. When the first stage is pre-
sent, the agreement is still fairly good for the Andersen impactor, but
a large shift and a decrease in sharpness of cut is seen with the Sierra
impactor.
O.I 02 0.3 0.4 0.5 06 0.7 0.8 0.9
Figure 10 Second stage collection efficiency curves for the Andersen
and Sierra high-volume impactors with and without stage 1
present (the curves are for liquid particles collected on a
smooth plate).
428
-------
The principle reason for this shift in the second-stage impartion ef-
ficiency curve of the Sierra impactor is probably due to a pronounced
vena contracta formation in the throat (T/W = 0.8) of the rectangular
jet when stage 1 was present. The velocity in the jet would thus be lar-
ger than for the case where the entrance is not obstructed, and smaller
particles would be collected, with a corresponding shift of the impaction
efficiency curve to a smaller Stokes number. This also means that the
pressure drop through stage 2 will be larger, when stage 1 is present,
than when it is not- •
Lateral flow into the nozzles of an impactor stage and the resulting
shift of the impaction efficiency curve does not mean it is detrimental
to the impactor design. In simply indicates that the theoretical pre-
dictions for this type of inlet condition have not yet been made, and
therefore, we have to rely on experimental calibrations.
High Velocity Effects
In using impactors, it is often desirable to be able to collect as
small a particle as possible. As equation 1 implies, small particles can
be collected by Increasing the nozzle velocity, V0. However,the theory
previously presented for impactors assumes that the fluid is incompres-
sible. The assumption of incompressibility^ is normally assumed valid
for air flows with Mach numbers,M, less than 0.2,for which the density
is in error by only 2%. However, M = 1/3, for which the density is in
n •>
error by about 5%, may be considered a practical upper limit^ .
High velocity through the nozzle also may bring about more particle
losses between stages and from the impaction plate due to reentrainment
(discussed later). These losses, however, can be minimized by the use
of a proper impaction plate adhesive.
Inlet Losses
As is the case with any size analyzing instrument, the efficiency
with which the impactor inlet is sampling the aerosol particles must be
known before accurate size distribution and particle concentration data
can be obtained.
Recent work^ has shown that large particles (up to about 100 pm)
can be drawn into an inlet from still air with efficiencies of nearly
100%. However, there may still be considerable losses on the internal
surfaces of the inlet due to impaction and settling. If sampling from
a moving air stream is required, isokinetic conditions at the inlet will
aid in efficient sampling of large particles.
429
-------
If the sampling efficiency of large particles becomes too low for
good data to be obtained, it may be necessary to use an impaction surface
on a rotating arm, since this device has no inlet.
Interstage losses
Particle losses in an impactor, generally referred to as wall losses
or interstage losses , is the deposition of particles on surfaces other
than the impaction plate. Currently, no theory exists to predict these
losses, and thus, they must be determined experimentally.
Recently, three commercial cascade impactors have been investigated
experimentally to determine the interstage losses. In a detailed evalua-
tion of the Lundgrem impactor (rotating drum impaction plate) and of
the Andersen viable aerosol sampler (28.3 1pm. 1 cfm, multiple round jets)
and also in a similar evaluation of the Sierra high-volume cascade impactor-'-"
(1133 1pm, 40 cfm, multiple rectangular jets), the interstage losses were
determined as a function of particle aerodynamic size, (i.e. equivalent
diameter of unit density spheres), and were expressed as the fraction of
the total number of particles entering the impactor. (Figure 11)
A study of Figure 11 shows that for the Lundgren and for the Andersen
Viable impactors, interstage losses are not a serious problem for partic-
les smaller than about 5 urn, but increase rapidly for larger particles.
The Sierra impactor, however, has appreciable losses for particles small-
er than 5 pm. It was found^" that the interstage losses occurred prim-
arily on both inner walls of the rectangular nozzle, apparently due to
lateral impaction of the entering flow. Tapering of the nozzle inlets
reduced these losses.
Particle Reentrainment (Bounce-off and Blow-off)
The limitations of impactors due to particles bouncing off the im-
paction plate or being blown off the impaction plate after collection are
essentially the same: in both cases, particles which should have been
collected are reentrained into the airstream. Thus, we shall use the
term reentrainment for both blow-off and bounce-off. For single stage
impactors, this means that the concentration of particles collected on
the impaction plate will be too small. For cascade impactors, the result-
ing size distribution will be shifted to smaller sizes, since reentrained
particles will be collected on stages intended to collect smaller particle
sizes. }
The degree of particle reentrainment is a function of the type of par-
ticle and the nature of the impaction surface. Liquid particles will im-
pact on any type of surface withf,verNy little reentrainment. The reentrain-
ment of solid particles, however" is\a strong function of the type of
\
\
i
43'Q
-------
o>
o>
o
60 j-
50
40 -
*OL
T
TTTT
Andersen Viable
(283LPM)
c \ Sierrj Hi Voi
^ 20 i- (! ^2 LPM)
10 r
AeocfyrufTHC Oiametar, ,aTs
Fig-art:-. 11 Total interstage losses fy- the Andersen (multiple round
ho'i t-'f ) , L
impaetors
'it-:-) „ Lundgren (ylnglc slots"' and S'.erra (multiple sJots)
e
; ion Burr'ac
;::, a dry, F
' 1 Lists sc
• used- The maxj^um arriount of retncrainmenf. is experienc
rooth surface and trie least with a verv sticky surface.
n::^ of tin1 sticky s'.irJ ;:cec; tnat have be-;-n used.
L detailed pxpt?:tmfi'tal study-'-' to J;-tPrmine the degree of reentrain-
'lioat f r'"-ffl several types of impact ion surfaces was made using single-stage
irapactors ("single round jets) designed according tc the theoretical cri-
teria described above. The results for one cf t'ifc ir.spactcrs aie present-
ed in Figure 17. Tt should be noted thai the to fit aeroiols were polytty-
re IP ]--.t:f>x ;,p',ieres, wuich are quilo icsili./nt. Thus, the curves of Figure
12 can b? considered to be approximate!'; tho ^orst obtainable.
Figure 12 «h.^ws that with an a4 \ coate:! glaas p" ~'te there was essen-
tially no r^e; '• tvi.im>iont , :-;ince the eff in'e icy c.jrve agreed weJ 1 wirh the
theoreti'-a] cr.rvt- ::•••: .attained ue;.M ly 100% efficiency. At the other
-------
Table I
Impaction Surface Adhesives Used in Impactor Studies
Reference
May30 (1945)
Davies et al.31(1951)
32
(1962)
33
(1963)
Stern et al.
McFarland and Zeller
Zeller34 (1965)
Lundgren20 (1967)
McFarland and Husar35 (1967)
Mercer and Chow (1968)
Hogan37(1970)
00
Berner (1972)
Wesolowski et al
40
39
McCain et al.
Walkenhorst
(1973)
(1974)
(1974)
Adhesive
- Fly paper mixture — one part of rosin
to three parts of castor oil (benzene
solution)
- Polyisobutane (chloroform solution)
- One part of methylated starch to three
parts of tricresyl phosphate (alcohol
or amyl acetate solution)
- Oil or vaseline
- Glycerin jelly
- Dow Corning Hi-Vac silicone grease
- Dow Corning silicone oil (DC-200 fluid,
2% solution in hexane)
- One part dibutyl phthalate, 10 parts
cellulose acetate butyrate and 90
parts cylohexanone
- Scotch tape and other films
- Viscous oils and sticky tapes
- Dow Corning Hi-Vac Silicone grease
- Minnesota Mining Kel-F Polymer wax
- 1000,000 cs. Dow DC-200 silicone oil
(1% solution in hexane)
- Dow Corning antifoam A
- G.E. silicone resin SR-516
- Vaseline dissolved in carbon tetra-
chloride or toluene
- Sticky film
- Dow Corning Hi-Vac grease
- 10 to 15% solution in benzene of high
vacuum silicone grease
- Several oils, CaCl~ solution
432
-------
o
c
Q>
UJ
e
O
O
O
100
90
80
70
60
50
Theory (Marpie.j9_7Q! -
Oil Coated
Glass Plate
Oil Coated Gloss / 7 7 /-Whatman No 4! Filter Paper
" Fiber Filter v //,/7/ /
/ / , /-Glass Fiber Filter
' /'
— Uncoated Glass Pia'e
/ /
/ Experimental Curves_Roo (1975)
Round Impacfor
Test Aerosol- I I urn PSL
045 05 055 06 1)65
'/STK
Figure 12 Collection efficiency of a typical single stage impactor for
various collection surface media
extreme, the uncoated glass plate had severe reentrainmenL for particles
larger than about /Stk = 0.47, and had a maxiinura collection efficiency
of about 30%. When a filter paper was used as a collection surface,
the efficiency curves were found to be between the two extreme cases of
the oil-coated and the dry glass plates, and tended to have poor sharp-
ness of cut characteristics.
Reentrainment in an impactor with several stages (Andersen viable)
was found to be similar to that in the single-stage impactor. As seen
in Figure 13, the best collection efficiency was with liquid particles
and the worst with solid particles impacted onto an uncoated plate. The
curve for collection on a glass fiber fJiter paper leached an efficiency
of about 90%. It appears that most of the particles get collected in Lh
void spaces between the fibers, while some always bounce off the top
433
-------
e
o.
<"-)
OO
QJ
J-i
ft!
n
0)
fa.
o
tfl
CO
O)
0)
>-i
QJ
T3
C
14-1
o
to
o
•r-l
4-1
to
•H
&-i
Ol
4-1
CJ
O
•H
O
0)
o
CJ
00
•H
-------
fibers so that 100% collection efficiency cannot be attained with glass
filter paper. Also, the cut-off characteristics are generally poorer
for the glass fiber paper than for a coated plate,17,19,21
DATA ANALYSIS
The particles collected on the impaction plates are analyzed after
tne sampling is terminated. In the most common type of analysis, the con-
centration and size distribution (by number or mass) of the aerosol par-
ticles is determined from the deposits by one of the several methods of
analysis of which only the more widely used ones will be discussed here.
The size distribution of the aerosol can be determined either from a
single stage impactor or from a cascade impactor. For a single stage i:n-
pactor the deposit must be analyzed with a microscope to get a number dis-
tribution. With a cascade impactor, the particle deposits of the various
stages can also be analyzed with a microscope, but are more commonly
analyzed gravimetrically to obtain a mass distribution.
The number or mass concentration of the aerosols can be determined
from the size distribution data. If the size distribution is found from
a portion of the deposit by microscope analysis, the concentration is cal-
culated from the quantity of particle laden air sampled and from the frac-
tion of the total deposit analyzed. Tf the gravimetric analysis is used,
only the quantity of air flowing through the impactor needs to be known.
Methods other than the gravimetric analysis are also used to deter-
mine the mass of particles collected on an impaction plate. In one
instrument the particles are impacted onto a piezo-electric crystal
which determines the mass by monitoring the change in the resonant fre-
quency of the crystal. Another instrument*-' uses beta attenuation to
letect the mass of particles collected.
Impactors also have been used in conjunction with an optical pars {en-
counter to obtain the number size distribution and particle concentra-
tion^. The aerosols were passed through an impaccor with a variable jet
width. The particles not collected on the impaction plate were counted
with an optical particle counter. By varying the jet width in a step-
wise manner, a size distribution was obtained.
435
-------
The elemental composition of the deposits can be determined by tech-
niques such as atomic absorption, neutron activation and x-ray fluores-
cence. The collection of the particles should be on a medium having a
very low level of impurities. For instance, Jt has been found that
;;lass fiber filter paper is unaccepteable for x-ray fluorescence analysis,
Vit membrane filters, made of mixed esters of cellulose and having 0.8 urn
pore size, is satisfactory.
In order to avoid the problems of jjarticle reentrainment, a dichoto-
fiious (or virtual) impactor may be use•
dynamic cutoff size.. <.rC D^Q. f-'i'U'.e the cutoff characteristic;-,
are fairly constant and s'uirp ov*;;: a Reynolds number range, fror-
-------
500 to several thousand, a Reynolds number of 3,000 may be as-
sumed for the calculations.
Rectangular - Use Figure 8 and choose a desirable Reunolds num-
ber (again Re = 3,000 is satisfactory) and determine the Q/L
value and jet width, W, for the desired aerodynamic cutoff size
vC D5Q. Special note must be made of the resulting value of the
jet aspect ratio, L/W, since end effects can become deter imental
to the sharpness of cut for nozzles with small L/W values.
Select a convenient size of jet diameter or jet width which is
close to the value found in step 1. For a round impactor this
would be a standard reamer size.
For this value of jet diameter or jet width, check the Reynolds
number using equation 4 or 9.
Determine the value of /Stk^Q from Figure 3 and calculate the
cutoff size, /C Dp, by using equation 5 or 10.
Determine the pressure in the impaction region, ?2 .by assuming
that the pressure drop in Che stage is equal to the dynamic
pressure of the jet. Thus
P2 = PI - 1/2 PV02 [12]
where Pj = static pressure at the impactor stage iriiet
?2 = static pressure at the impaction plate
p = fluid density
V0 = average fluid velocity in the jet
Determine the cutoff particle size, by calculating the Cunning-
ham slip correction36 from the equation:
, . 0.163 , 0.0549 , , ,, „ - , ri_,
C = 1 + -f-^-+ -Tr~p — exp <~6-66 Dpp2> I13J
Ur Ur
where ?2 = static pressure (atmospheres)
Dp = particle diameter (pm)
8. Repeat steps 1 through 6, for the next stage of the impactor.
Besides determining the impactor jet diameter at each stage, it
Ls also important to design the impactor so that the jet-to-plate dis-
tance, S/W, is greater than 0.5 and 1.0 for round and rectangular impact -
ors, resepctively. In order to provide for a margin of safety, the
criteria of
S/W = 1.0 (round impactor)
and S/W = 1.5 (rectangular impactor) [14]
437
-------
should be the minimum jet-to-plate distance used. Under this condition,
small variations of jet-to-plate distance will not effect the value of
As shown in Figure 3, the throat length, T/W, does not greatly in-
fluence the cut-off characteristics of the impactor. However, these
curves were calculated for an impactor with a tapered or conical inlet
section as shown in Figure 1. In the absence of the tapered or conical
entrance, a short throat may not allow sufficient time for the particle
to accelerate to the fluid velocity in the throat so that the efficiency
curves may be different from those shown in Figure 3. Also, particle
losses may be found at a sharp entrance. Thus, if possible, the entrance
of an impactor jet should be tapered and the T/W value should be at least
Design of the impactor between stages is also important since this is
where interstage losses of the aerosol occur. In general, the fluid ve-
locity must be kept large enough so that particles are not lost from set-
tling and yet not so large as to lose particles from impaction in the
Interstage space. Also, obstructions and sharp corners should be kept at
h minimum, since particles are deposited in the resulting turbulent areas
behind these obstructions.
COMMERCIALLY AVAILABLE CASCADE IMPACTORS
Some of the cascade impactors which are commercially available are
*:sfed in Figure 14. For each impactor the aerodynamic diameter at the
'•', cut point is given for each stage at a typical flow rate. Most im-
.;*-»'• tors, however, are not confined to the indicated flow rate and may be
-Derated over a range of flow rates which would shift the cut points to
.--v-T or smaller diameters. Particles vith a density different from
,--;-\f would also shift the cut points. This shift may be calculated from
• •".stflon 11. The variation in Cumii .i^ham slip correction is usually neg-
, :--ytb! „' for small diameter changes so that, equation 11 is reduced to
3p = D50/^PP [15]
.'' ?he 50% cut point. When not sampling at normal temperatures and
~t•-.:•. -ires, however, the slip correction factor tn equation 11 may have to
' -f r alcu lated,
,38
-------
Commercial Importers
•st
JR
*
8
1
Nominal
How Rate
20 - 40 cfm
(Hl-Vol)
0.5 - 1 cfm
(Stack)
0.6 - 1 cfm
n.05 ct
-------
The cut points indicated in Figure 14 should only be used as general
guidelines, since several of the stage constants were only theoretically
calculated but not experimentally calibrated. No measure of spread has
b'en listed for the collection efficiency curves. Also, some of the im-
purtors may also be purchased with fewer stages than indicated in the
Figure.
Impactors for s ,iilar purposes are quite comparable in price. Some
jf the costs indicateJ also include auxiliary equipment, e.g. flow con-
trollers and pumps.
The high-volume samplers are primarily used for ambient 24-hour sampl-
'.-.ig or in-plant short-term sampling. The samplers listed have 4 to 5
stages and differ from each other primarily through the shape and arrange-
ment of the openings in each stage. The BGI Hi-Vol (BGI Inc., 58 Guinan
Street, Waltham, Mass. 02154) has a single slot, the Sierra Instruments
inc., P.O. Box 909, Village Square, Carmel Valley, Calif. 93924) has 9
to 10 parallel slots, and the Andersen Hi-Vol (Andersen 2000, P.O. Box
20769, Atlanta, Ga. 30320) has 296 to 300 round holes arranged in several
concentric circles. Their sampling efficiency of large particles depends
m the sampling rate per overall inlet area, ind if housed in a shelter,
: n the sampling efficiency of the shelter inlet. The latter may be af-
f^ced by ambient air velocity and wind orientation with respect tc the
sampler inlet.
Cascade impactors used by industrial hygienists usually operate at
.-ou! 30 1pm. Cca.l cfm), if they are stationary, or at about 1 1pm (ca.
~>.{,'"> cfm) if they are carried by the worker.
In the Sierra and Andersen cascade Impactors the plates containing !~',e
i T orifices also support the collection medium for impaction from the nre
'• -is stage. The collection medium, usually glass fiber paper, has open-
-,-.- cut into il so that the flow may enter the orifices of the next stape
txception to this Ls the Andersen Viabio Sampler in which airborue
' ,c ro-organisms are classified by six stages. Each stage collects viablt-
• :io:i-viable aerosols into an agar containing Petri dish, while the un-
• tvictt'd aerosols flow around the Petri dish into the next stage.
When sampling effluents from industrial sources, the cascade impactor
; .. be inserted directly into the effluent stream, usually the vertical
iik. or a sample may be extracted for analysis outside the pollutant
',,:•', , Included in Figure 14 are several impactors used in stacks. The
\ -ces are generally contained inside a long cylindrical housing which is
•jv-illy less than about 8 cm. in diameter and is topped by a conical noz-
• - inieL. Tht inlet is usuill} dc-.rodynar.ically shaped and decelerates
-------
the effjuent flow for first stage itnpaction inside the housing. Several
nozzles are provided for isokinetic sampling from the effluent stream.
The flow rates generally range from about 14 to 30 1pm (ca, 0.5 to 1 cfw).
Most sampling is done in environments not exceeding 200°C (ca, 400°F).
Exposure to higher temperatures necessitates the use of stainless steel
in the construction. Particle bounce-off, reentrainment, and weight
changes caused by gas-to-solid conversions or chemical reactions with
the collection medium may make high temperature measurements more ambigu-
ous than ambient air measurements. Further studies are needed to find
suitable col Lection media for specific ranges of stack environments.
Stack samplers are presently produced by Sierra Instruments, Inc.;
Andersen Inc.; Meteorology Research Inc. (464 W. Woodbury Rd., Altadena,
Ca. 91001) and Pollution Control Systems Corporation (321 Evergreen Bldg.,
Renton, Wash. 98055). When effluent samples are extracted from the
stack into an impactor outside the stack, a survey of the entire stack
cross-section may be made more readily than with an instack device. How-
ever, particle losses in the sampling line may truncate the size distribu-
tion at the larger particle sizes.
Several cascade impactors have only one orifice per stage. Examples
are the BGI Hi-Vol, the May/Casella MKII and MKIIa (C.F. Casella & Col
Ltd., Regent House, Britannia Walk, London, No. 1, England; U.S. distribu-
tor: BGI Inc.). the Battelle Impactors DCI-5 and DCI-6 (Delron Research
Products Co., P.O. Box 95, Powell, Ohio 43065), the Unico Sampler
(National Environmental Instruments, P.O. Box 590, Fall Kiver, Mass.
02722), the Brink Model B Impactor (Monsanto Enviro-Chem Systems, Inc.,
800 N. Lindbergh Blvd., St. Louis, Mo. 63166) and the Aris impactors
(Aerosol Research Instrumentation Equipment and Services, Inc., P.O. Box
25541, Albuquerque, N.M. 87125).
In the Lundgren and in the Multi-Day Sampler (Sierra Instruments) the
aerosols impact onto rotating drums for time resolved analysis of the
deposits. The Centripeter (Bird & Tole, Ltd., Bledlow Ridge,High Wycombe,
Buckinghamshire, England; U.S. distributori BGI Inc.) is f> dichotomous
virtual impactor which splits the jet into a center flow with large par-
ticles and a laterally deflected flow of small particles. An impartor
with a similar basic design concept has been developed^ by the Envi-
ronmental Research Corporation (3725 N. Dunlap St., St. Paul, Mn. 55112).
441
-------
SUMMARY AND CONCLUSIONS
Inertial impactors have become a popular tool for the analysis of
the size distribution and concentration of aerosols.
The performance of impactors can be accurately predicted by the use
ol modern numerical methods which solve the equations governing the fluid
low and particle motion in this flow. This theory shows that impactor
^tages which are properly designed and operated will provide sharp class-
ification between the particles collected and those which are not. Also,
tise theory can be successfully used to aid in the design of multiple hole
j.ipactors. These results, presented in the form of design charts,
indicate the number and size of holes needed, for a specific flow rate,
* o keep the flow conditions (Reynolds number) near optimum.
Although impactors are capable of making a sharply defined cut be-
cween particle sizes, special precautions must be taken to insure reliable
•md accurate data. Such problems as particle losses at the inlet, par-
ticle reentrainment from the impaction surface, and particle losses be-
1 w<: -211 stages of a cascade impactor can result in the indicated particle
-fze distribution being considerably different from the one sampled. How
• vr, by the use of proper design and impaction surface coatings, these
, i •"'-.• iems can be minimized.
The current popularity of cascade impactors is evident by the large
. n..3er commercially available. The listiiv-> of some of these impactors
'.if U:ates a variety of intended uses, flow rates and cut-off sizes.
-------
References
1 Marple, V.A., B.Y.H. Liu. Characteristics of Laminar Jet Impactors.
Journal of Environmental Science and Technology 8:648-654, July 1974.
2 Marple, V.A. , and B.Y.H. Liu. On Fluid Flow and Aerosol Impaction in
Inertial Impactors. Journal of Colloid and Interface Sci. (1975).
3 Ranz, W.E. Principles of Tnertial Impaction. Bulletin No. 66, Dept.
on Engineering Research, Pennsylvania State University. (1956).
4 Jaenicke, R. and C. Junge. Studien zur oberen GrenzgriJsse des natUr-
lichen Aerosoles (Studies of the Upper Size Limit of Natural Aerosols.)
BeitrHge zur Physik der AtmosphHre. 40:129-143, 1967.
5 Noll, K.E. and M. J. Pilat. Size Distribution of Atmospheric Giant
Particles. Atmospheric Environment. 5:527-540, 1971.
6 Gruber, C.W. and G. Harms. Fugitive Dust Measurement by the Rotary
Ambient Adhesive Impactor. Presented at 67th Annual APCA Meeting,
Denver, Colorado. June, 1974.
7 Marple, V.A. , B.Y.H. Liu and K.T. Whi'tby. Fluid Mechanics of the
Laminar Flow Aerosol Impactor. Journal of Aerosol Science 5:1-16,1974.
8 Ranz, W.E., and J.B. Wong. Impaction of Dust and Smoke Particles.
I & E Chem. 44:1371, 1952.
9 Mercer, T.T. and H.Y Chow. Impaction From Rectangular Jets. J. Coll.
and Interface Sci. 27:75, 3968.
10 Mercer, T.T. and R.G. Stafford. Irapaction From Round Jets. Ann.
Occup. Hyg. 12:41, 1969.
11 Davies C.N. , M. Aylward and D.Leacey. Impingement of Dust From Air
Jets. A.M. A. Arch. Inc. Hyg. Occup. Med . 4:354,1951.
12 Marple, V.A. , A Fundamental Study of Inertial Impactors, Ph.D. Thesis,
University of Minnesota, Particle Technology Laboratory, Publ. No. 144
1970.
443
-------
13 Marple, V.A., B.Y.H. Liu and K.T. Whitby. On the Flow Fields of Incr-
tial Impactors. ASME Journal of Fluid Engineering. 96:394-403, Decem-
ber, 1974.
14 Gosman, A.D., W.M. Pun, A.K. Ruchal, D.B. Spalding and M.Wolfshtein.
Heat and Mass Transfer in Recirculating Flows, Academic Press. N.Y.
l~) Fuchs, N.A., The Mechanics of Aerosols, Pergamon Press, New York, p.
154, 1964.
Ib Jaenicke, R. and I.H. Blifford. The Influence of Aerosol Character-
istics on the Calibration of Impactors. J. Aerosol Sci. 5:457, 1974.
17 Willeke, K. and J..J. McFeters "The Influence of Flow Entry and Col-
lecting Surface on the Impaction Efficiency of Inertial Impactors.
J. Colloid and Interface Science, in print ,1975.
18 Schott, J.H. Jet-Cone Impactors as Aerosol Particle Separators. M.S.
Thesis, University of Minnesota (1973).
19 Willeke, K. Performance of the Slotted Impactor. Am. Ind. Hyg.Assoc.
J., Sept. 1975.
10 Lundgren, D.A. An Aerosol Sampler for Determination of Particle Con-
centration of Size and Time. J. Air Poll. Cont. Assoc. 17:225, 1967.
21 Rao, A.K., An Experimental Study of Inertial Impactors. Ph.D. Thesis,
University of Minnesota, Particle Technology Laboratory, Publ. No. 269
1975.
_ ','ison, R.M. , Essentials of Engineering Fluid Mechanics, International
Textbook Co. Scranton, Pa. p. 165, 1961.
5 s.ohen.J.J. and D. N. Montan. Theoretical Considerations, Design, and
/valuation of a Cascade Impactor. AI!1A Journal. 28:95-104, March-
/.pril, 1967.
_ • Fuchs, N.A. Sampling of Aerosols. Atmos. Env. 9:697-707, 1975.
' ""• Agarwal, J.K. Aerosol Sampling and Transport. Ph.D. Thesis, University
-.if Minnesota, Particle Technology Laboratory, Publ. No. 265, 1975.
in Cnuan, R.L. An Instrument for the Direct Measurement of Particulate
yass. Journal of Aerosol Science 1:1.11-114, 1970.
-------
27 Lltienfeld, P. and J. Dulchinos. Portable Instantaneous Mass Monitor
for Coal Mines Dusts. Am. Ind. Hyg. Assoc. J., 33:136-345, 1972.
28 Cooper D.W. and L.A. Spielraan. A New Particle Size Classifier: Variable-
Slit Impactor with Photo-Counting, Atm. Env. 8:221,1974.
29 Dzubay, T.G. and R.K. Stevens. Ambient Air Analysis with Dichotomous
Sampler and X-ray Flourescence Spectrometer, Env. Sci. and Tech.
9:663,1975.
K) May K.R.. The Cascade Impactor: An Instrument for Sampling Coarse
Aerosols.J. Sci. Instr. 22:187, 1945.
n Uavies, C.N., M. Aylward, and D. Leacey, Impingement of Dust from
Air Jets, A.M.A. Arch. Ind. Hyg. Occup. Med. 4:354,1951.
'\'l Stern, S.C., H.W. Zeller, and A.I. Schekman. Collection Efficiency of
Jet Impactors at Reduced Pressures., I & EC Fundamentals 1:273, 1962.
33 McFarland, A.R., and H.W. Zeller. Study of a Large-Volume Impactor
for High-Altitude Aerosol Collection, General Mills Inc., Electronics
Division, Minneapolis, Minn. Rept.No. 2391, Contract AT (11-1),
p. 401, 1963.
J4 Zeller, H. Large Volume Impactor Collector, Litton Systems Inc.,
Applied Science Division, 2295 Walnut St., St. Paul, Mn. 55113, Rpt.
No. 2893, Project No. 89125, 1965.
35 Mci-'arland, A.R. and R.B. Husar. Development of a Multistage Inertial
Impactor, Particle Tech. Lab. Publ. No. 120, Dept . Merh. Eng,, U. of
Mn. Mpls., Mn. 55455,1967.
36 Wahi, B.N. and B.Y.H. Liu. The Mobility of Polystyrene Latex Particles
in the Transition and the Free Molecular Regimes. J. of Colloid and
Interface Science 37:374-381, Oct. 1971.
37 Hogan, A.W.. Evaluation of a Silicone Adhesive as an Aerosol Collect--
ing Medium, J. Applied Meteorology, 10:592,1971.
38 Burner, A. Practical Experience with 20-Stage Impactor, Staub-
Reinhaltung Luft, Engl. Ed. 32,8"1, 1972.
39 Wesolowski, J.J. Ambient Air Aerosol Sampling, Proceedings of"Second
Joint Conf. in Sensing of Environmental Pollutants", Instrument
Society of America, 400 Stanwix St. Pittsburgh, Pa. (1973)-
•''•0 McCain, J.D., K.M. Gushing and W.B. Smith. Methods for Determining
Particulate Mass and Size Properties: Laboratory and Field Measure-
ments. J. Air Poll. Control Assoc. 24:1173, 1974.
445
-------
41 Walkenhorst, W. Untersuchungen Uber den Haftgrad von Staubteilchen.
(Studies of the Degree of Adhesion of Dust Particles). Staub-Reinhalt.
Luft (German ed.) 34:182, 1974.
446
-------
THE CYLINDRICAL AEROSOL CENTRIFUGE
1 ) "* )
Mohammad Abed-Navandi , Axel Berner und Othwar Preining^
1st Physics Institute, University of Vienna, A Io9o, Strudelhofgasse 4
1) since July 1974 Bundesstaatliche bakteriologisch seriologische
Untersuchungsanstalt, Abt. Lufthygiene, Wien
2,1 contact for reprints
ABSTRACT
A very important size parameter of aerosol particulates P is the
aerodynamic diameter D . The distribution of the P's of an aerosol
over D permits one to predict the mechanical behavior and is therefore
of special interest. Instruments sep rating the P's according to
!.• ^ permits one to derive this size distribution experimentally; aerosol
Centrifuges are much used instruments 10^ this purpose. Instruments
sampling cummulatively have been developed as well as sach .vith discrete
-ampling modes/only the latter permit one to derive the distributions
directly. The principle is: A laminar clean air flow is superimposed
by a thin sheet of aerosol flow in a strong centrifugal field.
"'s are deposited at different locations according to their D on
a removable foil for further analysis. The iorris of the flow chambers
differ widely, e.g. spiral channels and hollow cones have bc-en used
successfully. The instrument described in the following uses a hollow
cylinder, the aerosol is superimposed after the flow profile has
developed, the P's are sampled on the wall of the outer cylinder.
The aerosol inlet slit is perpendicular to the axis of rotation which
LS a.'.so the axis of the hollow cylinder. If the flow rate of the
.aerosol is kept small, size resolution can be increased to 0,5 %,
i.e. two particles with D 's of l,ooo um and I,oo5 urn are deposited
at different locations. Tne instrument works best in the size range
l:etwet-:n 1 um and O,l ^um. The desiqne and the perfoimance characteristics
are discursed as well as the possibility of absolute mea.suleweiits
•without calibration.
The paper covers partly the Ph.D. Thesis cf Mohammad Abed-Navandif
December 1973.
447
-------
THE CYLINDRICAL AEROSOL CENTRIFUGE
Mohammad Abed-Navandi , Axel Earner und Othmar Preining
;;-;t Physics Institute, University of Vienna, A Io9o, Strudelhofgasse 4
•') since July 1974 Bundesstaatliche bakteriologisch seriologische
Untersuchungsanstalt, Abt. Lufthygiene, Mien
Aerosols are complex systems consisting of particulatcs, P, i.e.
liquid and/or solid particles without reference to state, and a gas.
I'o predict the mechanical behavior of the aerosols the sizes of all
-he P's have to be known. This is accomplished by acquiring the size
distribution f(D ) experimentally. The size distribution is mathemati-
cally a probability density. The "sizes" of the P's have to be propeily
i ;fined. A very useful and realistic measure is the aerodynamic
diameter D , i.e. the diameter of a sphere of density one which assumes
the same velocity in the gravity or in a centrifugal field as the P in
q testion.
''-'he reliable conversion of other size parameters, particularly from
optical measurements, to D is very difficult and subject to great
>}'\.ors. Therefore the direct measurement of" the £ vs D relation
„•>..'!-ones necessary, at least for the purpose of calibrating aerosol
measuring devices. In the D -range below 1 urn the gravity field is
insuf.ficent to separate the P's from the gas, hence centrifuges were
-jonel subject to a strong centrifugal field which separates the
'' •; from the gas on a surface which can be further investigated. The
~ ~^a~i<~>n of deposition being a direct uj indirect measure of the I), of
•-.• P. The distribution of the deposited P's on the surface yieldsL the
• • ?.' distribution directly, if only a tl'±n sheet of .aerosol flow j.~
. ,r t?rirr;r>osed over a clean gas 51 or/'. I::-'- Jotvnstream location L of the
u ":;.>:-.• ted P is «ameas>j re of if'; D •. L being the distance from the aerosol
:>ii.<--t ' -• the location of deposition, called deposition length or
~" -,','jen-tation distance. So far quite a few different instruments have
;••••".n designed. The idea arose in England, Sawyer and Walton v The first
•ic-tr^ment commercially available was the Goetz centrifuge ' ' , which
:-' .:t:ll used, e.g. Gerber" ; in spite- of the (^raw-back of being a
u- _T;,..2.itiye sampling device. Kei th ,arjd. ;£ir''rick ', Kast . stober and
'( ~5'.i .:,< '" , Stober and Flachsbart" ~~' *' , .Tc3b^r et al , Stpber ,
,,;-./< ir.d f>ch&dling~ ', A'ochraider and BT~C,\ n ''. Hochgainer , Berner
.: ," K^ichelt ' , Burson et al. and i''/:ttc--son et al developed difieren-
. \
-------
One particular solution is discussed in Stober's contribution. Our
centrifuge, designed primarily by A. Berner, used the following
principle: The aerosol flow is superimposed on a laminar clean air flow
through a rotating slir. Aerosol, clean -lir and channel walls have the
r>ame rotational speed to minimize flow instabilities, A cross sort ion
if the centrifuge rotor is shown in fig. 1.
r$he cylindrical rotor is rotating around tie axis of symmetry driven
4y a high speed router motor (stanleii tools) via the shaft K. The
ylean air is produced internally by passing the aerosol through the
r.^iy-channel V. At the entrance in tr.e deposition chamber M only ! he
.lus JLS left. The aerosol enters into '.'he rotating slit S, which is
:^et pendicular to the axis, flows outward and is super irrposed on the
laminar clean aii flow at the entrance ring slit. The P's are deposited
.it- the outer wail of M at the distance I, from the aerosol inlet
depending on their D . The stream leaves rhe sampling head via the
ringsl.it 52, the rint^ channel F and six flow^detennining orifices (7.
To make any theoretical calculations one has to know rh> to>~al flow
in M. To measure it one has to make an air tight seal between the
rotating and the stationary parti, of the centrifuge - -i veiy difficult
problem. This has been solved by use of a teflon ring as shown in
iig. 2. The flow measurement was made using a calibrated 'Cyli
glass tube connected air tight to rhe inlet and observing the
of a soap film. If the flow limiting or if ires were leplaced by plui/i'.
no flow did enter the centrifuge proving such th-jt the system was
stiff icently air tight. One also has to know the rotational speed v^r'j
•accurately. The centrifuge head was supplied with ten small magnet.:-,
They produced an electrical pulse in a coil mounted on the stationary
hood of the centrifuge while passing. At, electronic cuu,?*>rr v^rmi 11 ed
rapid and exact measurement of the rotational speed.
If a mult^dispersed test aerosol {e.g. spray dried Dow J .-fct-'x aeros'yl
j-i'clvdinfj aggregates) is supplied to the instrument th,- r<,i .' shd'-j-: '3
Jr?po.'j/t in the form of lines, ea-rh line corresponding to :::>;ho-
iq'iregaie.?• photograph of such a ''jpectrum" is shown in f':^, 3. If the
.Dumber concentration of P's on the foil is measured as a function of L,
: C'j!i.,.entration profile as shown in fig. 4 is ob^aineJ, U:'-:nq different
sized la':ex aerosols the instrument can be calibrated, i.e. the I. vs
L>r curve cjr. be established experimental ly.
C'n the other hand we know from the flow meas a lament the. Linear
downward velocity of the air in the centrifuge, v. Since the flow
profile, in "he channel does not influence L, as has been shown by
r^ , the L vs D relation can oe calculated. The pa'tide velocity
perpendicular to the axis of rotation is given from riivplf- theory;
449
-------
ro.~s section of
rotor:
% ^I'.af:., V ring channels for producing L'Zean air, M deposition chamber,
,c aerosol inlet, rotating slit twidtli •:'-.,'< vim.:, 52 ringslit connecting
: ;,"; c;i channel E Dairying tL ( tofa1 flow T.lir .ugh M, E outlet channel
•.: ':-•;«..i.-tlng S2 with iht- fJuw J.iir,j. ti>:', or. .' i^es, C one of the six
•.:•;.-• I :•-] spaced flow limiting orif":'rec..
•50
-------
H
v\\X\\\\X\
rig. 2
cross section of the top of the rotor and the hood.
R rotor, H stationary hood, TG Teflon gasket facilitating the air
tight ora.Z between R and H, NS nylon spring pressing TG against F
451
-------
or the depoc.it from a spray dried multidispersed latex
-------
200- -
150-
100-
50-
o
10 20 30
DEPOSITION LENGTH [mm]
concentration profiles across the Jines shown in fig. ."?
centrifuge operating conditions: 3oov RPM, Q - 31, m
453
-------
= __ ND
P 9n
with
rotational speed of the centrifuge in rps
geometric diameter of a spherical P(cm)
density of P (I,o5)
viscosity of the air (O,183.1o cm s )
correction factor for the Stokes law.
B = 1 + 1,36 - + 0,70 ^ exp(-3,68 ^) (2)
mean free path of gas molecules (\ = 0,653. lo cm)
channel height (o,3 cm)
average channel (cylinder) radius (38.5 mm)
constant centrifugal force over y yields the residence time
«f the airborne P in the channel
(3)
L = vt = y — (4)
VP
"u'ng all numerical constants and entering the volume flow rate
>.h rough the channel Q(ml/s) one gets for
L = k Q/N2D2B p with k=0, 85325.10 (5)
iM/uation (5) yields directly the inverse squared aerodynamic diameter
.' , a normalised deposition length.
* I & J/
D D Bp k Q
P
. c m«. \jisjred L-vaiues and the o;;^rc3t-'nc? conditions give L and L
'.-, Z'.u~zi yermitq orx? f n ralcul.itt D .
P
",'u- Vci z i a r J or; of the centrifugal force accross the channel alters the
,;:startt in equation 5 only slightly (k >=o, 853t,6. lo'6 ) ,so one does not
. -.'.:.'dld<-ed values L , for the different operating conditions and
-ViaJjf/ neasured L values are given in table 2 together with the
as lerived from the L values.
M
,.' rt-sjj cs are summarized in fig. 5 shoeing the good agreement between
' ".' .. ry ir.d experiment. However more imporcant for the application of
.,•!-,".' 'o'' centrifuges is their h:gh size resolution and the stability
454
-------
1.5 H
1.0 -i
.7-
.5-1
l
.5
T
.7
1.5
I
3
T
4
,-j.g.
values vs L for the different
aerosols used.
Solid line, calculated relation using equation 6.
horizontal bar: range of measured L values, vertical mark: average L
v-ilue,. horizontal marl D calculated from manufactures date using
(, - I,o5 ,-ind B from .-guatio;? 2.
-------
°f operation , e.g. the calibration must remain constant during hours
or oven days of continuous operation. Since some warming up occurs,
the independence of the calibration from this warming up effect has
to be ensured experimentally. Fig. 6 shows the counted line profiles
for cold and warmed up conditions,- the coinciding curves prove
that additional cooling is not necessary with our instrument.
The flow rate through the centrifuge has to be kept below a critical
value, otherwise secondary parasite flows develop rapidly, first
producing regular wiggles in the lines indicating a cellular structure
of the parasite flow.
Still higher flow rates lead to turbulence in the entrance region
and the line structure of the deposit disappears. All these effects
can be avoided by restricting the aerosol flow by a limiting orifice.
finally the modes of operation which ensure highest resolution are
Jiscussed. Obviously one may expect sharper lines as the superimposed
aerosol sheet becomes thinner. Therefore the line width was measured
for different aerosol flow limiting orifices, leaving the other
operating parameter constant. Fig. 7 shows two line profiles as
obtained with different aerosol flow limiting orifices.
'The variances V and standard deviation a of the measured concen-
tration profiles, the lines, were calculated directly from the
'.-ounts using the equation
=J
5* = —I— i
G v . i=l
v: i r.h
>. particle diameter of the i-th interval
,0_ wean diameter of all P's
-2 2
number of P's in an area of lo mm located so that its L
corresponds to D.
7::-r,-?e tlie variance of "lines'1 produced in different experiments under
nt conditions were determined separately it was also possible
456
-------
250-
200-
150-
XXH
20 22
24 26 28 24 26 2£
DEPOSITION LENGTH [mm]
jt j. y. &
liiiti pr-^tjl^s f>roduceu 1:1 a cold (:.">,.:J line) and a warmed up
(disked line) centrifuge.
1 6000 KPM, 0 = Io9 ml/s,aerosol flow limiting orifice diam.2,o mm
? 5rt:o EP'A, Q = 74, 2ml/s, aerosol fl ••/> limiting orifice (liam.o, 35mn
-------
200H
150H
100H
50-4
UJ
(X
o
H
o
u
(E
UJ
03
ul
a
cr
0-"
20 2^2 24 26
20 22 24 26 28
DEPOSITION LENGTH [mm!
'.' -ie profiles obtained with aerosol limiting ozif'ices of
'c^rs Lit 2,o mm solid line and o,35 dashed line
woc P.PM, Q = Io9 ml/s
: ooo RPM, Q = 74,2 ml/s
458
-------
5-
2-
1-
E
3.
111
O
Q
UJ
HE
6
5-
3-
2
-I
__ — — O •
„--
1 Mill! 1 "" i
345 67891x1o° 2
1-
i U
AFLO CROSS SECTSOH fff/4
Fig. 8
measured variance V Cafcsolut value)
vs. the diameter of the aerosol flew limiting orifices (AFLO). Size (_
circa is corresponds approximately to '.he range of the experimentally
derived variances.
459
-------
to estimate the scatter of the variance values by calculating the
variance of the variance. The results are given in table 3;
V appears as a function of the inlet orifice of the aerosol flow.
This measured variance V consists of both the variance V
Describing the width of the size distribution of the aerosol and the
variance V describing the instrument influences.
V = V + VT (8)
w A I
Since V is not known with sufficient accuracy V cannot be
calculated but V is certainly an upper limit of V . Comparing the
V values with the V, given bu the manufacturer of the latex
27? A
suspension (Dow Chemical) one finds V and the smallest measured
V of about the same magnitude.
m
ACKNOWLEDGEMENT
1'he work reported was partly supported by the US-EPA-Grant R Sol 983
and RO1 AP OO468.
460
-------
Table l
Correction factor for the Stokes law according to equation 2
D B
I,3o5 I,o97
o,79o 1,156
o,557 1,214
o,357 1,316
Table 3
The measured variances V of tne "lines" and the variances of these
variances versus the diameters of tne aerosol flow limiting orifices
diameter of the
aerosol flow V variance of
m
limiting orifice , . V
1 4 2 * fflo
mm lo ym ,42
1 o pm
o, 35 o, 5'1 o,o9
o,5o o,SI 0,06
o,65 1,15 o,o3
lfoo 1,53 o,12
1, 5o 2, 79 o,o5
2,00 4, /(j 0,08
operating condition: 3ooo RPM, (,'- ~ 49,5 ml/s
Latex:
manufactures data mean geornetr.i." diameter u, ?9O }jm
standard cievj at;'en O,OO44 ^m
461
-------
1)
-c
4J
C
O
"2
p
O
Cn
E E
(£>
CO vj
6 o
II II
0> *Q-
Q O
*
CD _
£ ro-
in m
o o
o CM
o £
^t
gj
I 5 S
f>
s» UJ
s
in
in U>
in in
PO S 888
CO
CM
co
{5} 5
CM
CM
'E
S
o
o
o
in S2 »
m P 3
m
n
o'
n
-u
<0
in
C1
QJ
E £
a. ^
t- o
m co
in
ci
M
lp i".
CM
»
in
O 5
in m
CM CM
8
«si CD
T
o m vj
«* eo t>
in in -*
CM CM CM
§§ s§
3 g 55
III
in •* w
CM CM CM
C-~ *S «-'
^ £ a
o P
tvi t*-.
co o
vt 5
§ ?,
o g
<&.
o
II II
*e *Q.
i_l IQ
r-H
0
CU
HJ
4J
-3
4J
5
is
1 1
O iC> *QL *£
Q Q —I
E R S S 8 § S
-i ob co r; <£> en J2
f " s 5 s
°> :* CM ^
in ob
S8 S IS «
co in ob Jii
CM
'E E
in a?
CM Q
Q o
o o
O
o
n
'4-i
In
O
0)
'H
J
&i
3
~l C
"J ^
S. I.
in §
ro
S
o'
M n n
cn *Q
o a
in,
_r
io o
CM O
in in
o o' c>
o o en
R o> q
co in cri
3 8 S
ro lf» 0>
CO
in
S>
in
"- «"a
CD
462
-------
REFERENCES
1. Sawyer, K.F., and Walton, W.H. The Conifuge - A size-separating
Dimpling dovice for airborne particles. J. Sci. Instruments 27:
272, 195o
2. Goctz, A. An Instrument for the quantitative separation and size-
classification of air-borne particulate matter down to o.2 micron.
Geofisica Pura e Appl. Proc. II, 36, 1957a, 49
3. Goetz, A., Stevenson, H.J.R, and Preining, 0. The Design and
Performance of the Aerosol Spectrometer. A.P.C.A. Journal lo:
37S-383, October 196o.
4. Goetz, A. and Preining, O, Bcstimrnung der GrdBenverteilung eines
Aerosols rnittels des Goetz'scnen Aerosolspektrometers.
Acta Physica Austriaca, 14:292~3o4, 1961
5. Gerber, H.E. On the Performance of the Goetz Aerosol Spectrometer.
Atm. Environment 5: Ioo9-lo31, 1971.
6. Keith, C.ii. and Derrick, J.c. Measurement of the particle size
distribution and concentration of cigarette smoke by the
''conifuge". J. Coll. Science Is: 34o~3b6, 196o.
7. Kast, W. Neues StauhmeBgerat zur SchnelIbestimmung der Staub-
konzentration und der Kornverteilung. Staub, 21: 215-223,
April 1961.
8. Stober, W. and Zessac': U. Zuc Theorie einer konischen Aerosol-
zentrlfuge. Staub, 24: 235-3u5, August 1964.
9. Stober, N. und Zessack, U. Zur Messung von Aerosol-Teilchen-
ciroBenspektren nit Hilfe von Zentrifugalabscheidern. Zentralblatt
rur Biologische Aerosolforschung, 13: 1-19, Dezember 1966.
lo. Stober, M. and Flachsbart,H. Aerosol Size Spectrometry with a
Ring Slit Conifuge. Environmental Sci., 3: 641-651, July 1969.
11. Scober, W. and Flachsbart, H. High Resolution Aerodynamic Size
Spectrometry of Quasi-Monodisperse Latex Spheres with a Spiral
Centrifuge. J. Aerosol Sci. 2: lol-116, 1971.
12. Stober, N., Flachsbart, H. and Boose, Ch. Distribution Analyses
of the Aerodynamic Size and the Mass of Aerosol Particles by
Means of the Spiral Centrifuge in Comparison to other Aerosol
Precipitstors J- Coll. Sci. 39: Io9-12o, April 1972
463
-------
13. StoJber, W. Design and Performance of a Size-Separating Aerosol
Centrifuge Facilitating Size Spectrometry in the Submicron Range.
In: Assessment of Airborne Radioactivity. Vienna, International
Atomic Energy Agency, 1967.
14. Hauck, H. und Schedling, J.A. tiber ein modifiziertes Modell einer
Konifuge. Staub 28: 18-19, January 1968.
15. Hochrainer, D. and Brown, P.M. Sizing of Aerosol Particles by
Centrifugation. Environmental Sci. and Technology, 3: 83o-835,
September 1969.
16. Hochrainer, D. A New Centrifuge to Measure the Aerodynamic
Diameter of Aerosol Particles in the Submicron Range.
J. Coll. Interf. Sci., 36: 191-194, June 1971.
17. Berner, A. und Reichelt, H. Uber EinlaBspaltsysteme in Konifugen
Tell J: Das ROSL-System. Staub 29: 92-95, 1969
18. Burson, J.H., Keng, E.Y.H., and Orr, C.,Jr. Particle Dynamics in
Centrifugal Fields. Powder Technology 1: 3o5-315, 1967.
19. Matteson, M.J., Boscoe, G.F. and Preining, O. Design Theory and
Calibration of a Field Type Aerosol Spectrometer.
Aerosol Sci. 5: 71-79, 1974.
2o. Stober, W. and Boose, ch. Developing Flow and Particle Deposition
in Horizontal Elutriators and Semi-Dispersive Aerosol Centrifuges.
Atmospheric Env. 7: 119-13o, 1973.
464
-------
PART IV
AEROSOL MEASUREMENT
AND ANALYSIS
-------
-------
METHODS FOR DETERMINATION
OF AEROSOL PROPERTIES
RUPRECHT JAENICKE
MAX PLANCK-INSTITUT FtJR CHEMIE
D G5 MAINZ, GERMANY
ABSTRACT
In aerosol research, the measurement of the aerosol size distribution
today is in a transition stage from first generation measurements and
methods to the second generation. The first generation used instru-
ments arid methods which attempted to filter certain size fractions out
of the aerosol. From the data obtained and an assumed rectangular
filter function the size distribution was estimated. In the second gene-
ration, the size distribution is obtained by trial and error methods,
applying any filter function in its exact form without falsifiing assump-
tions. Typical methods - as they are applied in Germany today - are
summarized and additional jnethods for determination of the size dis-
tribution and physical and chemical properties of aerosols are sur-
veyed.
Finally the problem of monitoring aerosols is discussed. It is recom-
mended for monitoring purposes that instruments be selected with fil-
ter functions which simulate typical effects of the aerosol.
467
Preceding page Wank
-------
METHODS FOR DETERMINATION OF AEROSOL PROPERTIES
Ruprecht Jaenicke
Max Planck-Institut fur Chemie, D 65 Mainz, Germany
INTRODUCTION
One of the most essential properties of aerosols is the size distribu-
tion: the particle number concentration as a function of particle size.
It is known, that most observable effects of natural aerosols are in-
fluenced by the size distribution. This can be seen, for example, for
visibility, water adsorption, mass deposition, cloud formation, and
scattering of radiation. In general the effect E is caused by the aero-
sol size distribution n" (r) and some specific weighting functions w (r).
E = / w (r) • n' (r) dlgr (1)
Where r is the particle radius.
The integration over dlgr is prefered, because the size distribution
is usually expressed as the differential distribution of the particle
concentration N:
n'N(r) =
dlgr
In contrast to laboratory aerosols, the size distribution of atmosphe-
"ic aerosols is not only a function of the particle size (given here as
particle radius r), rather it is a function of time (this includes certain
meteorological parameters like relative humidity) and location (inclu-
ding altitude).
If the effect E changes from piece to place, with altitude, or with time,
i: is usually caused by a change of the size distribution n * (r). It,
herefore., was one of the most important aims to measure the size
distribution. All measurements M, however, can only be done the
following way
M = m (r) n ' (r) dlj;T (3)
Each measurement or method exhibits a typical filter function m (r),
d> only some information of n ' (r) is contained in each M. Until re-
463
-------
cently, the measurement of n ^ (r) was accomplished through the sui-
table selection of m (r). m (r) was assumed to be a rectangular func-
tion. Typical examples are cascade irnpactors. This kind of measure-
ments can bo regarded as belonging to the first generation. Today we
are in a transition stage to the second generation of measurements. In
these measurements the m (r) are not selected as rectangular, in-
stead they can have more or less any shape.
The evaluation of n ' (r) is then done by variing n '* (r) in (3) in a fit-
ting procedure, until the measurement M is reproduced. These second
generation measurements of the size distribution do not depend on fal-
sifiing assumptions which had to be made in the first generation. This
paper will review some of the new methods which are being used in
Germany today.
Until now, we know only some of the weighting functions w (r) which
relate effects of the aerosol with its size distribution. This paper will
review some new methods for determing w (r), mostly as physical or
chemical integral properties of aerosol samples.
Since it was realized that the measurement of the aerosol size distri-
bution is too difficult and time consuming to be used for monitoring
the atmospheric aerosol, single parameters have been selected for
this purpose. From equation (3) one can see that such monitoring gives
only limited information about the changes of the aerosol. It will be
discussed how monitoring can be done.
AEROSOL SIZE DISTRIBUTION
AITKEN PARTICLES
The particle size distribution of atmospheric aerosols covers 5 orders
of magnitude, from a radius of 0, 001 ^im to several TOO/am. This
wide range can be observed in polluted aerosols as well as in clean
aerosols. To measure any size distribution the entire radius range
has to be subdivided into small fractions which can be covered by sing-
le methods. As a rule of thumb a single method can cover one order
of magnitude in radius. This rule is valid for natural aerosols with
steep size distributions, but in cases where the size distribution is
comparatively constant, a wider fraction can be covered. Below a
radius of 0. 1 ,um the knowledge of the size distribution in natural aero-
sols is very unsatisfactory. This is because only two methods are
469
-------
available - diffusion and electrostatic precipitation- both of which can
be regarded as poor filters, based on the considerations of equation
(3).
Attempts with a method employing the charging of particles have
shown success for particle radii greater than 0. 01 ^um and concentra-
tions greater than 104 cm'3 (Willeke et al 29).
Integrated Diffusion Analyzer
O £ *3
For the concentration range from 100 cm to 10° cm" and the radius
range of 0. 001 ,um to 0. 1 ,um the problem has been undertaken by
Jaenicke^. An integrated diffusion analyzer in connection with an
electrostatic precipitator was used. The measurement of the particle
concentration is done with Condensation Nuclei Counters "). Whereas
diffusion channels have been used in the past, the present method has
some new features.
To avoid uncontrolled particle losses in connection pipes, the diffusion
system was designed very compactly. The losses that do occur in the
connecting pipes are integrated into the intented loss of particles in
the system. The evaluation of the size distribution from measured
values is done using an iterative fitting procedure of a minimum seech
program. Thus it is no longer required to employ narrow-banded filter
functions in equation (3). It is only necessary to know the weighting
function w (r) whether analytical or stepwise. The size distribution
n"1 (r) described by a number of parameters (whether as a set of nor-
mal distributions or as any other distribution) is varied until calcula-
ted values M^ agree to within the given accuracies with the measure-
CD ents.
Since only a number of 7 diffusion channels is used the evaluated size
distribution is given without fine structures. Size distributions in clean
air have been evaluated as a superposition of two normal distributions,
As opposed to classical ideas, these distributions showed two maxima,,
one KC a radius of 0. 1 um the other close to 0. 02 yum.
Siiniliar instruments have been recently used by A. Schultz,
Heidelberg (personal communication) and Maigne et nl 21.
470
-------
LARGE PARTICLES
Spiral Centrifuge
For the measurement of particles greater than 0. 1 um a spiral centri-
fuge has been developed (Stober et al ^') and will be presented in an-
other paper of the symposium. Spiral centrifuges have been used in
the past and the one developed by Goetz ^ being best known. The dis-
advantages have been discussed by Hochrainer 12. The new develop-
ment shows features avoiding these handicaps. For the best size reso-
lution, the aerosol particles are spread over a large surface. The in-
strument thus is well adapted for polluted or laboratory aerosols with
high particle concentration. In low concentration natural aerosols long
sampling times are required and its use is limited due to the variabili-
ty of the aerosol.
Laser Particle Counter
The size range covered by the spiral centrifuge is larger than that
covered by optical particle counters. Optical particle counters have
been used because of their capability to observe the particles in the
airborne state. In addition optical particle counters have been regard-
ed as absolute instruments in terms of particle concentration. Because
of the cross-sensitivity, however, this certainly is not the case.
(Jaenicke 14).
An optical particle counter using small angle scattering of a laser beam
has been developed by Gebh:a-dt et al ?. The laser beam provides ex-
cellent focussing and thus a sensitive volume as small as 0. 01 mm^
can be obtained. The small sensitive volume makes it possible to count
particle concentrations up to 10^ cm"1^ without dilution and only 10%
loss due to coincidence. Following Jaenicke 14 the size distribution
then can be measured in particle concentrations up to 1000 cm~3. This
is about a factor of 100 more than with other optical counters.
Scattered light is measured between 1. 5° and 7. 5° in the forward
scattering regime where the output signal becomes more or less inde-
pendent of the index of refraction of the aerosol particles. This method
has been used by earlier investigators, although it produces a certain
ambiguity in the range 0. 5 um to 1 ,um. Thus the upper limit of the in-
strument is at a radius of 1 um, while (.he lower limit of 0. 08 /um is
given by the scattered light of the partjple free sensitive volume.
Gebhardt et al ^ give the resolution of jhe optical counter as 10-20%.
-------
Sheet Air hupaetor
Impactors are instruments which are used widely in aerosol physics
and chemistry. The construction is usually very simple and thus the
instrument is well suited for field measurements. In impactors, aero-
sol particles are physically sampled on a substrate and are thus avai-
lable for studies of physical as well as chemical properties. Besides
being used for sampling, the impactor can separate particles due to
inertial characteristics. However, the size distributions derived from
specially designed set ups remained comparatively crude, although
sophisticated evaluations have been applied. (Jaenicke ^, Berner ^).
Zebel et al 31 have introduced a round jet impactor which seperates
the particles on the substrate according to their aerodynamical be-
havior. This is achieved through the use of sheet air. The sheet air
is released at the beginning of convergent part of the nozzle. The
stream is thus focussed hydrodynarnically and the aerosol particles
are separated and deposited a.; rings with he largest - 2 jam - in the
center and those with a radius of 0. 3 um ii a ring of 1. 4 mm diameter.
The resolution of this instrument is not gr\ en, but can be estimated
to be 20% or greater. As with most instru nents with sheet air. the
flow of aerosol to be measured is comparatively small - 3 cm'-'/s -
and thus the instrument is well sxiited for highly polluted or laboratory
aerosols. It was tested first in a coal mine.
Rotating Disc Impactor
Another impactor for size distribution measurements of particles with
radii greater 0. 1 .um has been developed by Schiitz 33 The sampling
plate is shaped like a disc and rotates under a rectangular slit. This
••.Ic-Kign produces high area densities in the center of the disc and low
concentrations at its border. This impactor eliminates the uncertainty
about the optimal sampling time for best evaluation, which usually is
a narrlicap with impactors. A location with optimal area density always
can be found somewhere on the disc. The impactor was designed for
studies of mineral dusi in clean air aerosols, so evaluation by electron
microscopy can be applied.
Optical Instrument Combinalion
The instruments described above determine the aerosol size distribu-
tion. The results obtained for limited size ranges can be compiled to
from the entire size distribution For clean air this has been done by
472
-------
Abel et al * and in polluted air Willeko et al "9 have published results.
The resulting size distributions usually show some discrepancies in
the overlapping ranges of different instruments, since different prope
ties of the aerosol (optical, inertial, geometric) have been measured
and transformed into radii. Attempts have been made to bring results
of different instruments together into a single evaluation program in
order to produce a consistent size distribution instead of a sum of ad-
jecent distributions. (Jaeriicke et al 19). Recently Heintzenberg H has
combined the results of 3 different instruments: an integrating nephe-
lometer (Ahlquist et al ^}, an optical particle counter) and a Concensa-
tion Nuclei Counter*). He expanded the fitting procedure described
above to 3 instruments. A size distribution of 7 normal distributions
(with given and unaltered average radii and standard deviations) is
varied until calculated results of the instruments agree with the mea-
sured values. Heintzenberg ^ showed that the resulting size distribu-
tion depends on the initial values which must be assumed in the fitting
procedure. This ambiguity however, remains within narrow limits, so
useful information of the size distribution in the range 0 03/um to l^
is obtained. The advantage of this method is, that all instruments em-
ployed monitor the aerosols continuously, thus the size distribution
can be calculated for any given time.
VERTICAL AEROSOL PROFILES
Twilight Polarization
Similar optical methods have been applied by Steinhorst ^", however,
he used them for remote measurements of atmospheric aerosols The
depolarization of sky light by aerosol particles has been known for a
long time. The Agung eruption in 1963 brought attention to the fact that
the degree of polarization of red light can be used to investigate the
stratospheric aerosol layer (Shah 24) From measurements of the
degree of polarization during twilight, Steinhorst 26 derived informa-
tion about the particle size distribution in the aerosol layer and the
) ROYCO 225, distributed by Royco, Menlo Park, Calif.
} distributed by General Electric, Pittsfield, Mass
473
-------
vertical profile of the large particle concentration. During twilight the
altitude of the atmospheric layer illuminated by tht sun depends on the
depression of the sun. The information of polarization therefore origi-
nates predominately from a narrow layer at known altitude. Since all
functions used are non linear, he applied a fitting procedure similar
to Jaenicke 15 anc| Heintzenberg 11. AS his analysis shows, the infor-
mation about the particle size distribution is rather vague, but the ver-
tical profile of particle concentration can be determined comparative-
ly accurately.
As opposed to lidar soundings, such measurements can only be perfor-
med in twilight, but the instrumental set-up remains comparatively
small. The evaluation methods applied by Steinhor.st 26 can be used to
derive more information from lidar soundings than is done today.
The most recent vulcanic eruption by the volcano Fuego late in 1974
could be monitored by Steinhorst 26 as the stratospheric layer arrived
over Europe.
From the reasons given above, one can consider the methods used by
Heintzenberg H, Steinhorst 2"} Jaenicke ^^> ^ to measure the size
distribution of aerosols as of second generation type compared to typi-
cal size separation measurements.
PHYSICAL PROPERTIES OF
AEROSOL PARTICLES
Vs equations (1) and (3) show, effects and .measured properties of
aerosols depend on the size distribution as well as physical and chemi-
cal properties of the aerosol. Measurements of physical and chemical
properties depend mostly on the aerosol mass density which is small
compared to that of trace gases and do not exceed 300 jig/m.3 as the
average. Until recently this tiny mass forced such measurements to be
restricted to informations on the total mass.
474
-------
BULK DENSITY
The density of the aerosol particles is of importance for a number of
processes and has influence on the collection efficiency of impactors
as collecting devices. Usually the aerodynamic density, however, is
needed. Measurements of optical properties and conclusions about the
origin of aerosol particles need the knowledge of the bulk particle den-
sity. Efforts have been made to deduce this parameter from other
measurements. (Hanel 1(^).
Gaspycnometer
Recently Thudium 2b attempted to determine the bulk density from the
two parameters defining it: Mass and Volume. The mass of a sample
of atmospheric aerosol particles can be determined very accurately
with commercially available microbalances. The determination of
volumes of the order of 10 mm3 remain difficult however. Thudium 2°
developed a gaspycnometer for small volumes. The instrument mea-
sures the pressure reduction of air in a confined volume due to the
volume of the sample. It was necessary to minimize all volumes used
and measure with :i very sensitive pressure sensor. Exact data are
not published but the confined volume of the instrument is of the order
o o
of 2 cm0. Volumes of 5 mm'5 could be determined within 1 10%. Im-
pactorsamples in atmospheric polluted aerosols showed bulk densities
of 2. 1 g/cni^ at 35% relative humidity. Respective values in continen-
tal clean air were 3. 3 gem . Both values confirm the usual assump-
tion of an aerodynamic density of 1 gcm"^ for irregularly shaped par-
ticles at mean atmospheric relative humidities.
MASS
Spiral centrifuge
The mass-distribution in aerosols was recently measured with a cen-
trifuge from Stober et al 27 ^y Monig et al 22 with a newly developed
direct method. They mounted quarze-crystalls at distinct, locations on
the foil usually used for precipitation of the particles. The change of
oszillatiori of the quarze-crystalls due to mass precipitation was moni-
tored. An accumulation-time of 30 h was estimated for atmospheric
aerosols with 100 ug/m^. The mass of 0. 25 (um particles then can be
determined with 20% accuracy.
475
-------
Intrrferom eter
Tho mass of single particles of 10" ^ g (equiv. of 0. 65 ,um at 1 gem'3
density) was measured by Hollander et al . The system uses an
aerosol particle beam brought into a vacuum chamber impacting single
particles on a thin membrane. The impact causes small vibrations
which are detectable because the membrane is part of an interferome-
ter. The vibrations are converted into deviations of the interference
pattern. This method has another advantage that optical properties can
be analysed simultaneously. The method, however, is restricted to
aerosols which remain stable in a vacuum.
LIGHT ABSORPTION
Sphere - Photometer
Fischer 6 has determined the absorption of light by aerosol particles.
Comparisons between measured and calculated intensities of sky
brightness indicated absorption by aerosol particles which had been
neglected in earlier assumptions. Fischer 6 collected aerosol partic-
les with impactors and measured the absorption in a sphere-photome-
ter. He compared the total scattered light with the light of the incident
beam to evaluate the absorption for wavelengths of 400 - 1000 nm and
found the imaginary part of the complex index of refraction to be
n . K 'v 10~". The aerosol samples showed approximately grey ab-
sorption. As was expected, the imaginary part of the complex index
of refraction is a function of the water content of the sample.
This result effects our understanding of the energy budget of the clo'io-
free atmosphere. It also forces us to assume absorption of energy by
elouddroplets, since the absorption of the cloud condensation nuclei can
be larger than that of the water in the cloud droplets.
SCATTER OF LIGHT
Measurements and calculations in the field of atmospheric optics al-
ways assume the aerosol particles to be ideal spheres. This certainly
is correct if the relative humidity is close to 100% and considerable
amounts of water are attached to the particles. For lower humidities
this certainly is incorrect It was assumed, however, that the huge
number of particles in the volumes under investigation are arranged
randomly, so despite their irregular shape, ideal spheres could be
assumed.
476
-------
Experiments with Microwaves
Recently Zerrull 32 has conducted scattering experiments of micro-
waves ( A = 8 mm) with single particles of a size, that the results can
be compared with Mie-scattering in the atmosphere. The dielectric
constant was selected to be comparable with an index of refraction of
1.5 - 0. 005 i similar to atmospheric particles. For test purposes
ideal spheres were investigated and showed scattering properties as
predicted by Mie-theorie. As nonspherical particles rough spheres,
octaeders, and cubes were investigated. The results showed that ran-
domly distributed nori-spherical particles in a scattering volume can-
not be treated as spheres.
For single particles, the scattering function of non-spherical particles
did not show the typical spikes observed for spherical particles, as
was to be expected. Randomly distributed non-spherical particles
comparable to a polydisperse aerosol did not show the scattering func-
tion of a polydisperse aerosol of spheres. The forward scattering up
to 45° of cubes agreed with that of spheres, but right angle scattering
was a factor of 5 larger than expected and backward scattering was
smaller. This result has large effects on all calculations in atmosphe-
ric optics. It also influences the interpretation of size distribution
measurements with optical particle counters.
CHEMICAL PROPERTIES OF
AEROSOL PARTICLES
In the discussion of the physical properties of aerosol particles, it
could be seen that the investigations have been done on single partic-
les and samples of particles,, depending on the information desired.
Analysis of a single chemical substance can usually be done on single
particles, if the particle can be observed under the microscope. If a
survey of chemical components is conducted, a sample of aerosol par-
ticles is usually required.
477
-------
SULPHATE
Single particle analysis
An example of the first kind of investigation is the work of Georgii et
al . They investigated SO4 ~ (sulphate) in aerosol particles with the
well known barium sulphate reaction in a gelatine-layer. They increas-
ed the sensitivity of the method, so that "sulphate"-particles down to
0. 26 jam radius could be detected. This radius is an equivalent value
and describes the size of an ammonium sulphate ( (NH^o SO4 ) sphere
which contains the same amount of sulphate as the particle under in-
vestigation. Georgii et al 8 collected impactor samples at various lo-
cations and various altitudes in the troposphere and determined the
size distribution of the sulphate particles. They found that about 50%
of the sulphate mass can be associated with the Aitken-particles
(<,0. 1 ,um)
ELEMENTS
X-ray fluoreszence of Aitken particles
Winkler 30 investigated the chemical composition of tiny masses of
Aitken nuclei samples. The samples were obtained with absolute filters
behind an impactor stage used as a precollection device. The impactor
collected all particles greater 0. 2 ;um so only smaller particles were
collected by the filter. The sample was analysed byX-ray fluoreszen-
ce for the elements S, Cl, K, Ca, V, Se and Br. The detection limit
for S was 0. 07 jug and 0. 6 ,ug for Cl. The standard deviation of the an,,-
lysis was around 15%. The results showed that sulpher compounds
- expected as (NH4)2 SO4 - accounted for 90% of the mass, the over-
whelming component in the investigated aerosols.
ORGANIC COMPOSITION
A \ery sensitive method to determine other soluble organic material
in aerosol samples was developed by Ketseridis et al 20. The method
is based on a combination of well known procedures: Classical separa-
tion and thinlayer chromatography together with gas chromatography
and mass spectrumetry. The method is sensitive enough for the deter-
mination of all organic compounds in 100 mg of aerosol particles (witn
about 10% organic matter). In samples from various locations approxi-
478
-------
Experiments with Microwaves
Recently Zerrull 32 has conducted scattering experiments of micro-
waves ( A = 8 mm) with single particles of a size, that the results can
be compared with Mie-scattering in the atmosphere. The dielectric
constant was selected to be comparable with an index of refraction of
1.5-0. 005 i similar to atmospheric particles. For test purposes
ideal spheres were investigated and showed scattering properties as
predicted by Mie-theorie. As nonspherical particles rough spheres,
octaeders, and cubes were investigated. The results showed that ran-
domly distributed non- spherical particles in a scattering volume can-
not be treated as spheres.
For single particles, the scattering function of non-spherical particles
did not show the typical spikes observed for spherical particles, as
was to be expected. Randomly distributed non-spherical particles
comparable to a polydisperse aerosol did not show the scattering func-
tion of a polydisperse aerosol of spheres. The forward scattering up
to 45° of cubes agreed with that of spheres, but right angle scattering
was a factor of 5 larger than expected and backward scattering was
smaller. This result has large effects on all calculations in atmosphe-
ric optics. It also influences the interpretation of size distribution
measurements with optical particle counters.
CHEMICAL PROPERTIES OF
AEROSOL PARTICLES
In the discussion of the physical properties of aerosol particles, it
could be seen that the investigations have been done on single partic-
les and samples of particles, depending on the information desired.
Analysis of a single chemical substance can usually be done on single
particles, if the particle can be observed under the microscope. If a
survey of chemical components is conducted, a sample of aerosol par-
ticles is usually required.
-------
SULPHATE
Single particle analysis
An example of the first kind of investigation is the work of Georgii et
al . They investigated SO4 ~ (sulphate) in aerosol particles with the
well known barium sulphate reaction in a gelatine-layer. They increas-
ed the sensitivity of the method, so that "sulphate"-particles down to
0. 26 ;um radius could be detected. This radius is an equivalent value
and describes the size of an ammonium sulphate ( (NtLjJo SO4 ) sphere
which contains the same amount of sulphate as the particle under in-
vestigation. Georgii et al 8 collected impactor samples at various lo-
cations and various altitudes in the troposphere and determined the
size distribution of the sulphate particles. They found that about 50%
of the sulphate mass can be associated with the Aitken-particles
(<0. 1 ,um)
ELEMENTS
X-ray fluoreszance of Aitken particles
Winkler 30 investigated the chemical composition of tiny masses of
Aitken nuclei samples. The samples were obtained with absolute filters
behind an impactor stage used as a precollection device. The impactor
collected all particles greater 0. 2 ;um so only smaller particles were
collected by the filter. The sample was analysed byX-ray fluoreszen-
ce for the elements S, Cl, K, Ca, V, Se and Br. The detection limit
for S was 0. 07 jug and 0. 6 ,ug for Cl. The standard deviation of the 3: ..-
lysis was around 15%. The results showed that sulpher compounds
- expected as (NH4)2 SO4 - accounted for 90% of the mass, the over-
whelming component in the investigated aerosols.
ORGANIC COMPOSITION
A \ery sensitive method to determine other soluble organic material
in aerosol samples was developed by Ketseridis et al 20. The method
is based on a combination of well known procedures: Classical separa-
tion and thinlayer chromatography together with gas chromatography
and mass spectrometry. The method is sensitive enough for the deter-
mination of all organic compounds in 100 mg of aerosol particles (\vi;:h
about 10% organic matter). In samples from various locations approxi-
478
-------
mately 150 - 200 organic substances could be seen. Of these, up to 50
substances have been identified and quantitatively determined. On the
average in clean air 1 ^ug/m ether soluble organic material was found.
MONITORING OF AEROSOLS
With the concern today about air pollution, attention has been turned to
monitoring and controlling atmospheric aerosols, which means making
measurements as a function of time. For practical reasons, only sing-
le effects or properties can be monitored. A good example for this is
the compilation of total particle concentration or total mass, done by
governmental agencies as part of their air pollution survey. However,
one has to ask what conclusions about the aerosol can be derived from
the monitored parameter. As discussed in connection with equations
(1) and (3), the measured parameter is always an integral over the
size distribution and a certain weighting function. In Jaenicke 16 it was
discussed that monitored parameters only contain information of limi-
ted particle size ranges of the size distribution. Under the assumption
of a continental aerosol size distribution, the total particle number is
affected by particles of 0. 001 ,um to 0. 1 jum radius only. The sampled
mass on the contrary is affected by particles from 0. 1 ^m to 50 jum.
The dry fall-out mass used as a very simple measure in a German
monitoring network, gives information for particles of 10 jum and grea-
ter. The electrical conductivity of air - recently cited by Cobb et al 5
as a measure of air pollution - over ocean areas gives information for
particles in the narrow range 0. 07 (um to 0. 1 um only (the range 0. 01
to 0. 1 jum was assumed by Smic ,
From this, we conclude, that the aerosol can only be meaningfully mo-
nitored by the simultaneous use of a number of different instruments.
If the procedure recommended by Heintzenberg H is used, the size
distribution at any time can be derived. Results of single monitoring
instruments have to be discussed very carefully to draw conclusions
about changes of the aerosol size distribution. Working with single mo-
nitoring instruments there is one possibility for application. Instru-
ments can be selected which have filter functions similar to the weigh-
ting functions of aerosol effects. One example can be given. Breuer et
al4 describe a dust photometer working with scattered light. The filter
function of the instrument is carefully selected, so the measured value
is proportional to the volume of respirable dust retainable or deposi-
479
-------
table in the aveoli of human lungs. The instrument is designed for ae-
rosols with mass concentrations greater than 100^ig/m3. Other in-
struments with filter functions corresponding to the weighting function
of aerosol effects are already known (i. e. visibility, Ruppersberg 23).
In this way, the aerosol and its effects are monitored and conclusions
can be directly related to the influence of aerosols on man and his en-
vironment.
480
-------
REFERENCES
(1) Abel, N. , R. Jaenicke, C. Junge., H. Kanter, P. Rodriguez
Garcia Prieto and W. Seiler. Luftchemische Studien am Observa-
torium Izana (Teneriffa). Met Rundschau. 22: 158-167, 1969.
(2) Ahlquist, N. C. , and R.J. Charlson. Measurement of the wavelength
dependence of atmospheric extinction due to scatter. Atm Env. 3:
551-564., 1969.
(3) Berner, A. Praktische Erfahrungen mit einem 20-Stufen-Impaktor.
Staub. 32: 315-320, 1972.
(4) Breuer, H. , J. Gebhardt, K. Robock and U. Teichert. Fotoelektri-
sches Mefigerat zur Bestimmung der Feinstaubkonzentration. Staub.
33: 182-185, 1973.
(5) Cobb, W. E., and H. J. Wells. The electrical conductivity of ocea-
nic air and its correlation to global atmospheric pollution. J Atrn
Sci. 27: 814, 1970.
(6) Fischer, K. Bestimmung der Absorption von sichtbarer Strahlung
durch Aerosolpartikeln. Beitr Phys Atm. 43: 244-254, 1970
(7) Gebhardt, J. , J. Bol, W. Heinze and W. Letschert. Ein Teilchen-
grofienspektrometer fur Aerosole unter Ausnutzung der Kleinwickel
streuung der Teilchen in einem Laserstrahl. Staub. 30: 238-245,
1970.
(8) Georgii, H. W. , D. Jost and W. Vitze. Konzentration und GrolBen-
verteilung des Sulfataerosols in der unteren und mittleren Tropo-
sphare. Bericht des Institutes ftir Meteorologie und Geophysik der
Universitat Frankfurt/Main, May 1971.
(9) Goetz, A. An instrument for the quantitative separation and size
classification of airborne particulate matter down to 0. 2 micron.
Pageoph. 36: 49-69, 1957.
481
-------
(10) Hanel, G. The real part of the mean complex refractive index
and the mean densitiy of samples of atmospheric aerosol particles.
Tellus. 20: 371-379, 1968.
(11) Heintzenberg, J. , Uber die Bestimmung der Grofienverteilung
luftgetragener Partikeln mit optischen Methoden. Dissertation
Universitat Mainz, 1974,
(12) Hochrainer, D. . On the reliability of measurements with the
Goetz aerosol centrifuge. Atm Env. 6:699, 1972.
(13) Hollander, W. , and J. Schormann. Mass determination of single
aerosol particles by optical interferometry. Atm Env. 8: 817-822,
1974.
(14) Jaenicke, R. . The optical counter: cross- sensitivity and coinci-
dences. J Aer Sci. 3: 95-111, 1972.
(15) Jaenicke, R. . Der Doppelstufenimpaktor, eine weitere Anwendung
des Impaktorprinzipes. Staub. 31: 229-236, 1972.
(16) Jaenicke, R. . Monitoring of aerosols by measurement of single
parameters. Stockholm Tropospheric Aerosol Seminar, University
of Stockholm, 29-31, 1973.
(17) Jaenicke, R. . Size distribution of condensation nuclei in the NE
Trade-wind regime off the African coast. J Rech Atm. Submitted,
1974.
(18) Jaenicke, R. . Comments on "Size distribution of atmospheric
particles" by C. N. Davies. J Aer Sci. Submitted 1975.
(19) Jaenicke, R. , C. Junge and H. J. Kanter. Messungen der Aerosol-
grofienverteilung liber dem Atlantik. Meteor Forsch Erg. B7: 1-54,
1971.
(20) Ketseridis, G. and J. Hahn. Bestimmung der organischen Be-
standteile von Aerosolpartikeln in Reinluft. Z Anal Chem. 273:
257-261, 1975.
(21) Maigne, J. P. , P. Y. Turpin, G. Madelaine et J. Bricard. Nouvel-
le methode de determination de la granulometrie d' un aerosol au
moyen d'une batterie de diffusion. J Aer Sci. 5: 339-355, 1974.
(22) Monig, F. J. , N. Schwarzer und W. Stober. Bestimmung der Aero
sol-Massenverteilung in einer Aerosol-Zentrifuge mit Hilfe von
Schwingquarzen. Gesellschaft fur Aerosolforschung, Jahresbericht
1973, p. 58-61.
482
-------
(23) Ruppersberg, G. H. . Principes et Precedes de Mesure automati-
que de la Visibilite. Bull, de 1'A. I. S. M. 31:11-19, 1967.
(24) Shah, G. M. . Enhanced twilight glow caused by the volcanic erup-
tion on Bali island in March and September 1963. Tellus. 21:
636-640, 1969.
(25) Report of the Study of Man' s Impact on climate (SMIC). Inadver-
tent climate modification. Cambridge, MIT-Press, 1971, p. 209.
(26) Steinhorst, G. . Inversion gemessener Polarisationsgrade zur Be-
stimmung der stratospharischen Aerosolkonzentration. Disserta-
tion Universitat Mainz, 1975.
(27) Stober, W., H. Flachsbart, and C. Boose. Distribution analysis
of the aerodynamic size and the mass of aerosol particles by
means of the spiral centrifuge in comparison to other aerosol
precipitators. J Coll Interf Sci. 39: 109-120, 1972.
(28) Thudium, J. . Entwicklung eines Gaspyknometers zur Volumen-
messung im Kubik-Millimeter-Bereich zwecks Dichtebestimmung
an Proben atmospharischer Aerosolteilchen. Diplomarbeit Univer-
sitat Mainz, 1973.
(29) Will eke, K. , K. T. Whitfey, W. E. Clark, and V. A. Marple. Size
distribution of Denver aerosols - a comparison of two sites. Atm
Env. 8: 609-633, 1974.
(30) Winkler, P. . Chemical analysis of Aitken particles ( < 0. 2/urn ra-
dius) over the Atlantic ocean. Geoph. Res. Let. 2: 45-48, 1975.
(31) Zebel, G. und D. Hochrainer. Zur Messung der Grofiehverteilung
des Feinstaubes mit einem verbesserten Spektralimpaktor. Staub.
32: 91-95, 1972.
(32) Zerullj R. . Mikrowellenanalogieexperimente zur Lichtstreuung
an Staubpartikeln. Forschungsbericht W 73-18, Bundesministeri-
um fur Forschung und Technologie, 1973.
(33) Schiitz, L. . Mineralische Komponente im maritimen Aerosol. In:
Arbeitsbericht des Sondorforschungsbereichs 73 "Atmospharische
Spurenstoffe", 1974, p. 82-85
483
-------
AEROSOL MASS MEASUREMENT
USING PIEZOELECTRIC CRYSTAL SENSORS
Dale A. Lundgren, Ph.D., University of Florida
Lawrence D. Carter, University of Florida
Captain Peter S. Daley, Ph.D., USAF
Piezoelectric crystals, when mechanically stressed, develop elec-
trical charges on certain crystal surfaces. Quartz, as a piezoelectric
material, vibrates at a very precise natural frequency, which can
easily be determined to one part in ten million when the crystal is
placed in an appropriate electronic oscillating circuit. When foreign
material, such as particulate matter, is deposited onto the crystal
active area, the vibrational frequency characteristics of the crystal
change in a predictable and measurable manner. This phenomenon led to
the development of two commercial instruments which were described in
the literature of 1970. These two devices, one based upon particle
deposition by impaction and one by electrostatic precipitation, were
revolutionary because of their extreme sensitivity and rapid, near real
time, response characteristics.
Many studies have been conducted and papers published on the use
and performance of piezoelectric crystal sensors for aerosol mass
concentration measurement. This paper carefully reviews these studies,
defines the capabilities and limitations of the piezoelectric technique,
and discusses recent developments and improvements made to commercially
available instruments.
485 Preceding page blank
-------
AEROSOL MASS MEASUREMENT
USING PIEZOELECTRIC CRYSTAL SENSORS
Dale A. Lundgren, Ph.D., University of Florida
Lawrence D. Carter, University of Florida
Captain Peter S. Daley, Ph.D., USAF
INTRODUCTION
Piezoelectric crystal mass monitors are devices which measure the
mass concentration of an aerosol by depositing particles from a sample
stream onto a vibrating crystal sensor and determining the change in
frequency caused by the additional mass. Since the sample gas volume
is known, the aerosol's concentration can be calculated manually or
electronically.
Currert interest in piezoelectric crystals as aerosol mass concen-
tration measuring devices is based on the properties of extreme sensi-
tivity and rapid response which are inherent in the piezoelectric
method. For example, in 1970 Leemhorst reported a first generation
quartz crystal microbalance that measured the mass of single particles
of lO"11 gram.1 The quantitative output of mass concentration from
crystal microbalances can be nearly real-time; electronic circuits
generally average the output over a 1 or 10 second interval.2 (The
desire for representative average concentrations may dictate aver-
aging readings over time periods of 1 to 10 minutes.)
Along with the potential for precision evident in piezoelectric
mass monitors, several investigators have reported that these devices
are capable of significant errors.3^7 Research, recently conducted at
the University of Florida, identified the sources of these errors and
suggested realistic operating limits for piezoelectric aerosol monitors
in order to reduce the effects of such error sources.6'7
This paper reviews the early developmental work of several
researchers, two commercial first generation devices, analytical work
at the University of Florida, and current work underway on second
generation piezoelectric devices.
486
-------
THEORETICAL CONSIDERATIONS
Due to the temperature and vibratory parameters associated with
various cuts of quartz crystals, type AT crystals are generally accepted
as most appropriate for mass monitors. Such a crystal vibrating in the
fundamental thickness-shear mode has a natural frequency of
a)
Where: N = frequency constant of AT-cut quartz, 0.166 MHz'cm
b = crystal thickness, cm.
Differentiating equation (1) shows how an incremental change in crystal
thickness affects frequency.
df =-2db (2)
b
Dividing equation (2) by equation (1) yields:
df db f^
— --— U)
By definition:
m = pAb (4)
Where: m = mass of vibrating quartz, g
p = density of vibrating quartz, 2.65 g/cm
A = Area of vibrating quartz, cm2
3
Since A and p are constant,
dm = pAdb (5)
so,
db = dm/pA (6)
Substituting equation (6) into equation (3) yields:
487
-------
df = _ dm
f m
Rearranging, we find:
dm m pAb
N
Recalling f = —, we see:
(8)
dm pAN .
introducing a new parameter, Cf, the layer sensitivity constant, such
lhat
c -li
f pN , (10)
it follows that
dm ~ A .
Replacing the differentials in equation (11) with finite deltas is
acceptable if Af and Am are both small relative to f and m. Since C
is a known quantity (C = 2.27f , for AT-cut quartz), we now have the
following working relationship for the change in resonant frequency of
a quartz crystal due to an incremental addition of mass to the vibrating
area
It should be emphasized that the area in equation (12) is the
vibrating crystal area and that practical piezoelectric devices limit
this area to the electrode area. Daley has shown that the crystal and
electrode can be designed to damp out both harmonics problems and
vibration beyond the electrode (facilitating crystal mounting, in the
process) . 7 Hence the vibrating area in a properly designed crystal is
the electrode area. It has also been demonstrated that the deposit
area may be substituted for the vibrating area if the deposit area is
greater than the vibrating area.7
If the deposit diameter is a significant fraction of the electrode
diameter, then theoretical calculations are generally based on the
assumption that the collected mass is evenly deposited over the deposit
area, that the deposit area is a constant, and that the layer of
deposit is thin enough to act as part of the electrode's surface.
Essentially, the latter is the assumption of Sauerbrey who initially
generated interest in piezoelectric crystals as mass monitors.1
488
-------
However, when particles exceed a certain size or when the thickness of
the layer is too great, the deposit begins to move relative to the
electrode surface and the theory does not apply.
Sauerbrey determined that the vibrational amplitude and, therefore,
the mass sensitivity varies across the active (vibrating) portion of
the crystal surface.10- This fact was verified by Daley, whose findings
are presented schematically in Figure 1, together with the dimensions
of a typical crystal.6 This mass sensitivity distribution character-
istic is of greatest importance for impaction devices in which the
deposit area is small compared to the sensitive area.
Mass Sensitivity
Distribution
ELECTRODES
Figure 1. Size and mass sensitivity distribution of a typical crystal,
489
-------
GENERAL STUDIES
Most of the work with crystal monitors conducted prior to 1970
focused on thin film measurement or gas concentration measurement.
King has noted a list of references which deal with quartz crystal
applications ranging from measuring dew points to detection of hydro-
carbons and sulfur compounds.11 He reports thickness gauges measuring
deposited films of 0.1-30,000 Angstroms. He suggests personal monitors
for measuring employee exposure to specific gases. These consist of a
coated crystal carried by the employee all day and plugged into a
circuit for frequency measurement before and after working hours. King
developed COz , H20, and dust monitors to accompany the proposed 1975-
1976 Viking Mars lander.12
Frechette and Fasching have investigated the use of crystals to
monitor 862 concentrations.13 A thin film of styrene-dimethylamino-
propylmaleimide 1:1 copolymer adsorbs SOa • Response was linear from
20-300 ppm and concentrations as low as 0.1 ppm were detectable.
Quite recently, Scheide and Taylor reported a quartz sensor
capable of measuring the concentration of mercury in air from about
0. 1 to over 3 ppb. llf
The work of Warner and Stockbridge is significant for its analysis
of temperature and crystal parameters as they relate to optimal design
of the crystal. l 5
Pulker and Schadler have reported on factors affecting the accu-
racy of thin film measuring devices.16 They also noted the higher mass
rens i tivity of the central portion of the electrode which, as mentioned
earlier, Daley has since confirmed.
WORKING MODELS OF PARTICULATE MASS MONITORS
The work of Chuan8' l 7~2 ° ' 26~2 B (Celesco Industries, Inc.), Olin
:ind Sem'"1"23 (Thermo-Systems, Inc.), and Carpenter and Brenchly3 '2I*
(Purdue University) is of particular importance to piezoelectric mass
sampling because these men constructed working models, two of which
490
-------
became commercially available.
A schematic of Chuan's quartz crystal microbalance is shown in
Figure 2. Particles are collected, by impaction on a sensor crystal.
A reference crystal is exposed to ths sample stream but does not collect
particles. The difference between SiflSQr crystal frequency and
reference crystal frequency is converted to a voltage which is elec-
tronically differentiated to give a mass concentration readout. The
original Celesco device had only one impaction jet. Because of the
relationship between mass sensitivity and position on the electrode, a
small misalignment of the crystal and £b§ impactor nogzle could cause
a significant error. This single nozsli diSlp was abandoned in favor
of a 4 nozzle design. The result was an increased mass response and a
decreased sensitivity to deposit position. The sample rate is 1.5 £pm
with 0.6 £pm flowing thru the impaction jets and the remainder being
by-passed around the crystal,28 When concentrations are low enough
that particles impact at intervals greater than 0.01 second, Chuan
reported the microbalance could detect a single particle mass of 10 ]1
gram.28
1. Air Inlet
2. Air Outlet
3. Test Crystal
4. Reference Crystal
5. Impaction Orifice
6. By-Pass Air Orifices
Figure 2. Schematic of the Celesco quartz crystal microbalance.
This impactor device was used to measure the mass concentration of
effluents from a power plant, mineral wool cupola and electric furnace
(all measurements downstream of collection devices).8' 7 Electron
microscope analysis of the impacted particles showed a size range of
491
-------
0.1-4 ym diameter for the power
the cupola and electric furnace.
size for this impactor is about
data if the upper size limit was
device or the control equipment.
for the electric furnace showed
tests showed collected particles
few exceed 10 urn diameter.
1 8> 19
plant and slightly larger particles for
It should be noted that the 50% cut -
0.55 ym.7 It is not clear from the
due to capabilities of the impaction
Nevertheless, parallel filter samples
comparable results.17 Data from other
as large as about 18 ym; however, very
2 0
Chuan also ran parallel tests of another aerosol with a Hi-Vol
sampler and reported comparable results.8 Unfortunately, size distri-
bution and environmental conditions were not reported with these results.
Volatile particles are known to cause rapid negative then positive
frequency shifts of the quartz crystal as particulate material is first
sensed and then lost by evaporation.18 Chuan reported in 1972 that the
volatile particulate mass concentration could only be measured quali-
tatively by the quartz monitor.
20
The piezoelectric mass monitor developed by Olin and Sem and sold
by Thermo-Systerns, Incorporated (TSI) differs from the Celesco device
in several respects. Most importantly, the TSI unit collects particles
by electrostatic precipitation and its deposit area is significantly
larger than that of the Celesco impactor. A schematic of TSI's Model
3200 is shown in Figure 3. The sample flow rate is 1 &pm and the
crystal's resonant frequency is about 5 MHz (versus 10 MHz for the
Celesco device). TSI reports their device has a particle sensing range
of 0.01-20 ym (diameter) and a concentration range of 1-200,000 yg/m3.2
l. Air Inlet
2. Air Outlet
3. Test Crystal
A. Reference Crystal
5. Precipitator Electrode
6. Aerosol-Corona Contact Orifice
'igure 3. Schematic of the TSI piezoelectric mass monitor.
492
-------
TSI indicates the frequency response to mass is linear until satu-
ration occurs; i.e., until an additional mass deposit produces no addi-
tional frequency response.2 The sensor crystal reportedly can operate
8 hours before cleaning is necessary, when sampling a 100 yg/m3
aerosol.2 The mass sensitivity of the instrument (change in frequency
per unit mass) was reported to be 180 Hz/yg.
TSI reports their precipitator is essentially 100% efficient on
tobacco smoke of 10,000-100,000 yg/m3 concentration and that the major-
ity of the mass of tobacco smoke is in the size range of 0.01-3.0 ym
diameter. Other aerosols in that size range were also reported as
collected with equal efficiency.22 Data for larger particles have not
been published by TSI.
A reported problem with the precipitator design is that the precip-
itator may "manufacture" aerosol by interaction of the corona with
certain vapors (as much as 20 yg/m3 of sampled air).22 Daley did not
find that this occurred. Another problem encountered with the precip-
itator design was inefficient collection of very dry particles, like
sand. In one experiment, poor sensitivity was reported on particles
of 10 ym diameter.22 In TSI's commercial device an inlet air stream
treatment apparatus was added to alleviate this problem. Olin
acknowledges this treatment may aggravate humidity problems (discussed
later).23
Carpenter designed, constructed and calibrated a 4-stage cascade
impactor which used quartz crystals in each stage as mass sensors. In
his calibration experiments, Carpenter noted that the frequency response
(Af/Am) decreased as the mass loading on the crystal increased.
Frequency changes associated with large mass loadings were not
considered accurate measurements of true mass.24
Carpenter was unable to obtain accurate data from his impactor's
first stage. He attributed this error to the fact that the impaction
area exceeded the crystal electrode area. Another possible problem was
the inability of a crystal to sense large particles (>20 ym) with high
efficiency.
EVALUTION OF COMMERCIAL INSTRUMENTS
An in-depth study of piezoelectric mass concentration monitors was
conducted at the University of Florida in 1973 and 1974.7 In this
study two commercially available quartz crystal mass monitors (TSI
Model 3200 and Celesco Model 37A) were the subject of carefully con-
trolled experiments to quantitatively determine the effects of temper-
ature and humidity changes on mass measurement accuracy, linearity of
response, particle collection characteristics, and mass sensitivity
493
-------
tot a number of specific aerosols.
The effect of fluctuating temperature is minimized by the choice
of an AT-cut quartz crystal with a cut angle of 35°14'+1'. (This
optimum cut angle may vary with crystal source). The +!' specification
is within the accuracy of commercial crystal suppliers. Both the
Celesco and TSI devices contained reference crystals to compensate for
temperature (and other) effects. However, the reference crystals
failed to correct for rapid temperature changes since these were effec-
tively damped out as the air passed from test to reference crystal
sensor. Daley found that the addition of metal tubing to the inlet
line provided a heat sink capable of reducing temperature related
errors to 5 yg/m3.7 This represents a temperature change rate of 0.3
and 1 C/minute in the Celesco and TSI units respectively (these are
relatively large change rates which would not normally be encountered
in ambient air sampling). The temperature behavior of beveled, lower
frequency crystals was found to be significantly more stable than
non-beveled crystals.
The Florida study found humidity affected the mass sensing ability
in two ways. First, the crystals increased in mass as they adsorbed
moisture from the air stream. Second, the accumulated particulate
deposit increased in mass by adsorbing moisture. Figure 4 shows the
change in crystal frequency with respect to humidity for the TSI unit
without a reference crystal (Note: TSI supplies platinum electrode
crystals).7 It was possible to maintain almost "new" crystal response
by cleaning the sensor crystal with detergent and distilled water.
The precipitator corona in the TSI unit apparently activated its
sensor crystal surface thereby increasing its hygroscopicity. Since
the reference crystal was not activated, humidity change compensation
was less effective than that theoretically obtainable.7 The crystals
of the Celesco unit were less sensitive to humidity changes; the
hydrophobic crystal coating used to help retain impacted particles was
partially responsible.
Error due to moisture adsorption by the aerosol deposit was found
tc be an order of magnitude greater than error due to crystal adsorp-
tion. This problem is basic to the method and not the device since
the reference crystal clearly cannot compensate. Curve A in Figure 5
shows that for ambient aerosols at high relative humidities, signif-
icant changes in relative humidity can cause large errors.25 Figure 6
indicates how use of a dirty crystal in an environment of changing
humidity may cause large errors.25 For an aerosol of constant concen-
tration the piezoelectric unit indicated high concentration (steep
slope) and no concentration (flat slope) as relative humidity increased
and decreased, respectively. This error was due to water addition and
removal from the deposited aerosol. Clearly, the rapidly
494
-------
40 60
RELATIVE HUMIDITY, %
80
Figure 4. Humidity response of various 5 MH? crystals.
495
-------
I.5--
1.0--
UJ
CO
<
u
CC
O
UJ
O
I
UJ
S 0.5-
cc
MATERIAL
A. ROOM AEROSOL
B. METHYLENE BLUE
C. DIMETHYL POLY-
SILOXANE
LOADING, HZ
2200
900
3000
2'0 4'0 60 80
RELATIVE HUMIDITY, PER CENT
Figure 5. Humidity response of ;latinum electrode, 5 llHz crystals
Traded with various materials.
496
-------
ZD
X
UJ
tU
(t
60 •{
(a) CLEAN CRYSTAL
64 H
62 H
-200 |
* 1
-175 x !|
UJ UJ
H50 £ >
60-
-125
-100
-75
-50
•25
0
UJ
tr
b
UJ
rj
o
UJ
ec
u,
58 -j
56 -
25 20 15 10 5 0
TIME, MINUTES
75
50
-25
0
25 20 15 10 5 0
TIME, MINUTES
HI
x
ul
in
<
UJ
a:
o
UJ
o
o
z
UJ
s
cr
u.
(b) LOADED CRYSTAL (Af%-3IOO H«),
Figure 6. The effect of ^elatjve humidity fluctuations on a piezoelectric
device used in ambient monltorinp.
497
-------
150
140
^ (NH4)2S04
Uronine
Foom Aerosol
CoS04
MeB./Uron.
Dimethyl
Polysiioxone
10 20
30 40 50 60
RELATIVE HUMIDITY, %
80 90
Fi'p.ure 7. Hygrcscopicity of various aerosol deposits.
498
-------
changing relative humidity was responsible. Figure 7 shows the varying
effect of humidity changes on indicated deposit mass.6 Suggestions for
reducing the effects of humidity are discussed later in this paper.
The linearity of response of the Celesco and TSI units was
analyzed by using them to monitor a constant aerosol concentration.
Variation of the readout from actual concentration by >20%, constituted
nonlinear response.
Linearity was found to be a function of particle characteristics.
For grease aerosols, frequency shifted 30 KHz before reaching nonlin-
earity on the TSI unit. However, for a methylene blue aerosol the
shift was only 600 Hz. Figure 8 shows the effect of increased mass
on linearity for a variety of deposits. Tables 1, 2 and 3 give the
linear response limits of several different aerosols.6 Note the
absolute mass limits for the Celesco's impactor (with small deposit
areas) are much lower than for TSI's precipitator, but the mass per
unit area is greater. Daley reports that the sampling periods dictated
by linearity considerations are 0.5 hour and 2 hours for the Celesco
and TSI units, respectively, when sampling a 100 yg/m3 ambient aerosol.7
When linearity is lost it is necessary to cease sampling and clean the
sensor crystal, a relatively simple operation.
01
c
m
u
c
0)
3
er
4)
EXAMPLE
0.2 urn uranine - T
0.3-35 um road dust - C
Outside air - T
0.5 pm methylene blue -
0.2 um methylene blue -
0.08 um polymer beads -
0.2 um NaCl - T
0.5 ym glycerol - T
(T = Thermo-Systems, C = Celescoi
Accumulated Mass
Figure 8. Typical crystal responses to various aerosols' deposits.
499
-------
( — 1
(U
_
id
H
•^
M
§05
3
LQ
1>
(f>
c
O
a,
CO
cu
a
rH
o
in
0
J-4
f3*
0)
CO
^i
CD
in
TJ
,_)
O
04
«•
M
O
•^
c
o
s
U)
CO
id
;£
i-i
CO
£ l~^» tr^ ro ^3* O^
O fNt * * *••*)
H trt g o o rH o o o
•d
M
-P N co o orHinm
(d B • .... i
to .id O fN ro rH ro ro
A
^j
,
g ^o rH ro in o in
• rH (N rH fN CM CM ro CO
trJ g o o o o o o o
(1) "T_l F- 71
10
c
O in in r^ in o m in
in 3.1 CM ro CM i1 in *!• r-
CD ii
a:
^1
rd Nin t— r^oioom
C .V 0 0 OOrHrHrH
•*H il
tr
i a
rH ^ *\
(d -H N
-P-Pffim o ominoo
in-H in o\ rooomw
>i U) «• CM rH (NCMCMfNrH
)-i C >
CJ 0) -P
•fc
-P i-l
•H 0)
in-p oo ro no mro
QiSgro in rrminrrin
(U IT)
Q -rH
Q
0 g
r-- in
•" V( K
•rHDp^g3r--H
M'O'-HrH TJfNn3Xc*p
d) ^3 " m S in
•Pi} cr> • TJ • >, »j
IOHJQ) rHldj-tldP
SOC^ OOOr~
M a)0)»rJL04J(N (1)
rH C K <>1 (0 C
. >,-H « • — M - -H
N£:C N jQrHC
•H4J(dop-HB:xiu nj
>-( CUMfN ±4 "Z,
r^ o
^O
v
VH •*
•H >,
(0 4J
Outside
visibil
in
•
ro
in
•
ro
CO
0
r-
CM
•-J
•«r
in
in
rH
T
vo
m
.
o
0)
c
fd
X fit
o u
rH
•H>n
tn o
>1rH
rH
O II
>,
Dimethy
viscosi
•
ro
A
A
»
O
CM
r-»
o>
o
,_J
CM
0
^
O
cr>
rH
CO
in
CM
.
o
c"e
•H \
c o^
ft) £
M
3 CM
1 •
0) 0
rH *
n 52
OS
(U
Q) CM
rH CM
.C »
-P rH
0) ••
yj
•
rH
ro
•
^T
in
rH
r-
rH
O
•
O
ro
CM
ro
rjl
CM
•
O
tn
e
Cn
e
LO
•H
•
o
^
OP
CM
rH
u
*
o
*
CM
in
•
CT>
00
rH
0
{&
o
•
m
in
rH
*»
vo
f\|
•
o
0)
c
(0
o o
•H<"
U) O
^
O H
Qt
>i
—i jj
Dimethyl
viscosil
(M
*
^i*
A
•
in
rH
A
0
ro
.
vo
o
•
rH
O
CTv
rH
ro
in
lO
.
o
»C
ct
•If
^
Uranine
a
4J
X
(U
•P
fl)
(U
U)
_
TJ
0)
4J
ro
M
Q)
C
Q)
Cn
C
o
U)
•H
rH
rH
O
u
(U
J_l
(U
^
in
rH
O
CO
O
^4
0)
Q)
t)
g
• rH
4-1
tfl
0)
*Ranges
500
-------
XI
id
c
O N
•p all
10
3
4J N
•d 33
10 X
i in
0 M
•P C 0>
"H O ^
e S id
•H rH
1 T
0)
0) CM
c trjg
a
in
01
tr
rH 3-
id
0)
c
•H
r-J N
33
O
tn
HJ -H N
-P -P D3
in -H
£*1 in *•
M C >"
U 0) -P
C/3 -H
^
-P M
•H Q)
in -P
O/I) C^
U/ c
o> nj
Q "H
Q
*•
C
0)
}_l
4-1
OT
>1
rH
o
IX
in
(N
^*
•
in
f~-
^
r-
m
o
m
•
r-
o>
•
rH
O
ro
CM
ro
T
^*
rH
O
10
c
0)
M
-P
01
>i
rH
O
cu
ro
rH
CM
•
*4*
0
TT
in
rH
r-H
m
^*
.
rH
r-
•
n
0
n
(N
^.
n
TT
rH
T
0
a)
c
0)
^|
4J
10
>~)
rH
0
(N
rH in
in
• •
i >i
rH rH
0 0
fXi fa
>1
rH
_jj
c
01
•H
U
-r)
IM
UH
id
HJ »a
id a)
Tl 4--*
u
O 0)
Z -P
0>
Tt
4J
o
z
in if)
V£) CO
IN (N
»* in
m CM
O> He
OO OO
rH P^
1
i-H
^
c
,^
>
rH -O
C 0» (1)
H C C
> 0) 0>
>1 3 rH
rH rH >|
O O ±>
(X -P W
VH
0)
C
£*•<
t—j
Q
a
o
u
Q)
0)
N
C
01
X)
>1
rH
C
0)
•H
u
MH
UH
•a
Q)
^J
u
0)
0)
-p
o
o
0
ro
in
rH
CM
rH
K*I
M
M
O>
XI
rH
rj
e
rH 0)
0) rH
id o
P] P!
>i
r-H
C
0)
•H
O
•H
*4H
ftj
•a
d)
-P
u
0)
•p
0)
n
4J
O
J5
o
o
rn
in
i-H
0
iN
C
01
rH
rH
O
a
•a
0)
0)
3=
id
T)
0)
-P
U
a>
r-t
r-H
0
O
-P
Z
•
4-1
a»
e
id
•*H
-a
c
id
0)
e
in
in
•d
e
x:
4J
-H
s
73
(U
_l_i
X)
•H
V-l
4-1
in
•rH
73
^
rH
r-H
tj
•—;
M
O
C
>i
rH
01
4->
id
E
•r4
X
o
a,
Ou
<
•K
^
O
>*-!
M
0)
4->
0)
E
id
•rH
T)
4J
•rH
in
O
a,
0)
T)
.
in
>
^4
Q)
-P
0)
e
id
-rH
r£)
0)
r-H
O
-M
id
a,
MH
o
4-)
O
r-H
a
E
2
n i
rrj
0)
C
•H
£
0)
*J
0)
T)
• 4J
3- 3
XI
rH
rr-J
>4H 0) •
O M in
3 0)
C W) N
O id -H
•H a> m
•r» 15
•r< 4-1 0)
> o x:
0) Z 4->
TJ+ O
501
o
-------
CU
r—l
x»
E-"
>-,
Li
t
|
3
0)
to
C
o
a
to
cu
rH
O
CO
o
d,
,dj
cu
u
C
(TJ
rH
n)
XJ
o
o
•H
s
J
r~i
-P
CO
>1
U|
CJ
N
-P
Vi
(TJ
3
O
O
U
to
cu
rH
0>
u
C ~
O crig
H al s
.P
(TJ
D
3 N
-P 33
(TJ
1 CO
O M
C CU
-P O >i
•H £ ,
-P
to
rH
o
ex
OA
•
0
in
rH
(N
o
rH
^J*
ro
ro
0
•
o
o
ON
-1
in
V£>
o
CO
rrj
(TJ
O
XJ
cu
C
cu
>_l
>-,
-p
to
rH
0
IX
-!
4J
CO
rH
0
(X
rH fN
A A
0 0
vo o
A ro
A
rH rH
0 0
in
0 rH
tT\ rH
0 rH
O O
• *
O O
in o
(N m
^J<
fN 00
rH rH
to
rrj
(t)
0)
XI
CO CU
'O C
(TJ CU
CU 3
XI rH
o
0) -P
C
CU rH
Ll >i
r^l C
4J -H
10 ^
^1 p^l
rH rH
O 0
f\. f\i
to
rH
(TJ
J_l
CU
-P
(TJ
s
0)
to
M
0)
a
to
•H
T)
rH
O
(X
ro
A
O
0
^O
A
•
.
in
o
33
04
OP
r-
1 fN
cu
3 *
rH rH
x> ••
cr»
CU
C *
cu cu
rH C
XJ G
P IT)
CU Lt
IS 3
CM r^
11 A
O O
in o
(*™l ^J*
A
• •
ic£ i
•H X rH
0) 0)
0) 10 -H
CO rH 4J
3 03 O
•H CU
tr LI a
a cu to
\-P 0)
N ro M
33 g
«».
O rH 10
O -H ^J
00 (TJ -H
fN 0)
Ll O
u-i o a
CJ *4~( fi)
g
•H g
> "T
•H • CO
-P O •
•H O
CO 4-1
C O T3
0) C
10 Ll (TJ
CU
rH P in
(TJ CU fl
•P £ •
CO (TJ O
>i "H
Li T3 T3
U (TJ
•P x:
•H
•• to x:
woo
O CU X3
z Q :*
502
-------
Daley found particle collection characteristics to be an important
consideration. Obviously, a particle which is not deposited on the
crystal will not be weighed. Less obvious, but also important, is the
need for particles to be uniformly deposited over the deposit area if
the deposit size is significant compared to the crystal size.
The Celesco device collects particles by impaction. A 50% cut
size diameter of 0.55 ym was calculated and empirically verified for
unit density spheres.7 This lower size limitation must be recognized
by users of the Celesco unit.
The TSI unit was found to have high inlet passage losses (20%) of
small particles (0.2 ym) because of precipitation by the field around
the precipitator electrode shaft. The actual deposit is somewhat
smaller than the active area resulting in a higher than expected mass
sensitivity. These factors, therefore, compensate for each other;
however, the user must be aware of and utilize these facts to obtain
accurate results.
If the deposit area is a significant fraction of the active area,
a change in deposit area causes a change in the sensor's mass sensi-
tivity. The mass sensitivity of a crystal is the change in frequency
per unit change in mass (Af/Am). Tables 1 and 2 show the change in
deposit area and mass sensitivity observed for various aerosols in the
TSI unit.6 Using a single value for crystal sensitivity, therefore, results
in error. This error can be reduced by actually measuring the deposit
area and calculating mass sensitivity.
It should be noted that large particles (>5 ym) collected by the
TSI unit interfered with the collection of smaller particles. Further,
these large particles are not reliably measured, as is seen in the next
paragraph. (Daley suggests eliminating them with an inlet cyclone.)7
Figure 9 shows the sensing ability of the two devices for spherical
particles with diameters between 0.1 and 12 ym.6 The sharp drop in
sensing ability for larger particles is evident. (Qualitative evidence
indicated that sensing ability for larger particles is improved in poly-
dispersed aerosols.) Hence the large particle problems of "bounce off"
and inlet loss are magnified by the inability of the crystals to detect
large particles even when they are properly deposited.
The California Department of Health tested the TSI Model 3200 A
(with auxiliary pumps and electronics) in parallel with high volume
samplers and found poor agreement for ambient air monitoring.5 Daley's
analysis of these results indicated that lack of agreement resulted
mainly from sampling beyond the linear limit of the quartz crystal
device.
503
-------
150 •
>; ioo
0
0
jo
iS
O)
c
g 50-
«
(0
0
0
O TSI
0 A TSI(coaUd)
0 O Celtsco
O
o D 8
\
\
\
\^
1 1.0 10
Figure 9.
Particle Diameter, pm
Particle sensing ability vs. particle size.
Tests by the EPA of a TSI device designed for measuring automotive
exhaust had similar results. The TSI monitor mass determinations
differed from filter measurements by +30%."* The possible sources of
error were several with the high moisture content of the gases most
suspect. Temperature changes and the presence of volatile organics
were other possibilities.
These three studies'*'5'7 may suggest that piezoelectric mass
monitors are too error prone to be of significant value. However, this
is not true. The latter two reports discussed above can be somewhat
discounted because they show only that the piezoelectric monitor cannot
do everything. Daley's work is more representative of reality. The
devices do work, and they are quite accurate, if they are not taken
beyond their ranges of linearity (short sample periods) or collection/
detection ability (particle size range), and if factors such as humid-
ity changes and the presence of volatile aerosols are considered and
minimized. These are a lot of if's, but the/ apply to first genera-
tion devices, which may simply be unsuited for many applications. We
shall next consider second generation devices being built by Celesco
and TSI which may alleviate some of the problems encountered with their
predecessors.
504
-------
RECENT DEVELOPMENTAL WORK
Two companies, Thermo-Systems, Inc. and Celesco Industries, Inc.
have or are now developing second generation piezoelectric mass monitors.
Celesco is developing a ten-stage impactor which will provide aerosol
mass concentration and mass distribution data over a reported size
range of approximately 0.05 to 100 ym diameter.26 A six-stage abbre-
viated version of the above device has been built and tested on ambient
California aerosol.27 Performance of the unit's first two stages, with
particle 50% cut sizes of 53 and 25 ym, may be suspect until the unit's
ability to efficiently collect and sense large particles is convincing-
ly demonstrated. Also, a low sampling rate coupled with a low concen-
tration of large particles results in statistical problems associated
with the large-particle collection frequency. However, discrete large
oarticle impacts on a given stage may be counted if particle mass cannot
be directly and accurately sensed because of poor particle adhesion.
The multi-stage impactor with near real time aerosol mass distribution
determination ability will certainly find many interesting applications,
TSI and a Japanese firm have developed an improved version of the
original TSI mass monitor.2 The new device continues to utilize an
electrostatic precipitation collector. Basic information about the
unit is still unpublished and proprietary, but the design appears to
have eliminated several earlier problems. For example, the device is
designed for a more specific purpose, measuring indoor particle concen-
trations, so the aerosol size distribution and concentration range is
somewhat restricted. This restriction enables a more accurate mass
sensitivity calculation. Large particles (>10 ym) which previously
interfered with small particle collection, and which were not accu-
rately detected anyway, are removed from the inlet air stream by an
inlet impaction stage. The dust monitor's metal tubing Inlet should
dampen out error causing, short term temperature variations. Short
sample periods (2 minutes) are recommended; hence, linearity problems
should be resolved. A built-in crystal cleaner is included, and should
prevent the need for frequent crystal changing. Other potential error
sources include the crystal's ability to accurately sense particles in
the 3 to 10 ym diameter size range, wall losses associated with
particles in the 5 to 10 ym size range, the humidity fluctuation
problem noted with earlier models and the precipitator's reported
ability to "manufacture" particulate matter.
A possible new error source may be associated with the crystal
wash solution leaving a volatile or hygroscopic residue, the effect of
which may be amplified by short sampling periods. A new crystal
mounting technique, epoxying the crystal's total perimeter to a base,
departs from previous "clip mounts", but should not contribute to error
if correctly designed.
505
-------
In six parallel filter vs. piezoelectric instrument tests of .1
laboratory aerosol, a prototype dust monitor was reported to have- nn
average error of about 10% as compared to a 0.8 vim pore size Nucleoporc
filter.29 The aerosol size distribution was not stated but the aerosol
concentrations were between 30 and 100 yg/m3. Obviously it is desirable
to conduct more extensive testing with various aerosols.
It should be noted that the model tested was a prototype for a
possible commercial model and the effect of any design changes is of
course unknown.
DIRECTIONS FOR FUTURE WORK
Piezoelectric crystal aerosol mass concentration monitors have
tremendous potential in research and in field operations. Several
improvements to early units have been made and were discussed earlier.
A few other basic design modifications or suggestions, perhaps simpler
in concept than execution, are set forth below.
With existing designs, the operator has only one means to compen-
sate for high aerosol concentrations, i.e., reducing the sample time.
(A noteworthy exception is the TSI accessory dilution system wbich
offers a 50% or 80% aerosol dilution ability.) With high dust concen-
tration, crystal response becomes nonlinear early in the sampling run.
A dilution apparatus, perhaps with fixed dilution ratios of 2 to 1,
5 to 1, 20 to 1 and 100 to 1, would facilitate a large operating range
for one measuring device so that the entire ambient air concentration
range encountered could be covered.
Ambient air dilution with a temperature and humidity controlled
air stream has several additional advantages such as reducing the
effect of humidity variations in the sample gas stream. For example,
diluting ambient air at 80% relative humidity (RH) with an equal volume
of dry air will produce a mixture at 40% RH. Referring to Figures 4,
5 and 6 it becomes apparent that the lower the relative humidity, the
less the tendency for water adsorption by the crystal or the deposited
sample, therefore the lower the relative error resulting from humidity
changes. Dilution can also be used to stabilize temperatures or to
reduce the temperature of hot effluent gases.
Humidity effects could also be reduced by utilizing a crystal with
a relatively flat temperature-frequency response at moderately high
temperatures. This technique could be used to study or determine the
volatility of aerosols by heating the sampled air stream to various
temperatures (in order to drop the relative humidity) and then measuring
the change in aerosol mass. TSI has utilized the principle of gas
sample heating for a source sampling instrument.
506
-------
Sampling of volatile aerosols, presently a rather difficult propo-
sition, should be considered via the piezoelectric approach. Electronic
circuitry which would filter the plus-minus signals of deposited and
evaporated volatiles, and then sum the magnitude of plus-minus frequen-
cy changes, could measure total volatile mass associated with larger
particles which were individually deposited on the sensing crystal.
Cascade impactors have several interesting possibilities. It may
be possible to calibrate large particle impaction stages with inlet
loss and sensing efficiency factors in order to accurately sense large
particles (5-100 ym). If a lower size limit of 0.05 pm is obtainable
(through low pressure impaction techniques or by other means) the
staged impactor would effectively cover the particle range of primary
interest to most aerosol investigators. The multi-stage impactor could
further be optimized by adjusting the number of impaction jets per
crystal for the anticipated mass loading for that stage. This would
maximize sensitivity vs. sampling time but would require prior knowledge
of the intended use of the monitor.
Apparently, the use of crystal sensors in other size selective
particle collection devices, such as those involving centrifugal force
or electrical force, has not been exploited. Crystal sensors also
have potential use in thermal precipitators, diffusion batteries, and
other special particle collection or sampling devices because of their
extreme sensitivity. Although operational problems have made the first
generation piezoelectric crystal mass monitors much less than univer-
sally acceptable, none of the problems encountered are insurmountable.
Therefore, it must be concluded that the greatest potential use of
crystal sensors remains to be found.
507
-------
REFERENCES
1. Leemhorst, J.W. Direct Measurement of Particulate Mass and Humidity.
Contamination Control. 4:11-14, July/August 1970.
2. Thermo-Systems, Inc. Particle Mass Monitor, Instruction Manual for
Thermo-Systems Models 3200(A), 3200(B), 3205(A), 3205(B). St. Paul,
Thermo-Systems, Inc.
3. Carpenter, I.E. , and D.L. Brenchley. A Piezoelectric Impactor for
Aerosol Monitoring. Am. Ind. Hyg. Assoc. J. 33:503-510, 1972.
4. Herling, R. , W. Karches, and J. Wagman. A Comparison of Automotive
Particle Mass Emissions Measurement Techniques. (Presented at
Central States Meeting of the Combustion Institute, University of
Michigan. Ann Arbor. March 23-29, 1971.) 23 p.
5. Imada, I., and P.K. Mueller. Evaluation of a Piezoelectric Quartz
Crystal Microbalance for the Continuous Measurement of Aerosol.
Air and Industrial Hygiene Laboratory, Calif. Dept. of Public
Health, Berkeley, Calif. AIHL Report No. 114. 1971. 17 p.
6. Daley, P.S., and D.A. Lundgren. The Performance of Piezoelectric
Crystal Sensors Used To Determine Aerosol Mass Concentrations.
(Presented at American Industrial Hygiene Conference, Miami. May
12-17, 1974.) 41 p.
7. Daley, P.S. The Use of Piezoelectric Crystals In the Determination
of Particulate Mass Concentrations In Air. Doctoral Dissertation,
Gainesville, Univ. of Florida, 1974. 189 p.
8. Chuan, R.L. Application of an Oscillating Quartz Crystal To Measure
the Mass of Suspended Particulate Matter. Celesco Industries, Inc.,
Costa Mesa, Calif. AT-159. 10 p.
9. Chuan, R.L. Particulate Mass Measurement by Piezoelectric Crystal.
Nat'l. Bureau of Standards, Gaithersburg, MD. Special Publication
412. October 1974. 12 p.
508
-------
10. Sauerbrey, G.Z. Verwendung von Schwingquarzen sur Wagung Junner
Schichten und zur Microwagung. Zeits. Phys. 155:206-222, 1959,
11. King, W.H., Jr. Using Quartz Crystals as Sorption Detectors, Parts
I and II. Research/Development. 20:29-33, April 1969, and 20:28-
, 30, May 1969.
|
t 12. King, W.H., Jr. The State of the Art in Piezoelectric Sensors.
In: Proc. of 25th Ann. Symposium on Freq. Control, U.S. Army
Electronics Command. Ft. Monmouth, 1971. 7 p.
i
13. Frechette, M.W. and J.L. Fasching. Simple Piezoelectric Probe for
. Detection and Measurement of S02. E.S.&T. 7:1135-1137, December
' 1973.
* 14. Scheide, E.P., and J.K. Taylor. Piezoelectric Sensor for Mercury
in Air. E.S.&T. 8:1097, December 1974.
15. Warner, A.W., and C.D. Stockbridge. Mass and Thermal Measurements
With Resonating Crystaline Quartz. Vacuum Microbalance Techniques.
2:71-92, 1962.
16. Pulker, H.K., and W. Schadler. Factors Influencing the Accuracy of
a Quartz Crystal as a Thickness Monitor for Thin-Film Deposition.
II Nuovo Cimento (Italy). 578:19-24, September 11, 1968.
17. Chuan, R.L. Measurement of Particulate Pollutants in the Atmosphere.
(Presented at Joint Conference on Sensing of Environmental
Pollutants, Am. Inst. of Aeronautics and Astronautics, et al.,
Palo Alto, November 8-10, 1971.) 6 p.
18. Chuan, R.L. Electron Microscope Analysis of Particulate Sample
From Rocket Launch at Cape Kennedy. Celesco Industries, Inc.,
' Costa Mesa, Calif. AT-155. November 9, 1972. 3 p.
, 19. Chuan, R.L. Aerial Sampling and Post-Flight Analysis of Partic-
ulates Over the San Juan Basin at Four Corners. Costa Mesa, Celesco
Industries, Inc., 1972. 8 p.
I
20. Chuan, R.L. Measurement of Vertical Temperature and Particulate
Distributions Over the San Gabriel Valley. Celesco Industries, Inc.,
Ccsta Mesa, Calif. AT-154. November 1972. 35 p.
21. Olin, J.G., and G.J. Sem. Piezoelectric Microbalance for Monitoring
the Mass Concentration of Suspended Particles. Atmospheric
Environment. 5:653-668, 1971.
509
-------
22. Olin, J.G., G.J. Sem, and D.L. Christcnson. Piezoelectric-Electro-
static Aerosol Mass Concentration Monitor. Am. Ind. Hyg. Assoc. J.
32:209-220, April 1971.
23. Olin, J.G. Design and Operation of a Piezoelectric-Electrostatic
Particle Microbalance for Automatic Monitoring of the Mass Concen-
trations of Air-Borne Particles, Paper 71-558. (Presented at
Instrument Society of America International Conference and Exhibit.
Chicago. October 4-7, 1971.) 10 p.
24. Carpenter, I.E. The Design, Construction, and Calibration of a
Piezoelectric Cascade Impactor for Monitoring Aerosols. Master's
Thesis, Lafayette, Purdue University, 1972. 104 p.
25. Daley, P.S. Real Time Aerosol Mass Concentration Measurement:
Capabilities and Limitations of the Piezoelectric Microbalance
Technique. (Presented at International Conference of Human
Environment Conservation. Warsaw. February 1974.) 26 p.
26. Chuan, R.L. An Active Cascade Impactor for Real Time Sizing of
Airborne Particulates. Celesco Industries, Inc., Costa Mesa,
Calif. AT-149. 3 p.
27. Chuan, R.L. Recent Test Results With an Active Cascade System.
Note accompanying correspondence to D.A. Lundgren. December 12,
1974.
28. Chuan, R.L. Application of an Oscillating Quartz Crystal to
Measure the Mass of Suspended Particulate Matter. (Presented at
165th National Meeting of the American Chemical Society. Dallas.
April 1973.) 37 p.
29. Tsurubayashi, K., and G.J. Sem. Personal Communication. Nihon
Kagaku Kogyo Co., Ltd., Osaka, Japan. Thermo-Systems, Inc.
St. Paul.
510
-------
MEASUREMENTS OF AEROSOL OPTICAL PARAMETERS
A. P. WAGGONER
R. J. CHARLSON
UNIVERSITY OF WASHINGTON
SEATTLE, WASHINGTON 98195
ABSTRACT
An aerosol is most usefully described in terms of its integral proper-
ties, in particular those integral properties that directly describe
the aerosol effect of interest. Atmospheric optical properties nor-
mally considered would include: visibility, scattering extinction co-
efficient, absorption extinction coefficient, hemispheric backscatter
coefficient and the ratio of hemispheric backscatter to absorption co-
efficients.
These parameters can be calculated via Mie solutions if the aerosol par-
ticle size distribution and complex refractive index are known or
assumed. These calculated values of integral aerosol properties will
contain errors that arise from an inaccurate description of the aerosol
particles. The calculations assume: (1) the particle size distribu-
tion is accurate, (2) the particles all have the same complex refractive
index, (3) the complex refractive index is uniform within each particle
and (4) the particles are spherical in shape. Most of these assumptions
are poorly justified.
Techniques have been developed at the University of Washington for dir-
ect measurement of the integral aerosol optical properties listed in
paragraph one. This paper describes these techniques and measured
values of these aerosol optical parameters.
511
-------
MEASUREMENTS OF AEROSOL OPTICAL PARAMETERS
A. P. WAGGONER
R. J. CHARLSON
UNIVERSITY OF WASHINGTON
SEATTLE, WASHINGTON 98195
I. INTRODUCTION
The aerosol is composed of particles that range in size from smaller
than 0.01 pm to larger than 10 urn diameter. The particles are of
various chemical compositions and each particle can be a mixture of
substances or a single substance. The integral optical effect of the
aerosol particles is dependent on all of these parameters. Atmospheric
optical properties normally considered would include those of interest
from a human impact standpoint, i.e., visibility and colored haze, and
those of scientific interest, i.e., scattering and absorption extinc-
tion coefficients.
II. ATMOSPHERIC OPTICS AND VISIBILITY
It is convenient to define several parameters commonly used to describe
atmospheric optics.
The extinction coefficient bext of a real atmosphere defines the change
in intensity of light traversing a pathlength Ax by the Beer-Lambert
law:
^ = -b Ax (1)
I ext
b is the sum of two terms:
ext
b = b (eases) 4- b (particles)
ext ext & ext ^
b (gases) = b,, + b , where
ext VB Rg ag'
b Ax is the fraction of incident light scattered into all directions
by gas molecules in Ax.
b Ax is the fraction of incident light absorbed by gas molecules
ag in Ax.
512
-------
Our interest is in b (particles) which can be broken down as follows;
GX L
b (particles) = b + b (2)
ext ap sp
where b Ax is the fraction of incident light absorbed by particles in
; aP Ax.
b Ax is the fraction of incident light scattered into all direc-
^ tions by particles in Ax.
' The observer visibility, or visual range, is that distance at which a
\ black object can be just discerned against the horizon. Koschmiederl
1 showed that a turbid media, such as urban air, reduces the contrast
(ratio of brightness of an object to the horizon brightness, minus one)
of distant objects as given by
P
-b Ax ,
pvf~ 2.
C = C e exL (Middleton ) (3)
o
where CQ and C are the contrast relative to the horizon of an object
at zero distance and at distance x. A black object has a Co of -1.
Experiments have determined that typical observers can detect objects
on the horizon with a visual contrast of 0.02 to 0.05. Assuming hori-
zontal homogeneity of aerosol properties and illumination and a 0.02
detectable contrast, the visible range is
"v - ¥-
ext
For a contrast of 0.05,
L -0- (5)
V
Usually the assumption is made that b = b
ext sp
b can be calculated from known or assumed aerosol particle size
distribution, concentration and refractive index, as discussed below.
III. PARTICLE OPTICS
The atmospheric aerosol is composed of particles that range in size
from smaller than 0.01 ym to larger than 10 ym diameter. The particles
are of various chemical compositions and each particle can be a mixture
of substances or a single substance. The integral optical effect of
the aerosol particles is dependent on all of these parameters.
513
-------
The scattering coefficient of a single particle divided by the particle
volume is shown in figure 1. The value of bsp is the product of the
curve in figure 1 times the particle volume distribution function. The
aerosol particle volume per log radius interval usually is similar to
that of figure 2, bimodal with the two volume modal diameters about
0.6 Mm and 10 ym. The calculated optical scattering per log radius
interval is predominately due to particle volume in the 0.1 ym to 1.0
diameter range, as shown in figure 2. In all the measurements we have
made, the particles in the 0.1 to 1.0 decade dominate scattering extinc-
tion in the visible spectrum although there clearly are cases in fogs,
rain, snow, clouds and dust storms in which large particles influence
or dominate visible extinction.
The correlation of bgp, measured with an MRI 1550 nephelometer, and 0.1
to 1.0 ym diameter particle volume, measured using an electrostatic
mobility and single particle optical counters from Thermo Systems, was
0.95 at various locations in the Los Angeles basin. These measurements,
shown in figure 3, are from the 1973 State of California Air Resources
Board ACHEX^ program.
The wavelength dependence of bsp depends almost exclusively on particle
size distribution'. If the wavelength dependence is described by a
simple power law:
b (A)«A~a (6)
sp
where a is an experimentally determined exponent, the usual range of a
is 0.5 to 2.5 with occasional measured values of a in the -1 to 0 range
at background sites. Rayleigh scattering always occurs simultaneously
and has a wavelength dependence that is similar:
bRg - A'4 (7)
As a result, blue scattered light (against a dark background) or red
transmitted light (from the sun or a bright white object) is no indi-
cation by itself of the presence of particles. Whether bsp or bRg
dominates is determined by the amount of particulate matter that is „
present. In remote, clean marine locations at sea level, Porch, et al
showed that bsp <_ bRg at 500 nm. In continental, low altitude sites,
bsp is usually larger than b^g, so that such hazes can often be assumed
to be dominated by bsp. However, clean arctic air intruding or air from
aloft subsiding into mid continent cities occasionally produce bsp
-------
SCATTERING COEFFICIENT PER VOLUME
12
CM
-------
o
\
J
o
LU
_J
O
h-
o:
2
cc
LU
CO
ID
o:
MEASURED
0.01
O.I
I
POMONA 21:40
10-5-72
1.0
10
cr
LU
0
CO
o
o:
LU
CO
5
Q
CC
CD
0.01
— CALCULATED
O.I 1.0
PARTICLE DIAMETER,
Figure 2. Top: Aerosol particle size distribution measured at Pomona
during 1972 State of California Air Resources Board 'ACHEX
program.
Bottom: Calculated optical scattering by particles, bsp,
for measured size distribution.
516
-------
00
H
S3
M
0
t-«
CK
(x<
W
8
§
M
S
O
CJ
o
O) 0)
t/> i.
O v- w H:
3 O)
I/)
o s-
1- CJ
o; -4->
ra <1)
•O rO
•r- O
S- •
o s-
4- Q
fd jQ
O
•o
4- O)
O T-
S- -O r-
a.
Q.
2
w e
(O 3.
a> o
E •
r—
lf_
O O
fO
•»->
to
W»
«4- (O
O 3
O i— i- •«->
r— • fO n3
a. o 0.0
o
o
CM
o
o
'un/A-
517
-------
Blue hazes - for example, in mountainous areas - may or may not be due
to scattering by particles, depending on viewing conditions (e.g., dark
or light background) and the distance from the observer to the back-
ground. If the product of b^R times distance to the background is much
above two, the blue haze has a significant input due to bg«. On the
other hand if the product of b times distance is of this magnitude,
then the haze is likely to be due to particles. Since b^g, 530 nm =
0.15 x 10~4m~ , if b ^ 0 mountains should not appear to be behind
a haze if they are within 10 km. or so. They will, however, appear
hazy if the distance is much more than 100 km. due to the omnipresent
scattering by gas molecules. Conversely, if such a distant mountain
is not visible at all, b » b^g and the haze is due to particles.
When viewing bright objects (the sun and moon, sunlit snow capped peaks
and cumulous clouds) hazes with 1 <_ a _<_ 2 of sufficient optical depth
cause the color to be reddened^'^. The color thus produced is remark-
ably similar to that observed through an optically thin layer of NC>2^
so that the presence of color thus viewed is no proof of the existence
of N02- To further complicate this issue, Husar-'--'- has shown that light
scattered in the backward hemisphere calculated from typical measured
size distribution is enriched in the red wavelengths also causing the
haze itself to appear reddened. In forward scatter this same haze
appears white. Charlsonl^ showed that, in perhaps 20% of the measured
cases during August, 1969 in Pasadena, CA, there was enough N02 to
influence the coloration of white objects viewed through the haze and
that the remaining cases particles dominated the wavelength dependence
of total extinction (bext).
IV. MOLECULAR COMPOSITION
The particle interaction with water, biological effects and complex
refractive index depend on the molecular composition. Therefore, it is
important that the composition of various aerosol systems be classified,
particularly insofar as this determines the imaginary part of the re-
fractive index and hygroscopicity. Unfortunately, this is an area in
which so far very little work has been done. Rasmussen^^ suggested
that organic materials (terpenes) are a major source of atmospheric
particles, but did not quantify their work adequately for application
to optics. The reaction products of S02 with water and ammonia have
been shown to play an important part in urban and rural aerosols by
Junge-^ although he did not attempt to relate quantitatively the com-
position with optical effects. We have data from rural Missouri sug-
gesting that continental aerosol optics is often dominated by t^SO^ and
the products of its neutralization with NH^-'-^' -"-".
518
-------
The molecular nature of individual particles is a function of the
source and removal mechanisms for these particles. The most important
observable effects of composition on particle optics are the relation-
ships of bsp and relative humidity and the complex refractive index.
V. RELATIVE HUMIDITY EFFECTS
The humidity effects in aerosol optics fall into three categories:
RH £ 100%: particles between and above water cloud
(including high RH hazes);
RH > 100%: unactivated particles in water clouds and fog;
RH > 100%: activated cloud droplets.
Our efforts have been limited to the first case and are discussed in
the following paragraphs.
Since a large fraction of submicrometer particles are hygroscopic or
deliquescent^-lo the size distribution of an atmospheric aerosol and
hence its optical or climatological properties, depend largely on
relative humidities, even at RH < 50%.
First, light scattering always increases with humidity, although for
relatively hygrophobic systems the increase may be very slight up to
extremely high RH. While for most aerosols such as ^SO^ droplets the
curve increases monotonically, definite inflection points due to deli-
quescent salts are seen at some locations indicating i;he dominance by
rather pure inorganic substances such as (NH/JoSO/. or sea salt
(NaCl)15,16,19.
A system has been designed and operated by this laboratory that (over a
period of about 120 seconds) sweeps the relative humidity of air contain-
ing aerosol particles from 30% to 95%. Changes in particle diameter
are detected as changes in the scattering coefficient of the aerosol
particles15 >16>19.
In the midcontinent region 30 km southwest of St. Louis, this system
detected H2S04/(NH4)HSO^/(NH4)2S04 as dominate materials in the 0.1 to
1 ym decade of aerosol size. Injection of sub ppm concentrations of
NH-j converted the bgt)(RH) response characteristic of ^$04 to that of
(1^4)2504. The (^4)2804 is detected by comparing the value of rela-
tive humidity at the deliquescence point for the unknown sample with
that of laboratory generated (Nlfy^SO Aerosol. 98% of the time either
H2S04 or (NH/KSO, was the dominant substance in terms of optical
effectl5,16. * ^ '
-------
VI. TECHNIQUES FOR MEASUREMENT OF RELEVANT OPTICAL PROPERTIES
In the past several years our efforts have been focused on design and
testing of methods to measure aerosol otpical properties that directly
determine aerosol radiative interactions. Methods for measurement of
these relevant integral aerosol optical properties; namely, bsp, bosp,
b (RH), and b , are described in the following sections.
A- bsp
Consider a small volume of thickness dx illuminated by a parallel beam
of wavelength A and intensity IQ A. For unpolarized light, the inten-
sity of light scattered into solid angle dfi at scattering angle 0 is
dIA
~ (0)d = IQ ABA(0)dx (8)
A visibility meter using the operator's eye as a detector was devised
by Buettell and Brewer^O that geometrically performs the integration of
8, (0) over solid angle to measure b 1.
A Sp , A
The geometric errors of the instrument have been studied by Middleton ,
Ensor and Waggoner^3> Heintzenberg and Quenzel^ and Rabinoff and
Herman ^ and are estimated to be 10% or less for the aerosol particle
size distributions normally found in the atmosphere.
21
Ahlquist and Charlson increased the original instrument sensitivity
by using a photomultiplier tube to detect scattered light from a xenon
flash lamp. Ahlquist, et al.^2 improved the sensitivity, stability and
dynamic range by substituting an incandescent lamp for the xenon flash
lamp and detecting the scattered light using digital photon counting
techniques. This insturment, called an integrating nephelometer, is
shown in Figure 4. Modern versions of Buettell and Brewer's device
have sufficient sensitivity to be calibrated in an absolute sense with
b^g, the scattering coefficient of particle-free gases such as He, C02,
CC,!2F2.
The modern high sensitivity instrument is alternately filled with ambi-
ent and particle-free air and the difference in scattered light inten-
sity is proportional to the scattering extinction coefficient due to
aerosol particles, bsp. The instrument calibrates itself by inserting
3 white object (end of a wire painted with Kodak optical white paint)
Into the scattering volume during a portion of the clean air cycle.
Three digital counts of detected photons are accumulated in three
memories under the operational conditions described.
520
-------
CLEAN AIR
PURGE
I NARROW BAND
J$l /OPTICAL FILTER
TUNGSTEN FILAMENT
LIGHT SOURCE
AEROSOL
OUTLET
I
O
COLLIMATING DISKS
AEROSOL
INLET
CLEAN AIR
PURGE
TUNGSTEN FILAMENT
LIGHT SOURCE
AEROSOL
OUTLET
j t
Partial/
Shutter
SCATTERING VOLUME
Figure 4» Diagram of nephelometer with enlarged view of the
partial shutter. Without the shutter, the instrument
integrates the particle scattering coefficient over
7° to 170° to measure b_. With the shutter in
place, the
to measure
instrument
bbsp'
integrates over ^ 90° to 170'
J.,
<&'
521
-------
Memory 1: Instrument Background plus air Rayleigh scattering, instru-
ment filled with clean air.
Memory 2: Memory 1 plus white object brightness, clean air plus
calibrate object.
Memory 3: Memory 1 plus scattering due to particles, instrument filled
with ambient air.
Memory one is sub tracted from m amories two and three, giving digital
counts proportional to the brightness of the particle scattering and
calibrate object with instrument background and air Rayleigh scatter-
ing subtracted. The particle scattering count is divided by the
calibrate object cou.it. This sequence, repeated several times per
hour, zeros and spans the nephelometer.
Two instruments of this type have been constructed and the span is
stable within + 2% over several months. Both instruments measure the
particle scattering coefficient at several wavelengths determined by
optical interference filters to measure the value of a, defined in
equation 6 as
b (A) <* A~a (6)
-3 -1
ikasured values of b in the atmosphere range from 3 x 10 m in
>io Lluted Los Angeles to 10~^m~ at Mauna Loa Observatory in Hawaii.
. ii£h sensitivity, multiwavelength instruments have been purchased by
institute fur Meteorologie, Mainz, Germany, Air Force Cambridge
Research Lab and the National Oceanographic and Atmospheric
Administration. Several hundred lower sensitivity, single wavelength
instruments have been produced and are in regular use for both research
cind monitoring. The draft version of volume I of the ACHEX final
report from Rockwell International to the Air Resources Board, State of
California, recommends the integrating nephelometer for both long term
monitoring and short term surveillance of aerosol properties.
B. b,
bsp
An optically thin aerosol layer over a dark surface increases the
albedo by scattering incident radiation backwards into space. The
albedo per unit thickness of an aerosol layer illuminated by a zenith
sun can be determined by integrating the aerosol volume scattering
function over the backward hemisphere of scattering angle. A partial
shutter, shown in Figure 4, can change the angle of integration of the
nephelometer so that the scattered light intensity is proportional to
the backward hemisphere scattering extinction coefficient b, due to
522
-------
aerosol particles, b^gp normally is in the range 0.1 to 0.2 times the
aerosol scattering extinction coefficient bsp.
The two aerosol parameters needed in simple radiative climatic models
are the particle backward hemisphere scattering coefficient, b^sp and
the particle absorption extinction coefficient, b . There are a num-
ber of ways of measuring bap, and none is entirely satisfactory.
_ / _1 _Q -J
Long path extinction cannot be used because bap is 10 m to 10 m
or smaller. Various techniques based on inverting angular scattering
information have been used by Eiden^o and Grams, e_t al. , etc., but
these methods require precise knowledge of the aerosol size distribution,
and contain errors of unknown size and magnitude, since the scattering
by irregular particles is calculated using Mie formulae for spheres.
The absorption coefficient of collected aerosol samples can be estimated
with low precision from measurement of the transmission of KBr pellets
containing dispersed aerosol^S. Lindberg and Laude^" measured aerosol
absorption by measuring the decrease of diffuse reflectance of a whit e
powder when a small amount of aerosol is dispersed in it.
All of the above methods, in our opinion, are poorly suited for measure-
ments in background locations. Measurement of the angular dependence
of the aerosol volume scattering function is difficult when molecular
scattering dominates. The methods of Volz and Lindberg require collect-
ing an aerosol sample over several days, scraping the sample off the
i ollecting surface and dispersing the sample in another media. Any
t reatment of the sample that alters the aerosol size distribution will
Qrj o 1
alter the optical absorption coeff icientJVJ»J . A different technique
for measurement of bap has been developed in our laboratory that we
lie] ieve is superior to those described above.
Atmospheric aerosol is collected by passing ambient air through a
N'uclepore filter. The filter consists of a 10 ym thick film of poly-
carbonate plastic with 0.4 pm holes etched through it. The holes are
etched along damage tracks from highly ionizing particles and are round
and perpendicular to the surface of the film. Individual particles
with a mean separation of several diameters are collected on the surface
of the filter. The filter and the particles are placed in an optical
system that illuminates the particles and the filter with a parallel
beam of, in this case, green light and collects both direct transmitted
and forward scattered light. The extinction or change in transmission
between a clean filter and the filter plus aerosol is assumed to be the
same as absorption by the same aerosol dispersed in a long column of
air. Knowing the volume of air passed throug1! the filter during
523
-------
collection of the aerosol, one can calculate the optical absorption
coefficient due to particles, b
ap
This Method has been checked for accuracy using laboratory aerosols
of known (including zero) absorption coefficient and is described by
Lin, et al.-^. The disadvantages of the method center on errors in-
troduced by sample alteration that may take place during collection,
but the sample alteration is probably much less than in the techniques
of Volz and Lindberg. The sample collection is simple and only requires
10 to 20 ug/cm of aerosol on the filter.
VII. ATMOSPHERIC MEASUREMENTS AND DATA
A. bsp AND VISIBILITY
As discussed in Section II, Koschmieder related b t to the distance
at which a black object is just visible when viewed against the horizon
sky. The distance of visibility is given by
V = ~~ (Middleton2) (9)
bext
assuming aerosol homogeneity, uniform illumination and a 0.02 detecta-
ble contrast. Commonly it is assumed that bext = bscat> i.e., ba^s = 0.
Measurements of bscat and observer visibility show good agreement with
the formula above.
33
Horvath and Noll conducted a study in Seattle between total light
scattering, bscat measured with an integrating nephelometer, and pre-
vailing visibility observed by two separate people. Their results were
in good agreement with the theoretical expression of Koschmieder when
data for RH >65% RH were excluded. Apparently the location of the
nephelometer in a heated room caused reduced RH in the scattering
measurements. In the cases where RH < 65%, the correlation between
'•'scat anc^ prevailing observer visibility was 0.89 and 0.91 respectively
with a coefficient in the Koschmieder expression of 3.5 + 0.36 and
3.2 + 0.25 respectively. This can be compared with the theoretical
value of 3.9, indicating a slightly lower prevailing visibility than
meteorological range. Since no ideal balck targets were used (only
trees, buildings, etc.) these would have caused just such a deviation.
4
Samuels, et_ al. conducted the most extensive tests to date of the
relationship of prevailing visibility to light scattering and various
mass concentration measures,
524
-------
They conclude that bsp as measured with the integrating nephelometer
is a good predictor of prevailing visibility and that the regression
analysis in agreement with Koschmieder's theory. These workers noted
that there was a smaller observed prevailing visibility than that pre-
dicted from theory and bgp measurement, which they suggested was due to
; non-ideal black visibility targets.
, B. bsp AND AEROSOL VOLUME OR MASS
As discussed in section III, b should be and is highly correlated
„ with particle volume in the 0.1 to 1.0 ym diameter range. Correlation
of bsp and total particle volume (or mass) is not expected unless the
two modes of particle volume happen to be correlated. Thus we would
not expect to find a particularly good correlation between bsp and
„ measured filterable mass concentration, for example, measured with the
high volume air sampler.
It is somewhat surprising, in view of this, that the measured correla-
tion coefficient between b and total aerosol mass concentration is as
high as the observed range between 0.5 and 0.9. While the former value
is not impressive nor particularly useful, the latter is sufficiently
high to allow inference of mass concentration from bsp. Table I sum-
marizes the various published correlations of bSD and mass. Included
in the table are correlation coefficients, r, and regression constants
A and B.
The correlation coefficient of 0.9 in New York City must be due to
either a correlation between the upper and lower volume (i.e., mass)
modes or an absence of the upper mode. The location at the 16th floor
of a Manhattan building suggests the latter since it was well removed
from sources of wind blown dust and other mechanically produced
p-irt ides.
in contrast, the low correlation coefficient in San Jose, CA, of 0.6
, wus obtained at a dusty athletic field, with the air intake at approxi-
mately 7 meters above the ground. In this case, the poor correlation
was likely due to a large and variable fraction of the aerosol in the
supermicrometer mode.
_ i'.. bsp AND EXTINCTION
In a cooperative experiment with Dr. John Hall, Lowell Observatory,
Flagstaff, AZ, bsp and long path extinction were simultaneously
measured at a number of wavelengths in the visible and ultraviolet.
The nephelometer measured bg at three wavelengths; 430 nm, 530 nm and
640 nm, and is described in section IV, A. Hall measured extinction by
525
-------
(-4
0)
O
•H
E£-I
CO
CO
CO
•H
0
r*;
60
•H
S3
e
o
m
•
CN
cfl
•H
-a
a
CJ
in
•
CN
rr
!t
]2
/
CO
•O -H
C r-4
•H r-l
43 cfl
01 EC
i-O 1
C 0)
^ 13 13
0) cfl 0
4-J 8 rH
rH O, O
•H -H >,
4H >J CJ
0) 01
S rH
3 P.
rH W B
O -H cfl
> < 0)
0)
O
p. J-l
01 01
• rH 4-1
CO CJ rH
•H 3 -H
TJ ^ 4n
* 01
01 ,0
CJ 'H
Cfl <4-l
U-l S-l
CO 01
C Cfl 4-1
0) CO rH
D. rH -H
O 60 4-1
m
<£
1-1
pp
<^
i-l
ca
o
^J
PQ
m
•
o
OS
1 —
•
0
o-
•
CN
rH
•
0
1
\Q
00
•
o
1
1
1
1
1
1
1
1
1
01
•
rH
O
vj"
(
o
1
o^
^o
•
o
^^
x^
v— '
^c
CJ
„
T^
c
CO
rH
y
CO
O
^
•
o
4-J
c
0)
e
cfl
j^
CJ
cfl
CO
,
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
O
•
m
ro
ro
•
O
1
CN
Os
•
O
1
1
1
1
1
1
1
1
1
In
l*~s
g
rj'
p4
O
PH
^
01
Z
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
^
•
T-H
LO
•
iH
*sO
uO
*
0
• 1
1
1
1
1
1
1
1
1
f S
m
s — '
CJ
OJ
en
o
n
(3
CO
CO
i
i
i
i
i
i
i
i
i
vO
•
ro
vO
CN
O
1
CO
p»^
•
o
i
i
i
i
i
i
i
i
i
m
•
en
00
O
C3
1
en
00
d
/— -s
m
<
S
„
0)
H
4J
4-J
Cfl
01
1
1
1
1
1
1
o
•
CN
in
rH
O
vO
00
•
O
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
/-^
vO
""•^
C
o
4J
CO
O
CO
0)
3
CO
C
O
•H
4-1
Cfl
r-l
4-1
C
01
o
c
o
CJ
CO
CO
1
e
0)
4-1
cfl
rH
3
CJ
•H
4J
V4
cfl
a.
01
rH
43
Cfl
1-1
01
4-1
rH
•H
4-4
1
60
13
•H
r-l
0)
4J
4-1
Cfl
O
to
4-1
60
•H
rH
4-1
0
p>^
r-l
§
g
3
CO
4-J
C
0)
CJ
•H
4-1
4H
0)
O
CJ
C
O
•H
4-1
Cfl
rH
0)
1-4
r-l
O
CJ
M
cfl
0)
c
•H
rH
II
>-l
• •
01
cfl
CO
^4
01
1 1
0)
a
CO
M
cfl
p.
0)
H
.
CO
e
60
a.
^^
w
CN
1
O
rH
^
<
II
I_[~N
|
a
i
o
rH
a
ca
^
fi
13
0)
C
•H
4-1
01
PQ
T3
c
cfl
•<
cfl
H
526
-------
2.0 _
; i.o _
M 6 -I
O _
H
M
Z
X
at
0.2 -
0.1 ._
300
* LONG PATH EXTINCTION
A NEPHELOMETER
i
400
500
i
600
WAVELENGTH, nm
Figure 5. Measurments of b (A) by long path extinction
and by nephelometer. Data taken near Flagstaff Az. on 6 Nov.
1974.
527
1 <"•'
-------
measuring the brightness of a regulated incandescent light source at
distances of 0.5 and 21 km. Extinction over the path was determined at
eight wavelengths from 330 nm to 660 nra during the night of 6 November,
1974 at a rural site near Flagstaff. Good agreement was obtained be-
tween nephelometer and long path extinction measurements as shown in
figure 5.
D. MEASUREMENTS OF SCATTERING PARAMETERS
Under support from the Environmental Protection Agency, National
Science Foundation and the California Air Resources Board we have
measured various aerosol scattering parameters in urban and rural loca-
tions in California, Colorado and Missouri. In all locations the in-
coming air was heated 5°C to 20°C above ambient to lower relative
humidity of the sample. The measured parameters were:
b - Scattering extinction coefficient of particles at 530 nm
— (Rayleigh at 530 nm = 0.15 x KrV"1)
o - Wavelength dependence of b parameterized as
sp
b = KA~a (10)
sp
Two values of ot were computed from Red-Green b and Blue-Green
b . Red is 640 nm. Blue is 430 nm. Green is 530 nm.
Scat, ratio - Ratio of half sphere back scatter to b from particles
at 530 nm.
The sites were:
Rijr:hinrrid_ - Northeast corner of San Francisco Bay in vicinity of petro
chemical plants.
Point Reyes - Coast Guard station on cliff 150 meters above the sea
surface, 50 km NW of San Francisco.
Fresno - Central valley of California, urban agricultural site.
Hunter Liggett - Rural California site 20 km inland from ocean. Local
elevation 400 m. Local vegetation consisted of dry grass and
sparce trees.
Ca_l. Tec. - Site on campus in Pasadena in Los Angeles basin.
Pomona - Site at county fairgrounds in inland area of Los Angeles basin.
528
-------
Washington Univ. - Campus site located in residential area of St.
Louis, MO.
Tyson - Rural area 25 km WSW of St. Louis.
- .St.' Louis Univ. - Campus site in industrial St. Louis.
f
' Henderson - Site 10 km NE of Denver.
1
Trout Farm - Site 8 km N of Denver.
I
Table II lists the measured values at each site. For each measurement
r parameter, the range of that parameter containing 63% of the data is
L specified. For b , the units are 10~^m and the range low to high
' contains 63% of data.
E. bap MEASUREMENTS
Using the technique described in section yj, C, measurements were made
of bap at two locations NE of Denver and three sites near St. Louis
during Fall of 1973. The measured values of the ratio of absorption
to extinction are presented in figure 6. In Denver, the absorption
to extinction ratio is very high, indicating that the aerosol heats
and stabilizes the lower atmosphere. At the three Missouri sites the
measured values are as one would expect - the rural area (Tyson) has
a less absorbing aerosol than the industrial site (St. Louis University).
Only the industrial MO site had absorption comparable to that measured
outside Denver.
The probable chemical species that produces the absorption is graphitic
carbon. Without chemical analysis for this material it is only possible
to speculate about the nature of Denver's very absorbing aerosol. The
absorption could result from:
i (1) high graphitic carbon content.
(2) large concentrations of graphitic carbon particles smaller than
; 0.1 ym.
i (3) lack of (Nlfy^SO/ as a major component of Denver aerosol when
compared to that found in rural Missouri.
5
, The ratio of absorption to filterable particulate mass can be used to
estimate an imaginary refractive index for the aerosol if a size
; distribution and chemical uniformity are assumed. We believe the
particles are not uniform chemically and prefer to report bgp rather
than \\2- With this warning, the average aerosol bap at Denver was
0.35 x 10~^m~ . The imaginary refractive index, n2, given the stated
assumptions was 0.035.
529
-------
•o
5
i.
•
DC
>^
IO
a:
•
+j
IO
u
LA
CM
1-^
^
in
r—
^^
CM
r—
"•^
CO
CM
^^
^^
^^
v^
f^
CO
1
f
1
1
1
1
CM
r-^
— .
IO
•^
in
CM
CO
CM
N^
f~~
f~
*«^
in
,—
CO
i
i
i
i
i
i
CM
f-»
^
«»
•i*^
OO
^.
cn
CM
^^
cn
cn
CM
CO
m.
*
CM
r-.
\
0
»•»
in
P_
• —
cn
CM
^^
cn
CM
cn
i
i
i
i
i
CM
t^
CO
CM
^^
O
CM
^^
O
**^
o
CM
cn
%3
?l
0
CSJ
K
*s^
in
•i^
t—
tn
0
r—
CNJ
*«^
i__
f_
'^^
^«
^s^
o
%^
f°l
VO
CO
t^
*^^
cn
f—
CO
--^
CO
CO
*«s^
CM
in
CM
CO
fe4
qTj
•—
CO
t^
^.
CM
f~~
***
r-.
CM
cn
CO
^^
cn
%^
CO
cn
»«
<*
•*
co
»s^
in
f^
•**.
•*
o
CO
^»
o
CM
1^
CM
cn
%4
+1!
*
ro
*n^
1
1
^^
«*•
^n,^
"£i
CO
*^^
t
1
^.^
o
•s^
r—
%^
?l
J^
CO
*1^
r™
^^
co
CM
_
co
"n,^
o
x^
LO
*«^
^~
**
•?!
00
VI
c c
•r- 0
•«—
I/I +J
I- 10
cu o
•!-> O
g"
IO r—
IO IO
CL
C
c
US
OJ-C
10 cu
0 +J
vi o
z
t-
o •
1- +*
J- CO •
0 +J OJ
t.
cu
intocof-. • • • • • ET-Q.
•—, I O. O. O. O. +1 +1 +1 +1 +1 I-T3 CO
,n +1 +| +1 +1 +| o in in in o 3 01-u
fK CM I I""* IO in OO O CM CM *"* CO I/I V)
•t: . ......... iai/i4J
° f- Ir— r- i— I— t— i— r-i— r- CU3C
E u a>
co ^^ ^^ co co r^» *i~ ^r* o
cn • ^~ co ^" *^ ..... o "O e
c? o +|o o o o +|+|-f|+l+| *
o^ +1 in +| +1 +{ +| r^ o in in in co s_ cu
cJ co CM o «j- in co ^" oo co 10 f^« 10 10 >
O
O O r" ^- r— r~ r— r— »-• r^ t— C7I I/I rt
O> f- •M
•r- i— "O -r- O
JT I 31/10
S «* i— in in oo •— o
Q. «*^-CnOOOOCM«l-CMCMin U T3 CM
01-0
••- cum
•o a; "a
«* CMOOOCOCM-r-ECU
Q. CMOfOCMOOlO"— CM'J'OCM i-IO-f^
j3IOCU
O O.-C
•a
•—• C CU V)
ro 4-> 10 I.
in i- cu • -r-
»—• C E VI IO
^3 CM OO CO r— r- VO C
c^ ........... o -r- r—
^1^— OOr-Oi— r- •— OOOO +J«3-
c u 10
•— O VI
•r- O 10 >-.
3 U1 U. >-.
O J-
C _l CU 4-> CU
O T3 3 r—
VI • C O JO
>, 4J eu s- 10
t- W» X I- I—
c
o
•f—
IO
u
o
_J
•a
c
r"
-------
I
S-
S-
3
O
u
o
Denver, Two Sites 10km
HE Of City.
11/10/73- 11/23/73
St. Louis University
9/28/73-10/4/73
Washington University
8/2-2/73-8/30/73
Tyson, Mo.
9/5/73-9/26/73
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
* "ao^sD + ba "
ap "sp ap
Figure 6. Ratio of absorption to extinction by particles.
531
-------
VIII. CONCLUSIONS
Comparisons can be made between our measurements at Denver and other
location. Deliquescent salts were not detected in the aerosol at
Denver and the b (RH) curves were at times quite hygrophobic. The
aerosol is less water soluble in Denver than at other sites.
The aerosol had somewhat higher backscatter to b ratio and much
higher t>ap/bext than values of the same parameter at other locations.
both measurements could be explained by a shift of the small particle
mode to smaller particles. The absorbing character of Denver aerosol
may enhance the brown or yellow color of distant white objects viewed
through the urban plume.
ACKNOWLEDGEMENTS
This research has been supported by Environmental Protection Agency,
National Science Foundation and California Air Resources Board funds.
REFERENCES
1. Koschmieder, H., Beitr. Phys. Freien Atm., 12, 33-55 & 171-181
(1924).
2. Middleton, W. E., Vision Through The Atmosphere, University of
Toronto Press, Toronto, Canada (1968).
3. ACHEX, Aerosol Characterization Experiment of the State of
California Air Resources Board. Prime contractor is Rockwell
International Science Center.
4. Samuels, H. J. et al., "Visibility, Light Scattering and Mass
Correlation of Particulate Matter," Report of California Air
Resources Board (1973).
5. Charlson, R. J., et al., Atm. Env. 2, 455 (1968).
6. Simmons, W. A., et al., "Correlation of the Integration Nephelometer
to High Volume Air Sampler," Mass. Dept. of Pub. Health (1970).
7. Thielke, et al., Aerosols and Atmospheric Chemistry, G. M. Hidy,
editor, Academic Press, New York (1972).
8. Porch, W. M., Science, 170, 315 (1970).
9. Horvath, H,, Atmospheric Environment, 5, 333 (1971).
10. Waggoner, A. P., et al., Applied Optics, 10, 957 (1971).
11. Husar, R. B., Private Communication (1974).
532
-------
I
12. Charlson, R. J., et al., Aerosols and Atmospheric Chemistry, G. M.
Hidy, editor, Academic Press, New York (1972).
13. Rasmussen, R. A., et al., PNAS 53, 1, 215 (1965).
14. Junge, C., J. Meteorology, 11, 323 (1954).
' 15. Charlson, R. J., et al., Science, 184, 156 (1974).
S '
1 16. Charlson, R. J., et al., Atmospheric Environment, 8, 1257 (1974).
| 17. Winkler, P., Aerosol Science, 4,, 373 (1973).
18. Hanel, G., Beitr. Z. Phys. Atm., 44, 137 (1971).
19. Covert, D. S., Ph.D. Thesis, University of Washington (1974).
20. Buettell, R. G., et al., J. Sci. Inst., 26, 357 (1949).
! 21. Ahlquist, N. C., et al., J.A.P.C.A., 17, 467 (1967).
22. Ahlquist, N. C., et al., Patent Application (1974).
23. Ensor, D., et al., Atmos. Env., 4, 48 (1970).
24. Quenzel, H., Atmos. Env., 9 (1975).
25. Rabinoff, R., et al., J.A.M., 12, 184 (1973).
26. Eiden, R., Applied Optics, 5, 56,9 (1966).
27. Grames, G. W., et al., J.A.M., 13, 459 (1974).
28. Volz, F. E., J.G.R., 77, 1017 (1972).
29. Lindberg, J. D., Applied Optics, 13, 1923 (1974).
30. Waggoner, A. P., et al., Applied Optics, 12 896 (1973).
31. Bergstrom, R. W., Beitr. Z. Phys. Atm., 46, 223 (1973).
32. Lin, C. I., Applied Optics, 12, 1356 (1973).
33. Howath, H., Atmos. Env., 3, 543 (1969).
533
-------
- '•£ g$jj» ^£#&\'-
-------
• I
i
i A REVIEW OF ATMOSPHERIC PARTICULATE MASS
P
* MEASUREMENT VIA THE BETA ATTENUATION TECHNIQUE
i
E. S. Macias
5 Department of Chemistry
Washington University
St. Louis, Missouri 63130
and
R, B. Husar
Department of Mechanical Engineering
Washington University
St. Louis, Missouri 63130
ABSTRACT
The mass of atmospheric aerosols can be determined by using the
attenuation of beta particles from radioactive sources. In this method
atmospheric particulates are removed by filtration or impaction and the
nass density of this deposit is determined from the decrease in the num-
ber of beta particles passing through the deposit. This paper reviews
the theory of beta attenuation, the strengths and limitations of the
method for mass measurement, the design of various instruments reported
in the literature, and the characteristics of these instruments for both
laboratory and field measurements. The criteria for selection of beta
sources, electron detectors, and pulse processing electronics are dis-
cussed with an emphasis on state of the art instrumentation. Several
aerosol collection systems are also described including a two stage on-
line mass monitor with aerosol size separator (TWOMASS). This instru-
ment independently analyzes the mass concentration of two particle size
fractions. Calibration and field testing of this instrument coupled
with aprticulate sulfate measurements for detailed aerosol characteri-
zation studies are also discussed.
535
Preceding page blank
-------
A REVIEW OF ATMOSPHERIC PARTICULATE MASS
MEASUREMENT VIA THE BETA ATTENUATION TECHNIQUE
E. S. Macias
Department of Chemistry
Washington University
St. Louis, Missouri 63130
and
R. B. Husar
Department of Mechanical Engineering
Washington University
St. Louis, Missouri 63130
INTRODUCTION
Air borne particulates, due to their effects on visibility, are
the most obvious air pollutants. For this reason, particulate emis-
sions were the first air pollutants to be controlled. In the past ten
years, gaseous pollutants have received major attention from control
agencies, but continuing deterioration of visibility in many urban
areas, along with strong evidence of health effects on humans, and pos-
sible effects on climate, has led to a renewed concern for controlling
atmospheric particulates.1
The present air quality standard for particulate matter is expres-
sed in terms of the total aerosol mass in yg/m^ as measured by the high-
volume air filter sampler. Recently, there has been concern among re-
searchers, as well as within control agencies, that total particulate
mass or any other single parameter is an inadequate and to some extent
misleading measure of adverse effects of atmospheric particulates.
There is an increasing body of evidence which suggests that particulates
found in the atmosphere that are less than 5 ym in size contribute sig-
nificantly to the adverse effects of air pollution, and in fact con-
stitute a large segment of the total air pollution problem. Accordingly
536
-------
the current thinking on a first step toward a proper characterization
of atmospheric particulates is to divide the total aerosol population
into two size classes, i.e., fine and coarse particulates. In study-
ing the characteristics of the Los Angeles smog aerosol, Whitby gt al.
discovered and substantiated by a variety of data, that most of the
atmospheric aerosol volume is distributed bimodally; the lower mass
mode is in the size range 0.1 to 1.0 lim, while the upper mode is over
5 urn as shown in Fig. 1. The saddle point between the two mass (or
I
1001—
i
90h-
|
I
8o[—
I
I
70 —
i I I I i i i:
LA
~i i i r ry ~ r'"T
MPLS (CLARK, 1965) 56 RUNS
MPLS (PETERSON, 1967) 45' 7 RUNS
COLORADO, 1970, 3 RUNS
SEATTLE(NOLL)
- •- JAENICKE S JUNGE0967)
OKITA ( !9-o5)
001
100
PARTICLE DIAMETER, Dp , >im
jj FIG. 1 - Typical atmospheric aerosol volume distribution data.
volume) modes is between 0.8 and 3.0 urn. The significance of bimodal
;; aerosol mass spectra is that the two modes have distinctly different
physical characteristics (particle shape, volatility) and chemical
| composition. Furthermore, the two modes are produced by different
sources, and they are associated with different effects. These findings
provide a scientific rational for the separate consideration of fine
and coarse particulates.
It is anticipated that within the next few years these findings
will be recognized by establishing new standards requiring the deter-
537
-------
mination of aerosol mass and composition as a function of particle size.
Independent determination of the mass concentration of only two size
fractions, divided at about 1 to 3 vim will probably be adequate for
monitoring purposes.
The high volume filter technique and other gravimetric weighing
methods currently in use for atmospheric monitoring are inadequate for
future monitoring purposes. Slow time response and tedious manual
operation are the principal disadvantages of these methods. Other
methods which are more easily automated have been proposed based on
indirect measurements such as acoustic attenuation, pressure drop across
a nozzle, pressure drop across a filter, unbalance of a centrifuge,
acoustic particle counting, hot-wire anemometry,decrease in natural
frequency of a vibrating band or wire, tape spot photometry, light scat-
tering photometry or nepholometry, lidar, single particle light scatter-
ing, light transmission holography, electrostatic bounce, electrostatic
probe-in-nozzle, electrostatic ion capture and electrostatic contact
charging. However Sem,Borgos and Olin3 point out that all of these
techniques have serious limitations for monitoring of particulate mass
emissions.
It has been known for some time that the attenuation of beta parti-
cles as they pass through an absorber can be used as a thickness gauge
or mass monitor. In a recent comprehensive study of potential tech-
niques for measurements in emissions from fossil fuel combustion sources,
Sem et^ _al. ^ identified beta attenuation as the most promising method
for the sensing of particle mass concentration based on an evaluation
of the basic sensing technique applied to a specific measurement. This
conclusion was not a recommendation or criticism of presently available
commercial beta instruments nor a recommendation of beta attenuation
for use in other specific applications.3
This paper is concerned with application of the beta attenuation
technique for automated atmospheric particulate mass measurements.
Another promising technique for atmospheric mass measurements, the pie-
zoelectric microbalance, has been discussed in a previous paper. A
review of the physics of beta absorption and its use as a mass monitor
is reviewed. The strengths and limitations of this method for use in
monitoring atmospheric aerosols are also discussed.
A _two jBtage cm-line mass monitor with Aerosol size separator (TWO-
MASS) employing the beta attenuation technique is described. The oper-
ating characteristics and calibrations are discussed as tested in St.
Louis during the summer 1974 and spring 1975. The application of this
instrument coupled with sulfur detection techniques for detailed char-
acterization of atmospheric aerosols is also described.
538
-------
PHYSICS OF BETA ABSORPTION
Although the absorption of beta particles has been used extensively
for mass and thickness determinations, the physics of beta emission and
interaction with matter is not obvious. The following is a brief dis-
cussion of the underlying physical principles which govern these pro-
cesses.
BETA DECAY
In the process of B decay, a negative electron is emitted from
the atomic nucleus and the nuclear charge changes from Z to Z + 1 in
units of the electron charge. This process transmutes the beta-active
element into the next heavier element in the periodic system. This
nuclear emission of negative electrons is not to be confused with the
emission of atomic orbital electrons as in internal conversion or
Auger processes. In many cases beta decay does not populate the ground
state of the daughter but rather populates an excited state which may
de-excite by the emission of gamma rays or conversion electrons.
Beta decay can take place only if the mass of the daughter is less
than that of the parent. The beta transition energy is given by the
AEfi_ = [M(Z,A) = M(Z+l,A)]c2 (1)
p
where M(Z,A) is the atomic mass of a nuclide with atomic number Z and
atomic weight A and c is the speed of light.
Beta particles are emitted with a continuous energy spectrum as
shown in Fig. 2 rather than a single discrete energy equal to the tran-
sition energy. This apparent contradiction of conservation of energy
and two-body kinematics is explained by a third mass less particle, the
anti-neutrino which carries off the missing energy. For example the
beta decay of 11+C can be depicted as follows.
1UN + 3~ + v (2)
539
-------
o
o
I I
I4C
BETA RAY SPECTRUM
(DETECTED WITH A SILICON
SURFACE BARRIER DETECTOR)
0
00
200
ENERGY (kev)
FIG. 2 - 11+C beta ray spectrum detected with a silicon surface barrier
detector used in TWOMASS.
The shape of the beta spectrum is not the same in all cases. However,
the average energy is at approximately one-third of the maximum value
for most beta emitters.
INTERACTION OF BETA PARTICLES WITH MATTER
Beta particles interact with matter through elastic and inelastic
scattering with atomic electrons and elastic nuclear scattering. For
low energy electrons (Eg < 0.5 MeV) ineleastic scattering (ionization)
with atomic electrons is the predominant mode of energy loss. The num
ber of beta particles passing through an absorber decreases exponen-
tially with absorber thickness, to a good approximation, given by the
following
I = IQe V (3)
540
-------
where IQ is the beta intensity without an absorber, I is the intensity
observed through an absorber of thickness x and um is the mass absorp-
tion coefficient. The exponential form of the curve is fortuitous,
since it also includes the effects of the continuous energy distribution
of the beta particles and the scattering of the particles by the absorb-
er.5 The range Ro is the distance traversed by the most energetic
particles emitted, and corresponds to the energy at the endpolnt of the
continuous spectrum. The exponential absorption of beta particles is
not expected to hold for absorber thicknesses near the beta particle
range.5
The absorber thickness, x, is usually given in units of mg/cm2
with the actual thickness multiplied by the density. For low energy
beta emitters, the mass absorption coefficient is nearly independent
of the chemical composition of the absorber. This has been shown exper-
imentally by a number of researchers.6'' This is because the absorp-
tion of electrons depends on the initial energy and the number of elec-
trons with which they collide in passing through the absorber. The
latter depends on the absorber electron density, i.e., the number of
electrons per unit mass. Therefore the absorption of particles depends
on the ratio of atomic number to the mass number (Z/A). Although this
ratio decreases from light to heavy elements, the effect of this varia-
tion on ym for low energy beta emitters is not large. Most chemical
compounds have Z/A ratios in the range 0.44-0.53. The absorption of
high energy radiation (Eg > 1 MeV) includes a radiative contribution
from Bremsstrahlung radiation (inelastic scattering with the nuclear
Coulomb field). This effect causes large variations in the mass
absorption coefficient with increasing atomic number8 and is largest
for heavy element absorbers. The ratio of energy loss due to ioniza-
tion to that due to radiation is given approximately by
(dE/dx)ioniz = 800 ,,,
(dE/dx)rad EZ ^ ;
where E is the beta-particle energy in MeV and Z is the atomic number
of the absorber.
541
-------
MASS DETERMINATION USING THE ABSORPTION OF BETA PARTICLES
HISTORICAL SURVEY
The exponential absorption oi beta particles and the implications
of this absorption for thickness measurements of thin films have been
known for some time.8 The work of Crowther8 and others has shown that
the absorption of beta particles from sources with Eg -* 1 MeV is strong-
ly dependent on the elemental composition of the source.
Clapp and Bernstein described a beta attenuation thickness gauge
in 1950. They used a ^°Sr-g°Y source which has high beta transition
energy (ER = 2.16 MeV) and therefore did not have an element independent
device. They used an ionization chamber to detect the beta particles
but did not discuss a linear relationship between the logarithm of
activity and absorber thickness. This device was intended to be cali-
brated for each particular absorber material. Peterson and Downing10
built an early device which used the attenuation of low energy beta
radiation for determining absorber mass. They use a l4C source and a
quartz fiber electroscope for detecting beta particles. It was found
that the relation between the logarithm of the discharge time and the
weight in mg/cm2 of thin cellophane films and aluminum foils is almost
linear.
Mandel11 showed that 147Pm (Eg = 0.23 MeV) was a more suitable beta
source than the more commonly used (in 1954) 2Q!+T1 (Eg = 0.74 MeV) for
high sensitivity thickness measurements of very thin materials. Later
Anders and Meinke12 used a 147Pm source and a Geiger counter for meas-
uring thin films. Pate and Yaffe13 reported measurements of the
thickness of thin films using a 61Ni source (Eg = 0.067 MeV) and a 2-"
beta proportional counter.
The adaption of the beta absorption thickness measurement technique
for use as a particulate mass monitor requires that particles be collect-
ed on a filter or other substrate. The mass determination is made by
determining the beta intensity transmitted through the substrate before
and after the particles are deposited. Mass concentrations are deter-
mined by also measuring the air volume from which the particles were
542
-------
collected.
An early application of the beta absorption technique for determin-
ing the mass concentration of atmospheric particulates was reported by
Nader and Allen.lk Several beta emitters, (14C, 137Cs, 2QLtTh) were
studied; a 14C source and a gas flow end-window proportional counter
were chosen. Investigating several relative source-absorber-detector
distances and source diameters, they found the optimal geometry with
the absorber relatively close to a small diameter source. They also
determined that the typical background due to radioactivity in the
aerosol deposit is negligible compared to the beta source Intensity.
This work also demonstrated the feasibility of the beta absorption
technique for use in automated tape samplers.
More recently many researchers have reported the use of beta atten-
uation mass monitors using a variety of experimental configura-
tions.6'7' *5> 2t+ The variations are mainly in the source, detector,
aerosol sampling and electronics discussed in detail below.
BETA SOURCE
The important characteristics of a beta source for beta absorption
atmospheric aerosol mass measurementsare long halflife (>1 yr), pure
beta decay (no gamma decay), and low beta transitions energy (<1 MeV).
The beta transition energy is directly proportional to the maximum
thickness which can be pentrated and inversely proportional to the
sensitivity of the measurement. At least 13 sources meet these cri-
teria.25 They are 3He, 10Be, lt+C, 26C1, 39Ar, ^Ni, 79Se, 85Kr, 87Rb,
107pdj l<+7prj 20UTlj and 228Ra- Qf these nuclides, 1UC, 79Se and ll+7Pr
have beta transition energies ^0.2 MeV which optimize sensitivity
and penetration of the absorber (6-12 mg/cm2) in a typical mass monitor
(due to filter paper, air, deposit detector window, etc.). 11+C is pre-
ferred because it has a longer halflife than 147Pr (5730 yr vs. 2.6 yr)
7 Q
and is less expensive than Se.
DETECTORS
A desireable beta detector for atmospheric mass measurements must
have high stability, portability, and efficiency for detecting beta
particles. Detectors in use for beta particle mass measurements which
meet these requirements include gas flow proportional counters, geiger
counters, scintillation detectors, and scjLid state detectors. However,
Geiger counters and gas flow proportiona^ counters have slow time re-
sponse limiting the counting rate to aboi',: 1000 cps without large dead
time. Scintillation and solid state detectors are much faster and allow
the use of counting rates in excess of l(i 000 cps. The efficiency for
full energy events is somewhat less in su_id state detectors than in
543
-------
the other detectors. However the efficiency for simply scoring an
event as is needed in beta attenuation measurements is very high even
with fairly thin detectors.26 Silicon surface barrier solid state
detectors have recently been made light tight, rugged and with very
thin windows (40 ygm/cm2) and appear to meet all the important cri-
teria for beta attenuation mass monitors. The use of such a detector
is described in a later section of this paper.
ELECTRONICS
The simplest electronic components needed for a beta detector in-
clude a preamplifier, amplifier, discriminator and counter. Geiger
counters can be operated without a preamplifier. The NIM modular nu-
clear electronics operate from a single power supply located in a bin
are recommended because of their flexability and reliability. This
equipment is available from several commercial suppliers.
AEROSOL SAMPLING
Ideally, the physical and chemical characterization of atmospheric
Particulates should be performed 'in situ'. Unfortunately,'in situ' mass
measurement of suspended particulates at present does not seem possible.
Beta attenuation, for instance, requires the deposition of particulate
matter onto an impactor, filter or electrostatic precipitator substrate
placed between the beta source and the detector. In the past all three
aerosol deposition mechanisms have been used by various investigators.
Filtration is the most convenient sampling method since it can be
easily automated for quasi-continuous operation.7'14' 6> 17' '23 Such
an instrument typically consists of a filter tape configuration similar
to the commonly used tape sampler, A variation of this approach is the
use of a casette-type system similar to those used in slide projectors.
Inertial impaction as the mechanism has also been used.17'23'27
SAMPLING CONSIDERATIONS
The problems associated with aerosol sampling may be divided into
the following categories.
Losses in the Inlet System
Inertial or gravitational deposition in the inlet system before
particles reach the sensing area prevents the effective sampling of large
atmospheric particulates above 30-50 ym in diameter. In systems with
1 Q O 0 '
impaction sampling ' the large particle cutoff size is smaller be-
cause of the interference of the beta source or detector. Design cri-
teria for minimizing the losses in the inlet system are discussed by
544
-------
Fui-hs.28 In beta attenuation mass units that we are aware of, the upper
size cutoff was taken for granted and no special effort was made to In-
crease the cutoff size.
Filtration Efficiency and Bouncing
The physical retention efficiency of the aerosol substrate must be
high for meaningful data interpretation. In the case of filtration,
high (>98%) efficiency for any size particles may be attained by conven-
tional glass fiber filters. Other criteria for filter choice such as
low flow resistance, low tare weight, adequate mechanical strength,
adequate capacity before clogging, inertness to water and chemicals,
are generally much more difCicult to satisfy.17
The use of impaction as the aerosol deposition mechanism imposes
the problem of large particle bounce-off following impaction. Although
the details of the bounce-off mechanism are not well understood it is
generally accepted that it can be minimized by using a sticky substrate
surface. Crystaline mineral particles, impacted on an uncoated teflon
substrate has been shown to be the poorest combination.29 Methods for
bounce-off reduction include the use of sticky (greesy) or bushy glass
fiber filters as the impaction substrates.
Sample Loss or Gain During and After Deposition
Atmospheric aerosols consist of solid or liquid particles. Tne
large particle mode is typically composed of solid mineral material,
while the mass in the snail particle mode is contributed by liquid
droplets consisting of an aqueous solution of salts and organic matter.
Upon deposition, hygroscopic and deliquescent particles may gain or
lose volatile matter in accordance with their own properties and the
thermodynamic conditions of the surrounding air. Therefore, aerosol
stability on a filter is an inherent problem, common to all measurement
methods in which the detection is not performed 'in situ'. Pres.ently,
there is little quantitative information available on this subject,
mainly because of experimental problems. One of the virtues of beta
attenuation is that it can be used as a tool for the investigation of
aerosol volatility at the moment of deposition. An example of such
results is shown in Fig. 3. A fraction of a cigarette smoke puff was
deposited in a beta mass monitor (described in the next section) and
the mass concentration as a function of time was monitored in one second
increments. As shown in Fig. 3 there is 30% loss in the deposited mass
within about 20 seconds. Turning the sampling pump off immediately
after deposition of the smoke puff, reveals that the smoke is stable on
the filter with no air flow. Immediately after turning on the pump, a
fraction of the deposit volatilized, indicating that the change in the
pressure and/or flow conditions caused the loss of volatile compounds,
545
-------
800 i-
NO FLOW THROUGH FILTER
RESIDUE MASS
j L. ____ J ___ -1
8 12 16 20 24 28 32 36
TIME, sec
FIG. 3 Loss of volatile cigarette smoke mass after deposition on a
glass fiber filter. The dashed line is for continuous pump
operation. The solid line is for interrupted pump operation
as indicated.
probably water and volatile organics. Unfortunately, such experiments
can only be performed when aerosol concentration is high, i.e. the
aerosol deposition rate is higher than the evaporation rate. According-
ly, the actual behavior of atmospheric aerosol upon deposition on a sub-
strate can only be inferred from laboratory simulation experiments.
It is well established that a large fraction of atmospheric fine
particulates is composed of hygroscopic or deliquescent particles.13
Their mass, therefore, depends on the relative humidity of the ambient
air. Although extensive studies have been carried out on the relative
humidity (RH) dependence of the aerosol light scattering coefficient31
we are only aware of one work in which the RH dependence of particle
mass was studied in detail.32 Here again beta attenuation provides a
546
-------
simple and convenient means of detecting the mass dependence on RH.
A typical result on the RH dependence of ammonium sulfate aerosol as
detected by beta attenuation is shown in Fig. 4.33
2 5
2.0 -
tr
CO
to
1.5 -
1.0 -
\ 1 1-
AMMONIUM SULFATE MASS
AS A FUNCTION OF RELATIVE HUMIDITY
5% (NH4)2S04
M -- 14 7 fq
40 50 60 70 80
RELATIVE HUMIDITY (•/.)
90
FIG. 4.
Ammonium sulfate mass determined with a beta attenuation mass
monitor as a function of relative humidity.
Recently, concern has been raised that particulate matter may form
a filter media as a consequence of catalytic gas-particle conversion.34
It has been speculated that the catalytic effect may be caused by the
filter or by the deposited particulate matter itself.35
In summary, the following have been identifidd as essential for
the proper use of the technique. A source with low beta transition
must be used for mass measurements to be independent of chemical com-
position. A collimated source with small source-detector distance is
optimal. The total abosrber mass including the aerosol deposit must
be a small fraction of the range of the beta particle. The source
547
-------
must have a long lifetime (>1 yr)/and decay without the emission
radiation. The pressure drop and/flow rate should not change over the
sampling period. Probleris such |s losses in the inlet system, filtra-
tion efficiency and bouncing, an/i aerosol stability on the collection
substrate must be minimized in t/ae sampler design.
548
-------
TWOMASS
OBJECTIVES
The present section describes a j:wo stage cm-line mass monitor
with a.erosol j^ize sampler (TWOMASS) which was designed to monitor atmos-
pheric aerosols in the two size ranges described previously, namely
fine particles with submicron diameter and coarse particles with super-
jr.icron diameter. This instrument was optimized for high sensitivity
over the shortest possible measurement time. TWOMASS shown in Fig. 5,
is fully automated so that continuous on-line determinations of par-
ticle mass are available every ten minutes over long periods of time
(1-10 days). TWOMASS was developed for use in ground based ambient
aerosol monitoring stations but it could be adapted for other uses.
INSTRUMENT DESIGN
Aerosol Collection System
The flow system separates particles into two size fractions.
Coarse particles (diam >3 ym) are impacted on a glass fiber filter
with a cellulose backing; submicron particles are collected on the same
type of high-efficiency glass fiber filter tape. This system is shown
schematically in Fig. 6. The single stage impactor head has a 3-mm
diam inlet aperture with a 1-mm jet to plate distance. Various impac-
tion materials have been tested, the glass fiber filter with thin cellu-
lose backing was found to be most satisfactory because of its low mass
density (3.2 mg/cm2), efficient impaction properties, and uniform
thickness. The impaction collection efficiency was designed to approx-
imate the removal efficiency of the human respiratory system with 50%
^efficiency for 3 vim particle collection.
The second filter stage retains all remaining particles. An
earlier study by Lilienfeld and Dulchinos17 indicated that Pallflex
E70/207W has high particle collection efficiency at high face veloci-
ties, low flow resistance, low tare weight, adequate mechanical
strength, and adequate filter capacity before clogging. The mass den-
sity of this filter is 3.2 mg/cm2 and the efficiency is over 99% for
549
-------
Reproduced from
best available copy.
FIG. 5 Photograph of TWOMASS.
550
-------
TWOMASS
Si
DETE
:CTORX
0
0
\1
H
i.,
W
""•*"
14C
D/
/IMPACTION
S STAGE
£>
^
f>
TO PUMP
PREAMP
*.
AMP/
DISC.
t \
P
R
1
N
I
, p
COUNTER
> *•
|
TIMER C
ROGRAMMABLEL-
CALCULATOR 1^
OUNTER
» \
|
-Sj DETECTOF
-FILTRATION
STAGE
-I4C
AMP/ PREAMP
DISC.
/ V f
i — *T TAPE 1
1 — <\ PLOTTER |
FIG. 6. Schematic diagram of TWOMASS including the data acquisition
and analysis system.
both cigarette smoke and 0.312 ym diameter polystryrene latex spheres
at the maximum face velocity obtainable, 6.4 m/sec. The flow rate
through the TWOMASS was set at 12 &/min using a CAST model 1531 rotary
vane pump.
Beta Attenuation Mass Monitor
Both the impaction and filtration heads of TWOMASS were fitted
with independent source-detector systems which were run simultaneously.
The beta source chosen was 3 mCi of 1^C deposited as BaC03 mixed with
Krylon spray adhesive in a 6-mm diameter brass disk covered with 800
yg/cm2 of mylar film. This source, a pure low energy beta emitter
with 5.7 x 103 year half life, 'E^^ = 156 keV, and range Rg = 28.26
mg/cm2 was chosen for the reasons outlined previously. The total mass
density of all absorbers (air, detector window, filter deposited mass,
551
-------
etc.) in the system is about 6 rag/cm2 or about 21% of the range of 14C
betas which is optimal for measurement sensitivity. The cross sec-
tional area of the collection spot is 32 mm2 and 7 mm2 in the filtra-
tion and impaction stages respectively.
The beta detector is a solid-state ruggedized silicon surface-
barrier detector with 50-mm2 surface area, 40-ygm aluminum window and
noise width of <11 keV (manufactured by ORTEC, Inc.). This detector is
light tight and rugged, thus it is suited for field operation. It was
chosen because of its high count rate capability, low noise character-
istics, low cost and simplicity of operation. As shown in Fig. 6 the
output of each solid state detector is sent to an ultra low-noise (<5
keV) preamplifier with very fast rise time (<20 nsec) and then to a low-
noise gaussian-shaping amplifier (timing constant 1.2 psec) with a lower
level discriminator set to remove noise. The energy spectrum of *4C
beta particles measured with this system is shown in Fig. 2.
The outputs of the two discriminators are sent to a Tectronix 31/53
calculator instrumentation system with two DC503 counters. This pro-
grammable calculator determines the mass with the following formula
M = ^- £n(I0/I) (2)
m
where M is the mass concentration in units of Mgm/m3, A is the cross
sectional area of the collector, f is the flow rate, ^ is the mass
attenuation coefficient in units of cm2/yg, IQ is the count rate of the
previous time interval and I is the count rate of the current interval.
This mass concentration data is plotted with an x-y point plotter,
printed, and stored on a magnetic tape casette.
On-line Operation
TWOMASS was designed for mass measurements over extended time
periods with a minimum of operator maintanence. By measuring the beta
intensity transmitted through the filter paper as the particulates are
being deposited, mass concentrations are obtained at the end of each
counting period. Furthermore, because the initial beta intensity, I,
is always the intensity measured in the previous counting interval,
there is no error introduced due to irregularities in the filter
medium. This counting arrangement also minimizes the effect of elec-
tronic instability on the mass measurement. The system is automated
with two motors moving both the filter and impaction tapes at program-
med intervals. In order to avoid clogging the filter paper which can
cause decrease in the flow rate, the tape is moved every two hours.
In sampling tests of atmospheric aerosol with TWOMASS using 2 hour
periods, the flow rate changed by less than 5%. A related effect, the
552
-------
change in the pressure below the filtration-stage filter tape with
increasing filter loading, was also examined. This pressure decrease
during on-line operation results in a decrease of the total absorber
mass between the 14C source and the beta detector. The geometry of the
filtration stage in TWOMASS was re-designed to minimize this effect
which was experimentally determined to be <6% of the mass of all
aerosols studied. The automated tape movement is controlled by an auto-
matic timer and therefore this interval can be adjusted to any experi-
mental requirements. Note that with this arrangement the tape is not
moved while particulates are being deposited or measured and thus, the
tape movement does not effect the mass measurement.
OPERATING CHARACTERISTICS
Sensitivity
Emphasis in the TWOMASS design has been placed on low noise and
fast timing characteristics in order to minimize instabilities in the
counting system due to noise drift and to maximize the counting rate
without large dead time. Typical counting rates are 10,000 cps with
the unloaded filter tape in place. This high counting rate was poss-
ible with minimal dead time (<1%) because of the fast timing charac-
teristics of the solid state detector and associated electronics. The
total sensitivity of the instrument is a function of the stability
of the system and the counting statistics. The sensitivity S, based
only on counting statistics using the 2a error limit given25 as
follows
2A ,„,
c — "• •-• (3)
min(2o) p ft\fc"
m
where t is the counting time and C is the detected count rate. With A,
pm, f, and C fixed for a given instrumental design, the statistical
sensitivity becomes a function of counting time only, i.e.
S . ,_ ,t = K . (4)
min(2cr) o v
For TWOMASS with the conditions described above, this constant is KQ =
2000 ygmsec/m3. From the above considerations it is clear that statis-
tical sensitivity increases while stability decreases with the counting
time. For sampling times of 600 sec the total sensitivity of TWOMASS
is 4 ygm/m . The system has also been run with subsecond time inter-
vals for laboratory experiments using high aerosol mass concentrations
as described in the previous sections.
553
-------
Calibration
Ttie instrument was calibrated against a Cahn Microbalance with
laboratory and St. Louis atmospheric aerosols. The mass of laboratory
aerosol added to the filter was in the range of 1-7 mg/cm2 as shown in
Fig. 7. The mass of atmospheric aerosols was in the range of 0.2-1
I I I T I
GRAVIMETRIC CALIBRATION OF
TWOMASS
0.6
0.4 -
0.2 -
x GLYCINE
A (NH4)2S04
D ZnCI2
O NaCI
0
8
GRAVIMETRIC MASS (mg/cm2)
FIG. 7. Gravimetric calibration of TWOMASS using laboratory aerosols.
mg/cm2 as shown in Fig. 8. These calibration data yield an exponential
relationship between mass loading and beta attenuation (I/IQ) as expect-
ed from equation 3 within the accuracy of the measurements. Sem and
Borgos24 used an uncollimated ll*C source and various filter absorbers
to simulate aerosol loading and found some deviation from the exponen-
tial absorption law particularly for high masses near the range of ll+C
beta particles. The operation characteristics and source-detector con-
figuration of TWOMASS are quite different irom the Sem and Borgos
arrangement e.g. use of a highly collimated lkC source and relatively
low total absorber mass, and a filter of uniform mass. Therefore the
554
-------
T
T
GRAVIMETRIC CALIBRATION OF
TWOMASS
ATMOSPHERIC AEROSOL
0.7 -
GRAVIMETRIC MASS (mg/cm2)
FIG. 8. Gravimetric calibration of TWOMASS using St. Louis atmos-
pheric aerosol.
exponential calibration curve determined for TWOMASS is not inconsistent
with their results.
As mentioned previously the mass attenuation coefficient, um, is a
weak function of the chemical form of the aerosol. The values of pm
extracted from Figs. 7 and 8 are plotted versus the atomic number
divided by atomic weight (Z/A) of the various aerosols is shown in Fig.
9. Because the Z/A ratio for the atmospheric aerosol is not known the
ym value for the atmospheric aerosol' is given in a shaded horizontal
band. This data indicates that somo variation in ym does exist over
the Z/A range for most chemical compounds (0.44-0.53). Thus we suggest
that determination of pm is necessary when using a beta gauge for study-
ing a specific aerosol. However ou) data indicate that, at least for
St. Louis, the atmospheric aerosol jas a value of um = 0.260 ± 0.025
55
-------
o
Z
UJ
o
u,
u_
UJ
o
o
z
o
UJ
t-
(f>
-------
60
ro 40
20
0
40
20
0
n ] 1 _
TWOMASS MASS MEASUREMENT
ST. LOUIS AEROSOL
23 AUGUST 1974
COARSE PARTICLES-
FINE PARTICLES
04
FIG. 10. On-line mass concentration data in two particle size ranges
determined with TWOMASS during the August 1974 St. Louis
RAPS experiment.
be seen from this data, both the fine and coarse particulate mass concen-
trations exhibit strong temporal variations; much more than expected
from earlier low time response measurements with filters. In addition
it is evident from these data that the mass concentrations of each
size fraction are each roughly half of the total aerosol mass and at
times vary independently. It is clear from these results that TWOMASS
gives more detailed mass concentration data than previously available
which should lead to a better understanding of atmospheric aerosols.
A more detailed characterization or the ambient aerosol using TWO-
MASS with 10 min sampling intervals aruf moving the filter tapes every
two hours was carried out in April 197/ in St. Louis at Station number
112 of the Regional Air Monitoring System (RAMS). In these experiments
55/|
-------
the light scattering coeffic
particle mass and relative h
These parameters measured dm
ent, (hscat)» sulfur content of the fine
midity were also measured simultaneously.
ing a five day experiment are shown in
i, Ll^-WJ ^, £^ U J. ULll*i_ l~ ^, J- fcJ lll*»-UU VA J_ V~t_t U U .*-*.&£2 U. *- -»- v %v W»%A_^ •w«kfr ^~ *- -kiit.^*.*. V. %«*. v w »* w» b- —.••
Fig. 11. These data again eAhibit a rather striking temporal variation
of mass over the sampling perjiod. Data from 18 April indicate the arri-
val of a polluted air mass, cjrastically changing both the mass concen-
tration and light scattering/coefficient within a one hour interval.
A rain storm at 1600 on that/day caused the sharp drop in fine particu-
late mass concentration. /
/
It should be noted that such observations on this time scale were
not possible with conventional gravimetric measurements or beta attenu-
ation instruments. The fine particulate mass concentration, after the
arrival of a clean air mass on 18 April dropped to an extremely low
value (3-5 yg/m3) as shown in Fig. 11. Even at these low concentrations,
meaningful results can be obtained with TWOMASS operating with 10 min
counting intervals.
Analysis of the data in Fig. 11 shows a strong correlation between
fine particulate mass and light scattering coefficient with a correla-
tion coefficient of 0.96 for two hour averages. The ratio b /fine
particulate mass extracted for these data was 0.2 gm/m2.
The TWOMASS two hour filter deposits were analysed for sulfur using
flash vaporization-flame photometric detection method.36 The data in-
dicate that sulfur in the fine particulates expressed as sulfate (ug/m3)
constitutes 15-50% of the fine particulate mass concentration. It is
interesting to note, however, that this fraction exhibits substantial
temporal variations.
The results of this study show that TWOMASS with high time resolu-
tion, ability to separate fine and coarse particulates, high sensitiv-
ity, and ability to analyse the chemical composition of the
deposit, is a significant advance over existing atmospheric
particulate mass instrumentation.
CONCLUSION AND RECOMMENDATIONS
Based upon the above discussion of the physics of beta absorption,
beta source characterics, detection methods and sampling considerations,
we conclude that the problems of precise determination of ambient aero-
sol mass concentration with high sensitivity are essentially solved.
We recommend that future research effort be focused on a better under-
standing of the processes ir.fluencing the aerosol sampling.
558
-------
40
20
20
I 0
100
60
20
ST LOUIS AEROSOL
APRIL 1975
FINE PARTICLE MASS
SULFATE MASi ( //g /m3)
RELATIVE HUMIDITY ( "/,
4 '17
4/18 4/19 4/20 4/21 4/22
TIME (days)
FIG. 11. Typical data for the characterization of the St. Louis aerosol
by simultaneous determination of fine particle mass, light
scattering coefficient, sulfate mass and relative humidity.
559
-------
ACKNOWLEDGEMENT S
The authors gratefull acknowledge the assistance of Robert Flet-
cher, Larry Friedman, Roland Head, Connie Rutledge and Pamela Stubits
in this work. Part of this work was supported by the U.S. Environmental
Protection Agency, Research Triangle Park, North Carolina, Division of
Chemistry and Physics, Field Methods Branch, under grant number
R803115-01-0. Special thanks are due the Environmental Protection Agen-
cy and the St. Louis Regional Air Pollution Study for their cooperation
in a portion of this work.
560
-------
REFERENCES
1. Abatement of Particulate Emission from Stationary Sources. National
Research Council, National Academy of Engineering, COPAC-5, 1972.
2. Whitby, K.T., R.B. Husar, and B.Y.H. Liu. The Aerosol Size Distri-
bution of the Los Angeles Smog. J Colloid Interface Sci 39:117,
1972.
3. Sem, G.J., J.A. Borgos, and J.G. Olin. Monitoring Particulate
Etnmissions. Chem Eng Prog 67:83-89, October 1971.
4. Sem, G.J., J.A. Borgos, J.G. Olin, J.P. Pilney, B.Y.H. Liu, N.
Barsic, K.T. Whitby, and F.D. Dorman. State of the Art 1971:Instru-
mentation for Measurement of Particulate Emissions from Combustion
Sources. Thermo-System Inc., St. Paul, Minn. Prepared for the U.S.
Environmental Protection Agency of 1971.
5. Kaplan, I. Nuclear Physics. Reading, Mass. Addison-Wesley Pub-
lishing Co., 1962. 770 P.
6. Dresia, H. , P. Fischotter, and G. Felden. Kontin.iierlich.es Messen
des Staubgehaltes in Luft und Abgasen mit Betastrahlen. VDI-Z
106:1191, August 1964.
7. Husar, R.B. Atmospheric Particulate Mass Monitoring with a B
Radiation Detector. Atm Env 8:183-188, 1974.
8. Crowther, J.A. On the Coefficient of Absorption of the 3 Rays From
Uranium. Phil Mag 12:379-392, 1906.
9. Clapp, C.W. and S. Bernstein. Noncontacting Thickness Gauge Using
Beta Rays. Elec Eng 69:308-10, April 1950.
10. Peterson, J.H. and J.R. Downing. Film Thickness Measurements for
Infrared Spectrophotometry. J Opt Soc Am 41:862-862, November
1951.
561
-------
11. Mandel, L. The B-lay Absorption Spectrum of ll+7Pm and its Applica-
tion to Thickness Measurement. Brit J Appl Phys 5:287-9, August,
1954. |
12. Anders, O.U. and W.W. Meinke. Beta Gauge for Localized Measurements
on Thin Films Rev JSci Inst 27:416-17, 1956.
13. Pate, B.D. and L. Y',iffe. A New Material and Techniques for the Fab-
rication and Measurement of Very Thin Films for use in Air-BCounting.
Can J Chem 33:15-23;i 1955.
14. Nader, J.S. and D.R. Allen. A Mass Loading and Radioactivity Ana-
lyzer for Atmospheric Particulates. Am Ind Hyg Assoc J 21:300-307
(1960).
15. Izamilov, G.A. Measuring the Gravimetric Concentration of Dust in
the Air Using 3-Radiation. Ind Lab (Eng Tran of Zavodskaya Labora-
toriya 27:40-43, January 1961.
16. Benarie, M. and A. Loverdo. Pesee automatique des polluants atmos-
pheriques par jauge beta. Mesures, Regulation et Automatisme.
32:90-95, September 1967.
17. Lilienfeld, P. and J. Dulchinos. Vehicle Particulate Exhaust Mass
Monitor. Final Report to Environmental Protection Agency, Durham,
N.C. Contract No. 68-02-0209. March 1972. 41 p.
18. Horn, W. Process for Continuous Gravimetric Determination of the
Concentration of Dustlike Emissions. Staub-Reinhalt 28:20-25, 1968.
19. Lilienfeld, P. Beta-Absorption-Impactor Aerosol Mass Monitor.
Amer Ind Hyg Assoc J 31:722-29, November, 1970.
20. Dresia, H. and F. Spohr. Experience with the Radiometric Dust
Measuring Unit "Beta Staubmeter." Stuab-Reinbar. 31:19-27, 1971.
21. Duke, C.R. and B.Y. Cho. Development of a Nucleonic Particulate
Emission Gauge. Industrial Nucleonics Corp., Columbus, Ohio. Pre-
pared as final report under U.S. Environmental Protection Agency
Contract No. 68-02-0210, 1972.
22. Dresia, H. and R. Mucha. Registrierendas Radiometriscb.es Messerat
zur Kombinierten Messung der Immissioner von Staub und Radio-
aktivitat in Luft. Staub-Reinhalt 34:125-128, 1974.
562
-------
23. Macias, E.S. and R.B. Husar. High Resolution On-Line Aerosol Mass
Measurement by the Beta Attenuation Technique. In: Proceedings of
the Second International Conference on Nuclear Methods in Environ-
mental Research, Vogt, J.R. (ed.) USAEC, Oak Ridge, Tenn., 1975.
24. Sem, G.J., and J.A. Borgos. An Experimental Investigation of the
Exponential Attenuation of Beta Radiation for Dust Measurement.
Staub-Reinhalt 35:5-9, January 1975.
25. Lilienfeld, P. Beta Absorption Mass Monitoring of Particulates—
A Review. Paper No. 71-1031, Joint Conference on Sensing of En-
vironmental Pollutants. Palo Alto, California. November 1971. 30
P-
26. Instruction Manual for Surface Barrier Detectors. Ortec, Inc. Oak
Ridge, Tenn. 1974.
27. Denzel, P. and W. Horn. Ein empfindliches Messverfahren zur Konti-
nuierlichen Bestimmung der gravimetrischen Konzentration von
Staubfbrmigen Emissioner. Elektrotechnische Z PT. A., 87:311,
1969.
28. Fuchs, N.A. Mechanics of Aerosols. 1 ergamon Press, London, 1964.
29. Hidy, G.M., ejt £il. Characterization of Aerosols in California.
Interim Report for Phase 1. Science Center Rockwell International.
Submitted to Air Resources Board, State of California, April 1973.
30. Junge, C.E. Air Chemistry and Radioactivity. Academic Press, N.Y.,
1963.
31. Covert, D.S., R.J. Charlson, and N.C. Ahlquist. A Study of the
Relationship of Chemical Composition and Humidity to Light Scatter-
ing by Aerosols. Submitted to J Appl Met, 1975.
32. Winkler, P. The Growth of Atmospheric Aerosol Particles as a
Function of the Relative Humidity-II. An Improved Concept of Mixed
Nuclei. Aerosol Sci 4:373-387, 1973.
33. Yamamoto, K. Measurement of Aerosol Mass as a Function of Relative
Humidity. MS Thesis, Dept. of Mechanical and Aerospace Engineering,
Washington University, St. Louis, Mo. 1975.
34. Novakov, T., S.G. Chang, and A.B. Harker. Sulfates as Pollution
Particulates: Catalytic Formation on Carbon (Soot) Particles.
Science 186:259-261, 18 October 1974.
563
-------
35. Lee, R.E. Jr. and T. Wagnuh. A Sampling Anomaly in the Determina-
tion of Atmospheric Sulfalje Concentration. Amer Ind Hyg Assoc J
27:266-271, 1966.
36. Husar, J.D., R.B. Husar, /and P.K. Stubits. Determination of Sub-
microgram Amounts of AtmcUpheric Particulate Sulfur. Submitted to
Anal Chem. May, 1975. \
564
-------
DETECTION OF ULTRA-FINE PARTICLES BY MEANS OF A CONTINUOUS
FLUX CONDENSATION NUCLEI COUNTER
ABSTRACT
A new continuous flux condensation nuclei counter is described.
Gases containing these nuclei go through an alcohol vapor saturator,
then over-saturation required to generate condensation is produced
in a Peltier cooling nozzle. Droplets cross a well lit measurement
spacing of variable sizes so that the probability to emerge with a
single droplet is equal to 0.95. The luminous flux scattered at 45°
by each droplets is picked by a photomultiplier. A device enables
to extend the field of use of this system to low pressures and low
temperatures.
Thesecounter's operating characteristics are given and the stu-
dy of particle detection as these were produced through photolysis
of gaseous impurities in air within a pressurized environment vary-
ing from 760 to 60mm of Hg is discussed.
565
-------
DETECTION OF ULTRA-FINE: PARTICLES BY MEANS OF A CONTINUOUS
FLUX CONDENSATION NUCLEI COUNTER
by Messrs- J. BRICARD, Professor, University of Paris VI,
France
P. DELATTRE, PhD, Engineer, French Atomic
Energy Commission, Fontenay-aux-Roses,
France
G. MADELAINE, PhD.Sciences, "
M. POURPRIX , PhD Sciences, "
+ + T+ + + + -t--t--t+ + -i +
In order to count very fine particles suspended in a carrier
gas, we use artificial means wherein suspension is easily satura-
ted with readily condensable vapor, then such vapor condensation
is generated over these particles. Each give rise to a droplet and
their sizes are adequate enough to al]owfor their detection. The
concentration of droplets, that is of particles, can be derived
either by measuring a luminous beam attenuation through a given
thickness of the mist represented by all the droplets, or by measu-
ring the luminous flux scattered by a given volume of said mist,
once properly illuminated.
In general, when using presently marketed measuring instruments,
the oversaturation required for vapor condensation is obtained by
566
-------
by lowering the temperature of the previously steam saturated gase-
ou:; media through adiabatic release, and in which the aerosol to
study is suspended. The use of such particle counters results in
two major difficulties : they do not allow for the detection of
loosely concentrated aerosols, on the one hand, and they operate
only by means of intermittend samplings of samplets in a media
involved in a study, on the other hand, this resulting in media
interferences.
So that both of these drawbacks are eliminated, we have devised
a continuous flux counter with which we count particles whose radius
-7
is somewhere above 10 cm and whose concentration varies between
35 3
I particle per cm and 10 particles per cm .
I. INSTRUMENT DESCRIPTION
The particle-charged gas first goes through a saturator contai-
ning ethyl alcohol or n-butylic alcohol (FIGURE!). This saturator
is a inner-lined cylindrical box, with spongy walls inhibited with
liquid to obtain condensation. It is fitted with a 4mm 0 and 8 cm
in length calibrated nozzle coolled by a Peltier effect module and
a constant flow intake device. Condensation appears at nozzle outlet
and generates the formation of droplet mists whose droplets radius
is between 0 y 5 and 1 u. They can be seen with the naked eye. The
observed droplet volume V which will now define is located at 2mm
beyond the nozzle. The droplet velocity through it is even and
567
-------
o
• f"
o/
-O
C
o
CJ
o
3
C
C
•3
O
Oi
cd
-^
-d
5
Q>
3
568
-------
adequately reduced to be considered as evenly distributed. It is
equal to 35 cm s
The measuring volume is defined in the following manner
(FIGURE 2). A source of light S illuminates a diaphragm D1 evenly.
The latter is set against a condenser lens shown by L • The conden-
ser L creates the image of diaphragm D at A in the measuring
field of photometer those axis is angular by 50° with that of the
condenser. Lens L in this photometer creates the image of D at
o 2.
A. Behind DQ, a lens L creates the image of lens L on the photo-
£. H* O
cathode of a photomultiplier. Under these conditions, the measuring
volume has the shape of a parallelepiped defined in space by the
images of diaphragms D and D ; it is evenly illuminated because
D is also illuminated evenly.
The particle carrying air .current is directed at right angle
to the plane of FIGURE 2. When a particle crosses the field, the
light scattered within the photometer field is received by the
photomultiplier. It delivers a current impulse that is proportional
to the luminous flux scattered by the droplet, which can also be
me as ure d •
Knowing the number of impulses delivered during a given time
interval, the straight surface of the measuring volume that is
perpendicular to the air current charged-with droplets and its
specific velocity, we cans derive the droplet concentration.
569
-------
Kl
B
0)
*j
en
>,
to
I—I
cd
-------
To effect an individual count of the droplets while mini-mizing
the simultaneous presence of several particles in a single field,
we can modify the sizes of diaphragms D. and D . Nine combinations
are possible, GO that, theoretically, it is possible to detect
O C q
at least 1 particle per cm and at most 10 particlet per cm , with
a probability of double presence of 5 %. The experimental checkout
of such forecast is actually underway and should enable to eventual-
ly change the sizes of diaphragms in order to make sure that coin-
cidences are of riegJ.igeable value.
By comparing the present counter findings and those of a General
Electric counter, we realize that there is a linear relationship
between concentrations given by both instruments when they do not
5 3
exceed 10 ^articles per m • Impulse amplitude enabled us to deter-
mine the size of droplets whose diameter exceeds 0.3 p , and to
check the operation of our instrument. With the photometer axis
tilted by 50° over the collimating lens, this amplitude is about
independent from the droplet index. Moreover, if the cooling ry.~-
tem is shut down, the unit operates as a conventional granulorneter
for the study of particles with diameters over 0.3 p.
We have designed a simplified version of this instrument to
effect measurements in sity, when less accuracy is satisfactory.
Diaphragm D is then removed and the photomultiplier is replaced
by a photo-transistor which directly gathers all the light diffused
571
-------
by all droplets.
II. OPERATING CHARACTERISTICS
The continuous flux counter operating characteristics have
been defined as follows : an ultra fine aerosol was generated
through radioiysis of airbone gaseous impurities, introduced, after
filtration, into a previously dust-freed vessel under the action
210
of a rays emitted by a source of P0 of 12 mC. The mean diameter
of particles thus produced - practically monodispersed and determi-
ned by a Whitby analyser - is in the order of 0.1 ^ . FIGURE 3 showt,
results obtained by varying the temperature difference AT between
alcohol and the cold wall of the Peltier effect module. Of course,
this difference in temperature is endowed with a qualitative signi-
ficance only and relates particularly to the geometric configuration
of the instrument.
Curve 1 represents the determination of homogeneous nucleation
threshold and of the nucleation thresholi over ions generated by
a source of C0 of 10 mC placed near the counter- These thresholds
were determined in the absence of any and all foreign particle
in the counter. They correspond to these differences in temperature
AT = 32° C and 30°C, respectively. The presence of ions lowers nu-
cleation threshold, as we had expected.
Curve 2 shows the same curve obtained in the presence of a
3
prior concentration of about 3000 particles per cm . We noted the
572
-------
I05
I I I I I I I I I I I I I :
Figure 3. Nuclei concentration as a function of AT of counter in
different conditions.
573
-------
existence of a level, thereby indicating that we had detected ac-
tually all present particles, and as well a slight shift of nuclea-
ticn threshold corresponding to the air cleansed of particles, or
to ionized air, and this can be explained by the prior condensation
of steam over present particles. We noted that homogeneous nuclea-
tion was obtained for a difference in temperature AT = 32'JC. This
would correspond, were we to allow that vapor temperature was 0°C
inside the module and that the air was previously saturated with
steam in the saturator, to an oversaturation of 700% of ethyl alco-
hol. The theoretical value which corresponds to homogeneous nuclea-
tion (formation of 1 particle per cm and by sec.), is of 230 %.
This results would imply that either saturation was not reached
in the saturator, or that condensation takes place on the walls
of the cooled nozzle.
Curve 3, obtained with the same diaphragms system in the presen-
b 3
re of a very high concentration of particles (about 10 per cm')
still shows the existence of a counting top level ; however, it is
not possible to p]ace in evidence the nucleation over ions, nor
homogeneous nucleation, and this can be explained by the catching
of ions over particles and by vapcr condensation available over their..
OF BRAT I CM OF THE LOW P R'-".S_SUR£ COUNTER
During our preliminary experiments which were intended to arrive
at the development of o counter whose purpose was to effect conden-
574
-------
sation nuclei counts in the stratosphere, we have? found that niotv
and more important count losses were encounted in the proportion
in which the carrier gas pressure decreased, in corroboration of
findings already obtained by SCHLARB (3) and JUNGE (4) with relief
counters.
In order to eliminate this drawback, we carried out a series
of experiments using the following device (FIGURE 4) :
- a reactive vessel of 200 1. in capacity, in stainless steel,
linked to a pumping system to reduce the pressure in this vessel.
- a continuous flux counter linked to the vessel by means of a
1 cm 0 hose and just a few cm in length, to reduce aerosol diffu-
sion losses.
We worked at laboratory temperature,. After emptying the reaction
vessel, we filled it at atmospheric pressure p0 with air taken
inside and charged with the atmospheric aerosol, hence with an
aerosol generated by heating a platinum wire through the Joule
effect. An experiment was also conducted by directly generating
in the vessel a photolytic aerosol obtained by rrradiation of
air gaseous impurities.
Counting is done under such conditions. Supposing at C0 the
concentration of particles. We then reduce pressure through pumping
inside the vessel to the value p. Supposing at C' the pressure
that we then measure. We see that the value C' is systematically
575
-------
cr
UJ UlUJ
2 <0
CO
-
$2 K
X UJ
UJ I O
_) 0- 00
Q. CO .-.
(T
UJ
H
(VI
CC
xl
o a:
z uj
UJ
o
cc
g
00
UJ
o
«
1
>
1
*
0.
Z
3
0.
E><3
0 -J J
H UJ 0
o w o
< CO ~
Ul UJ
o: >
V
A
V
A
\7
-*-
CONTINUOUS FLUX
NUCLEI COUNTER
oc
UJ
N
<5
0<
UJ
EC
3g
UJ?
a.
i
+j
(U
to
nl
•M
S
I
•3
0)
x
oc
•H
U
•T-l
4-
vt
3
576
-------
iower than the value C = Cc (*- ) that we should obtain by ,1) lowing
•p Po
that aerosol particles followed the air flow during pumping. By
bringing the pressure in this vassel to the value .->Q by introducing
filtered nitrogen, we found that the particle concentration, measu-
red under these conditions, assumes the value C . Errors in counting
which were discovered at low pressures cannot be attributed to
evaporation, nor to losses resulting from diffusion over walls,
nor again to aerosol coagulation. Therefore, we can conclude that
a portion of aerosols, although existing at low pressure, is not
counted under limited pressure.
FIGURE 5 show counting losses obtained with the three aerosols.
They practically reach 100 % at a pressure of 60 torr. Curves 1,
2, and 3 correspond to atmospheric aerosol, platinum aerosol and
photolytic aerosol, respectively. Counting loss is even more pro-
o
nounced in the 3c.se of photolytic aerosol whose particles (100 A 0)
o
are even smaller than those of platinum aerosol (600 A 0) and of
atmospheric aeros.' •
The measurement of signals delivered by the photomultiplier
shows that their amplitude lessens wi^h decreasing pressure, and
that below a given value of it they cannot be differentiated from
background noise. Thus, it would seem that at lew pressure a certain
number of droplets were not able to developp sufficiently and the
light diffused by each became too weal to be detected.
577
-------
SS01 %
V-i
D
05
M
01
J3
4-1
c
o
•rl
*J
O
3
u-i
en
rt
o
o
W-i
O
ca
to
o
(U
S-,
(JO
1-t
fs,
578
-------
We have compensated for the reduction in the velocity in which
droplets are developing under low pressures by increasing the over-
saturation in the counter by heating the alcohol in the saturator
and by maintaining as constant the temperature of the cooling
nozzle to a value of 0° C ± 1° C.
Thus, we are able to obtain a sufficiently high over-saturation
at cooling nozzle outlet to cancel out the counting loss up to
a pressure reaching about 60 torr.
A model specifically developed for the study of stratospheric
particles, insensitive to ionic nucleation, has enabled us recen-
tly to carry out a primary test up to an altitude of 17 km. The
analysis of measurements is presently underway and it should ena-
ble us to evaluate a concentration of at least 100 particules per
3
cm of air at that altitude.
5J3
-------
BIBLIOGRAPHY
1) J. BRICARD, G.MADELAINE, P. REISS, P. Y. TURPIN,
C. R. Acad. of Sciences , PARIS, 275 Series B 387 (1972)
2) J. BRICARD, P. DELATTRE, G- LEBEAUPAIN & G. MADELAINE,
C. R. Acad. of Sciences, Paris, 278 Series B 191 (1974)
3) G. SCHLARB, Bi^klirn Beiblatter 7.86 (1940)
4) C. E. JUNGE, Journ. Meteor. 18-181 (1960).
580
-------
ELECTRICAL MEASUREMENT OF AEROSOLS
by
Kenneth T. Whitby
Mechanical Engineering Department
University of Minnesota
Minneapolis, MN. 55455
ABSTRACT
This paper reviews the basic principles of electrical size distrib-
ution and concentration measuring methods. Electrical aerosol measuring
instruments are discussed under three headings: charging, classification
and detection.
Recent work by Liu and his students shows that the particle charging
rate, based on the Boltzmann Law, can best describe bipolar and unipolar
diffusion charging of aerosol particles, over the size range and charging
conditions of most interest to electrical aerosol measurement. Working
equations for calculating diffusion charge are presented and the latest
Several charged aerosol classifiers or mobility analyzers are de-
scribed, including the latest differential mobility analyzers developed
at the University of Minnesota.
Particle charge versus size curves for bipolar, diffusion and field
charging are integrated with some of the latest aerosol size distribution
models, in order to calculate the response characteristics of the dif-
ferent kinds of electrical instruments.
Three late model electrical aerosol instruments developed at the
University of Minnesota are described. These are the Electrical Aerosol
Analyzer, the Differential Mobility Analyzer, and an Electrical Aerosol
Concentration Meter. ,'
581 \.
-------
NOMENCLATURE
a - particle radius, cm
Di - diffusion coefficient of ion, cm^/sec
Dp - particle diameter, urn or cm
e - charge per electron - stat coul per charge
f(r)- particle size distribution function
fc(r)- fraction of particles charged
g(r)- characteristic response function of the instrument
h - dimensionless parameterK|k-l| ,fr> , ,
TrM IT \ I \*-&K-*- )
K i^k+lj
I - current, amperes or pico amperes.
Jc - charging rate in the continuum regime, number/sec
Ji<. - charging rate in the free molecular regime, number/sec
K - dielectric constant, dimensionless
k - Boltzmann's constant
l(r)- instrument sampling efficiency
mi - mass of ion, g
m.: - mass of gas molecules, g
N - particle concentration, no/cm
Ni - ion concentration, no/cm-'
NT - total number concentration of particles, no/cm
np - number of unit charges per particle
npf - final particle charge, elementary charge unit
npi - initial particle charge, elementary charge unit
P - pressure, atmospheres
ro - distance from center of particle where the image force of
attraction is equal to the columbic force of repulsion
SG - geometric standard deviation
ST - total surface area concentration, pm^/cm-'
s(r)- sampling efficiency of aerosol transport system
T - temperature - K°
t - time, sec.
U-j - mean velocity of ions, cm/sec
Ui - mean thermal speed of ions^ cm/sec
V - voltage, volts
VT - total volume concentration, ym-Vcm
Z - function of charge parameter, npe2/akT (See Fig. 2, Gentry &
Brock, 1967)
Zj - electrical mobility of ions, cm^/volt - sec
Zp - particle electric mobility, cm2/ volt - sec
a - fraction ion captured by the particle from the limiting sphere
(See Table I, Fuchs, 1963)
[_- electrical aeroso] analyzer sensitivity, pa/(!06 part./cc)
- a factor of the order of unity (sec Eq. 3, Natanson, 1960)
- constant in the equation Di = Bi U^ Aj_
582
-------
6 - radius of limiting sphere, cm (see Eq. 5, Fuchs, 1963)
n - =6/a, dimensionless
^ ~ - \l?t /\.\f—^—-1 » dimensionless
ir1/"1 (a/Ai) I m-^
X^ - mean free path of ion, cm
A. - mean free path of gas molecule, cm
p - =ro/a» dimensionless
$ - electrical potential, statvolt
-------
ELECTRICAL MEASUREMENT OF AEROSOLS
by
Kenneth T. Whitby
Mechanical Engineering Department
University of Minnesota
Minneapolis, MN. 55455
INTRODUCTION
The high electric mobility of aerosol particles in an electric field
makes it possible to separate and classify aerosol particles electrically.
If the electric mobility is a single valued function of particle size,
it is possible to use mobility classifiers as size analyzers. Also the
net electrical charge carried by an aerosol can be measured by an electro-
meter and used as a measure of aerosol concentration.
Although these principles have been known for a long time it seems
that Rohman-'- in 1923 was the first one to attempt the measurement of an
aerosol mean size by electrical means.
The rapid advances in electronics during the past few decades com-
bined with an increasing need for fine particle measurement, has resulted
in a number of electrical measurement techniques and instruments being
brought into general use.
It is the purpose of this paper to review the basic principles of
the various electrical techniques in the light of the latest knowledge
about electric aerosol charging and the size distributions of aerosols
and then to discuss in more detail several of the latest techniques and
instruments developed at the University of Minnesota for the classifica-
tion and size distribution measurement of aerosols.
Emphasis will be on the developments of the last decade and on those
techniques which have either demonstrated their usefulness or have the
greatest potential for future development.
BASIC PRINCIPLES OF ELECTRIC AEROSOL MEASUREMENT
An electric aerosol measuring system can logically be broken down
into three parts. These are:
* aerosol charging
* precipitation or classification by electric mobility
* aerosol detection
584
-------
The following discussion will be organized in this way.
Particle Charging
There are three different ways of charging aerosols:
* electrification of particles which results from the contact
and separation of a particle from a surface.
* the induced charge which results when a liquid drop separates
from the bulk liquid in an electrical field.
* charging by coagulation of one charged particle with another.
The charged particle may be an electron, a small ion or another
charged particle.
Contact Electrification - Contact electrification is of concern to aerosol
technology for two reasons. First, it produces enough unwanted and un-
controllable charge on the aerosols produced by dust generators so that
all such generators should include some provisions for removing or re-
ducing this charge. Secondly, contact electrification may also be used
to charge aerosols for aerosol concentration measurement.
Because of its industrial and scientific importance there have been
periodic conferences to review the state of knowledge of particle electro-
statics. Harper^ (1967) is a good review on static electricity. Static
Electrification 1971, is a good recent conference review as is Moore^
(1973).
There have been numerous instruments developed, a few of which have
been developed for industrial dust measurement applications. Among these
are the "Konitest", the deveopment of which is described by Prochizka^
(1966) and the IKOR Air Quality Monitor5 (1974).
John in another paper in this symposium has reviex^ed the application
of contact charging for particle monitoring. In most of these instru-
ments the particle charge generated by collision with the walls of the
plumbing leading up to the sensor or with special impact surfaces in the
sensor, generates an electric charge which is then collected by a final
impact of the particles with a probe connected to a current measuring
circuit. Although the relationship between aerosol mass concentration
and current may be linear and reasonably reproducible for a given material
and given humidities, John5 (1974) using the IKOR instrument found that
the sensitivities varied over a range of more than 20 to 1. The insula-
tors he tested had a mean sensitivity of 0.2 ycoul/g, semiconductors 0.62
ycoul/g and metallic conductors 2.25 ycoul/g.
The variable and low sensitivity of contact electrification has so
far limited its use to the measurement of high concentrations of aerosol
I
\
185
.1
-------
from industrial sources.
Guichard and Chauvelier (1973) have described a novel instrument
in which the aerosol is first charged by passage through a fluidized
bed and then the charge measured by a second fluidized bed connected to
a current measuring circuit. It is claimed that this instrument is less
sensitive to variation in material properties.
Charging With Small Ions - The most important method for charging aerosol
particles, prior to measurement, is by the coagulation of the particle
with a charge carrier. The charge carrier may be an electron, a small
ion or another charged particle. Because of the importance of this
method of charging, it will be discussed in much more detail than other
charging methods.
In the most common and useful approach, small ions generated by a
radioactive source or by a corona discharge are allowed to charge the
aerosol to be detected or collected.
Making small ions for aerosol charging is not difficult. However,
making an appropriate and constant concentration and exposing the aerosol
to it for the correct length of time is much more difficult. Before
discussing some of the latest advances in methods of precision for
aerosol measurement it is necessary to review the basic .principles of
partible charging by small ion clouds.
There are three basic kinds of small ion charging which may be
identified, each of which results in a different magnitude of charge and
a different relationship between particle electric mobility and size.
The first kind of charging exists when a particle is immersed in a
bipolar mixture of small particles in the absence of an electric field.
Under these conditions of bipolar diffusion charging, individual particles
fluctuate in charge as the particle acquires + or - ions as a result of
collisions of ions with the particle.
The second kind of charging exists when the ion mixture is unipolar,
but there is no electric field. Under these conditions ions diffuse to
the particle until the repelling force of the charge on the particle
reduces the probability of acquiring further charge to a small value.
The third kind of charging occurs when an electric field is super-
imposed on the unipolar ion cloud. The additional velocity of the ions
along the field lines which pass through the particle can increase the
equilibrium charge for given ion concentrations and charging times above
those for unipolar diffusion charging, especially for partizles larger
than a few tenths of a ym.
586
-------
Apart from such operational problems as charging stability, uni-
lorinity of charging in different parts of the charger, etc., there are
thrp.i characteristics of ion charging which affect usefulness of a given
charging method for aerosol sizing by electric methods. These are:
.1, The relationship between electric mobility, Z , and particle
size, D . The average charge for the different kinds of
charging are shown in Figure 1, and the resulting mobilities
in Figure 2. Note that the 2 vs. D curves are single valued
for the Boltzmann bipolar diffusion and the unipolar diffusion
hut not for the field charging. Also, note that the slope of
the unipolar diffusion curve is essentially zero above 1 ym. Thus,
sizing methods using unipolar diffusion charging will have
\"?ry poor resolution above about 0.6 ym at atmospheric pressure.
?. fraction charged. Particles which do not acquire a charge
during their passage through a charger, cannot be influenced
by subsequent electric fields and, therefore, cannot be
measured electrically. Recently, Liu and Pui-^nave verified
that Equation 1 is an accurate description of the fraction
charged during unipolar diffusion charging.
fc = 1 - exp(- I Dp2 u± N±t) (1)
In Figure 3 the fraction charged, f , calculated from Equation
L is compared to the loss fraction for the University of
Minnesota electrical aerosol analyzer. At an Nt = 10' the
Flection charged decreases rapidly below 0.02 um to a value
of «nly 0.01 at 0.002 ym. From Figure 1 it is seen that for
bipolar diffusion charging the fraction charged is only 0.003
at 0.01 ym. For any electrical instrument measuring aerosol
concentration (neutral as well as charged particles), the
Indicated concentration in a given size range must be cor-
rected by dividing by f . Since the error in the corrected
concentration will be proportional to lAf when Af is the un-
cfMrainty in f , errors in corrected concentrations become un-
a^coptably large at values of f < 0.03. Thus, fraction charge
in combination with aerosol losses, becomes the principal
factor which limits the lower useful size of an instrument.
H. Discrete nature of electric charge. From Figure 1 it can be
snan that the average charge goes from less than unity to
aiore than unity at about 0.04 ym for unipolar diffusion
charging and at about 0.3 ym for unipolar diffusion charging.
Tl:-' discreteness of the charge at these sizes limits the
resolution that can be achieved. Husar^ estimated this as
corresponding to a charge a = 1.3 for unipolar diffusion
87
-------
1
o
.e
§10
II
g*
c
"«J
O
II
9
O
.C
O
9
O
1
.01
.0
Field -=—7 '
Nt=l07 / >
E = 5kV / X
/' /
— Diffusion x / /
E Nt = I07 \ / /
: \S
Unit — , /
" Charge 1 / Unit ^ ^-x'
/ Charge J.-"'
: / .x'
/ X
/ ^
/ X
/ -x'
/ /-/
/ / Bipolar
J •
i i
i
i
, 1 IMII/ , .1 ! , , 1 MM
03 .01 .1
Dp , /Am
Figure 1 Average charge of particles charged in four different ways
For sizes smaller than that at which the maximum charge is a
unit charge, the curve for smaller sizes corresponds to the
fraction charged.
588
-------
DTIITT'T ~jTTTTrTT~'|
—
"' '
--
f
.
jfX&W^
— -p N ^\ •"ttaag.-^
= ir ,- X^^^xr-
wa .-5 ^^^ "" .
W4. **f ^^^A
- "x.
— "cC l! ^s^.
x ~ ^
UJ •?- ^*
J-LLJ.~L.LJ _J III 1 1 1 L I 1
III 1
g
H
z
— -^e
"""""• ~^^,
d
JJJ1_
0 n o
0 i
3
111 =
—
—
—
-
.
rH
4J
CO
O
B
^
cn
cu
CO
CO
O
•H
CU
60
M
cd
42
O
CU
U
cd
ft
CO
B
~-i
,— i
o
•
o
s
o
rH
CU
43
O
4-J
O
4-4
CU
CJ
3
T3
cu
V-i
*
l-i
01
00
1-1
cd
42
CJ
cu
42
4-1
C
•H
CO
cu
cn
CO
O
rH
cu
CO
01
42
H
*
cn
cu
3
rH
cd
>
rH
cd
O
•H
4_J
cu
\-t
o
cu
42
4-1
CU
42
4-J
60
C2
•H
N
•H
CO
1-1
CU
5
O
rH
CU
42
4-1
4-J
•H
B
•H
rH
>-,
H
CU
•H
4-1
O
cu
4-1
4-i
01
•
B
3.
r~-
o
0
•
o
4-1
3
O
43
cd
O
4-1
^
4-J
•H
rH
•H
43
0
g
CJ
•H
H
4J
CJ
CU
rH
CU
CU
A
H
cd
42
4-J
cu
4-1
O
J3
•
CO
!^
cd
&
4-J
C
CU
r>i
CU
U-l
iw
•H
T3
l-i
3
O
14-1
(2
•H
t3
CU
63
l-i
cd
42
CJ
4-J
3
O
42
cd
4J
cd
!>,
4-)
•H
rH
•H
42
O
B
£
3
B
•H
C
•H
B
CU
42
4J
4-1
O
CU
cn
3
cd
a
cu
43
UJ
CU
N
•H
CO
o
s
4_j
fl
60
t2
•H
60
}-l
cfl
42
CJ
-d
rH
cu
•H
4-1
0)
42
4-J
1-1
0
4-1
0
3-
rH
•
O
•
(^
4-1
•H
rH
•H
43
o
B
CU
e
cd
CO
CU
42
4-1
CU
>
cd
42
>^
B
O
60
•H
tn
589
-------
charging at 0.04 ym and similar calculations by Knutson *ori
bipolar diffusion charging give ag = 1.3 at 0.3 ym. Knutson
has developed a calculation procedure that corrects for
this effect on the measured size distribution. Cantrell
has also developed a correction procedure for the electrical
aerosol analyzer that includes a correction for discrete
charge. Although these correction procedures have not been
thoroughly evaluated yet they appear promising enough so that
resolution of monodisperse aerosols with geometric standard
deviations of less than 1.1 at the worst size may be possible.
Of course if measurements are made using only the singly
charged particles, then the resolution is as good as the
resolution of the mobility analyzer. This requires, however,
that the fraction of aerosol carrying unit charge be known
accurately.
Diffusion Charging Theory
Since most existing and potential electrical methods for the measure-
ment of aerosols utilize diffusion charging, the theory of this charging
method will be examined in more detail than contact, induction or field
charging.
In a recent thesis Pui , has examined various diffusion theories
and compared them with experimental measurements. The various theories
are listed in Table I along with brief comments on their range of ap-
plication. Only those which have the most experimental confirmation will
be discussed in detail.
14
Liu and Pui found that the bipolar charge equilibrium reached by
aerosol particles immersed in a bipolar ion atmosphere is best described
by Boltzmann's Law, Equation 9, Table I, over the size range from 0.02
to 1 ym. for which they did experiments. Liu and Pui-'--' also found that
the Nit product required to reduce a highly charged aerosol particle to
the Boltzmann equilibrium charge to be best described by Equations 8a
and 8b, Table I. For a 1 ym particle N^t values on the order of 10^
cm^/sec are required to neutralize it with small ions.
13
For unipolar diffusion charging Pui found Equation 10, Table I,
agreed best with his experimental data over the size range of his exper-
iments, 0.0072 to 5.04 ym. This modification of the White Equation,
equation 1, Table I, contains the Liu-Bademosil^ interpolation formula
between the free molecular and continuum region. Why this relatively sim-
ple formula which contains no correction for image force should agree with
the experimental data so much better than some of the more inclusive
theories, is not known at this time.
590
-------
tn
OJ
•H
VJ
o
a)
60
C
•H
60
t-i
tfl
rC
u
o
•H
(fl
3
•H
O
5§
•H -H
n) • C O
CL 0) O
•3-8
U CO
00 *i
01
TJ
• B CO 'H 1H
01 CO J2 rH
00 O rH J=
Lj B O *J
p UJ O iH
o
i M O
• iw • OOt
O
00 CO
0) 0)
U 00
§!»
,c
I 4J
) -H
: 3
o>
C rH
o o
3 M rt
CJ 0) d
01 T>
i e K a c
O CO -H O
I U J3 T(
I 0) " 0) B
i o rt oo rt
V4 M
; O y-4 ccj in
I u-i O J5 O
•oS
OS
00 01
01 O
V C.
3 O,
o
z
•H
13
•H 41
u CJ
§-3
s
Jl
3
to
rt
H
b o^S
-• U 3 ^
ai
B rt "
rt rH O
0) 0) B
CW >W CO
0) 01 iH
01 01 O
us *: a
jn o
u co -
3 H
591
-------
00 M
01 01
Li TJ
O u
M -H
•H S
3
Xi
00-H
o> tn
M C
O
C CJ
o
•H O)
4J O
•H P
en o
rt
w a)
rj
60 M -H
U 4J
U< 00 U
C 0>
c *H a
o oo to
•HMO)
•H £•
to o *
R) 01
n x: -o
4J *J C
n]
C 0)
-H M -*
rt
01 X
Xi l"i O
CQ 0)
O TJ -C
•H C *J
rH Oj
x: M
U MH
a g
rH Z 3
co s
•H 01 O
U DO -H
•H Li J->
B CO B
•H f O
0 0
00 rH 00 B
•rt (0 B -rH
x: B -n
iH 00 01
0 CO XI
> x: to
B o u u
O rH •H
4J O cfl CL
O W rH Q<
3 on
3
CJ
01
0
B
01
1-1
B
01
XI
CO
CJ 01
rH TH
CL 00
a oi
CO U
CO 00 >
^S °
4J 0 rH 00
CO O. CX B
01 01 3
LI a. 111
•H X 10
4-1 01 01
01 rH CL
rH
B CO »
TH 6
£ rH
01 4-1
rH -rH 00
XI 9 B
s-
01
X
01 0)
Qj 4J
cn-o
T3 I
a
x
01
/I,
".J
I
I •>
I O
3
•H
rJ
N-i
U Xi
H O
592
-------
r|1'ie average number of charges per particle, np,calculated from
Kq'iation 10, Table I, is shown in Figure 4 for three N.t values in the
r.iiMV1 of practical interest. The average number of charges for values
of Up less than 1 have been calculated from the fraction charged
equal ion. It can be seen that N.t values on the order of 10? cm or
higher should be used in order to obtain a sufficient fraction charged
for :;izes smaller than 0.02 ym and to obtain sufficient charge for the
larger particles.
Pui also found that the ions from the unipolar charger used ap-
peared to have an electric mobility of about 1.4 cm^ volt~l sec~l
co<-> <>sponding to a molecular weight of 109. amu. These values have
hi.-nn used in the calculations for Figure 4.
31
f'fi cntly, Pouprix (1973) has developed an arrangement in which the
ion.<; produced by alpha particles near their maximum range in air, can be
UKC'J to produce a unipolar diffusion charge that is nearly identical to
thru, for an N.t -- 10^ as shown in Figure 1.
if aerosol particles immersed in a unipolar ion cloud are also sub-
ject to an electric field, particles larger than about 0.1 ym acquire
additional charge above that acquired purely by diffusion. The magnitude
of the charge acquired by field charging is a function of the dielectric
constant of the particles and the electric field as well as the variables
sv,dificant in diffusion charging.
The electric mobility of aerosols charged by bipolar, diffusion and
i i - !.d charging for a particular set of conditions is shown in Figure 2.
i-Yii particles smaller than a 0.1 ym, the electric mobility is essentially
i'! '"pendent of the method of charging because the particles either have
a "u.t charge o\- no charge. For particles smaller than about 0.1 ym the
pi :'>riplc charging mechanism is diffusion and an external electric field
m.11- •;; little difference. For particles larger than a few ym, the charge
it: IP electric Field is essentially proportional to Dp2 with the result
Ll'.iC the. electric mobility, Z , is proportional to D .
M. •;•; LiiJLJ^lilkL^'?53 and Precipitators
K!metrical acror.ol concentration measurement requires some method of
fl '-cting the charged aerosol. In addition, electrical sizing methods
f'jiure a precipitator or classifier by which the particles of different
i>! "M ric mobilities are. separated before detection. The various approaches
i;.; shown in Figures 5 to 10. The characteristics and advantages and
c'' s.-idvan-fcages ul each are discussed below.
Ion Counter - Figure 5 - The condenser of this type counter
593
-------
UJ
_J
O
h-
CE
o:
UJ
o.
UJ
cc
UJ
UJ
o
cr
<
X
o
o
cr
LU
m
1000 cr
100 =
001
Figure 4. Number of unit charges calculated from Equation 10,
Table I for diffusion charging at three different JLt
values. The fraction charged f calculated from Equation
1 is also shown for comparison.
594
-------
may consist cither of parallel plates or of concentric cylinders. The
current collected from one of the electrodes is measured as a function
of applied potential difference by means of a pico ammeter. Although
this instrument is deceptively simple, to obtain AI and hence the number
of partic.les falling within a given mobility range requires that the
s urrent versus voltage curve be differentiated twice. This seriously
limits its accuracy for size distribution measurement purposes. It has
mostly been used for measuring small and intermediate ions of high
mobility because having the pico ammeter connected to one electrode of
the precipitator also limits the maximum voltage that can be applied to
about 700 volts and hence limits the lowest Z that can be measured.
P
Israel " (1969) has made a comprehensive review of the design, applica-
tion and application of the condenser ion counter. Mohnen33 (1966) used
.' variation ol the cylindrical condenser that is similar to that shown in
I•'.gure 'J j 11 that the aerosol is introduced as a layer around a sheath of
•'lean air. Measurement was from the collected deposit on the center
electrode,.
Parallel Plate Sheath Air Precipitator - Figure 6 - By introducing the
charged aerosol as a thin sheet between layers of particle free air
differential deposits may be obtained. Instruments of this type have
k"..-n developed by MeGaw and Wells34 (1964) and in a miniature form by
Yeh, Raabe and Wood^ (1973) and used for precision measurement of aerosol
charge distribution. Its greatest disadvantage is the great amount of
labor necessary to evaluate the deposits on the plate in order to obtain
good si.7.e distribution data.
!tenuder_ - Figure 7 - The Denuder developed by Rich et al. (1959), uses
n condensation nuclei counter as the detector at the end of a precipitator
lo measure the fraction of the aerosol charged. If the fraction charged
is related in a unique manner to the particle size as it is for a Boltz-
•nann e-.uiiJ ;K rium bipolar charge distribution, then the instrument can
i e use! ns a particle size distribution analyzer. However, Figure 7
'hews thai Ihis technique is limited to the size range from about .02
'o 1 inn, !f the aerosol to be measured has a large number of particles
.•nailer '_h;m ,02 |jm, then the change in nuclei count as the voltage is
varied If-, : co small to provide useful estimates of aerosol size distrib-
ution. This is often the case near sources of aerosol from relatively
••loan conil iv l ion. Nevertheless, this is a useful technique which because
'•' Lt-j.,simp i ic i t.y, has been used in atmospheric field studies, Fluger
i: al. J/ ('"W',
1 on JJjipUm- Aerosol Concentration Measurement - Figure 8 - If essentially
neutraL acrcyol particles are passed through a corona charger as shown
MI Figure- 8, the aerosol particles will be charged by the flow of ions
' rom the- rr»u?r io the outer electrode. The capture of the ions by the
/ 595
-------
_ 1
8 o
o
o
•o —
^-
g I:
X
i X
•••^^ —• •""
o
o «
&S
•0 Q.
to
o.
c
o
JS
4J
t8
0)
J2 •
CD ^1
O
01 4J
4-1 n)
n) -u
rH -H
p. 0.
•r-l
rH O
0) 01
.H >-l
^H ex
CO
1-1 >J
cfl -H
CU
J-i
bO
•H
fn
o
*J
CO
•H
ex
O
0)
tO
0)
N
to
c
>>
4-1
•H
01
t-l
00
•H
fllff
o
V)
o
0)
N
tO
C
CO
>>
4-1
•H
O
Ol
o.
Ol
en
d
0)
•o
C
o
u
Ol
u
60
•H
596
-------
:u:rosol part Lcles reduces the effective electrical mobility of the
charges no ('hat they are carried out of the device instead of reaching
I he nuter olectrode. This reduction of current is a measure of aerosol
Concentration. The exact relationship between I and the aerosol con-
ccntratior doponds on the method of charging, and the aerosol size
distribution. The relationship between I and aerosol concentration
(called sensitivity), will be discussed and illustrated later for several
typical r.i/,e distributions.
PRINCIPLE OF THE DENUDER
AEROSOL AT
BOLT /MANN _____
FOUILIBRIUM
n (fr rv-~
I
1
CONDENSATION
NUCLEI ^
COUNTER
FRACTION
CHARGED
0
0.01 0.1 I
PARTICLE DIAMETER, MICRONS
Deuuder
- nuitablp i".liter charge collector. Then I becomes an increasing function
;«L the coiu untration, Mohnen and Holtz^S (1968).
!il!Lili MojiiHt/y Analyzer - Figure 9- To overcome the limitations of the
previously described types of mobility analyzers, Whitby et al. (1966)
a'l. C.liarj-c particle collection for current measurement is accomplished
i>-' a sepj'Ptc carefully shielded and isolated filter connected to the
j/ii'oanmu:! 01 , Separation of the precipitator and current collection
'unctions r 'n»its the use of precipitator voltages near breakdown and,
i !;vrefoi.. . ' ho collection of particles with electrical mobilities less
i him 10"/l ""i,'«tc per volt/cm. With good shielding, the precipitator
597
-------
voltage may be varied without shorting of the electrometer input. This
permits completely automatic operation as in the latest commercial
instruments, Liu, Whitby, and Pui^° (1974).
PRINCIPLE OF ION CAPTURE
AEROSOL CONCENTRATION MEASUREMENT
NEUTRALO
1
CHARGED
AEROSOL
AEROSOL CONCENTRATION
Figure 8. Ion capture detector of the type in which the current is
reduced by the presence of particles.
With this appraoch, obtaining the current increment in a given particle
size range requires differentiating the I vs. V curve once. This will
be discussed later.
Differential Mobility Analyzer - Figure 10 - For high accuracy research,,
on particle charge distribution KnudsenlO (1971) and Knudsen and Whitby
(1975) developed a precision mobility analyzer that retains all of the
desirable features of the Whitby mobility analyzer, but in addition is
capable of classifying out and delivering a suspended, charged aerosol
having less than 2% spread in electric mobility to either a current
collector or to an external device. Liu and Pui-^ (1974 have further
developed this analyzer into a precision monodisperse aerosol generator
and have applied it to the calibration of condensation nuclei counters
and other instruments. Its features and application will be discussed
in more detail later.
598
-------
M CO rH
01 -H
N 60 X>
>. C Q
R)
C
CO
•H
t-H
•H
O
CO O
jr -H
>-i
c
CO
to
o> u e
t-i cu ai
u rH x;
•H 0) 4-1
n)
o at
i-H 01 -H
cfl CU 60 rH
•H 4-i e a>
4J CO CO T3
C u )-<
cu co Td i—t
vj a 3 c o
o cu o co en
u-i en M o
14-1 t-i en V4
•H O cfl 01 01
O J-1 C -H CO
60
•H
0)
60
cc
1_
(-
d
s-. ,r>
cu
N T3
>, (1)
to
cu
rH !J
U O
4-) Cfl
!-l 4-1
Cfl -H
D, D.
•H
>~, U
J3 CU
>,
co
o
CU
SC T3
M Oi
CO 4-1
U
•H
J3 .
O O CU
CU rH
^ i-i O
X 3 <->
4J CO
•H cfl 4J
JZ CU O
sec
C^
cu
3
60
599
-------
Time of Flighr Ion Mobility Analyzers - For the measurement of the elec-
trical mobility of ions and particles having mobilities greater than about
0.1 cm /volt, sec, time of flight mobility analyzers have been developed
which have two pairs of grids spaced a few cm apart. With the proper
application of combined AC and DC fields to the grids, the ion mobility
spectra can be measured with high accuracy and resolution. Typical of
the instruments of this type used for atmospheric ion mobility analysis
is that of Huertas et al^ (1970).
Electrical Aerosol Instrument Response - The response of an aerosol
measuring system is rather complicated. The response is generally a
function of the characteristics of the instruments, the characteristics
of the aerosol transport system and the characteristics of the aerosol.
For an electrical aerosol measuring instrument the current contribution
dl to the radius range, dr, is given by
dl = f(r) g(r) l(r) s(r) dr (2)
where fc(r) is the current contribution of the total instrument output due
to particles in the radius range dr, f(r) is the particle size distrib-
ution function, g(r) is a characteristic response function of the instru-
ment l(r) is the instrument sampling efficiency (ratio of aerosol con-
centration in the instrument sensing volume to the true aerosol concen-
tration at the inlet to the instrument), and s(r) is the sampling ef-
ficiency of the aerosol transport system bringing the aerosol from the
sampling point to the instrument inlet. For electrical aerosol analyzers
the instrument response function g(r) may be further expressed as:
g(r) = fc(r) np(r) (3)
where f (r) is the fraction of particles charged and np(r) is the number
of unit charges per particle. The fraction charged f (r) is given by
Equation 1 and D (r) depends on the method of charging, e.g. Figure 1.
The sampling efficiency of the aerosol transport system up to the
inlet of the instrument determines s(r). Because instruments intended
for atmospheric aerosol measurement respond primarily to particles smaller
than 1 ym radius, s(r) may be assumed equal to 1 except for the particles
smaller than 0.007 pm, where diffusion might be significant if very long
sampling lines are used.
Instrument Response for Several Size Distributions - To illustrate the
response characteristics of instruments which utilize bipolar diffusion
or field charging, the average charge in Figure 1 has been integrated
with four model size distributions who's parameters are tabulated in
600
-------
IH
M
01
rH
,0
co
H
10
4J
C
111
E
3
M
4J
0!
C
-rl
rH
0
1C
o
,i<
CO
ca
rH
tfl
O
•H
l-i
*J
O
CU
rH
CU
OH
O
Ci
to
c
a
to
3
II)
g
VI
•U
3
.0
•H
Vi
4J
10
•rl
•o
ft!
N
TH
to
rH
01
13
0
£
H 0
> rH
r*.
r*.
VI rH
VD
Z rH
H
>
H
«
H
Z
Z
O
1/5
z
U
C5
CN
.
H 0
> rH
r-
H VO
t/3 rH
\D
Z rH"
•
43
4J
vl
w
•rl
H
Z
vf
^^
ja
4J
•ri
X
^i
U
01
3
rH »
CO C
> O
•H
O 4J
vH 0)
rl 3
01 42
4= S
O. O
' w o
o
e c
4J CO
CO 01
rH
O> U
00
CO !»>
kJ rH
0) 01
> >
CO -rl
4J
o a
4-> rH
CJ
rH VJ
«
3 e
cr o
CU M
ilH
C
O OJ
0) -O
0 0
o E
f.
O -H
o>
55 rH
TJ LI 0
3 t« 3
CO C
rH -O
C IT)
Vi CO
(J f~.
» Z rH
o o e
R- a u
s
rO
U
-H
g
rH
O
to
0
rl
O>
CO
n
CO
42
rl
3
01
00
CO
M
O>
>
CO
O
U
rH
CO
3
tr
Ol
r-,
rH
Ol
U
!
X
o
)-l
ex
ex
a
a
Ol
CO
o
o
x:
o •
m
Z 01
(J 3
in rH
to
•o >
e
CO C
CO
Z 43
0 M
C 3
43
vO
0)
rl
a
M
iH
If
c
•H
g
O
4=
(D
C
0
v(
U
3
43
•H
rl
4J
to
•H
•O
01
,i!
0
1
01
c
o
U
01
0
<
01
X
4-1
00
•H
U
4J
TH
UH
f>»
43
•O
01
c
•rl
CO
.U
43
O
CO
k*
Ol
±1
01
e
a
Vi
«
(X
C
o
•H
•U
3
42
iH
rl
U
in
•H
O
U
rH
It)
3
of
01
4J
3
O
43
re
o>
43
0
J-J
e
01
tn
o
o
f.
o
g
1
*^T
}
o«
d
^
•tf
o
n
a
o
•o
^
•o
r.
E
3.
O
rH
O
4J
rH
O
01
00
C
CO
rl
.
01 C
H O
iH -rl
n u
3
CU 43
4= -rl
U M
U
C 01
•H -H
•o
c
O rH
2.3
»o s
t-l 43
4^
n 01
Tt X
•O "
> M
CO O
i-H i*-i
h H
01 >
3
O 0
•o
601
-------
Table II. Whitby^3'^ has recently shown that atmospheric aerosol size
distributions may be modeled by adding three log normal distributions,
two of the modes being in the submicron size range. It has also been
shown that the smallest of these submicron modes (names the nuclei mode)
originates from the primary particles produced in combustion. The second
submicron mode (named the accumulation mode) results from coagulation and
condensation.
The submicron size distribution of combustion aerosols also show the
same modes. This is illustrated in Figure 11, where the surface distrib-
utions of several combustion aerosols is shown. It has also been found
that these combustion size distributions can be modeled well by two log
normal distributions.
AS
3-
2r
S A log Dp
005 0.01
0.05 01
Dp-/i
0.5
Figure 11. Normalized submicron surface area size distribution of
several combustion aerosols produced by diffusion flames
compared with an atmospheric aerosol size distribution observ-
ed in a relatively clean site in Ft. Collins Colo. The numb-
bers are the surface area in ym2/cm3 for the aerosol as meas-
ured. Note that the nuclei modes and the accumulation modes
for the combustion aerosol correspond to those observed for
the atmospheric aerosol.
602
-------
The nuclei mode distribution in Table II represents the kind of
distribution that would probably result from relatively clean combustion
aerosol that is diluted rather rapidly. It is not visible because it
contains practically no particles in the light scattering range.
The accumulation mode distribution is representative of a well aged
aerosol that is fairly dilute. It is highly visible because its surface
mode corresponds well with the peak of the light scattering extinction
curve.
The bimodal distribution was obtained by fitting an actual acetone
smoke distribution from a diffusion flame that had undergone moderately
rapid dilution. Because it is fresh, the nuclei mode is still apparent,
but hetrogeneous coagulation of the nuclei mode with the accumulation
mode has transferred most of the aerosol mass to the accumulation mode.
The power law model distribution has been included because it is a
fair lit to the number distribution over the size range from about 0.1
to 10 pm.
The resulting sensitivities expressed as the pico amps, current that
would flow if the charged aerosol was collected at 1 cm^/sec at 106
particles/cm , 1000 um^/cm^ and 100 ym-Vcm^ respectively are given in
Table III. The last two columns give average sensitivities and the %
standard deviation of the sensitivities for the four distributions. A
number of conclusions may be made from this table.
1. An electrical concentration instrument using diffusion or field
charging would be from 7 to 8 times as sensitive as one using
bipolar charging.
2. The variability of sensitivity with variations in size distrib-
ution is about the same for all three methods of charging,
about 140% for number, 90% for surface area and 180% for volume.
Real smoke distributions would be the closest to the bimodal distrib-
ution so that the sensitivity figures for that distribution would be the
best choice for predicting actual instrument performance.
The number distribution, dN/dlog Dp, and the number distribution
multiplied by g(Dp) from Figure 1, is shown in Figure 12. It is seen that
all of the charging methods weight the accumulation mode most heavily.
This is especially true for diffusion and field charging. This would be
desirable for a fire detector because the nuclei mode results mostly from
clean efficient combustion whereas a heavy concentration of smoke would
result in a large accumulation mode.
603
-------
M
M
41
rH
,0
1C
H
1
"O
r-l
41
•H
(•V
•C
C
B
r
c • •
•r C
t: o
3 -H
IK 4J
n-, 3
•H .0
T -H
rl
I* 10
ft -H
rH TJ
O
O. O
•r. N
CQ -rl
10
H
O rH
14-1 41
•o
41 g
•rH
»-> 0)
TH 0)
*rl ,rt
4J 4J
to rl
C 0
4' V*-|
10 to
4J C
C 1H
41 60
rl rl
£.3
O 0
^
o
*-}
H
«
1
}-4
O
id
c
c
,
r,
°
4-1
SNS CO
>,
•rl
I!
•H
C 1C
c c
O 41
!r; to
3
rH
^
Q)
0
P-
1
rH
nj
'O
o
K
to
I
c:
o
•H
i_i
o
rH
3
E
3
O
O
1
•rl
01
U
3
Z
1
rH
U
•H
*>
•H
W
4)
to
o o a*
I-H r-4
m co
a. c vt
rH C C
n in r^
I^ 0 C
1/1 O O
• • •
CM u~
i-H
(0
01 /-> /^ II
o E E 4J
•H O O O
n cMn ^
0 E E
a. 3 3 n
o
ij v^: r^ f^i -H
n o o o IP
r^ rH rH rH 3
O ^-^ ^-^ ^^* **->
a *^ ^-- '^ '^
— t n *o rj -H
rq o. a. c, Q
ON -a- ro
f-H rH
^
in m -t
rH 0 .'-)
* • •.
t-H
rt --N ."^
rHro n
o E E
•H U O
JJ "^^ "*1*^
ta E E
a 3. 3
VO fr; C-J
O O O
rH i-H f-H
*—* ^^ ^^
ro (8 re
O. Q. O.
y
/
^••^
c
u
to
rH
0
n
0
i-H
M
in
n
u
t>
p*.
0
n
^j
c
z
•o
r-H
•H
UH
-» «*
•» r-
CM
CO
j3
3
4)
CO
X
>,
rH
(M
g
O
e
a
41
n
to
41
3
rH
41
n
41
£
4>**N
fl
t
^-^
vD
VO
cr.
"^
v
i_(
rt
H
u
•o
e
a
j^
•rl
S
f«
O
•H
U
.D
•H
IH
4J
10
•H
"O
rH
Kt
rl
•rt
lu
O
5
Q
JC
n
C
01
41
.£>
n
s
*J
fl}
rH
VJ
41
g
D.
4)
JS
U
•H
X
3
|^
O
M-l
6
-i
0
r-l
U
41
n
co"
B
U
n
n
o
u
(rl
JJ
^j
to
4J
C
01
VJ
3
u
^M^
^
604
-------
a
Q
O
<3
a.
Q
a»
x
a*
3
o
4xl05
3x10"
2x10'
UIO5
Field
Diffusion
Number
.002
.01
Figure 12. Typical bimodal smoke number - size distribution
integrated with the average charge from Figure 1.
Note that all three methods of charging weight the
accumulation mode (largest size mode) the most heavily.
ELECTRICAL SIZE DISTRIBUTION MEASUREMENT AND CLASSIFICATION INSTRUMENTS
In the previous sections the basic principles of electrical size
distribution and classification measurement instruments have been
presented. In the sections below, instruments for size distribution and
classification measurement developed at the University of Minnesota will
be described and the performance of latest instruments discussed.
Differential Mobility Analyzer and Aerosol Genejcafor - The principle of
the differential mobility analyzer first constructed by Knudsen '"* and
later developed into a high accuracy monodisperse aerosol generator by
Liu and Pul^is shown in Figure 10. Details of its internal construc-
tion are shown in Figure 13.
From a careful theoretical and experimental study, Knudsen10 was
able to show that the electric mobility of the particles drawn through
the slot could be calculated with an absolute accuracy of better than
2% and that the mobility spread of the aerosol was also less than 2%.
60 =
-------
Thus, if the aerosolj being classified consisted of spheres (oil drops
for example), and if: only singly charged particles were separated, then
monodisperse aerosols of almost any material that could be prepared as
an aerosol in the proper size range, could be prepared with size spreads
of less than 2%. Furthermore, if the aerosol has a unit charge, measure-
ment of the current carried by the aerosol is then a measure of the
absolute concentration. Thus, this technique can be used as an absolute
standard for both particle size and concentration.
T
Figure 13. Internal details of differential mobility analyzer.
When used as an aerosol size and concentration standard the system
consists of the following components:
1. Aerosol Generator. Among those used are the Collison atomizer
aerosol generator [Liu and Pui^-5], condensation,.aerosol
generator producing DOP aerosols, [Liu and Pui ], fluidized
fluidized bed coal dust generator, Willeke, Lo, and
606
-------
Liu, Marple, Whitby, and Barsic^' and Nichrorae wire aerosols
Liu, Pui, Hogan and Rich^°. To he useful the generator must
produce a large number of particles in the desired size range.
2. Aerosol Charger. The .method of charging should produce the
maximum number of singly charged particles at the size which
it is desired to separate. For most purposes a Krypton 85
neutralizer containing about 5 me and producing a Boltzmann
bipolar charge distribution has been used. Occasionally the
natural charge produced by the aerosol generator is satisfactory
and under some conditions diffusion charging may be used. For
larger size particles from a few tenths up to above a micron it
is often impractical to eliminate the multiply charged
particles. If bipolar charging is used it is possible to
make corrections to the concentration Liu and Pui
3. Mobility Classifier. Details of the mobility classifier are
shown in Figure 13. It has been found that near perfect inter-
nal axial symmetry is essential to good classification. The
concentricity between the two electrodes should be kept below
0.002 in. and the surfaces must be smooth and the edges free
from burrs.
4. The standard operating condition of the mobility.classifier is
2 Jl/min of aerosol. Since the output is very stable, it is
possible to measure the concentration by switching the charged
aerosol stream to a filter collector connected to a sensitive
and well calibrated electrometer.
5. The classifier is capable of generating monodisperse aerosols
in the size range from about 0.01 to 0.5 ym depending on the method
of charging. The output concentration depends on the number con-
centration in the feed aerosol in the size range of the classifi-
cation. To obtain high output concentrations it is necessary to
feed, an aerosol that is initially fairly monodisperse with a mode
near the classification size. The highest output concentrations
obtained are about 6 x 10^ per cm3 obtained using a collision
atomizer. Higher concentrations could probably be obtained with
an optimized condensation generator.
Electrical Aerosol Analyzer - The electrical aerosol analyzer is a size
distribution measuring instrument with in situ measurement capabilities
over the 0.007 to 1 ym diameter range. The operating principle of the
device is that of electrical charging and mobility analysis, a principle
first described by Whitby and Clark . Recent advances (Liu, Whitby,
and Pui^O and in Liu and Pui in charger and mobility design and the use
of all solid-state electronics have resulted in an improved instrument
607
-------
that is portable (about 30 Kg in weight) and considerably more versatile.
Following is a brief description of this more recent device.
Figure 14. is a schematic diagram of the most recent commercial instru-
ment, Thenno-Systems Inc., 2500 Cleveland Avenue North, St. Paul, MN 55113,
showing its major components: the aerosol charger, the mobility analyzer,
the current sensor, and associated electronic and flow controls. The
instrument samples aerosols at the rate of 4 liters per minute with an
additional 45 liters per minute of clean air needed to operate the mobility
analyzer.
In the instrument the aerosol is first sampled into the charger where
the particles are exposed to unipolar positive ions and become electrically
charged. The most recent instrument differs from the one described by
Liu, Whitby, and Pui^O IB having the charger located in the upper end of
the same tube containing the mobility analyzer. Besides its mechanical
advantages, close proximity of the charger and mobility analyzer reduces
aerosol space charge losses.
The diffusion charging curve in Figure 2 shows the relationship
between the electrical mobility and the size of the particles under
different charging conditions. It is observed that there is montonic
functional relationship between particle mobility and size. Using this
relationship, the size distribution of the aerosol can be calculated
from the mobility distribution measured by the mobility analyzer.
The mobility analyzer shown in Figure 14 is in the form of a cylindri-
cal condenser with clean air and aerosol flowing down the tube in a lam-
inar stream. The charged aerosol are deflected through the clean air
core by the voltage applied on the center electrode. For a given voltage
on the center rod, particles above a certain critical mobility and size
are precipitated, while those with lower mobility and a larger particle
size escape precipitation and are sensed by the electrometer current
sensor. By changing the voltage on the center rod and measuring the
corresponding electrometer current, the mobility and size distribution
of the aerosol can be determined. The standard operating condition of
the instrument provides a total of 11 voltage steps dividing the size
range of the instrument into 10 equal geometrical intervals of four
intervals per decade in particle size. The size interval boundaries are
located at 0.0032, 0.0056, 0.01, 0.0178, 0.032, 0.056, 0.1, 0.178, 0.32,
0.56, and 1.0 ym. The complete voltage sequence can be scanned in about
two minutes, thus allowing a complete size distribution analysis to be
made in the same time period.
Calibration of the Electrical Aerosol Analyzer - The absolute calibration
of the EAA for both aerosol concentration sensitivity for particle size
and for size distribution resolution has been a difficult task because
608
-------
MASS-FLOW
TRANSDUCER
AEROSOL
NEEDLE t-Y
VALVE **
BALL £7
VALVE IA
AEROSOL
ROTAMETEl
SHEATH AIR — -- IS?
KILTER
PLASTIC
METAL
1
R
MOBILITY
ANALYZER
tL
>*
' £
o
CD
0
V)
I!1
»M or-
\
j »
ii.
M
5.9
'l
II
MOBILITY ANALYZER
CLEAN AIR
ELECTROMETER
CURRENT
SENSOR
ELECTRICAL SIGNAL
I -AEROSOL FLOWRATE
. CORONA VOLTAGE
.CHARGING CURRENT
'SCREEN VOLTAGE
H V
POWER
SUPPLY
--ROD VOLTAGE
VACUUM
-—c£j-
MASS-FLOW
TRANSDUCER
ELECTRO-
METER
BALL
VALVE
ELECTROMETER
CURRENT
--.TOTAL FLOW
Figure 14. Schematic of the Minnesota - T.S.I. Model 3030 Electrical
Aerosol Analyzer,
609
-------
of the lack of suitable aerosol standards in the size range from 0.003 to
1 ym. However, the recent development by Liu and Pui^^ of the electrical
classification generator described earlier in this paper has now made it
possible to make such calibrations with acceptable accuracy over most of
the size range of the aerosol analyzer, Liu and Pui^.The results to
date from these calibration studies are summarized here.
«•
45
1. Liu and Pui studied the losses and mobility versus size curve at
several charging conditions and concluded that N^t = 1 x 10' was the best
for the commercial instrument. It was also found that although the com-
mercial instrument had actually been originally designed for this Nt,
the actual N^t was only 0.6 x 10'. It is possible to adjust the commercial
instruments to an N^t of 1 x 10' by means of a few simple component changes
and major adjustments.
2. The current sensitivity in pico amps/10 particles per cm , for mono-
disperse aerosols of known size and known concentration generated with
the monodisperse generator described earlier has been determined. These
are shown at the top of Table IV. These range from 0.105 for 0.0075
ym to 285 for 0.75 ym particles. Also, from these data it is possible to
calculate the fractional charge and loss factor for the instrument,
Figure 3.
It can be seen by comparing the theoretical loss fac.tor for sizes
smaller than 0.02 ym in Figure 3 with actual loss factor, that the com-
bined loss factor is much smaller than the fraction charged. Note that
at 0.01 ym the fraction charged is 0.3, but that the combined factor is
only 0.075 at the same size. Recent studies of the instrument indicate
that most of this difference is caused by space charge and diffusion los-
ses in the charger. If these losses are reduced so that the overall loss
factor approaches that for fraction charged alone it should be possible
to obtain useful size distribution measurements down to at least 0.003
ym. These results indicate that at the most desirable operating N,t =
the loss factor is so great below about 0.006 ym that the lowest size range
0.0032 to 0.0056 ym, would only be usable under special conditions and that
for most purposes the size range of the present correctly adjusted com-
mercial instrument should be considered to be from 0.0056 to 1 ym.
3. rUzp rn1 ibrnt \ on of the i tinl rmnrnl WMH ftrromp! I nlu''l by tmuinil t sprr-Mr
deioHolu liuvt'.Lug Hl^t'H eoj respond iuy Lo the geometric ninan dlanittter of
each channel on the EAA and then measuring the fraction of the total cur-
rent that was measured in each channel of the instrument. The results
are shown in Table IV. The row at the top gives the actual size of the
monodisperse aerosol, the second row gives the sensitivity, the main part
of the table gives the fraction of current in each channel and the last
two rows give the geometric mean channel size and geometric standard de-
viation of the response to each size, respectively. In making these cal-
610
-------
/^
o
HI
to
^v
o
O rt u
r- e
(L (Q )>4
— 0 0
1C >-
H *-• w
O r-l
0) O
fH CO
01 O
0) 01
X ™
Ol
U-l Cfi
O l-i
01
01 P.
CO U
0 -0
a o
01 C
01 O
OS E
m
r*.
0
CM
•»
C
in rH CN CN CO
00 COCCOOC'--'
CM O C C C
>— '
.^»
O ?H \c oo in
in O O O C C C C (O fO rH rH
rH ....
o o c o
v-'
CM 00 O VO -3
O CM -3- CM O C
rH O O C O O
00 '-'
CM O O^
in -* -* o
0 C C C C • • O O
r-l 0 C O
-»
*3- in r^ CM CM
U1 OOCCr-liDr-IOCC
CM C C C O C
CM
in -a- o -3-
• c o c • • • c c c o
r-l CCO
1-1
C 00 CM
>c o o
c
-3- OOrHOOOOOCO
CM
in
C OrHOOOOOOOO
r-l
o
_J ^ ^
'rtt "^
CO) ^ ^ co r^ CM
CD. ^JI^cococMinror^cMO
-C^O., ^-* ^ ^ C O r-l CM *3- r^
^ oodoooodod
u
^7
M O r-l
- 0) C£ O
X 0. 1 >
J-NO U
T-l O O * *"*
>r-l uotC^ O% \D 00 O O O O O O O O
U^ aJoJrHir'SS»S?Sco§SS
lOQ.rHi-1 rHCMOO
COO r-l r-l
01 t O > A
CO ^^
B
-- B
>,CM %O O CO »O CM
"JUrtOoSrHCOinoSrHSoOO
CT3OO O C COr-^rHfOinOf^
-c o
u «
01
c « °
C rH CM (O "^ in ^O ^ 00 &* tH
eg
C- r-l
VD 00
rH
m
VO rH
-* 00
rH
SvD
CM r-
rH
c
rH <•
t .
rH
rH
00 -»
O in
r-i
rH
O f1^
rH
CO
CN r^
C r-l
rH
CO
CO
^
O rH
in
O rH
O
e
00
o
a.
r K r f\
O U
O en
611
-------
culations the responses shown in parenthesis were assumed to fall in the
1 to 1.78 }im size range.
The geometric means of the response versus the actual aerosol size is
shown in Figure 15. It will be noted that in the size range from 0.0056
to 0.561 urn, the correlation between the actual and indicated sizes is quite
good at these operating conditions. In the size interval 0.56 to 1 jam,
36% of the current that should be indicated in this channel is not.
E
=1
Q
I
•I
01
001
O Including fraction between I S 1.75
A Including fraction only up to I
001 01 I I
Op, , Geometric Mean of Channel Boundries
Figure 15. Comparison of the instrument indication with the response to
monodisperse aerosols for the TSI model 3030 electrical
aerosol analyzer. Data from Table IV.
4. The geometric standard deviations (SG), tabulated in the last row of
Table IV, show that the resolution varies from that corresponding to less
than one channel for the two smallest channels, to SG=1.81 for the twc
upper channels.
The sizing resolution is essentially a function of three factors:
the inherent mobility resolution of the instrument, which coi responds to
an SG < 1.1, the discrete nature of the electri charge and the slope of
the Zp versus Dp curve, Figure 2. The Liu and Pui^ studies indicate
612
-------
that mobility analyzer is functioning very close to theory with respect
to its mobility resolution and with respect to what can be accomplished
considering the discrete nature of particle charging. It appears t1 at
some improvement in the optimum Zp vs Dp curve is possible if the instru-
ment were to be operated at a variable pressure, Liu, Marple, and
Yazdani^". However, if the Nt were to be reduced to obtain more slope
and better resolution at the larger sizes, then the fraction charged at
the smaller sizes might decrease thereby reducing the instrument use-
fulness below 0.01 urn. In short it appears that the maximum size range
over which such an instrument can be made to operate with useful res-
olution is about two orders of magnitude. As Whitby and Liu have
previously pointed out, thi& is probably the maximim range of sizing
usefulness for any single measurement technique cr instrument operation
at a given set of conditions.
Table IV constitutes the calibration matrix of the instrument and
it is possible to use these data to correct the indicated size distrib-
utions to obtain results much closer to the true distribution. Initial
attempts by Cantrell to do this have bten reasonably successful and as
soon as the techniques are more fully developed and sufficient examples
have been evaluated, the results will be reported. Using the matrix
inversion technique preliminary indicatic ns are that it should be pos-
sible to resolve an aerosol with SC - 1.4 for sizes up to 0.56 i-m.
Most fine particle aerosol emissions and most atmospheric aerosols
have SG's greater than two and WnitLy^-5 has shown that the size distrib-
ution modes occuring in the size range below 1 urn, have SG's on the order
of two. Thus, the EAA appears to be able to resolve most of the signif-
icant aerosol size distribution structure found in smoke and atmospheric
aerosols.
In the size range below 0.03 urn, where the particles ar> singly char-
ged, by operating the EAA in the manual precipitator voltage control mode,
it is possible to resolve size distributions with SG - 1.1.
It will be noted also from Table IV that SG - 1.74 at 0.042 rm. The
resolution at this size is caused by the transition from singly charged
to multiple charged aerosols. This is a fundamental problem that can-
not be eliminated. From Figure 1, it can be seen that this transition
would occur at about 0.2 ym for bipolar charging.
5. The precision of aerosol concentration measurements as a function of
size is determined principally by two factors; short term noise fluctua-
tions of the electronics and the longer term stability of the instrument
and aerosol during the approximately two minutes required to complete a
measurement cycle. Liu, Whitby and Pui^" evaluated the precision due to
electronic fluctuations by the following procedure.
613
-------
In the University of Minnesota smog chamber, Clark and Whitby ,
0.23 ppm of S02 was illuminated for 17 minutes, the lights turned off,
and the aerosols allowed to coagulate until practically all of the par-
ticles smaller than a few hundredths of a pm had disappeared by coagula-
tion with larger particles and the changes of aerosol surface and volume
were small and linear with time. The AI's about the line of regression
were calculated. By multiplication by the appropriate sensitivity, a,
the standard deviation in AI can be translated into a standard deviation
in particle number, surface area, and volume. These results are shown
in Figure 16 as a function of particle size.
Figure 16. Variability in electrometer current, particle number,
surface area, and volume due to inherent instrument
variability.
It will be noted that the standard deviation of AI (OAI), ranges from
a maximum of about 0.01 pa at 0.004 ym to a low of about 0.001 pa at 0.4
Vim. This is approximately equal to the electronic noise levels at the
voltages used for measuring the particular sizes. For these measure-
ments a Doric data acquisition system (DorLc Scientific Corp., San Diego,
California) was used having a filter with an effective averaging time of
about 0.25 sec. By using computer averaging of the electrometer output
as was done on the Air Resources Mobile Laboratory in California during
the California aerosol characterization project, Whitby et al5 , the
614
-------
effect of the electrometer noise, can be. reduced even further to values
on the order of O.OQ05 pa.
From Figure 16, it will be. noted that the effect of the variability
in AI is quite different for particle number, surface, and volume. Plus,
or minus one standard deviation in number at 0.0075 ym corresponds to
10^ particles/cm . It is seen that the variability in number decreases
sharply with size, the variability of surface is relatively constant, and
the variability for volume increases with size. Thus, the instrument is
able to measure surface area distributions with approximately the same
precision over the full range of sizes. It has been found that the vari-
ability due to noise is only significant for the number distribution
below 0.02 ym for number concentrations less than I03/cm3 and for the
volume distribution above 0.5 ym. Therefore, for most field work, data
from the instrument has only been used for calculating the volume for
sizes smaller than 0.422 ym. An optical particle counter has been used
to obtain the surface and volume distributions for larger sizes. When
measuring atmospheric or smog chamber aerosols having a significant
number of particles smaller than 0.02 ym, it is preferable to measure the
total number concentration with a properly calibrated CNC instead of with
the EAA.
The effects of stability of the instrument and aerosol during the
measurement cycle is more difficult to predict and control. The current
increments, AI, represent the first derivative of the I vs. V curve.
Therefore, any changes in the aerosol concentration as a function of time
tend to cause errors in the calculated AN's. In field studies where
ambient aerosol concentrations can fluctuate significantly, a 20 to 50
liter buffer chamber has been used to damp the aerosol fluctuations. In
such applications only the aerosol sampling stream need go through the
buffer chamber. The clean air stream need not. Also, an automated bag
sampling system has been used aboard an aircraft to grab a sample and
hold it for the two minutes required by the EAA measurement cycle.
In addition to the above aerosol stability effect, other non-ideal
current drift effects have sometimes been observed that so far are not
completely understood. These seem to vary from instrument to instrument
and are probably associated with insulator charging effects in the mobility
analyzer. One manifestation of these effects is negative AI's for the
smallest several size ranges when measuring aerosols which contain neg-
ligible numbers of particles in these size ranges. Use of gold-plated
brass parts in the mobility analyzer instead of oxidized aluminum seems
to reduce these effects.
615
-------
Electrical Aerosol Analyzer Applicationi - The EM was originally developed
as an instrument to measure in situ atmospheric aerosol size distributions
in the size range below that which could he measured by any other in situ
instruments. Through development it haq become smaller, lighter, more
automatic, more reliable and its useful ('sizing range has been extended.
Incorporated into instrument systems witih one or more optical single
particle counters (OPC) and condensation nuclei counters (CNC), it has
been used in a number of large aerosol field studies, Whitby et al.5 ,
Willeke et al.55, Sverdrup et al .
Figure 17 shows a sample size distribution obtained in St. Louis by
a large mobile air pollution laboratory owned by the U.S. Environmental
Protection Agency. In this laboratory, aerosol size distribution data
from the EAA, and two optical particle counters is processed on line by
a minicomputer and the data recorded. Although Figure 12 was produced
later, plots similar to Figure 12 can now be produced on line as data is
collected. Note the excellent agreement between the EAA and the Royco
220 OPC data in the size range from .561 to 1 tun where they overlap. These
data were processed using the latest calibrations for both the EAA and the
OPC's.
RUN NO
WD 133 8 BS
MSP 3 2 CNC
RH P0 4 ST
ST LOUIS AEROSOL CHARACTERIZATION STUDY
SITE RAPS 103 DATE 08^14,74
13 TIME 21 06 TO 22>00 HR CST
2 45E-04 U3- 20.796 HOX. 0 074-
8 73E 04 S3 457 026 THC 0 00000
683 565 S2 203 927 03 0 00654
48 370 NT 1.12E05 S02- 0.02117
1 -
002
100
Figure 17. Volume-size distribution obtained with the large EPA mobile
laboratory in St. Louis, No. on a site located in an area
with industrial sources in all directions.
616
-------
Within the last year the EAA along with, an OPC have been used success-
fully in several aircraft-^ . For near urban use, where ambient aerosol
fluctuations are too large to be tolerated, the aerosol is iirst forced
into a 50 liter sampling bag by ram air pressure, the bag is closed, the
measurement cycle made and the bag emptied.
This bag sampling system has also been used for the sizing of aerosol
emissions from a Diesel Engine.
Electrical Aerosol Size Distribution Measurement at Low Pressures - From
Figure 2 it can be seen that if a sufficiently high NT is used to obtain
a satisfactory fraction charged at 0.01 ym, the Zp vs Dp curve has only
a very small slope above 0.5 ym. Liu, Marple and Yazdani showed that '*
if the pressure at which the mobility classification is carried out is
reduced below 1 atmosphere, then the whole Zp vs Dp curve can be shifted
to the right, the amount being determined by the product Dp P, where P
is the pressure in atmospheres. For example, if P = 1/2, the mobility
minimum would occur at 2 ym instead of 1 ym. This approach can also be
used to correct the data from the standard EAA when it is operated at
pressures below 1 atm. as in aircraft.
This approach is only useful for aerosol that are stable with respect
to pressure. It is not satisfactory for atmospheric aerosol measurement
because a substantial fraction of the total mass of these aerosols is
usually water or other vo .atile matter. However, it has considerable
potential for precision measurement of stable laboratory aerosols.
Electrical Aerosol Samplecs - Quite a number of electrostatic aerosol
samplers have been developed. These range from a wire and tube samplers
that are in essence scaled down versions of large industrial electrostatic
precipitators, to sophisticated samplers for special purposes. The chief
advantages of an electric sampler are:
1. low pressure drop compared to a filter.
2. can sample particles to below 0.01 ym as compared to an impactor
which is rarely useful below 0.3 ym.
Its chief disadvantages are:
1. complexity.
2. classification of the sample unless special care is taken to
obtain non-classified deposits.
3. orientation of the particles by the electric field prior to
deposition. This effect can be significant for very irregular
particles such as chain aggregates of very small particles.
617
-------
58
Linpann has reviewed many of the electrical precipitators for
aerosol sampling that were available up to 1971, including the historic
wire and tube designs of Drinker and of Barnes and Penny. These samplers
as well as later designs marketed by Mine Safety Appliance and Western
Precipitation and Del Electronics were designed for low pressure drop,
high volume sampling for mass determination. They are not readily used
for obtaining samples for microscopy or chemical analysis.
The special requirements of samplers for electron microscopy has
lead to the development of samplers which can obtain a satisfactory
sample onto an electron microscope grid or stud. This application re-
quires that the sampling efficiency as a function of particle size be
reasonably constant and be known if less than 100%.
Several point- to -plane samplers in which a corona needle opposite
the collecting surface is used have been developed"^' ^l, 62 < fiaum""
seems to be the first one to..have developed this type of sampler while
the developed at Rochester , seems to be the most widely used.
/; o
Reist has evaluated the sampling efficiency and other characteristics
of the point-to-plane precipitator and found that the efficiency was only
a few percent and varied with size. Thus, this type of sampler although
convenient because of its ability to concentrate a sample directly onto
a small area, should be used with caution if more than rough estimates
of the aerosol size distribution are to be made.
In an attempt to develop a sampler for both light and electron micro-
scopy, which would not significantly classify the deposited aerosol, Liu
Whitby, and Yu°^ and Liu and Verma"-' developed an electrostatic sampler
in which the particles charged by a corona are precipitated in a pulsed
electric field in such a way that classification is eliminated. Its
principal problem however is that the sampled aerosol is spread over an
area several orders of magnitude larger than for the point-to-plane
samplers. This makes it less suitable for sampling low concentrations.
It cannot be emphasized too strongly that the biases of sampling
methods must be known especially for sizes smaller than 0.01 and larger
than 5 ym if accurate assessment of size distribution are to be made.
Miscellaneous Electrical Aerosol Measuring Techniques - Meyers et al.
has recently described a technique in which the ions produced when a
particle strikes a hot surface, are detected in order to determine particle
size, something about chemical composition and particle concentration.
The ionized molecular fragments are analyzed in a continuous mass spectro-
meter.
618
-------
ACKNOWLEDGEMENT
This paper was prepared with the support of EPA Research Grant No. R
800971 Aerosol Research Section, Division of Chemistry and Physics,
Research Triangle Park, North Carolina.
619
-------
REFERENCES
1. Rohmann, H. , Method of Size Measurement for Suspended Particles, Z.
Phys. 17, 253-265, 1923.
2. Harper, W. R, Contact and Frictional Electrification, W. R. Harper Co.,
Oxford (1967).
3. Moore, A. D. (Editor), Electrostatics and Its Applications, John Wiley &
Sons, N.Y. (1973).
4. Prochazka, R., S taub, 26:22 (1966).
5. John, W., Investigation of Particulate Matter Monitoring Using Contact
Electricity, EPA Grant No.Final Dept. (Sept., 1974).
6. John, W., Contact Electrification Applied to Particulate Matter-Monitor-
ing, presented at the Fine Particle Symposium, University of Minnesota,
May 27-30 (1975) .
7. Guichard, J. C. and Chauvelier, Nouveau Precede de Mesuew en Continu de
la Concentration en Poussieres de Particles Solids. Proceedings of the
Conference on Aerosol in Physik, Medizin und Technik, Gesselschaft fur
Aerosolforschung, e.v. Bad Soden, Germany (1973).
8. Liu, B. Y. H., and D. Y. H. Pui, Aerosol Science. 5:465 (1974a).
9. husar, R. B., Recent Developments in 'In Situ' Size Spectrum Measure-
ments of Submicron Aerosols, APRL Comm. #2, paper presented at the Con-
ference on "Instrument Monitoring for Ambient Air", Boulder, Colorado,
Aug. 14-16, (1973).
10. Knutson, E. 0., The Distribution of Electric Charge Among the Particles
of an Artifically Charged Aerosol, Ph.D. Thesis, University of Minn-
esota, Mechanical Engineering Department, Minneapolis, Mn. 55455(1971).
11. Knutson, E. 0., Extended Electric Mobility Method for Measuring Aerosol
Particle Size, presented at the Fine Particle Symposium, University of
Minnesota, May 27-30 (1975).
12 Cantrell, B. C., unpublished procedures developed in the Particle
Technology Laboratory, University of Minnesota.
13. Pui, Y. H., Experimental Study of Diffusion Charging of Aerosols, Ph.D.
Thesis, University of Minnesota, Mechanical Engineering Department,
Minneapolis, Mn. 55455 (1975).
620
-------
14 Liu, B. Y. H. and D. Y. H. Put, J. Colloid and Interface Sci., 49:
305 (1974b).
15 Liu, B. Y. H. and D. Y. H. Pui, J. Colloid and Interface Sci., 47:
155 (1974c).
16 Bademosi, F., Diffusion Charging and Related Transfer Processes in
Knudsen Aerosols, Ph.D. Thesis, University of Minnesota, Mechanical
Engineering Department, Minneapolis, Minnesota 55455 (June 1971).
17 Liu, B. Y. H., K. T. Whitby and H. H. S. Yu, J. Appl. Phys., 38:1592
(1967).
18 Keefe, D., P. J. Nolan and T. A. Rich, Proc. Royal Irish Academy,
Section A, 60:27 (1959).
19 Keef, D., and P. J. Nolan, Geofis. Pura Appl. 50:155 (1961).
20 Keef, D. P. J. Nolan, and J. A. Scott,Proc. Roy. Irish Acad. 66A:2
(1968).
21 Natanson, G. L., Sov. Phys. Tech. Phys., 5:538 (1960).
22 Brock, J. R., J. Appl Phys., 41:843-844(1970).
23 Brock, J. R. and M. S. Wu, J. Colloid and Interface Sci., 33:473
(19.70).
24 Fuchs, N. A., Investiza Acad. Nauk. USSR, Ser. Geogr. Geophys., 11:
341 (1947).
25 Pluvinage, P., Etude Theorique et Experimental de la Conductibilite
Electrique duns les Nuages non Orageuz Thesis, Faculte des Sciences
de 1'Universite Paris (1945).
26 Pluvinage, P., Ann Geophys., 2:31 (1946).
27 Bricard, J., J. Geophys. Res. 54:39(1949).
28 Fuchs, N. A., Geofis. Pura Appl. 56:185-193(1963).
29 Marlow N.H., and J. R. Brock, J. Colloid Interface Sci.50:32-38(1971).
30 Liu, B. Y. H. and F. Bodemosi, Diffusion Charging of Kundson Aerosols.
II. Theory, Publication No. 156, Particle Technology Lab., University
of Minnesota.
621
-------
31 Pourprix, M., UnNouveau Precipitateaur Electrostatique - Applications
a L'etude de la Charge des Aerosols par Diffusion d'ions Bipolaires,
Thesis, University of Paris IV (1973).
32 Israel, H., Atmospheric Electricity - Fundamentals, Conductivity, Ions,
Vol I, Israel Program for Scientific Translations, Jerusalem 1970.
33 Mohnen, V. Investigation of the attachment of neutral and electrically
charged emanation decay products to aerosols- Measurement of the aaro-
sol size spectrum and of the attachement coeffocoemt pf meitra; decay
products. Thesis to the Natural Science Faculty, Ludwig-Maximilians
Universitat, Munich,(Sept. 1966).
34 Megaw, W. J. and A. C. Wells, A High Resolution Charge and Mobility
Spectrometer for Radioactive Submicrometer Aerosols, J. Scientific
Instruments (J. Phys. E.), Series 2, Vol. 2:1013-1016 (1969).
35 Yeh, H. C., 0. G. Raabe, and B. E. Wood, Development of a Miniature
Aerosol Charge Spectrometer, Report No. LF-46, Annual Report by the
staff of the Inhalation Toxicology Research Institute, Lovelace
Foundation, Dec. (1973).
36 Rich, T. A., L. W. Pollak and A. L. Metnieks, Estimation of Average
Size of Sub-micron Particles from the Number of all and Uncharged
Particles, Geofis. Pura. Appli., 44:233 (1959).
37 Flyger, H., K. Hansen, W. J. Megaw, and L. C. Cox, The background
Level of the Summer Tropospheric Aerosol Over Greenland and the
North Atlantic Ocean, J. Applied Meteor., 12:161 (1973).
38 Mohnen, V. A. and Holtz, J.A.P.C.A., 18:667 (1968).
39 Whitby, K. T. and W. E. Clark, Tellus, 18:573 (1966).
40 Liu, B. Y. H., K. T. Whitby and D. Y. H. Pui, J.A.P.C.A., 24:1067
(1974).
41 Knutson, E. 0. and K. T. Whitby, Aerosol Classification by Electric
Mobility: Apparatus, Theory and Applications, J. of Aerosol Science,
6:6 (1975).
42 Huertas, M. L., A. M. Marty, J. Fontan, L. Lanegrasse, and D. Blanc,
A Method of Measuring the Mobility of Radioactive Ions, Rev, of
Scientific Instruments, 41:1567-1569 (1970).
43 Whitby, K. T., Modeling of Multimodal Aerosol Distributions, Presented
at the GAF meeting Bod Soden, Germany (Oct 17, 1974).
622
-------
44. Whitby, K. T., Trimodal Log Normal Model for Atmospheric Aerosol
Size Distributions, to be submitted to J. of Aerosol Science (1975)»
45. Liu, B. Y. H., and D. Y. H. Pui, On the Performance of the Electrical
Aerosol Analyzer," J. -Aeroso-1 Sci.. 6:249(1975).
46 Willeke, K., C. S. K. Lo, and K. T. Whitby, J. of Aerosol Science,
5:465 (1974).
47 Liu, B. Y. H., V. A. Marple, K. T. Whitby, and N. J. Barsic, American
Industrial Hygiene Association Journal, 8:443 (1974).
48 Liu, B. Y. H., D. Y. H. Pui, A. W. Hogan, and T. A. Rich, J. of Applied
Meteorology, 14:46(1975).
49 Liu, B. Y. H., V. A. Marple, and H. Yazdani, Env. Sci. and Tech., 3:
381 (1969).
50 Whitby, K. T. and B. Y. H. Liu, Advances in Instrumentation and
Techniques for Aerosol Generation and Measurement, presented at the
Harold Heywood Memorial Symposium, Loughborough, England (Sept 17-18
1973).
51 Clark, W. E., and K. T. Whitby, Measurement of Aero'sols Produced by
the Photochemical Oxidation of S0? in Air, J_. of Colloid and Inter-
face Sci., 51:477 (1975).
52 Whitby, K. T., B. Y. H. Liu, R. B. Husar, and N. J. Barsic, "The Minnes-
ota Aerosol Analyzing System Used in the Los Angeles Smog Project", J.
Colloid and Interface Seicnec, 39:136-164 (1972).
53 Cantrell, B. C. and K. T. Whitby, Airborne Minnesota Aerosol Analyz-
ing System, Program Report in preparation EPA Grant No. R800971
(1975).
54 Whitby, K. T., R. V. Husar, and B. Y. H. Liu, J. Colloid and Interface
Sci., 39:211 (1972).
55 Willeke, K., K. T. Whitby, W. E. Clark, and V. A. Marple, Atmospheric
Environment, 8:609 (1974).
56 Sverdrup, G. M., K. T. Whitby, and W. E. Clark, Characterization of
California Aerosols, II. Aerosol Size Distribution Measurements in
the Mojave Desert, Atmospheric Environment, 9:463 (1975).
623
-------
57 Dolan, D. F., D. B. Kittelson, and K. T. Whitby, Measurement of Diesal
Exhaust Particle Size Distribution, to be presented at the Winter ASME
Meeting, Houston, Texas, Dec. 1975.
58 Lipmann, M., "Electrostatic Precipitators," Air Sampling Instruments,
ACGIH, P.O. Box Box 1937, Cincinnati, Ohio 54201 (1972 ed.)
59 Mercer, T. T., Aerosol Technology in Hazard Evaluation, Academic
Press, New York, N.Y. (1973).
60 Baum, J. W., "Electrical Precipitator for Aerosol Collection on Electron
Microscope Grids," USAEC Rep. HW-39129 (1955).
61 Billings, C. E. and L. Silverman, "Aerosol Sampling for Electron
Microscopy," J.A.P.C.A., 12:586 (1962).
62 Morrow, P. E. and T. T. Mercer, "A Point to Plant Electrostatic
Precipitator for Particle Size Sampling," Amer. Ind. Hyg. Ass. J., 25:8
(1964).
63 Reist, P.C., "Size Distribution Sampling Errors Introduced by the
Point to Plane Electrostatic Precipitator Sampling Device," Proc.
A.E.G. Air Cleaning Conference, 9th CONF-66094, p. 613 (1966).
64 Liu, B. Y. H., K. T. Whitby, and H. S. Yu, J. Colloid Interface Sci.,
23:367 (1967).
65 Liu, B. Y. H. and A. C. Verma, "A Pulse-Charging-Pulse Precipitating
Electrostatic Aerosol Sampler", Analj_t^cal_Chemis_try, 40:843-847,
(1968).
66 Myers, Richard L. and Wade L. Fite, "Electrical Detection of Airborne
Particulates Using Surface lonization Techniques," Environment a 1
Science and Technology, 9:334-336 (1975).
67 White, H. J., "Particle Charging in Electrostatic Precipitation", AJLEE
Transactions, 70:1186-1191, 1951.
624
-------
RECENT DEVELOPMENTS REGARDING THE USE OF A FLAME
IONIZATION DETECTOR AS AN AEROSOL MONITOR
Lawrence (Tom) L. Altpeter, Jr. (North Star Division, MRI),
J. P. Pilney (NS/MRI), L. W. Rust (NS/MRI), A. J. Senechal (Hennepin
Technical Center), D. L. Overland (Digital Systems)
North Star Division, MRI
3100 38th. Ave. So.
Minneapolis, Minnesota 55406
ABSTRACT
High speed waveform studies were performed with the output signals
of individual aerosol particles as they were consumed in the flame of a
flame ionization detector. These output signals were shown to consist
of modulated current pulses with mean amplitudes which varied with
particle size from 10~g to 10~6 amperes and with durations from a fraction
of a millisecond to several milliseconds.
Aerosol measurements were performed with the FID centered in an
aerosol test chamber. Sampling conditions of size, number concentration
and composition were carefully controlled. Repeatability of the flame
ionization detector was estimated to be 5 to 15 percent. Response of
the detector was shown to be linear throughout the tested size range of
0.5 to 10 vim. Size limit of detection was 0.15 to 0.2 ym with evidence
of ways to improve this limit significantly. Sensitivity varied with
particle size as the 3/2 power. There was evidence of limited selecti-
vity when the detector was used with parallel plate electrode geometry.
Considerations of the structured current pulses observed with the
conventional FID led to design and testing of a novel FID. All features
of the novel FID were identical to the conventional model with the
exception of the collection electrode which has been segmented into a
vertical array of horizontal strips. Each strip was a separate and
individually addressable electrode. The detector was referred to as a
segmented-plate FID, or SFID. Preliminary results with this device
suggest that the current pulse observed with the conventional FID is a
composite of two signals, one of which is the time-of-flight delayed
current of the charge carriers from the flame,
625
-------
RECENT DEVELOPMENTS REGARDING THE USE OF A FLAME
ION1ZATION DETECTOR AS AN AEROSOL MONITOR
Lawrence (Tom) L. Altpeter, Jr., (North Star Division/MRI),
J. P. Pilney (NS/MRI), L. W. Rust (NS/MRI), A. J. Senechal (Hennepin
Technical Center), D. L. Overland (Digital Systems)
INTRODUCTION
Although the flame ionization detector (FID) can detect single
particles on a real time and continuous basis,1"5 there has been rela-
tively little interest in its use as an aerosol monitor for several
reasons. First, the response of the FID varies with particle composition
which means that accurate size information is not possible when the
device is challenged with typical heterogeneous aerosols. Second, the
coincidence of two or more particles in the flame cannot easily be
deciphered so that the FID is count-rate limited. With appropriate
sample-conditioning, the count-rate limitation can be minimized or
eliminated. However, the variation of response with particle composi-
tion has remained as a drawback of the FID.
We wish to report the results of high speed waveform studies in
which the output current pulses of the FID, produced by single particles,
were examined on an expanded time base. If a high speed amplifier is
connected to the collection electrode of an FID, then the output signal,
produced by a single particle as it is consumed in the flame, can be
observed as a pulse, 10~9 to 10~6 amperes in amplitude, and, from a
fraction of a millisecond to several milliseconds in duration. If
parallel plate electrode geometry is employed, this current pulse is
observed to possess an irregular pattern of modulation. The modulation
pattern is repeatable for a given aerosol species. The signal amplitude
is repeatable for a given particle size. Further, during preliminary
studies at this laboratory there was evidence that for certain aerosols
the waveform of the current pulse might serve as an indicator of specie.
Because relatively little detailed data has been reported for the
FID, a general performance study was undertaken to evaluate analytical
parameters such as precision, response, sensitivity and size limit of
detection. A variety of aerosols were examined under controlled condi-
tions of size, number concentration and composition.
626
-------
EXPERIMENTAL
APPARATUS
FID Detectors
The FID employed throughout this study is shown schematically in
Figure 1. Significant features of this configuration included plane
parallel plate electrodes, a pre-mixed stoichometric air-hydrogen flame,
a ground-side amplifier, and removal of combustion products by evacua-
tion. An exploded view of the actual configuration is shown in Figure 2.
Significant dimensions and specifications are listed in Table 1. A
polarizing voltage of 500 VDC was applied throughout most of this study.
Hydrogen is premixed with the sample-laden air in a stainless steel tee
below the burner. The stoichmetric air-hydrogen mixture burned at the
orifice of the burner. Secondary air entered the enclosed space of the
FID chamber at the base of the rear wall (not shown). A copper exhaust
line, 0.6-cm OD (0.25 inches), lead to an aspirator (not shown). The
latter was required to develop sample air flow and to direct hydrogen
and secondary air into and through the FID.
In order to minimize electronic interferences from pickup, the
burner was enclosed in a Faraday-like cage. The cage was fabricated
from panels of copper-laminated epoxy board (Table 1) which were soldered
together.
EXHAUST
SIGNAL OUT
_BIAS
-5-VOLTAGE
Figure 1.
Schematic Illustration of Conventional
FID Aerosol Monitor (Parallel Plate)
627
-------
Copper Laminated
Epoxy Board Case.
Lava Stone Chamber
,Exhaust Line
(2) Mounting
Washers
Platinum Wire
Igniter Assy
Teflon Gasket
,Teflon Washer
Viewing
Port
S S Polarizing
Electrode
Sohd State
Amplifier
Teflon Gasket
Ho In
Slotted Opening
For External
Igniter Control
Figure 2. Exploded View of Experimental FID
Table 1. Dimensions and Specifications of an FID
with Plane Parallel Electrode Geometry
1. Copper Laminated Epoxy Board Case—15-2 by 10.2 by 7 6 cm (6 by 4 by
3 inches); fabricated from panels of copper laminated epoxy board
0.32-cm (0.125-inch) thick; composed of two compartments sized to
accept the lava stone FID chamber and the solid state amplifier with
its associated discrete components.
2. Lava Stone FID Chamber (with burner assembly)—outer dimensions: 14,6
by 7 by 7 cm (5.75 by 2.75 by 2.75 inches); inner dimensions: 7-6 by
1.9 by 1.9 cm (3 by 0.75 by 0.75 inches).
3. Lava stone Burner Orifice—1 mm (0.04 inches); projects above lower
edge of electrodes by 0.6 cm (0.24 inches),
4. Electrodes—plane parallel, stainless steel, 3.8 by 1.9 by 0.3 cm
(1.75 by 0.75 by 0.125 inches); separated by 1.6 cm (0.625 inches).
5. Premix fitting—1/4-inch Swaglok Tee with one 1/8-inch NPT.
6. Igniter Assembly—platinum wire, 0.08 cm (0.030 inches), bent in a
loop; attached to rotatable shaft which is sealed into FID housing,
powered by 2.5 VAC stepdown transformer.
7. Applied Voltage—500 VDC.
8. Gaseous Flow Rates—100 cc/min hydrogen, 250 cc/min aerosol-laden
air, 300 cc/min secondary air.
*
t
628
-------
With the indicated structural features and 500 VDC•-"'tilled, the rtns
background current from the FID was approximately 50 pA.
Segm?nted-Plate (SFID) Detectors " '?-*' '
A simple functional sketch of the SFID is shown in ^Egfore 3.
Basically, the SFID is identical to the FID with the dxc$p£ion of the
segmented collection electrode. The latter was designe$|;i& produce an
electric field as nearly identical to that of the convefit-iooal FID as
possible. Each segment is insulated from its neighbors?ito.d individually
addressable. Those segments not in use are grounded. Thje area of each
segment of the SFID collection electrode should be phyStcfa*lly identical
to the corresponding area of the conventional FID collection electrode.
By means of the SFID, it should be possible to measure indirectly, the
vertical spatial distribution of ion currents impinging ®JV the surface
of the collection electrode of the conventional FID.
An exploded view of the SFID developed at North Star is shown in
Figure 4. The same basic three-pieced construction of the conventional
FID is maintained (Figure 2). The burner housing and Igniter assembly
are identical to that of the conventional FID. The polarizing electrode
assembly is also identical. A cut-away view above the rectangular
viewing port of the main housing shows the location of the polarizing
electrode. Partial details of the segmented collection' electrode can
be seen in the indicated blow-up on the right side of the figure.
EXHAUST
SELECTABLE
INPUT
BUS
i BIAS
VOLTAGE
-------
Lava Stone
Side Plate
1/16" Coiled Spring
Electrode Segment
Leads
Lava Stone Chamber
Polarizing
Electrode
Platinum
Wire
Igniter Assy
Copper Laminated
Epoxy Board Case
FID Solid
State Amplifier N
Also See Fig 5
" O.D. Exhaust Line
1/8" O.D. Secondary Air Line
— Segmented
j Collecting
/ Electrodes
Amplifier Contact!
For Spring Leads
Hinged Front Cover
Plate of (Copper
Laminated Epoxy Board)'
Figure 4. Exploded View of Experimental SFID
Experimental System
A schematic layout of the experimental system employed in this study
is shown in Figure 5. The FID was centered within an atmospheric simula-
tion chamber which had a volume of 3,68 m^ (128 cubic foot). The chamber
was made from plywood sheets, 2.5-cm thick (1 inch) and 1,22 m by 2.44 m
(4 by 8 feet). The inner walls were treated with two coats of sealer and
three coats of epoxy paint with sufficient time allowed for complete
curing. The polarizing voltage for the FID was fed to the detector from
an external power supply through a BNC cable. The output signal was
carried through a BNC cable to an oscilloscope and Visicorder oscillo-
graph, also outside the chamber. The oscilloscope provided a convenient
means for visually monitoring the FID output, while the Visicorder pro-
vides a permanent record on command. Further details regarding equipment
are listed in Table 2.
A 100 CFM exhaust blower drew air into the chamber. With such a
flow rate, it was possible to fill or purge the chamber in a conveniently
brief period of time. The exhaust blower permitted the chamber to be
maintained at a slight negative pressure. Intake air passed through an
630
-------
EXHAUST
AIR
SCOPE VISICORDER
©
-U.
te
44...
ISIGNAU
I
'2 (TRIGGER)
FOR
AUXILLARY
BENCHTOP
RD
STUDIES
INTAKE AIR
Figure 5.
Schematic Layout of Experimental System
for Evaluation of FID
Table 2. Experimental Equipment and Specifications
Aerosol Generator - Berglund-Liu Monodisperse Aerosol Generator, Model
3050, Thermo-Systems, Inc., equipped with 5, 10, and 20 urn orifices
for liquid solution samples and 100 pm orifice for suspensions of
large particle size.
Oscilloscope - Techtronix Type 535 operated typically at ordinate sensi-
tivities of 5-200 mV/cm and time-base of 0.2-1 msec/cm.
Oscillograph - Honeywell Model 1806 Fiber Optics CRT Visicorder,
operated with ordinate and abscissa scales identical to those
of oscilloscope.
FID Power Supply - John Fluke Precision DC Supply, Model 301E, 0 to +
too VDC/300 mA.
FID Amplifier - Analog Devices, Model 48K Current Feedback Amplifier,
15 MHz bandwidth (at unity gain), open loop gain of 105 (500 Q DC
load), settling time of 500 ns to 0.01 percent (at unity gain),
input impedance of 10
11
and 3.5 pf, powered with Analog Devices
15 VDC Power Supply.
Chamber Exhaust Blower - Dayton, Shaded Pole, 115 VAC, 126 W, rated at
100 CFM.
Chamber Intake Filter - Flanders, Model 7C10-L-N2C2 absolute filter rated
at 600 DFM maximum.
Rotameters (H2- air) - Brooks, tube size R-2-15D CO equipped with needle
valves.
Rotameter (sample) - Brooks-Mite Model 2051, 2 SCFH (sample line normally
open).
631
-------
absolute filter. Aerosol samples were mixed with this clean air as it
entered the chamber. If desired, a sampling probe was available for
auxiliary benchtop studies of the FID.
Purge time for the aerosol chamber was anywhere from 1/2 to 4 hours,
depending on previous state of the chamber.
AEROSOL SAMPLE PREPARATION
Atmospheric aerosols can be divided into three general categories:
inorganic, organic, and biological. Sodium chloride was used extensively
as an example of an inorganic aerosol. Monodisperse latex particles
served as an organic aerosol. For biological aerosols a set of commonly
occurring microorganisms were prepared. A complete list of the aerosols
used during these studies appears in Table 3.
Inorganic Aerosols
Organic Aerosols
Microbiological Aerosols
Other Aerosols
FID Gases
Table 3. Experimental Materials
Csl, NaCl, BaCl2-2H2), NiCl2, FeCl3, U02
(N03)2-6H20, Na2WOt/2H20, Fe (NHU) 2 • 6H20,
AR grade quality.
Latex, monodisperse, 0.8 pro, 3.4 pm, 9.5 um,
Coulter Diagnostics, Inc., Miami Springs,
Florida.
Serratia marc^sc(-:ns,, Proteus vulgaris,
Azotobacter agilis, Bacillus cereus,
Bacillus subtilis, Escheriahi-a coli,
StapMoQoccus alb us 3 Streptococcus lact-is,
Saccharomyaes cerevisiaa, Aspergillus niger,
obtained from private sources or from the
American Type Culture Collection (12301
Parklawn Drive, Rockville, Maryland 20852).
Coal Dust (200 mesh), Arizona Road Dust,
Stack simulants (alumina, silica).
Hydrogen, ultrapure (99.9995 percent, Air
Products and Chemicals, Inc.), Air zero
grade (less than 1 ppm total hydro-carbons)
(Air Products and Chemicals, Inc.).
632
-------
Inorganic and organic aerosols were prepared with deionized and dis-
tilled water which was doubly filtered through Wattman 40 paper and
Millipore 0.45 urn paper. Test cells of biological aerosols were removed
from their nutrient broth by centrifugation and then resuspended in tap
water which was dechlorinated, sterilized and doubly filtered as described
above. Biological aerosols were measured promptly to minimize the extent
of lysis.
Aerosols were generated with a Berglund-Liu Model 3050 Monodisperse
Aerosol Generator (Table 2 ).
The number concentration produced initially in the aerosol generator,
N° (v), was readily calculated for a liquid solution as N (v) - f/Qg, where
f is the frequency of the orifice vibration which defined the rate of
droplet formation, and Qg is the flow rate of the dilution gas (air) which
carried away the aerosol. For a suspension, the number concentration of
the particles in the liquid phase was adjusted so that, statistically,
the mean number of particles per emerging aerosol droplet was one or less.
Knowing the liquid phase feed rate into the aerosol generator was
sufficient to define the gas phase aerosol number concentration.
Aerosol particle size could be predicted for the case of a non-
volatile solute in a reasonably volatile solvent such as water. Assuming
a spherically-shaped particle and typical solute weight.fractions of less
than 1 percent, one can derive the relation:
66Q
d = —4
S TIP 1
s
where ds is the diameter of evaporated solute particle, 8 is the original
solute weight fraction; Q£ is the liquid feed rate from a driven syringe
through the orifice, and ps is the solute density.
633
-------
RESULTS
FID OUTPUT SIGNAL
A typical amplified signal is shown in Figure 6 for an organic or
microbiological aerosol of approximately 1 um in diameter. Brief studies
with many materials showed that the waveform of Figure 6 was generally
produced by organic and microbiological aerosols in the 0.5 to 5 um range.
A number of exceptions were noted among the tested materials such as
NaCl, Csl, and certain of the larger biological aerosols. In the case
of NaCl, only a single shoulder was observed behind the leading spike
of the output waveform. For Csl, no shoulders were observed (see
"Selectivity" section for further details).
FID RESPONSE, SENSITIVITY AND DETECTION LIMITS
The response of the FID to particles in the size range 1.5 to 10 pm
was measured for selected inorganic, organic and biological aerosols.
Solution and suspension samples were prepared as described above and
dispersed from the aerosol generator. The FID was centered in the
atmospheric simulation chamber. Approximately twenty output signals
were recorded for each particle size of a given aerosol' specie. Output
100--
0.5
i.o 1.5
TIME (MSEC)
Figure 6.
Typical Current Pulse Produced by a Particle
of Approximately 1 ym as It Burns in the
Flame of a Parallel Plate FID
634
-------
signals were measured with a planimeter. The mean area under the curve
was taken as a measure of detector response.
Sodium chloride served as an example of an inorganic aerosol. The
results of two runs on separate days are shown in Figure 7 for a size
range from 0.5 to 10 vim. The error bar at 4 ym is the result of six
separate runs and is discussed below in the Precision section. The
response for this aerosol appeared to be linear over the tested size
range. A linear regression analysis of the data, transformed to a log-
log format, yielded the equation:
Iog1()y = 0.723 + 1.54 log^x
(2)
with a correlation coefficient of 0.996.
1000
.2 3 .4 4 6.7.891 2 3 4 5 6 789 K>
AERODYNAMIC SIZE ( jim)
Figure 7. Response of Conventional FID (Parallel Plate)
as a Function of Aerosol Particle Size
635
-------
The results for latex particles, an organic aerosol, also appear in
Figure 7. Monodisperse samples were dispersed from the aerosol generator
in three sizes: 0.8 ym, 3.5 ym, and 9.5 ym. The response for this
aerosol appeared, also, to be linear, though only three points were
developed.
The data for latex and sodium chloride can be fit to straight lines
which are slightly displaced and of very similar slope. A 2 o noise limit
was estimated as 0.44 cm2. By extrapolation the lower limit of detection
for NaCl and latex corresponded to be a particle size of 0.2 urn and
0.15 ym, respectively. The value for NaCl agreed well with that found
by Crider and Strong.14
Several microbiological aerosols were selected to cover the size
range 0.9 to 5.5 ym. These included 5. maraesaeris, P. vulgari-x, B. verfutSj
and 5. cerevisiae. The results for microbiological aerosols are summarized
in Table 4. The data points for these aerosols have not been included
with those of NaCl and latex. Among six common, linearly transformable
functions, a best least-squares fit was obtained with the exponential
function:
... .04x ,_,
y = 14.8e (3)
While the difference in response for the microbiological aerosols is
not completely understood, it should be recalled that these particles
are composed of 70 to 80 percent water, to which the FID does not respond,
and an equally uncertain percentage of non-aqueous mass. As shown in
Table 4, the inherent variation in the mean volume of each specie is
significant.
The slope of the response curve, or the sensitivity, was nearly 3/2
in the cases of the organic and inorganic aerosols (Equation 2). For
small particles, in the limit, it might be expected that the sensitivity
would relate to the cube of the particle size, i.e., with volume, or mass.
For larger particles, it might be expected that this correlation would
shift toward the square of the diameter, i.e., to the surface area.
Table 4. Summary of Microbiological Mass Response Data from the FID
Mean Volume7
Microbiological
Aerosol
S. mavoescens
P. vulgaris
B. ceveus
S. aevevisiae
0.4 + 0.4
4.1 + 3.3
15.0 ±7.7
90.1 + 74.5
Mean Area
Under The
Curve (cm2)
15.1
19.0
23.0
420.0
Relative
Standard
Deviation
15%
18%
20%
25%
636
-------
o
ttl
43
4-1
01
a.
c
o
a
cj
E
O
CO
c
o
>,
^
•3
a
-n
r-H
Xi
i-
tu
OJ
E
fO
•r—
Q
O
in
O
i.
3~ o
0 rH O O
in • CM CM
f- O
r-t
O
rH ^f
X (^ O O -tf" o
O rH O O
ON • CM CM
rH O
CM
^^
1 g ^
03 ^-s u EC
v/ a) w K
0) M U ^ CJ
C 4J Q) -H ^. t/J
O rt 4J l*H N M "— '
•H pd x^ (I) -H IB -H
iJ C S Vj ,^_ ^ Q t>0 >. O
fa tu o C O -H C
O 01 -H C to o
4-1 T3 O 4J 41 I-l 1-1
X 'fH * *H C3 3 0) 4-1
oO 3 O" *4H V-i D* CX 3
•H CJ* s^^ 'H i-i QJ CO rH fl
11 -H I-l -H I-i -rl tH M
SJ O>tnQQ fn
O O O O
o o m o
H co CM m
§000
o m o
H ro CN m
o o o o
o o m o
rH CO CN in
o o o o
o o m o
H co CM m
o o o o
o o m o
rH ro CM m
o o o o
o o in o
rH CO CM in
/_s
.9
a ^N
•--. u
^0 Q
CO >
"a ^ 'o' ^
~-> S -rl CJ
0 O a 60
O rH -^ cfl
s^ [J4 O 4J
O rH
& M ^ O
O i-l >
rH 41 cd O 4>
X t/2 c/3 (X OS
O CM
O
rH O
O CN
m
o
O CM
m .
o
o CN
rH .
o
in CN
o
in CN
CD
*B
~. 8
a ^
o o
•^- CD
> 01
3 3
4J CO
rd W
a -H
•H O
"O W
o <
637
-------
A summary of experimental conditions appears in Table 5.
FID PRECISION
The 4 Mm sodium chloride aerosol was selected for a precision study.
The usual techniques of sample preparation and aerosol generation were
employed. A set of at least 20 output waveforms was measured each day
by planimeter for a total of six days, Precision data is listed in
Table 6. This data was subjected to an analysis of variance to separate
analytical and sampling errors. Relative standard deviations of 16.8
percent and 2 percent, respectively, were obtained. The small sampling
error suggests that good control was achieved over the sample generation
and sample transfer systems. Built into the relatively large analytical
error (16.8 percent) were certain undefined sources of error which
included electronic settings and planimeter data reduction (see
Discussion). It would seem reasonable, therefore, to assume that a
representative figure for the precision of the FID alone should lie in
the range of 5 to 15 percent with the experimental arrangements employed
in these studies. The 4 pm data point in Figure 7 is enclosed by 2 o
error bars.
FID SELECTIVITY
In the following section real FID output data is presented for a
number of aerosols. This data, photographed from the original Xerox
copy, consists of one or more oscillographic recordings, in which each
recording corresponds to the pulse produced by a single particle. The
oscilloscope was set to unblank and record only signals above the noise
level; thus, leading edges of the waveforms are missing. Superimposed
on the waveform data is a calibrated grid in which each horizontal line
is separated from the next by one centimeter and each vertical hash mark
is separated from its neighbors by one centimeter. Waveforms do not
necessarily begin at the first vertical hash mark. The ordinate and
abscissa for each waveform are mV/cm and sec/cm, respectively. Specific
values are listed in the caption of each figure along with particle
dimensions in parentheses.
The NaCl data (Figure 8) reveals a sharp leading peak with one
shoulder. Several extraneous smaller peaks can be seen, occurring,
perhaps, due to memory effects from the experimental system. Data for
Csl (Figure 9) shows a single peak with no shoulders. An inorganic
aerosol with a divalent metal ion, barium chloride, produces an output
638
-------
Table 6. Analysis of Variance of FID Data from 4 pm NaCl Aerosol
Relative Standard Deviations
Day
1
2
3
4
5
6
Mean
Area Under
the Curve (cm2)
46.2
44.7
45.5
44.9
50.0
47.6
Variance
64.0
47.6
49.0
38.4
100.0
68.9
Analytical
16.8%
Sampling
2.2%
Figure 8.
FID Output Signals from
4 ym NaCl Aerosol.
Oscillograph Sensitivity
100 mV/cm. Time Base
1 msec/cm.
Figure 9. FID Output Signals from
4 urn Csl Aerosol.
Oscillograph Sensitivity
200 mV/cm. Time Base
1 msec/cm.
639
-------
Waveform consisting of a leading peak followed by two shoulders
10). Other multivalent cation salts produced data like that of
Data for latex, the only organic aerosol tested, reveals the more
commonly observed type of waveform, namely that of a leading spike with
two trailing shoulders (Figure 11),
All but two of the biological aerosols that were tested produced
FID waveforms much like Serrati-a marcesoens (Figure 12) ^ i.e., a leading
spike followed by two shoulders. Unique waveforms were obtained from
a yeast {Saacaromyces aerevisiae) and a mold (Aspergitlus nigev, as
shown in Figures 13 and 14). Data for Aspergillus niger produced the
largest signal of all species tested.
-.
;v
Figure 10. FID Output Signals from
4 um BaCl2 Aerosol.
Oscillograph Sensitivity
20 mV/cm. Time Base
1 msec/cm.
Figure 11. FID Output Signals from
0.8 ym Latex Aerosol.
Oscillograph Sensitivity
20 mV/cm. Time Base
1 msec/cm.
640
-------
i
l
X
I I
XL
^
V
V
Figure 13. FID Output Signals from
S.a. Aerosol (4 to 5 ym
approximately). Oscillo-
graph Sensitivity
200 mV/cm. Time Base
1 msec/cm.
Figure 12. FID Output from S.m.
Aerosol (0. 5 to 1.0 urn
by 1.0 pm. Oscillo-
graph Sensitivity
20 mV/cm. Time Base
1 msec/cm.
Figure 14. FID Output Signals from
A,n. Aerosol. Oscillo-
graph Sensitivity
5')0 mV/cm. Time Base
1 msec/cm.
641
-------
SFID Signal
Typical data, gathered from the SFID are shown in Figure 15. The
current pulses are recorded here as negative signals-* A series of wave
forms are presented, beginning with electrode segment 1 of the SFID (the
lowermost electrode of Figure 3) and running down the page with successjv
segments until the signal amplitude becomes trivial. These data were
obtained by connecting the SFID amplifier to each electrode segment,
separately, and recoi ding a series of signals from individual aerosol
particles of the sanu size and composition. During this period, all
other electrode segments were grounded so as to maintain the electric
field undisturbed. Tae hand-traced data of Figure 15 are a tableau of
representative signals collected at each electrode. The abscissa
(milliseconds/cm) and ordinate (millivolts/cm) are shown.
The signal, shown for a given segment of the SFID collection
electrode, should be equivalent to that which would have been developed
from the corresponding area of the collection electrode in the conven-
tional FID. As such, each SFID signal was effectively a part of the
total FID signal. At first glance, however, the SFID data did not appeo
at all related to the FID data. The SFID data was uniquely characteri70.
by the presence of bipolar features in the waveforms. These signals
predominated at the fringe electrodes furtht-st from the flame. They
appeared to develop continuously from the start of the waveform. A
closer inspection of the SFID data revealed the presence of a second
signal component which was most apparent at those segments directly
across from the flame. This component can be seen most clearly, in the
case of the Serratia maroesc&ns aerosol, at electrodes 2 to 4, as a
trailing second peak. A similar signal can be observed for the 4 urn
NaCl aerosol at electrodes 2 to 6—although not as clearly.
*An instrument malfunction prevented recording the data in conventiona
form as positive signals. Note that the ordinate is actually on
millivolts—the tranduced readout unit of the oscillograph.
642
-------
SODIUM CHLORIDE
S. MARCESCENS
(20/
(20)
(10*)
(10)
(10)
0 05 I 2 3
— Time(msee)-*
vertical sensitivity >n mV/crn
—
-HIcmH-
Figure 15. Tableau of Representative Output Signals from Co 1 U-
Electrode Segments I tolSof the SF1D, as Produced
Individual particles from Separate Aerosols of KaC'
S. marcescans (hand traced from original data)
643
-------
DISCUSSION
For organic and inorganic aerosols, the response of the FTD was
to be linear over the tested size range. The sensitivity appeared to
related with the particle diameter as the 3/2 power rather than the c
as might have been expected. It has been reported that the combustior
hydrocarbon aerosols, in the 1 to 10 ym size range occurs in 1 to 100
microseconds.5'8 The mean residence time in the flame of this study
approximately 500 microseconds. It may be thaf, with a larger flame,
cube relationship between particle size and sensitivity could be aclri
The biological aerosols, tested in this study, displayed a difft
response. It is not clear what effect the presence of 70 to 80 peroei
water would have upon the particle integrity during the vaporization
excitation processes.
Remarks, above, regarding the relationship between particle sizi
sensitivity may also have a bearing on the precision of the FID. The
significant error bars shown for some measurements in this study may
reflect the incomplete production of chemiions brought about by the
residence time limitation of the flame size. Nevertheless, the mean
values reported should be reasonably accurate for the conditions empj
Of particular interest during these studies was the selectivity
the FID, i.e., the extent to which the waveform of the output current
pulse could be uniquely related to composition. Of the tested electri
geometries, only from the parallel plate electrodes were signals obta
with discrete information which might serve potentially as a "fingerp
for identification. These signals were composed, generally, of a le;
spike with one or two trailing shoulders. Organic and biologic aero:
produce pulses with two shoulders while certain inorganic aerosols 1i
NaCl of Csl produce pulses with marked diminuation or disappearance (
one or both shoulders. Of possible practical significance were cert
biological aerosols like yeast (Saccharomycds Gerevisiae) and a mold
(Aspergillus n-Lgeic) which produced unique waveforms. Presence of thn
aerosols in the air might be detected by an FID equipped with a suit-
pattern recognition system.
The electrodes of the FID, described above, form a parallel pla
capacitor with the flame centered between the plates. As a particle
consumed in the flame and the resultant charge carriers are separate
the field, there occurs, effectively, a transient, localized change .
the permittivity of the medium between the electrodes. The result c
such an "upset" event is a transient fluctuation in the field and, a;
in the steady state charge accumulated at each electrode, or plate.
During, or following this process, the created charge carriers then ;
644
-------
at their respective electrodes. The resultant output current pulse would
he the algebraic sum of these two effects. If one takes a typical
mobility of positive ions in ambient air as 1 cm^/volt-sec, ^ and allows
for the effect of the elevated temperatures within the FID enclosure upon
the mean speed, then at 500 VDC and a flame-to-plate distance of approxi-
mately 7 to 8 mm, a time-of-flight of less than 3 milliseconds is reason-
able. This is consistent with the duration of the FID pulse.
The net current from the positive electrode of the FID can be des-
cribed, for the simple one dimensional case, by the expression:
I(t) = Ae0
3E(o,t)
—
- Aq~b~E(o,t)n(o,t)
(4)
where I(t) is the net output current at any time "t", "A" is the area of
the collection electrode (cm2), e0 is the permittivity (coul/volt-cm),
E(o,t) is the electric field component normal to the positive electrode
surface at time "t", b~ is the mobility of the negative charge carriers
(cm?/volt-sec) and n is the number concentration at the surface of the
positive electrode (cm ). The first term refers to the current associ-
ated with the changing field at the electrode surface. The second term
arises from the neutralization of the gaseous charge carriers as they
arrive at the electrode surface. Consistent with the principle of
electronuetrality, the net current at the collection electrode, I(t) ,
must be equal to that at the polarizing electrode, I(t) .
If one calculates the variation of the electric field with time at
the surface of the electrode, then the first term in Equation 4 can be
plotted as shown qualitatively in Figure 16.
-FIELD-INDUCED CURRENT
-COMPOSITE CURVE
-GASEOUS ION CURRENT
Figure 16. Sketch of Component Current Output Signals
from a Parallel Plate FID
645
-------
Also shown in Figure 16 is a qualitative sketch of the charge
carrier current (dotted line), which is the second term in Equation 4.
The arrival of the gaseous charge carriers at the positive plate would
be delayed because of the time of flight factor.
Although qualitative, the presentation in Figure 16 is consistent
with actual data from both the FID and the SFID. Thus, the composite
curve of the overall current in Figure 16 can be compared with the FID
data of Figure 6. The bipolar waveform of Figure 16 for the field-
induced current can be compared with those observed in the SFID data
of Figure 15. In the SFID data of Figure 15 these bipolar waveforms
were observed most at the peripheral electrodes, i.e. 3 those above and
below the flame. Many of the electrodes closest to the heart of the
flame showed waveforms with a second peak. This is readily observed for
5.77?. , less so for NaCl. Consistant with the previous discussion, the
second shoulder could be assigned to the gaseous charge carrier current.
Thus, the field-induced current predominates at the peripheral electrodes
while the ion current is restricted to the central set of electrodes.
This is reasonable since one should expect the charge carrier current
to be greatest at those electrode segments in a direct line of flight
from the flame.
It should be noted that an enormous amount of data is produced by
the SFID from a single particle. The sum total of all these data could,
itself, serve as a fingerprint for particle composition.
646
-------
BIBLIOGRAPHY
1. Ohline, R. W., "Hydrogen Flame lonization for the Detection and
• Sizing of Organic Aerosols," Anal. Chem., 37,, No. 1, 93 (1965).
2. Ohline, R. W., Thall, E., and Oey, P. H., "General Considerations
Concerning Atmospheric Aerosol Monitoring with Hydrogen Flame
lonization Detector," Anal. Chem, 41, No. 2, 302 (1969).
3. Frostling, H., and Lindgren, P.H., "A Flame lonization Instrument
for the Detection of Organic Aerosols in Air," J. Gas Chromato-
gvaphy, 43 243 (1966).
4. Crider, W. L., and Strong, A. A., "Flame lonization-Pulse Aerosol
Particle Analyzer (FIPAPA) ," Rev. Sci. Instrum., 38, 1772 (1967).
5. Bolton, H. C., McWilliam, I. G. , "Pulse Characteristics of the
Flame lonization Aerosol Particle Analyzer," Anal. Chem., 44,
No. 9, 1575 (1972).
6. Berglund, R. N., and Liu, B. Y. H., "Generation of Monodisperse
Aerosol Standards," Env. Soi. Technology, 7, No. 2, 147 (1973).
7. Breed, R. S., Murray, E. G. D. , Smith, N. R. , Sergey's Manual of
Determinative Bacteriology, 7th Edition, Williams and Wilkins,
Baltimore, (1957).
8. Essenhigh, R. H. and Fells, I., "Combustion of Liquid and Solid
Aerosols," Discuss. Faraday Soc., 30, 208 (1960).
9. Von Engel, A., Ionized Gases, 2nd Edition, Oxford-at-the-Clarendon
Press, (1965).
10. Altpeter, L. Jr., Senechal, A. J., Overland, D» L., Performance of
a Flame lonization Detector as an Atmospheric Aerosol Monitor*, IT:
The Multi-Electrode Flame lonization Detector, (SFID),
647
-------
CONTACT ELECTRIFICATION APPLIED TO PARTICIPATE MATTER-MONITORING
Walter John, Ph.D.
Air and Industrial Hygiene Laboratory
California Department of Health
Berkeley, California 94704
ABSTRACT
The theory of the charging of aerosol particles by contact electrifica-
tion is reviewed, as well as the development over the past decade of
monitors based on this phenomenon. Such monitors offer the advantage of
continuous measurements over a wide range of concentrations. Moreover,
the total electrical charge collected has been found to correlate well
with the total mass determined gravimetrically. The lack of widespread
use of these detectors may be due to some contradictory reports on their
performance as well as uncertainty arising from the presently incomplete
understanding of the operating principle.
The major factors which influence the instrumental response are discus-
sed, including the recent finding by the author that the sensitivity is
strongly dependent on the electrical resistivity of the material sampled.
It is also found that the condition of the probe's surface has an impor-
tant effect on the sensitivity. At present, theoretical understanding
of the contact charging is semiquantitative for particles of metals but
only rudimentary in the case of insulators.
649
Preceding page blank
-------
CONTACT ELECTRIFICATION APPLtt . TO PARTICULATE MATTER-MONITORING
Walte ohn, Ph.D.
Air and Industrial Hygiene Laboratory
California Department of Health
Berkeley, California 94704
INTRODUCTION
The transfer of charge between aerosol particles and a probe via contact
electrification can be used as the basis of a monitor for particulate
matter. ~" Important advantages of this type of monitor include the
ability to make real time measurements with a response time of less than
10 seconds, sensitivity over a wide range of concentrations and an elec-
trical signal convenient for data collection. Moreover, the simple
probe can withstand the high temperatures encountered in-stack. Most
importantly, the electrical charge collected by the instrument has been
found to correlate accurately with the mass determined gravimetrical-
ly.4,7,8,9
Despite more than a decade of development the contact electricity monitor
has not seen widespread application. The principal reason appears to be
uncertainty attributable to the lack of a quantitative theory for the
contact electrification process. While there is a qualitative theory
for metal-particles, there is no theory available for insulating mater-
ials. An experimental difficulty, as yet incompletely investigated,
concerns changes in instrumental sensitivity depending on the condition
of the surface of the probe. »°»*
In the present paper the theory of the charging process will be discus-
sed. The development of contact electricity monitors will be reviewed
including data on the experimental characteristics. Finally, the present
status of the monitor will be assessed.
650
-------
THEORY OF CONTACT ELECTRIFICATION OF AEROSOL PARTICLES
Contact Charging
Contact electrification refers to electric charge transferred between
1117
two bodies as a result of pure contact with no sliding; or rubbing. il »iz
By contrast, according to the strict definition, triboelectric charging
involves rubbing. The latter may include the transfer of material and
local heating.H In the case of the charging of aerosol particles there
is little evidence on the relative importance of the various types of
•interactions. The present discussion will be limited to contact charg-
ing, which is better understood than the other charging mechanisms. It
should be recognized that the other mechanisms may also be present.
For contact charging of large metal spheres good agreement between
theory and experiment has been achieved by Harper. When two metals
touch, the difference in the contact potentials of the two metals will
cause a flow of electrons to take place. The buildup of charge will
continue until the resulting electrical potential equals the difference
in the two contact potentials. Described in terms of energy levels,
charge will flow to equalize the two Fermi levels (Figure 1). The
resulting difference of potential of the two metals will be equal to the
difference in the work functions of the metals. 1>13,14,15
This simple picture of the charging is complicated by two additional
effects. First, some charge can flow bv quantum-mechanical tunneling
even when there is a small gap between the surfaces. Second, upon
separation of the surfaces, the air can break down, resulting in an
appreciable back flow of charge.
The theory for metal-metal contact can be generalized to include semi-
conductors . ^ > *•-*> *" ' 1'»1° Whereas the charges reside entirely on the
surfaces of metals, for semiconductors there will also be a flow of
charge from the interior. There will be an associated time constant
given in seconds by 8.85.10~1Lf Kp, where K is the dielectric constant
and p the resistivity in ohm-cm. This implies that the amount of charge
transferred will depend on the duration of contact. In the case of
semiconductors having high resistivity there will be no appreciable flow
of charge during a brief contact.
Insulators have an even higher resistivity so that there is no flow of
charge from the bulk material. Surface states become the important
source of charge for transfer. Such surface states are complicated,
depending on the detailed physical and chemical condition of the sur-
face. Adsorbed ions giving rise to electric double layers play an impor-
tant role. According to Harper, the charge transferred consists
of ions rather than electrons.
651
-------
T
?2
1
BEFORE CONTACT
AFTER CONTACT
Figure 1. Energy level diagrams for two metals having xrork functions
and
'2'
After contact, the Ferni levels equilibrate
resulting in a contact potential 4> 2 ~ $i-
The foregoing discussion has concerned ideal surfaces; in actuality, the
charging may be drastically altered by contamination of the surfaces or
by the physical condition. The surfaces of metals will be inevitably
oxidized. According to Harper^- such oxide layers behave as metals with
work functions of about 5.5eV. Most oxides are semiconductors; if the
surface layer is not too thick, charge can be transported either by con-
duction or tunnelling. For insulators particularly, charging mechanisms
other than by contact may have to be considered, such as electrolytic
effects for moist surfaces and frictional effects resulting in transfer
of material.
Charging of Aerosol Particles
Review of Cheng and Soo Theory - The charging of an aerosol particle is
a dynamic process involving the impact of the particle with the surface
of a probe. The theory developed by Cheng and Soo°» " considers the
transfer of charge during a collision between two elastic spheres of
radii &i and a 2- During the impact there will be a mean area of contact
A2i lasting for At, both of these quantities depending on the kinematics
of the collision and the elastic properties of the materials. While the
spheres are in contact, the current density \tfill be proportional to the
voltage difference between the two spheres. If the current is integrated
over the period of impact, the charge transferred is obtained:
652
-------
where Cx 2 are the electrical capacities of the spheres
9i 2 are the work functions of the materials
and <* = A21h21 . 1 ^ g
h21 the charge transfer coefficient, is given hy
with CT, o the electrical conductivities and d' .. the charge
? '" .L y t-J
transfer lengths.
The transfer lengths are defined by the equations
where -J^i. is the current density and Vc is the potential at
the point of contact. Thus d' and d^ are the effective dis-
tances in the material over which current is transported.
The following expressions for AHI and At are derived from Hertzian
theory:
a2
At
Au21 cos
where a, « ^ TT ( H^) (A^ } ^^s 9 '
L 16 V 2 / V <-w .
(note: Q?t is distinct from a)
with r* the ratio of the rebound speed to the incoming speed
Aii2i is the velocity of approach
6 Is the angle of impact
m^ 2 are tne masses of the spheres
653
-------
and k-, :, are elastic parameters given hy
and > is Poisson's Ratio, E the modulus of elasticity.
Numerical Evaluation for Typical Parameters - It is instructive to
evaluate numerically some of the above quantities using reasonable
values of the parameters. Take sphere 1 to be the aerosol particle
and sphere 2 to be the probe. The a1 « a2-
Using kl = k2 = 3.10"13 cm2/dyne, 6=0,^= 8g/cm3 (density),
BI = 10~5 cm, Au 2j = 1.5-103cm/s, and r* =1,
we find that
«1 = 3-10-?
At = 6-10~10s
— 1 1 2
A2l — TajCX} = 1*10 cm
The short duration of the impact imnlies that the theory applies only
to metals or good semiconductors (p < 10 3 ohm- cm) since -only then will
there be appreciable current flow. The area of contact has a diameter
roughly 1/20 of the particle diameter.
Q 2i is the charge transferred per particle. The total charge trans-
ferred to or from the probe is given by
T = ! 21
where NI , the number of aerosol particles, is given in terms of the
total mass of particles, M1 , by
"1
For pj <10 ohm-cm, (1-e " ) =1, and Q 21 = ^eoal ( ij) 2 ~ i)« Tnen
i • Thus the theory predicts that the total charge transferred
to the probe is inversely proportional to the particle radius squared.
Also, for p1 < 10 ohm- cm, the charge transferred is predicted to be
independent of the velocity of the particles relative to the probe.
For P! > 10 ohm-cm, Qx.-C a^"n, where 1
-------
The magnitude of the charge transferred can also be evaluated. Con-
sider metal particles with a radi_us_ of 0.05 ym. Assume that - 103 ohm-cm, which includes most semiconductors
and all insulators the charge transferred is that already on the sur-
face in the area of contact. In the absence of detailed knowledge of
the physical processes involved, it is not possible to carry out the
analysis for the charging of insulating materials.
CONTACT ELECTRICITY MONITORING INSTRUMENTS
Review of Instrument Development
In monitoring instruments based on contact electrification, a flow of
the aerosol is directed around or through an insulated probe. The
transfer of charge from particle-probe collisions results in a current
which is continuously monitored with an electrometer.
An early version of an Instrument employing a spherical metal probe was
described by Schiitz.^ A later version of his instrument, shown in
Figure 2, utilized a probe-in-nozzle technique.
rt i
Prochzaka ' has described the development of the Konitest, an instru-
ment which was produced commercially for a time. In the Konitest, the
electrode is a tube of a semiconductor, steatite (magnesium hydrosili-
cate). The gas is introduced radially and the resulting helical path
cause5 the particles to impinge on the walls of the tube. In a second
version, the tube is shaped as a Venturi nozzle (see Figure 3).
655
-------
Measuring head Flowmeter Filter
Dust-laden
air
Blower
Indicator
Figure 2. A dust monitor based on the contact electricity principle
using a probe in a nozzle according to Schutz (Ref . 2) .
Soo and his collaborators
1
constructed instruments using a spheri-
cal metal electrode and also a tubular electrode. In the USSR, Kisler
has described some contact electricity instruments including a monitor
using a wire probe. The IKOR Air Quality Monitor was apparently inde-
pendently developed and is now commercially available. It utilizes a
bullet-shaped Inconel probe in a pipe.
Characteristics of the Contact Electricity Monitor
A few experimental investigations have been made of the characteristics
of the contact electricity monitor. Most were carried out under field
rather than laboratory conditions. On some of the characteristics there
are contradictory reports. Some of the findings on the important charac-
teristics are reviewed below.
Electrical Current vs Mass Concentration - In most cases the instru-
ments are to be used as an indirect measurement of the particulate
mass concentration in the gas monitored. It is most important to
establish the correlation between the electrical current from the probe
and the mass concentration measured gravimetrically.
the following relationship for quartz dust:
I = akb
Schutz reported
where I is the electrical current in amperes, k is the mass concentration
in mg/m , the exponent b ranges between 1.26 and 1.30, and a is inversely
proportional to the particle diameter.
656
-------
Distributor
measuring chamber 'insulation-
Exciter tube-
! Pressure
measuring chamber.!!
Exciter tube
Figure 3. Tv/o versions of the Konitest dust sampler are shown; in (a)
the electrode is a cylindrical tubulence chamber, in (b) a Venturi
nozzle is employed (from Prochazka, Ref. 4).
651
-------
On the other hand, Prochazka4 found the Konitest to give an accurately
linear indication for concentrations from 0 to 3 g/m3. Samples of his
data are plotted in Figure 4. The dusts were sampled at various indus-
trial sites. Ito, et al23 also reported a linear relationship for a
Konitest monitoring cement kiln exhaust at the exit of an electrostatic
precipitator for concentrations up to 10 g/m3. Schnitzler, et al »
found a correlation coefficient of 0.92 between the gravimetric mass and
the Konitest reading for effluent from a coal-fired plant. The perfor-
mance of the Konitest compared very favorably to the best response
obtained from optical transmissometers and beta radiation attenuation
monitors at the same location.
/»
EMI
W •*/>
t*
oo"
5 »
O'lW
w
(0
w
»
0
/
/
/
•
1
/
1
J
/
i
I
,^
i
!
•1^
r\
,
1
y
s
• L
—
«/
Ki
1 !
!
• Nov. 64
• Dec. 64
i
1
H
/
i»f»»ttn*w«««fl
M/Afa
ifW/rnV
:
w
TV
0
w
»
7D
to
a
to
X
u
m
a
\
\
-,-L -
i
i
_•_
t
t
'
1
*
1 |
-
>
^
!
yT
'.^5
lS"7 j
[5^f* *
/Ti* 1
XI
1
/
«J :
«J :
•7 :
•8J
X
,^
—
^
i
1'i Nov
24 Feh
25 Feh
2fi Feb
27 Feb
10 June
^TH
j
i
i
! 5^
7-f-t-
1
i
— »--< — •
—
.r>i
r. 65 r f"
. Go
r f* '
r. 65 ,
: Go_j
T
i
x?
—
"
.--,
^
r
?j
::
I--
I
ar w aj o» 05 w a? oj as u ?.? u u iv 15 u v u t» M» it it i
IDust content b
Figure 4. Calibration curves showing the Konitest current vs the mass
concentration for (a) blast furnace gas and (b) flue gas, from ^roch-
azka (Ref. 4).
Cheng and Soo^'2^ obtained plots of mass flow vs probe current using
coal dust. The plots show small deviations from linearity, being con-
cave upward.
Field tests of the IKOR Air Quality Monitor show that the integrated
current (total charge) has a reproducible ratio to the gravimetric mass
for aluminum oxide.' Good agreement was also obtained between the total
charge and the mass concentration obtained with an EPA sampling train.
These tests did not span a wide range of concentrations. The author has
investigated the IKOR AQM using laboratory-generated aerosols.9 The
658
-------
total charge was found to vary linearly with gravimetric mass for alu-
minum, a conductor, and aluminum oxide, an insulator (see Figure 5).
ioor
E?
I
c*
s
O O
-00
10
100
Mass, Mg.
Figure 5. Charge vs mass of aluminum oxide particles measured with an
1KOR monitoring instrument by the author (Ref. 9). The line is drawn at
45° corresponding to a linear dependence of charge on mass.
Composition of Dust Sampled - Although field tests have been conducted
with a variety of dusts, there has been no reported correlation of the
instrumental response to some characteristic of dust sampled. Schiitz^
concluded that the influence of the properties of the material was far
less decisive l.han the particle size. Ito, et al^3 stated that the
current obtained for a given mass concentration depended on the mater-
ial, but that for cement dust the electrical resistance had little
659
-------
effect on the output. Most users of the contact electricity monitor
have simply empirically calibrated the monitor for each type of dust
sampled.
The author has measured the sensitivity (charge/mass) of the IKOR AOH
for laboratory-generated aerosols of a number of substances." The data
are listed in Table 1 in order of increasing sensitivity. It is to be
noted that this is also generally the order of increasing conductivity.
In Figure 6, the sensitivities are plotted against the log resistivity
of the materials.
3.0
1 Al
• Cu
2.0
— ••
Ag
Stainless Sleel
1.0
i Graphite
Glass
(Soda-lime)
I
Insulators
-10
-5
5 10 15
Log a , Ohm-Cm
20
25
Figure 6. Sensitivity (charge/mass) of the IKOR AQM vs the log of the
electrical resistivity of the material sampled, from the author's work
(Ref. 9).
For the glass particles the resistivity was quoted by the supplier. The
insulators have been lumped into a single point with a vertical bar
covering the range of sensitivities. The horizontal bar covers the range
nominally found for insulators; the individual resistivities of the sam-
ples used are not known. The correlation between sensitivity and resis-
tivity is evident except for U and Mo. These two metals have resistivi-
660
-------
TABLE 1
SENSITIVITY OF THE IKOR AOM
I. Insulators yC/g
Titanium dioxide 0.051 ± 0.004*
Red iron oxide 0.06 ±0.01
Silicon dioxide 0.09 ±0.02
Fly ash 0.148 ± 0.005
Aluminum oxide 0.20 ±0.01
Glass beads 3-8y 0.41 ±0.03
Glass beads 0.5-3u 0.46 ±0.01
II. Intermediate conductors
Molybdenum** 0.51 ±0.04
Tungsten** 0.67 ± 0.13
Carbon black 0.68 ±0.09
III. Metallic conductors
Stainless steel 2.0 ± 0..5
Silver 2.0 ±0.2
Copper 2.1 ±0.2
Aluminum 2.9 ±0.4
*standard deviation
**nonspherical metal particles
ties approximately three times that of the other metals used but the
decrease in sensitivity was disproportionate. One possible factor was
the nonspherical shape of the particles which could reduce the effective
contact area. The other metal particles used in this study were spheri-
cal .
The ranking according to electrical resistivitv can be understood in
terms of the characteristic time for charge transport. 'For good insu-
lators, this time is large so that only the area in the vicinity of t ,e
contact is involved in charge transfer. The area is larger for semicon-
ductors and charge can also be transnorted from the interior. For
metallic conductors the charge can probably be transported from the
entire particle. The oxide layers on the metal particles are nresumablv
thin enough so that they can be penetrated either by electron tunnelling
661
-------
or by conduction with a snail time constant. The theory of Ghent*, and
Soo" predicts a dependence of the sensitivity on the resistivity, how-
ever, the agreement with the experimental data is not auantitative.
These data provide a basis for predicting :he sensitivity for other sub-
stances, at least for dry dusts. It is likely that they will fall within
the range spanned by the table. The wide range in the sensitivities
(factor of 60) implies that a small amount of an insulator would be
difficult to detect in the presence of metals or good semiconductors.
Dependence of the Current on Particle Size - As pointed out above,
Schtitz found the constant a to be inversely proportional to particle
diameter. Thus, higher sensitivitv is -predicted for snail particles.
On the other hand, Cheng and Soo predict a current independent of
particle size for one experimental configuration. They report that this
was verified by Min. Ito, et al report the Konitest to be little
affected by particle size when sampling cement dust.
The author's experiments with the IKOR AOM showed that the sensitivitv
was not strongly dependent on particle size." A pclydisperse Al and a
sized Al sample had nearly the sane sensitivity and two sizes of glass
beads gave similar results (Table 1).
On the other hand the IKOR AOM. v/as found to barelv respond to high levels
of smoke from a filter cigarette. This shows there is an effective cut
off in the response to very snail particles. The experimental sensiti-
vity depends not only on the contact charging parameters but also on the
probability that the particles impact on the probe. Small particles tend
to follow the streamlines around the probe. This smaller probability of
impaction for small particles may offset the increased charging expected
theoretically.
Effects of Temperature, Humidity, and Precharge - Schu'tz^ stated that
the contact electricity monitor should not be operated above 70°C due to
the presence of thermal enf's. However, a modified IKOR AQM has beer-
operated at 593°C7.
Water layers on surfaces usually contain dissolved impurities. Some
authors believe electrolytic ions to be important for contact charging.
However, Harper ^ does not, and tried unsuccessfully to obtain an exper-
imental correlation between humidity and charging.
rt
Schiitz^ reported no effect on his instruments' performance for relative
humidity up to 99%. He did warn that droplets of water cause erroneous
indications. Cheng and Soo" point out that the conductivitv of mineral
particles increases with the relative humidity. Ito, et al^3 state that
humidity caused no problems in their tests of the Konitest.
662
-------
states that the humidity of the gas should be below the dew point. The
author found humidity up to 69% not to affect the 1KOR AOM, but droplets
caused the instrument to break down."
The author did not observe any significant effect on the IKOR AOM pro-
duced by pre-existing charge on the aerosol particles. Also, if pre-
charging effects existed, humidity would be expected to affect the
response. However, Kolar ^ suspected charge was interfering with his
Konitest measurements on dust from an electrostatic precipitator.
Probe Material - Uith the exception of the Konitest, most instruments
have used metal probes. According to Schiitz,2 the use of metal for the
probe ensures that the charging will always be positive. It seems clear
that this is a dubious assumption; in fact in field tests of the IKOR
the sign of the current has been found to vary with material and the
standard IKOR AOM is equipped with an automatic polarity-reversing
circuit to maintain positive indication on the recorder.
Dynamic Response - The author compared the response of the IKOR AQM
to an opticle particle counter." The time constant (1/e) of the IKOR
AQM is approximately 10 seconds. Since the Climet optical analyzer is
orders of magnitude faster, an R-C network was added to increase the
Climet time constant to 4 seconds. Figure 7 shows the superimposed
chart recordings of the two instruments for polydisoerse aluminum oxide
dust. Some of the differences in the records, particularly the more
extreme highs and lows of the Climet trace, can be attributed to the
2.5 times shorter time constant of the Climet.
On the basis of nany such runs, it was concluded that the correlation of
the current records was satisfactory. There is one difference which
should be mentioned. Sometimes at the very beginning of a run, the IKOR
AQM was much less sensitive. The sensitivity increased by an order of
magnitude in the first minute or so. This sometimes happened even if
the instrument had been previously running on the same dust, sometimes
even within the hour. This sensitizing effect will be discussed further
below.
Surface Condition of the Probe - Direct evidence was obtained by the
author that the condition of the surface of the probe of the IKOR
instrument has a very important effect on the sensitivity. It was
mentioned above that there is sometimes a sensitizing effect at the
beginning of a run. A long term change in sensitivity was also obser-
ved during a series of measurements, the sensitivity increasing by a
factor of 5 to 10 over a period of about one month. Between runs with
different substances, the probe was simply wiped clean with a dry cloth.
On the other hand, the sensitivity was found to decrease by an order of
magnitude when the probe was scrubbed with abrasive cleanser and deter-
gent.
663
-------
IKOR AQM
OPTICAL ANALYZER
100
200
300 400
Time, Seconds
500
600
700
Figure 7. Dynamic response of the IKOR AOM compared to that of an opti-
cal particle counter for aluninum oxide particles, from the author's
measurements (Ref. 9).
There would seem to be two possible explanations of the surface sensiti-
zing effect. One would be a cleaning effect or exposure of fresh surface
produced by the bombardment by particles, analogous to sandblasting.
The other possibility is that a coating of particles builds up on the
surface, altering the charging conditions, since particles would then
encounter other particles rather than the surface itself. ~"»25 ^t present
the evidence is insufficient to allow one to decide between these two
possible explanations.
664
-------
SUMMARY
Experiments confirm that the total charge collected by the contact elec-
tricity monitor correlates well with the mass determined gravimetrical-
ly. No evidence has been found for disturbing effects of humidity below
the dew point. The instrumental response cuts off for very small par-
ticles. The sensitivity for various materials depends on the electrical
resistivity, ranging over a factor of 60 from metals to insulators.
Also the sensitivity depends on the condition of the surface of the
probe.
The theory of Cheng and Soo for particles of metals or good semiconduc-
tors overestimates the contact charge by several orders of magnitude
but the discrepancy may be due to effects such as oxidation of the sur-
faces of real particles. No theory is available for the charging of
insulators due to lack of understanding of the physical mechanism.
In view of the evidence reviewed here it would appear that the contact
electricity monitor holds considerable promise for stationary source
monitoring, expecially for non-sticky effluent of relatively constant
composition. It is not suitable for ambient air because of insufficient
sensitivity and ambiguity in the response to complicated, varying mix-
tures of materials.
Additional work is necessary to clarify the nature of the probe surface
sensitization effect or to control it within limits. Better theoretical
understanding of the charging mechanism is desirable but the complica-
tions imposed by actual particle surfaces are formidable.
ACKNOWLEDGEMENTS
The author's work on the contact electricity monitor was supported by
Grant No. R-802726-01 from the U.S. Environmental Protection Agency and
sponsored by California State College, Stanislaus.
I thank Mr. John Nader for his encouragement and helpful suggestions.
Thanks are also due to Mr. Arnold H. Gruber of IKOR, Inc. for supplying
valuable information.
REFERENCES
1. Schu'tz, A. Eine Anordnung zur Registrierenden Kontactelektrischen
Staubmessung. Staub 24: No. 9, 359-363, 1964.
2. Schiitz, A. A Recording Dust-'leasuring Instrument Based on Electric
Contact, with Logarithmic Indication. Staub 26: No. 5, 18-22,
1966.
665
-------
3. Prochazka, R. Neueste Entwicklung des auf Kontaktelektrischer Basis
Beruhenden Staubgehaltsmessgerates Konitest. Staub 24: No. 9, 353-
359, 1964.
4. Prochazka, R. Recording Dust Measurement with the Konitest. Staub
26: No. 5, 22-28, 1966.
5. Schiitz, A. Technical Dust Control Principles and Practice. Staub
26: No. 10, 1-8, 1966.
6. Cheng, L. and Soo, S. L. Charging of Dust Particles by Impact.
J. Appl. Phys. 41: 585-591, 1970.
7. IKOR, Inc., unpublished reports, and Grubtr, A. H., private commu-
nication.
8. Schnitzler, H. Messtand fur die Prufund und Kalibrierung von Regis-
trierenden Staub-und Gasmessgeraten in einem Steinkohlengefeuerten
Kraftwerk. SchrReihe Ver. Wass-Boden Lufthyg, Berlin-Dahlem, V. 33,
Stuttgart, 1970.
9. John, W. Investigation of Particulate Matter Tlonitoring Using Con-
tact Electrification. Environmental Protection Agency, Research
Triangle Park, M.C., Technology Series Report Number EPA-650/2-75-
043, February, 1975, 45p.
10. Sem, G. J., et al. Instrumentation for Measurement of Particulate
Emissions from Combustion Sources. Thermo-Systerns, Inc. Vol I:
Particulate Mass, Report Number APTD-0733, April 1971 and THIS PB
202 665.
11. Harper, W. R. Contact and Frictional Electrification. Oxford,
Oxford U. Press, 1967. 369 p.
12. Loeb, L. B. Static Electrification. Berlin, Springer-Verlag, 1958,
240 p.
13. Hendricks, C. D. Charging Macroscopic Particles. In: Electrostat-
ics and Its Applications, Moore, A. D. (ed.), New York, John Wiley
6, Sons, 1973, p. 64-67.
14. Harper, W. R. The Volta Effect as a Cause of Static Electrification.
Proc. R. Soc. A 205: 83-103, 1951.
15. Vick, F. A. Theory of Contact Electrification. Brit. J. Appl. Phys.
London, Suppl. No. 2: S1-S5, 1953.
666
-------
16. Harper, W. R. llov; Do Solid Surfaces Become Charged? In: Static
Electrification, Proceedings of the Conference organized bv the In-
stitute of Physics and the Physical Society, Static Electrification
Group, London, May 1967. Inst. of Phys. and Phys. Soc. Conference
Series Number 4. p. 3-10.
17. Krupp, H. Physical Models of the Static Electrification of Solids.
In: Static Electrification, 1971, Proc. of the Third Conf. on
Static Electrification organized by the Static Electrification Group
of the Institute of Physics held in London, May 1971. Conf. Series
No. 11, The Institute of Phys., London and Bristol, p. 1-16.
18. Soo, S. L. Fluid Dynamics of Multiphase Systems. Walthatn, Mass.,
Blaisdell Publ. Co., 1967. p. 413-416.
19. Soo, S. L. Dynamics of Charged Suspensions. In: Topics in Current
Aerosol Research, Vol. 2, International Reviews in Aerosol Physics
and Chemistry. Oxford, Pergamon Press Ltd., 1971. p. 71-73, 90-93.
20. Soo, S. L., Stukel, J. J. and Hughes, J. M. Measurement of Mass
Flow and Density of Aerosols in Transport. Envir. Sci. and Tech. 3:
No. 4, 386-393, 1969.
21. Cheng, L., Tung, S. K. and Soo, S. L. Electrical Measurement of
Flow Rate of Pulverized Coal Suspension. J. of Eng. for Power,
Trans, of the ASME, 92A: No. 2, 135-145, 1970.
22. Kisler, S. YA. Monitoring of Concentration of Disperse Phase of
Aerosol Flows by Electric Contact Method. Mekh. Avtomat. Proizvod.,
26: No. 9, 27-28, 1972. (In Russian)
23. Ito, T., Saito, H. and Furuya, N. Continuous Dust Content Measure-
ment Konitest. Proc. Japan Soc. of Air Poll., 13th, Nov. 1972. p.
247. (In Japanese)
24. Kolar, J. The Electrostatic Dust Measuring Device "Konitest" and
its Functioning in a Remote Heating Plant. Tech., Ueberwach, Dues-
seldorf 10: No. 6, 188-190, 1969.
.,25. Soo, S. L. and Trezek, G. J. Turbulent Pipe Flow of Magnesia Par-
ticles in Air. Ind. and Eng. Chem. Fundamentals 5: No. 3, 388-392,
1966.
667
-------
OPEN CAVITY LASER
"ACTIVE" SCATTERING PARTICLE SPECTROMETRY
FROM 0.05 TO 5 MICRONS
Robert G. Knollenberg
and
Robert Luehr
Particle Measuring Systems, Inc.
1855 South 57th Court
Boulder, Colorado 80301
ABSTRACT
An Active Scattering Aerosol Spectrometer is one that uses the
active open cavity of a laser as the source of particle illumination.
The interferometric aspects of the oscillating radiation illuminating
the particles produces both forward and backward scattered radiation
at all collecting angles. This, coupled with the fact that the
collecting optics solid angle can be considerably greater than two
steradians, results in a system with great particle sensitivity and
reduced sensitivity to refractive index.
The particles are illuminated with a source of radiation many
times greater than that possible with the hottest incandescent source
known. This results in a system fully capable of sizing particles
several hundred Angstroms diameter using photomultiplier detectors
or 0,1 microns diameter using solid state silicon detectors. The
fundamental limit of size detection is governed by particle evaporation
rather than the noise component of the scattering from the molecular
By using a beam splitter and double pulse height analysis, a
system has been constructed which determines where a particle passes
through a beam in the sampling volume, and reduces background light,
The entire sampling volume has unrestricted flow precluding problems
with evaporation of wet aerosols and the effects of plumbing walls,
Preceding page blank
-------
OPEN CAVITY LASER
"ACTIVE" SCATTERING PARTICLE SPECTROMETRY
FROM 0.05 TO 5 MICRONS
Robert G. Knollenberg
and
Robert Luchr
Particle Measuring Systems, Inc.
1855 South 57th Court
Boulder, Colorado 80J01
1. INTRODUCTION
Techniques that employ the active open cavity of a gas laser ;JG
a particle illumination source for light scattering; measurements are
what we designate as active scattering techniques. Such an instrument
is particularly applicable to aerosol measurements. The open cavity
laser was first used by Schleusener as a particle illumination source.
He performed particle extinction measurements to size particles of
several tens of microns in size. Procter^ has reported a similar in-
strument to measure the surface area of dusts. Schuster and Knollenberg
detailed the open cavity extinction process with particular attention
given to the laser gain dynamics.
The open cavity extinction process is known to be quite insensi-
tive to refractive index and particle shape, however, there are certain
difficulties in achieving practical instrument designs to measure in
the sub-micron range. The problem generally becomes one of resolving a
very small loss in the presence of large signal background in the
presence of noise.* A single frequency laser does have considerable
* Very few lasers are thermally quiet. Most exibit high frequency
(lOOKHz-5MHz) components of 0«1 - 1% noise modulation in spite of
current regulation» At low frequency these are attributed to axial
mode competition with considerable FM. At high frequency is the
characteristic anode associated modulation above 1MHz. The theoretical
extinction size detection limit associated with the shot noise on the
transmitted light is about 0.3u.
670
-------
advantage in such applications.
Recognizing the lower limit of detectivity of the open cavity ex-
tinction process to be in the vicinity of O.Ju we turned to open
cavity scattering or active scattering in 1971 to extend to smaller
sizes. Some work was also done with interferometrically enhanced
scattering wherein a standard laser was output coupled to a secondary
total reflecting mirror. This method is still under study but has
measurable disadvantages over active scattering. The first operational
active scattering aerosol spectrometer (ASAS) was designed as an
airborne instrument for measurements of the aerosol population in the
urban plume of St. Louis as a part of Metromex (Metropolitan Meteoro-
logical Experiment). It was developed by Knolleriberg and flown aboard
the University of Chicago's Cloud Physics Laboratory aircraft. Figure
1 and 2 show some results of multilevel measurements using this instru-
ment which operated multimode and had a useful size range of about
0.25 to 5 microns.
4.000r
UJ
lu 3.0QO
t 2,000
K
-1
1
1.OOO
AEROSOL CONCENTRATION (Ncm-'MOO)
FLIGHT 106, RUNS 3-7
[28 JANUARY 1972]
INVERSION LEVEL
KE St Loun-Allon- 6rc°"""-|- SI Louii —
Wen Suburban -1
ME
! I I I
TROY 246
VOR
NAUTICAL MILES
sw
I I I I I I I i ! I I I I
8 10 12 1* 16 18 20 22 Z4 26 26 30 32
Figure 1. Vertical cross section of ASAS measured aerosol
concentration in St. Louis urban plume. Note high density
contours resulting from power plant plume.
671
-------
4.00O
I-
Ul
u 3,000
g
t 2,000
1.000
AEROSOL DENSITY (/tgm m-» x 10)
FLIGHT 106, RUNS 3-7
[28 JANUARY 1972]
INVERSION LEVEL
|—E si LOU* -
NAUTICAL MILES
NE
sw
TROY 24 6 8 10 12 14 16 18 20 22 24 26 28 30 32
VOR
Figure 2. Aerosol density corresponding to Figure 1.
In early 1972 work was also initiated on ASAS instruments for
NASA space related measurements in a hard vacuum (see Knollenberg4 for
a description of such a device). A number of active scattering aerosol
spectrometer instruments have also been manufactured by PMS,Inc.
Such an instrument is also described by Schehl, et.al^.
The present work is a description of a significantly improved
ASAS suitable for laboratory and ground based field measurements. An
airborne fully "in situ" instrument is also under development. These
instruments have ultimate sensitivities well below 0.1 microns and
have several unique features to be described.
672
-------
2. THE LASER CAVITY
AS A PARTICLE ILLUMINATION SODBCE
There are perhaps five distinct advantages of using the laser
cavity as a source of particle illumination rather than classical
laser scattering approaches. These are according to decreasing
importance.
1. Greatly increased illumination levels
2. The active interferometer reduces uncertainty due to
refractive index and particle shape.
3. Available illumination reference exactly proportional to
particle illumination.
4. Enhanced light collecting configurations with reduced
stray light possible.
5. Available Doppler signals for velocimetry applications.
There is little doubt that our initial attraction to the active
cavity was the available high energy density. As a source of illum-
ination the laser cavity is unmatched in available intensity. The
following exercise is worth following. Shown in the optical diagram
of Figure 3 is "the typical cavity configuration for ASAS production
instruments. A hybrid laser tube is constructed with a 35cm high
reflectivity (99«9+%) mirror sealed at one end of the plasma tube and
a Brewster's window termination at the other. A second plane high
reflectivity mirror completes the laser cavity. This results in a hem-
ispherical cavity operating in TEM . mode with a beam having a spot
diameter of about 300u at the plane mirror and 800u at the curved
mirror.
Now consider the fact that the manufacturer uses a 1% trans-
mitting mirror (generally the curved mirror) in this same tube marketed
as a 2mW device. It becomes obvious that in such a device, the
radiation progressing towards this output mirror must be 100 times
as great as the output or roughly 200mW. Now substitute a highly
reflective mirror for the 1% transmitting mirror. Further intensity
buildup is limited only by internal cavity losses. Measurements
generally show an additional increase of a factor of 5 to 6 indicative
of approximately 0.2% fixed losses at the mirrors and Brewster's
673
-------
Window, Thus, we have achieved a source of approximately one watt of
radiating power illuminating a particle from both directions. At the
sample volume location the laser beam is approximately 500 microns
diameter. Assuming a Gaussian intensity distribution, a brightness in
excess 1000 ty/cm is computed. We have measured values as high
as 4000 W/cnf.
ACTIVE SCATTERING AEROSOL SPECTROMETER OPTICAL SYSTEMS
REFRACTING SYSTEM
MASKED APERTURE
PHQTODIODE
PRE-AMP
PLANE MIRROR
OX TRANSMISSION
SCATTERING
PHOTOQETECTOR MODULE
50% lor unpolariitd H«M
HIGH COLLECTION EFFICIENCY REFLECTING SYSTEM
Figure 3« The refracting optical system is generally used.
The high efficiency reflecting system is reserved for maximum
resolution applications.
674
-------
Probably the most intense incoherent source available is the
PEK LABS 107/109 high pressure mercury arc lamp. It's corresponding
brightness over it's entire visable spectrum approaches 250 W/cm .
In any practical optical condensing system, it is nearly impossible to
collect more than 20% of the light resulting in an illumination source
more than one order of magnitude less intense than the laser just
described. It is also no small matter that the laser as described is
collimated to 0.5mr while the arc lamp would have a measured diver-
gence of more than one radian.
It should be pointed out that a highly focussed laser beam can
achieve even greater intensities than computed here. However, the
resulting beam affords an impractically small sample volume. It
should be apparent that the natural intrinsic beam width in the cavity
is on the same order as the sample "view volume" dimensions of more
standard optical counters. One might ask whether or not an argon or
other more powerful laser might offer additional advantages. 1'ossibJy,
however, the illumination levels of the He-Ne laser are already
approaching potentially high enough levels to vaporize particles. For
instance, a O.O^u diameter particle with an absorption cross section
equivalent to it's geometric cross section and a heat of vaporisation
of water would vaporize in the above described beam for transit times
longer than 3^0u sec. In our instruments transit times are generally
an order of magnitude faster.
Scattering within an open cavity laser is a quite distinct phen-
omenon from classical scattering. It is most closely approximated by
scattering within a standing wave. Various experiments have contributed
to a growing understanding of the phenomenon. Probably most important
is the fact that the particle itself determines to a certain extent
the intensity of radiation illuminating it. Specifically, it can be
shown that when a particle interacts with the internal laser beam,
that the extinction loss always exceeds the scattering loss for a
non absorbing particle. This results in the non-linear response firr.t
reported by Schleusener for extinction within the cavity. Schuster
and Knollenberg5also predicted that the extinction process itself
would not be expected to follow classical behavior showing rather
suppressed resonance behavior due to interferometric cavity effects.
More will be said in this regard in Section 5.
* The absorption cross section to mass ratio is the significant para-
meter here and typically has a maximum value in the vicinity of 500A .
The Qabs (efficiency factor) can be 10 times the Qsca at these sixes
(see Van de Hulst£ page 180).
675
-------
A collimated laser beam is probably the most manageable light
source available. The beam internal to the cavity has further advan-
tages since it does not require dumping in a light trap and background
light is more easily controlled. It is possible to design systems
with nearly 4TT steradian collecting solid angles (see the hemispherical
mirror configuration of Schehl et.al for example).
The sealed mirror of the laser cavity is a, protected surface
whose transmission characteristics do not vary with time. The leakage
out this mirror is thus directly proportional to the power illuminating
the particles at all times. It's measurement affords a means of auto-
matically compensating for changes in illumination level.
The internal laser cavity provides an internal standing wave of
radiation with predicted Doppler phenomenon. Particles passing down
the laser axis scatter forward and backscattered waves of frequencies
differing in Doppler content. Such scattering pulses have high
frequency ripple related to velocity. The modulation depth (index)
of such pulses varies with size since the ratio of forward to backward
scatter is also size dependent. The importance of the Doppler com-
ponent may eventually lie in intrinsic measurement of flow rate or
trajectory analysis. However, it is only easily observed in the sub-
micron size range.
3. INSTRUMENT OPTICAL DESIGN CONSIDERATIONS
It was previously mentioned that a laser cavity allows enhanced
designs for stray light shielding as well as collecting geometries. It
should also be pointed out that the laser is itself a monochromatic
source which allows spectral filtering. Spectral filtering is used in
all ASAS optical designs. Two variations of optical systems are used
in ASAS instruments. They differ only in the use of a parabolic
mirror instead of refracting lenses in the primary light collecting
optics. The reflective system is used for high efficiency light
gathering for applications below O.lu. The refracting system is a
high resolution system which is used for particle trajectory analysis
as well as light gathering. Because of it's unusual operating mode
it will be described in detail.
Since we were certain that the light level would be adequate
for O.lu with refracting optics we chose to study designs that offered
advantages in the sampling problem area. In this regards, one has gen-
erally only two choices. You either force the particle stream through
676
-------
a region of uniform intensity or by secondary measurements determine
where each particle passes through the beam and reject or accept the
measurement depending on whether it passes through the accepted part of
the "view volume" which we will hereafter refer to as the sample volume.
With regards to the laser beam in question we are dealing with a
Gaussian intensity distribution which can be considered as sufficient-
ly uniform in intensity only over it's central 20% of diameter. A
monodispersed particle spectrum would then result in a Gaussian
distribution of scattering pulse amplitudes without limiting the
measurement to the center of th-? Gaussian. There are of course well
known techniques involving hydr;>dynamic focusing coupled with sheath
air flow for entraining particles into small laminar jets. However,
under the best of conditions it is quite difficult to reduce the dia-
meter to less than 100 microns and there is a comparable uncertainty
in the positioning control of such a jet. In short the natural beam
diameter is smaller than optimum. One could of course reverse the
mirror configuration and double the beam diameter. This would also
reduce the intensity of the beam by a factor of four. However, one of
the primary goals was also to eliminate the plumbing and aerosol
interaction with it. This requirement was the determining factor in
the optical system design of ASAS instruments. The method used to
define this volume requires a high resolution imaging system and the
use of a masked beam splitter to derive two signals. The signal de-
tectors in conjunction with double pulse height analysis provide a
means of determining if a particle position is in the desired sample
volume.
The optical system design involves a high resolution optical
system wherein the particle trajectories at the object plane are
magnified 10X or greater, (see Figure 3). The laser used is a 2raW
He-Ne Hybrid tube tuned to TEM_n mode. The center of the sampling
volume is about 35~40mm from the adjustable mirror surface where the
beam is about 500 microns diameter. Particles passing through the
laser beam in the sampling aperture scatter energy into the optics.
The collecting optics include 10 spherical refracting elements in
3 cells, an interference filter, and a beam splitter. The first lens
cell contains four elements and is an F/1.4 55n™ fl lens. The first
element of this lens has the plane mirror of the laser cemented at,
it's center and has a central stop to exclude light collection over
it's central aperture. This lens gives a 0.37 numerical aperture and
collecting angles of 4° to 22° (backscatter collection at 176°-158°
is also implied).
A dielectric interference filter is immediately behind the F/1.4
lens. This filter has 60% transmission at 6328A° and has a lOOi°half
width. There is collimated transmission by particle scattering at the
object plane through the interference filter.
677
-------
The second lens is identical to the first lens and reproduces
the particle image at unity magnification. The third lens in the
collecting optics is a standard microscope objective which forms a
secondary image in the detector plane at 10X or higher magnification.
The scattered energy is relayed from the object plane collected by the
scattering photodetector module. A 50% beam splitter produces two
image planes for two detectors. The reflected image prism face is
masked with a 0.78mm diameter vertical slit to block central trans-
mission. The transmitted laser beam is dumped at a central stop
behind the mirror on the first lens element.
The relative size of the sample volume cross section with respect
to the laser beam is depicted in Figure 4. The sample volume is seen
to include only the region near the center of the laser beam. The
sample volume cross section is noticeably diamond shaped. Tho center
of this diamond cross section coincides with the object plane of the
collecting optics. The points of this diamond cross section define
the limiting depth-of-field of the ASAS. Both the width of the cross
section and the depth-of-field vary inversely with the magnification
used in the collecting optics. Why this is the case will now be
explained.
The image plane forms at the two exit faces of the beam splitter
in Figure 5. The light transmitted on axis through the beam splitter
is the signal used to size the particles. The light reflected at 90
forms it's image on a circular aperture with a central opaque vertical
slit. Shown in Figure 5 are the size and positions of images formed
by particles at various positions in the illuminated volume. It is
important to understand that the image size is only a function of it's
axial displacement from the object plane. The image size is linearly
related to the numerical aperture of the collecting optics and is given
approximately by: Image Size = N. A. x Displacement from Object Plane.
It is apparent that only images that are near the object plane
and the center of the sample volume form images with light concentrated
on the opaque slit on the masked detector as illustrated in Figure 5«
Thus, such images transmit little signal through the masked aperture.
By comparing the signals at the masked aperture with that from
the unmasked aperture,it is possible to determine whether they had
their origin within the diamond shaped sample volume cross section.
Figure 6 depicts the size and position of images in the image plane
originating on the line defining the diamond sample volume cross
section. These images are seen to be fully enclosed within the
envelope defined as the masked aperture slit width. It is apparent
that if we used a very high gain or extremely sensitive detector on the
masked aperture, that we would be able to see the edge of these images
and the sample volume would be defined as shown if one rejected parti-
cles whose images extended beyond the mask.
678
-------
OBJECT PLANE
LASER
BEAM
SAMPLE VOLUME
CROSS SECTION
I/E* INTENSITY
(BEAM EDGE)
GAUSSIAN
INTENSITY
DISTRIBUTION
TEMOQMODE
— II
-XX-
MAXIMUM VARIATION
IN INTENSITY OVER
SAMPLE VOLUME
Figure 4. Relative size and position of sample volume
cross section.
MASKED APERTURE
flD
O.78mm
*-2.50 mm
APERTURE AT
PRISM FACE
SIGNAL APERTURE
X APERTURE AT
\ PRISM FACE
1 1 IMAGE
PLANE
Df?pth-Of-FinlH and wit-
Sample Vojump ft'irtth
ParticIP Imaqp nr>ar
Object rianp and rpnt^r
of Samplp Volume ^iirffh
Depth-Of-Field but
out.sidr. Sample Vol i!
Width
Particle Imane near
Oblect Plane but
nut side sample Volume
TO OBJECT PLANE
Figure 5. Image sizes and positions on masked aperture detector.
679
-------
It is not practical to amplify signals beyond limits because of
signal-to-noise ratio deterioration. Also because of the large range
of signal levels encountered, it is more desirable to compare the
amount of signal received by the masked aperture detector with the
signal aperture detector to make more accurate measurements of particle
image position. What is normally done in the ASAS is to use a gain
ratio of masked aperture to signal aperture detector gain of 2X.
Particles whose pulse amplitudes as seen by the masked aperture
detector are greater than those seen by the signal aperture detector
are rejected. This defines the volume and image relationships shown
in Figure 7.
The gain ratio of 2X was selected as the best practical choice
for several reasons. First of all, the gain ratio must be sufficient
to reject images so diffuse that they fill the entire aperture. This
also places limits on the percentage of the aperture thai can be
masked (not greater than 50% in the case of 2X gain ratio). Second,
if one makes the gain ratio extremely high, there is less tolerance to
slight misalignment and reflected light. Thirdly, low gain ratios do
not sufficiently limit depth-of—field. The sample cross section is the
product of the sample volume width and 1/2 the depth-of-field.
The parabolic mirror in the optical system used for high effici-
ency light collecting is not astronomical quality, hence, it does not
lend itself to the rejection methods just described. The cost of a
mirror of comparable resolution would be prohibiting. Since the appli-
cation is limited to particles which may be adequately plumbed and the
measurement restricted to a small axial segment of the laser beam it
is not necessary to provide depth-of-field limiting. The measurements
are restricted to the center of the Gaussian by transit time analysis
of the pulses. Pulses of maximum width correspond to particles
passing through the center of the beam and are accepted. Pulses of
short duration correspond to trajectories through the weaker intensity
regions and are rejected. A 4si rejection ratio guarantees particle
trajectory acceptance only over the central beam region having
intensity variations less than 10%. This rejection method has been
used exclusively in the ASSP (Axially Scattering Spectrometer Probe)
manufactured by PMS for airborne cloud droplet sizing.
680
-------
OBJECT PLANE
MASKED APERTURE
4/SLIT WIDTH
LIMIT OF DEPTH-OF-FIELD
Figure 6. Image size as a funct.ion of sample volume position
for the case of high gain ratio of masked-to-signal
aperture.
OBJECT PLANE
MASKED APERTURE
1 SLIT WIDTH
LIMIT OF DEPTH-OF-FIELD-
Pigure ?• Image size as a function of sample volume position
for the case of 2X gain ratio of masked-to-signal
aperture.
681
-------
4. NOISE ANALYSIS
One of the most important aspects of any instrument designed bo
measure transient phenomenon is the consideration of noise. The de-
sign of optical counters most often involves photomultiplier tubes.
However, as was previously pointed out our baseline instrument design
involves solid state photodiodes rather than photomultiplier detectors.
It is a well known fact that the ultimate sensitivity of photomultipliers
allows for light level measurements several orders of magnitude lower
than with solid state photodiodes. This could be compensated by the
fact that the ASAS instrument produces orders of magnitude greater
light signal than conventional counters. However, our choice of the
solid state photodiodes is primarily a result of noise analysis per-
formed. It will be shown that the signal-to-noise ratio is in fact
higher for solid state devices than for photomultiplier tubes for the
bulk of our measurement range.
The appropriate starting point of such a calculation is to es-
timate the expected scattering signal as a function of particle size.
For a 1000 watt/cm illumination level a square micron of scattering
cross section produces about lOuW of scattered energy. The percentage
of light collected by the optical system can be as high as 50% for
particles whose central diffraction lobe (Airy disc) just fills the
collecting aperture. This occurs for a particle of about 1 to 2
microns diameter when the 0.37 numerical aperture is maintained. For
larger and smaller particl es it is substantially reduced although even
our 4 ~ 22 system generally collects more than 15% of all the light
scattered. A conservative estimate of 10% results in luW of scatter-
ing signal per square micron of scattering cross section. The conver-
sion of the photon flux to photocurrent in the detector can be as high
as 50% in solid state devices but is at best a few per cent in photo-
multiplier tube cathodes. In the case of the solid state detectors
we are thus dealing with roughly luA of photocurrent for a particle
slightly less than one micron in size. A typical response curve as
measured from an instrument is shown in Figure 8. In terms of our
noise analysis we will compare these with noise currents to define
signal—to-noise ratios and noise equivalent particle size which de-
fines the limit of detection. For a solid state detector the optimal
detector configuration uses an operational amplifier in a current-to-
voltage converter mode. The equivalent circuit is shown in Figure 9.
682
-------
It is most convenient therefore to express the associated noise in the'
form of current noise referred to the amplifier input.
Three noise currents were calculated:
Shot noise Ign = J 2 eIQ •>• ^ , (l)
Johnson noise I,_ = .IT? —- ; — /2)
V 0
Transister noise I, = T (l + 0 ) 1 (5)
tn o •Q n "
r F
—19
Where: E = 1.6x10 ' coulumbs
K = 1.38 x 10"25 eV deg "1
T = Absolute temperature
R_ = Feedback resistance
CL, = Feedback capacitance
I = Sum of all detector leakage current, amplifier
bias current and signal current
< - R r
1p - ^F °F
^T •— T? O
I A ~ KA LA
¥„ = Input transister voltage noise
C = Sum of all input capacitance
The above equations represent a 2 pole networth transfer
function analysis. The time constant R. C. is that associated with the
amplifier output, an output filter section or a second stage amplifier.
The ratio of/ft,/T/. for minimum transister noise is about 1.6.
683
-------
10,000 r
MEASURED SIGNAL PHOTOCURRENT
LATEX SPHERES (M=l.58)
FOR
1000 U
CO
LU
CC.
LU
Q.
LU
CC.
DC
Z)
o
o
Q_
<
S2
CO
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8
DIAMETER (MICRONS)
Figure 8. The measured photocurrents were with a laser generating
approximately 2.5 W/cm energy density.
684
-------
SILICON PHOTODIODE-PREAMP NOISE
0.001
10KHz
100KHz
I MHz
10MHz
3ANDWIDTH
Figure 9. The above noise analysis represents a low noise input
FET pre-amp. The following assumption was made:
'K = 1,6'Y (condition for minimum transistor noise)
V-, = 2.5 nVV~Hz(low noise input FET 2N55&4)
685
-------
An important aspect of equation 3 is that the transistor noise Is
amplified by the ratio of Co/C. Clearly this noise would be minimal
for Co=0. The actual minimum value one can obtain is about 5pf» while
25pf is measured in our system. Also the Johnson noise and transistor
noise decrease with increasing Rf. For computational purposes, the
minimum bandwidth requirement sets the value of R- and the stray feed-
back capacitance (typically 0.5 pf) generally limits the maximum
value of R-,. If the laser beam at the sample volume is 0.5mm diameter
and the flow rate is approximately 5m/sec , the required amplifier
rise time is lOu sec or an R^, (L, of 411 sec. Thus a maximum value of
R^ would be 8 megohms. For computational purposes the bandwidth from
10KHz to 10MHz were used. The results are shown in Figure 9. The
total noise curves represent the predicted minimum detectable signal
that can be realized. There are two total noise curves which differ
only in the total input capacitance. For most of cur instrument designs
the required bandwidth is in the neighborhood of 25KHz. It is quite
clear from the above results shown that the detector electronics would
be shot noise limited in the bandwidth range of interest. This may be
a curious result to some who are accustomed to the notion that only a
photomultiplier tube would be shot noise limited in this low current
measurement range. This would in fact be true if the background light
levels could be arbitrarily suppressed below the 100 nA baseline.
However, since at 100 nA background we are governed by shot noise it
should be clear that at even higher signal photocurrents,the overall
signal-to-noise ratio,will be entirely governed by the quantum effic-
iency of the detector. Since most silicon photodiodes have quantum
efficiencies of 50% versus 5% for "the multialkalic red sensitive
photonw.ltipliers it's advantages are obvious. Figure 10 shows the
resulting signal—tc—noise ratios for the silicon photodiode and the
photomultipliers of 0.2% and 0.5% quantum efficiency. Note the advan-
tage of the photomulitplier exists only at low signal levels where
background light is infinitesimally small.
Finally it must be stated that the above design is much more
sophisticated than that required by a photomultiplier. The advantages
of the photomultiplier lie in it's low noise amplifier qualities. The
resultant anode current is at a high enough level where induced noise
pick up is comparatively insignificant. The background light level
can also be reduced to less than the 100 nA baseline by increased
optical magnification. Under such conditions the photomultiplier has
increased advantage. We use the photomultiplier for sizing below
0.1 u where further magnification is desirable for reducing sample
volume size.
586
-------
1000
100
S/N
10
,>•,•" /
***-' /.
0001
0.01
O.I
10
10'
I03
COLLECTED SCATTERING SIGNAL (nW)
Figure 10. The signal-to-noise ratios presented, clearly
demonstrate the importance of quantum efficiency when stray
light is non-zero.
5. PERFORMANCE AND CALIBRATION CONSIDERATIONS
The exact calculation of the scattering response for an active
laser cavity is closely analgous to electromagnetic scattering in a
standing wave. Recent solutions to this problem will be illustrated.
The solutions are not strictly correct because a particle traversing
through the beam inside the cavity modifies the light field reducing
the cavity HQM and the laser gain. Schuster and Knollenberg^found
effective extinction cross sections that were roughly a factor of 100
times the classical cross sections. More recent measurements in our
laboratory of the relative extinction and scattering efficiency
factors are shown in Figure 11. Here one sees that in the absence of
absorption,that QscafQext (Q = efficiency factor and Q.'flr = cross
section) but in fact differs by a factor of 10,000 in current instru-
ment high "QM cavities. The shape of the curves is not easily
687
-------
X
UJ
O
o
t/5
O
.2 .4 .6 .8 |.o 1.2 1.4 1.6 1.8 2.0
LATEX SPHERE DIAMETER (MICRONS)
Figure 11. The above measurements are indicative of the role
the cavity"Q" plays in the extinction process. The scattering
efficiency factors are reasonably close to computed values.
explained but the decrease for larger size probably results from
localized laser gain reduction in the near vicinity of the particle,
(large particles can effectively burn holes in the laser beam). What
is observed in the extinction process is not true extinction at all but
largely cavity de-tuning and cannot be truly separated from it. The
scattering phenomenon is however generally explained by scattering
within a standing wave. It is only necessary to measure the light beam
illumination and normalize the scattering signal by the amount of cavity
detuning produced by extinction.
The theoretical calibration curves calculated for scattering in
a standing wave are shown in Figure 12 and 15 for the two collecting
geometries involving collecting angles of 4 - 22 and 5 - 90 • Eight
response curves are shown for real refractive indices of m^=1.3» m-1.4,
m=1.5, m=1.6, m=1.7, nu=1.8, m=1.9, m^2.0. The response function is an
effective cross section for the particular collecting geometry. To
calculate collected power,one simply multiplies the illumination level
(W/cm ) by the effective cross section.
688
-------
LASER CAVITY SCATTERING EFFICIENCY
ID
'7
E
o
CM
o*
10-*
10'"
, 10
-"
10
-u
10
-u
.010
.10
1.0
10
RADIUS
Figure 12. Theoretical instrument response for the 4°-22°
collecting geometry for m = 1.3, m= 1.4, m = 1.5, m= 196,
m= 1.7, m= 1.8, m = 1.9, m= 2,0
These results in general show finer structured resonance be-
havior than the computed response from superimposed HIE solutions for
the forward and backward scattered components. They also show con-
siderably less sensitivity to refractive index. The ASAS has less
sensitivity to refractive index than most white light counters (see
Cooke and Kerker? to compare). Outside of the reduced refractive
index sensitivity in the 1 - 2u diameter range for the 5°- 90
collecting geometry the primary advantage of the hemispherical collect-
ing geometry over the 4-22 collecting geometry is the significantly
larger response at the small particle sizes.
Figure 14 shows the calculated sensitivity of the instrument to
absorption for m= 1.5 and imaginary components of i = 0.5, 0.1, 0.08,
0.06, 0.04, 0.02 and 0.00. In general the greater the absorption the
smaller the resonance. Experimental measurements have verified the
predicted insensitivity of the ASAS to refractive index and to the
accuracy we can measure result in monotonic calibration curves. The
ASAS has less predicted sensitivity to complex refractive index than
any white light instrument in the submicron range(compare with results
of Cooke and Kerker '). Additional measurements on nonspherical
689
-------
particles show the hemispherical collecting geometry to be less senr.i-
tive to shape factor as one might predict.
Some general remarks on the associated resonance phenomenon
associated with refractive index sensitivity are perhaps worthwhile.
There are factors which also reduce the observed resonance. The
cavity gain dynamics have already "been mentioned. In other measure-
ments of classical scattering we have found ourselves only able
to reproduce the full theoretical resonance response if the beam was
truly uniphasal which generally necessitated spatial filtering. The
lack of spatial filtering or the presence of multimodes always sup-
presses the predicted resonances. These phenomenon may all be attri-
buted to the lack of phase front coherence implicitly assumed in such
calculations. In most cases the arguments would extend to scattering
within a standing wave.
LASER CAVITY SCATTERING EFFICIENCY
10
10
.10 1.0
RADIUS (/am)
10
Theoretical instrument response for 5-90
Figure 13.
collecting geometry for m = 1.3, m = 1.4> m = 1.5» m = 1.6,
m = 1.7, m = 1.8, m = 1.9, m = P.O.
690
-------
LASER CAVITY SCATTERING EFFICIENCY
10
-'
o
e
S 1(j-»
o*
. 10~n
x
O
10
-u
.010
,10 i.O
RADIUS 0*m)
Figure 14. Theoretical instrument response for absorbing
particles for 4°~22 for ns=l»';>, and imaginary components
of i = 0.5S 0.1, 0.08, 0.06, 0,04. Q,,0;> and 0,00.
6. GENERAL INSTKUMEM'
Our AS A3 instruments are desJgaed to siae particle;" into fifteen
linear size intervals, with a range- change switch to provide up to
four independent size ranges within a single system. Other features
include trialkalia photonralitplier detectors for ultimate sensitivity
and "free flow aspiration" for absolute "in
sampling. It is
it's optics
packaged in two separate enclosures hone^ug the probe
and pre-amps and an electronics console (see Figure 15) «
The laser used is a hybrid He-Ne (6>3^8A ) manufactured by
Coherent Radiation of Palo Alto, California, The laser has optimized
parameter selection for the AS AS, The op'.
system is mtu:h as des-
cribed in Section % All glasa optical elements are A-R coated for
minimal light loss. Optical transcispion is better than 90% without
the filter.
-------
The pulse height analyzer is capable of sizing pulses as short
as lOOn sec. Because it's reference voltage is derived from the source
of illumination, the entire system has an effective automatic gain
control (AGC). To accomodate the large dynamic range of the instru-
ment,a programable amplifier is used to gain switch and provide several
size ranges. Logarithmic amplification is provided in the tenth micron
size range.
The data acquisition system within the electronics console has
an active memory and is designed to decode, store and display particle
size information. The memory capacity is 16 addresses. Fifteen of the
addresses are used for particle size distribution storage. The remain-
ing address is normally used for other housekeeping information such as
selected size range and elapsed time. Each address has 16 "bits storage
capacity for 0-9999 counts BCD. A selectable digital display and a
graphical CRT display are provided for real-time data monitoring
requirements of precise particle counts or distribution functions.
The mechanical design of the probe section of ASAS instruments
utilizes extruded aluminum sections for mechanical stability of optical
bench quality. Laser and detector alignment is achieved with spring
loaded #10-80 x-y screw adjustments. The probe section is approximate-
ly J>6" long, 8" square, and weighs 25 pounds.
Two types of aspiration are used. The first type used primarily
with the reflecting optical system involves a pair of diametrically
opposed orifices of approximately 2mm inside diameter. A small vacuum
pump buffered by a plenum chamber draws the aerosol sample through the
laser beam. Since the sample volume is centered in the flow and
constitutes a few percent of the flow cross section,wall effects are
considerably reduced and sheath airflow is not required.
The second method of aspiration is unlike any used in convention-
al aerosol counters. As shown in Figure 16 the airflow is essentially
unrestricted. A 10:1 accelerator produces a flow rate of 6.15 meters
per second intersecting with a Jem length of the laser beam. Again
the sample volume is positioned at the center of flow. This system
relies entirely upon optical definition of the sample volume. Problems
associated with small bore tubing are eliminated. This makes the
technique particularly useful for sizing volatile materials such as
natural smog.
692
-------
Reproduced from
best available copy.
ELECTRONICS CONSOLE
PROBE;
Figure 15. Photograph of ASAS rju
shown has free flow aspiration.
i electronic consol
L'robp
693
-------
AIRFLOW DIAGRAM
EXPANSION SECTION
COLLECTING OPTICS
AXIAL FAN
•SAMPLE TUBE
I /
LASER
A
ACCELERATOR
Figure 16. The free flow aspiration system depicted above has
a minimum sample tube diameter of 3cm.
694
-------
7. CONCLUSIONS
Active Scattering Aerosol Spectrometers are fully capable of
particle size measurements down to 0.05 microns. As such they provide
an important capability for studies in the submicron range. -4ie
extensive use of solid state detectors in this kind of system is a
result of the high light levels developed and the design of preampli-
fiers pressing theoretical noise limits. However, photomultipliers
are still required for maximum sensitivity.
Probably the single most important aspect of the ASAS as designed
is the optical system and electronics which together allow for sample
volume control without plumbing. Of further significance is the fact
that the background light level can be reduced at will by increasing
optical magnification. The ultimate sensitivity of photomultiplier
tubes can be realized.
It is our opinion that the refractive index sensitivity of the
ASAS is of the same order of magnitude as the sensitivity to non-
spherical shape and need not be pursued further at this time. Of
considerably more importance is the possibility of utilizing both
extinction and scattering measurements to study the absorption pro-
perties of particles.
Some other aspects of the scattering phenomenon are also worthy
of further investigation. One that was observed early in our work
was the Doppler modulated pulses produced by particle trajectories
with axial components. For the collecting angles used in the ASAS,
the Doppler modulation increases with decreasing size since the back-
scattered radiation is in greater proportion to the forward scattered
radiation. Thus the modulation depth is itself a size sensitive
parameter. Also the frequency is a measure of particle velocity.
Other optical geometries are also useful in certain applications,
Certainly collecting 2l7Nsteradians provides for a system more insensi-
tive to particle shape. Ratio detection at certain paired angles can
also be useful. However, for sizing speed sensitivity and size
resolution the ASAS as currently designed would appear to be as
satisfactory as most uses for such information demand.
695
-------
ACKNOWLEDGMENTS
I would like to thank Jan Emming of Ball Brothers Research
Corporation for the use of his amplifier circuit model. Also the
solution for the electro magnetic scattering in a standing wave
which was supplied by Ron Pinnick of the National Center for
Atmospheric Research.
REFERENCES
1. S. A. Schleusener, J. Air Poll. Control Assoc. 19, 40 (1969),
2. T. D. Proctor, J. Sci. Instrum. 1, 631 (1968).
3. B. G. Schuster and R. G. Knollenberg, Detection and Sizing of
Small Particles in an Open Cavity Gas Laser, Applied Optics,
Vol. 11, page 1515, July 1972.
4. R« G. Knollenberg, An Active Scattering Aerosol Spectrometer,
Atmospheric Technology, No. 2, June 1975.
5. R. Schehl, S. Ergun and A. Headrick, Size Spectrometry of
Aerosols Using Light Scattering from the Cavity of a Gas Laser,
Bev. Sci. Instrum., Vol. 44, No. 9, page 1193, September 1973.
6. H. C. Van De Hulst, Light Scattering by Small Particles (Wiley,
New York, 1957).
7. D. D. Cooke and M. Kerker, Response Calculations for Light-
Scattering Aerosol Particle Counters, Applied Optics, Vol. 14,
No. 3, page 734, March 19'5.
696
-------
Single Particle Optical Counter: Principle and Application
Klaus Willeke and Benjamin Y. H. Liu
Particle Technology Laboratory
Mechanical Engineering Department
University of Minnesota, Minneapolis, Minnesota
ABSTRACT
Single particle optical counters (OPC) are used for in-situ measurements
of aerosol concentrations and size distributions. This paper examines
the optical configurations of several commercial OPC's and then the basic
considerations for the theoretical calculation and experimental calibra-
tion of a counter's response to a particle of given size, shape and re-
fractive index. Furthermore, detailed analyses are made of the influence
on the counter's response by coincidence, resolution, statistical count
accuracy, pulse processing and electrical noise.
697
-------
Single Particle Optical Counter: Principle and Application
Klaus Willeke and Benjamin Y. H. Liu
Particle Technology Laboratory
Mechanical Engineering Department
University of Minnesota, Minneapolis, Minnesota
INTRODUCTION
Single particle optical counters (OPC) have found widespread use
in clean room monitoring, pollution research and laboratory aerosol stud-
ies because of their ability to make in-situ measurements of the con-
centrations and size distributions of particles suspended in the air. A
beam of light inside the instrument is focused onto a "view volume"
through which the airborne particles pass one at a time. The amount of
light scattered or absorbed from each individual particle is measured by
a photosensitive detector. The signal amplitudes registered by this de-
tector are then stored in the channels of a Multi-Channel Analyzer (MCA)
from which the particle size distribution is determined. The range of
particle sizes extends from about 0.5 ym to about 10 ym in diameter for
most commercial OPC's.
Aerosol photometers, not discussed here, measure light scattering
or extinction from a cloud of particulates, and do not resolve the indivi-
dual particle sizes. These measurements will provide accurate concentra-
tion measurements provided the size distributions remain the same.
OPERATING PRINCIPLE
The light extinction system, shown in Figure la, will be used to ex-
emplify the operating principle of the single particle optical counter.
Light from a lamp, usually an incandescent lamp, sometimes a laser, is
condensed onto an aperature with the filament of the light source focused
698
-------
in the plane of that aperture. The emanating light cone is condensed into
the view volume where a sharp image of the defining aperture is formed.
The aerosol flow is ducted to the light beam in this plane of focus. A
view volume may then be defined as the region bounded by the cross-
sectional area of the aerosol stream and the height of the aperture image.
The light emanating from the view volume is collimated, condensed and
focused onto a photosensitive detector, such as a photomultiplier or a
photodiode.
In a single particle optical counter with an incandescent light
source, a light extinction system, such as shown in Figure la, is only
practical for particles larger than several micrometers in diameter.
Since most particles in an ambient or industrial environment are smaller
in size, this light extinction technique is rarely used for aerosols, but
is frequently used for the characterization of particles in a liquid.
Most commercially available single particle optical counters measure
the light scattered out of the incident beam rather than the light inten-
sity reduction of that beam. The optical system of the common scattering
instruments is schematically represented in Figures Ib and Ic. The illum-
ination is, in general, as described in Figure la, the details of light
collection and detection, however, may differ in how the scattered light
is focused onto the photodiode or photomultiplier. Each commercial system
has its own arrangement of illumination and acceptance angles. Specifica-
tion of the following four angles (graphically represented in Figures Ib
and Ic), fully describes the light interaction in the view volume: y is
the half angle of the illuminating cone, a, the light trap half angle, p,
the collecting aperture half angle, and ip, the inclination between illum-
inating and collecting cone axis.
The optical system in an OPC is sometimes characterized by the scat-
tering angle limits. Referring to Figure Ic, scattering of light from il-
luminating angle y to collecting angle B gives an upper limit of y + B;
scattering from y to light trap angle a gives a lower limit a - y. For
instance, an instrument with y = 15° and viewing angles ct = 35° to B = 90°
has a scattering range of 20° to 105°.
When particles are illuminated by a monochromatic light source,
such as a laser, the Mie scattering curve of scattered light flux shows
periodic amplitude oscillations^">*-*• with respect to particle size para-
meter a = TrDp/A, where Dp is the diameter of the particle (assumed spher-
ical) and A is the wavelength of illumination. The counter response may,
therefore, not be a single valued function of particle size. Incandescent
light usually washes out these oscillations and is, therefore, used in
most commercial OPC's. When a powerful light source is used, e.g. a high
intensity laser light, volatile particles may partially evaporate during
the time they remain in the view volume.
699
-------
View Photo
Lamp Aperature Sample Volume Diode
a) Light extinction system
Light
Traps
c) Co-axial scattering system
b) Off-axis scattering system
i Collecting
Aperture
Light Trap
Figure 1 Geometries of illumination and collection in conventional
single particle optical counters.
COMMERCIAL INSTRUMENTS
The significant characteristics of several commercial OPC's are
listed in Table I. All of these instruments have incandescent light
illumination.
The Royco 220 (Royco Instruments, 41 Jefferson Drive, Menlo Park,
Calif. 94025) is a right angle scattering instrument with the axis of
700
-------
cn
)H
cu
4-J
e
3
O
CJ
H
CO
U
•H
4-1
P.
O
CU
•H
CJ
»_J
r*l
4-J
cO
fx.
cu
H
60
•S
C/5
rH
CO
•rH
0
CU
3
3
o
QJ
*-M
o
^s
CD
CO
U
•H
4-J
cn
•H
j_j
CU
4J
O
CO
VJ
cfl
XI
.,
M
0)
rH
Cfl
H
9"
MH
co a
o* ^x
o
0
•
o
rH
1
X""N
d
•H
6
CO
co 6
O1 cj
s_ x
o
r**.
H
/— \|
co in
> s§ °
''-'i
x— s
x-s CU
01 H
8 60
01
rrj
s— x
ro
CO
x-s
a)
X-N CU
U V*
>- 6C
*rj
s«x
CO
rH
rH
1
O
,
60
O
H
O
CJ
W
r\
O
0)
H
>-,
rH
C
O
tn
0)
•rl
H
CU
T3
•H
^
60
cn
cO
cu
CO
—J
}_|
O
ft
Q)
CO.
B
•H
^
O
t-i
P!
P<
CO
CU
}H
CO
C
cu
•H
60
CO
0)
3
rH
CO
^— s
CO
CO
•H
CO
0)
C
O
0
60
•H
4_)
O
cu
rH
rH
O
o
rrj
c
Cfl
60
C
-H
4J
Cfl
C
•iH
^
rj
rH
rH
•H
c
CU
CU
4-1
01
r^
G
O
•H
4-1
cO
C
•H
rH
a
c
•H
II
-3-
^—x
X>
0)
rH
60
C
CO
UH
rH
r^
01
C
O
o
bo
r*
•i-f
4J
Cfl
C
•H
e
3
rH
rH
•H
II
)>-
X-N
CJ
CJ
rH
60
C
cfl
UH
rH
CO
CO
rH 0)
60 H
C 3
Cfl 4-1
CO
UH M
rH CU
co a.
x: co
a, 60
cfl C
J-t -H
4-J 4J
O
4-J CU
XI rH
60 rH
•H O
rH O
II !!
-------
the collecting aperture inclined at angle \|> = 90° to the axis of the il-
luminating cone (Fig. Ib). All other instruments, listed in Table I and
schematically shown in Figure Ic, have a common illumination and collect-
ing aperture axis, i.e. i|> = 0°, and collect the light scattered into a
hollow cone.
The Royco 218 and 245, TechEcology 200 and 208 (TechEcology, Inc.,
645 N. Mary Ave., Sunnyvale, Calif. 94086) and Climet CI-250 (Climet
Instruments Co., 1620 W. Colton Ave., Redlands, Calif. 92373) collect
light from the forward scattering lobe and may be termed forward or near
forward scattering instruments. Their half angle of illumination is
about 5° to 12°,their light trap half angle ranges from about 8° to 18°,
and their collecting aperture half angle ranges from about 20° to 30°.
The scattered light in these instruments is collimated, condensed and
focused onto a photosensitive detector by lenses with a central dark
stop. The light collection system is therefore much like a dark field
microscope in which essentially all of the illuminating beam is rejected
after passage through the view volume and the scattered light is focused
onto a detector. The Climet CI-201 and the Bausch & Lomb 40-1A (820
Linden Ave., Rochester, N.Y. 14625) have viewing angles which extend
from about 30° to as high as 90°. In the Bausch & Lomb instrument the
view volume is located at the focal point of a parabolic mirror with a
light trap at its center. The mirror reflects and collimates the scat-
tered light onto a lens which condenses and focuses the light onto a
photomultiplier . In the Climet CI-201 instrument, the use of lenses for
light collection is avoided by locating the view volume at one of the two
focal points of an elliptical mirror which reflects the scattered light
to a photomultiplier at its other focal point.
3
The view volume, v, varies by about one decade from 0.25 to 4 mm
and the sampling rate Qs varies by over two decades from 170 cm3 min"-'-
(0.006 cfm) to 28,300 cm3 min"1 (1 cfm) for the tabulated counters.
Battery operated portable instruments are generally limited to pumps
handling low flow rates.
A counter which utilizes the light extinction techniques is manu-
factured by HIAC (4719 West Brook Street, Montclair, Calif. 91763). The
Company claims that the instrument sizes airborne particles from 150 ym
to 5 ym, and from 60 ym to 2 ym, depending on the sensor used. In a
counter12»13 developed by Particle Measuring Systems (1855 So. 57th Court,
Boulder, Colo. 80301) the particles are injected into the radiation field
of an open cavity He - Ne gas laser. The light extinction by the par-
ticles perturbs the resonance in the cavity with a resulting power loss
of the laser, i.e., the laser behaves as a nonlinear amplifier of per-
turbations in the resonant cavity. It is claimed that this counter,
designated an Active Scattering Aerosol Spectrometer, sizes particles
702
-------
from 5 urn to 0.05 pm. Its cost, is on the order of $10,000. Convention-
al OPC's sell at $2,000 to $6,000. Peripheral equipment may increase
this cost.
THEORETICAL RESPONSE CALCULATIONS
The response of single particle optical counters, which typically
classify particle sizes from 0.5 ym to 10 ym in diameter, can be cal-
culated through the use of the Mie - scattering theorylO-H > 1^-27 . This
assumes that the particles are spherical and are illuminated by light of
unit flux per unit beam cross-sectional area, and calculates the scatter-
ing intensity as a function of scattering angle 6 (measured with respect
to the incident ray) , particle refractive index m and particle size para-
meter a = irDp/A. The results of Mie theory calculations are usually pre-
sented as angular intensity functions i^ (ct,m,6) and i2 (o.,m,9), the com-
ponents of scattered light polarized in and normal to the plane containing
the directions of illumination and observation. The quantity (A2/8 if2")*
(i^+i2) is then the flux per unit solid angle scattered in the direction
0 by a particle of given size and refractive index. The intensity func-
tions are represented by infinite series whose terms are functions (if
spherical Bessel and Legendre polynomials. The scattered angle 8 Is
calculated with respect to the incident ray which is inclined at angle $
to the axis of illumination. The response calculation for aach particular
optical design, therefore, includes a geometrical factor F(0,4>). An ad-
ditional factor f(A) accounts for the wavelength distribution of the
emissive power of the light source and the spectral sensitivity of the
photosensitive detector. The relative counter response R for spherical
particles, in the absence of coincidence and cross sensitivitv, is then
R(m,D) = I ! I (A2/8TT2)(ii + 12) F(8,<|>) f(X) dAd0d«£ (1)
p
I (A2/8TT2)(ii + 12) F(8,<|>) f(X)
14
As an example, the relative, theoretical amount of scatter of one par
ticular commercial counter is shown as a function of particle size in
Figure 2. Calculations for other commercial counters are given else-
where-^ »H» 1^-19. An opc ampiifies the photoelectric signal. If the
amplification is linear, the counter response will closely follow the
functional relationship given by these calculations.
703
-------
10
to
c
o
CL
(A
0)
cc
o
O
O.I
1.70
1.54
O.I
10 20
Particle Diameter,
Figure 2 Theoretical response of the Bausch and Lomb 40-1A particle
counter (adapted from D. D. Cooke and M. Kerker, 1975).
The counter response may also be represented as light flux received
per unit particle projected area for unit luminance in the source (e.g.
Ref. 16). The resulting curve shows the flux increasing by about 4 de-
cades per decade increase of a when a « 1 (Rayleigh or dipole scattering)
The curve reaches a peak at about a = 2 to 5, oscillates (Mie scattering)
and settles to an approximately constant flux value with small or no
oscillations when a > 50, i.e. the flux per unit particle-projected area
is approximately constant for large particles (geometric scattering).
The rapid decrease of flux for a < 2 imposes a lower size limit on con-
ventional OPC's, ca 0.3 um to 0.5 ym for white light illumination.
Stray light from optical elements such as lenses and light traps,
and Rayleigh scattering of the air molecules results in background scat-
tering intensities which should be lower than the scattered light obtain-
ed by passage of the smallest particle through the view volume. In a
well-designed optical system, stray light scattering will be negligible
with respect to Rayleigh scattering of the air molecules. The Rayleigh
scattering intensity is given by the product of the mean illumination
704
-------
intensity, the number density of the air molecules, the scattering cross-
section of a single air molecule, and the volume of scattering air in
the viewing field.H The latter is generally larger than the view vol-
ume, i.e. sampled particles can scatter only within the view volume, but
air molecules can scatter into the receiving aperture from the entire
viewing field. OPC's designed to detect small particles should, there-
fore, have a very small illumination angle which can best be achieved
through laser light illumination^. However, such illumination may
result in the aforementioned non-single-valued response of the counter.
Lowering the air pressure reduces the number concentration of the air
molecules, and therefore, the Rayleigh scattering intensities from the
air, but may evaporate liquid droplets or liquid coatings on solid par-
ticles, thus giving a particle size distribution which is different from
that of the ambient air. In unusual circumstances,where the particles
may be suspended in a gas such as Helium, the particle size limit can be
lowered, because Helium atoms have a lower scattering cross-section than
air molecules.-*-^ The lower particle size limit for special optical
counters is of the order of 0.1 ym in diameter.
INSTRUMENT CALIBRATION AND MODIFICATION
Relative, theoretical calculations may be used to determine how
monotonic the response of a counter will be with respect to particle
diameter, and how much the response will depend on particle refractive
index within the size range of interest. However, an experimental cali-
bration-'-~°>2°~32 ±s always necessary to establish the actual voltage
output of the counter as a function of particle size for a given aerosol.
As a first step, one needs to generate aerosols with a high degree
of monodispersity. The least expensive technique is one in which poly-
styrene latex (PSL) or polyvinyl toluene latex suspensions (Dow Chemical
Company, Midland, Michigan) are nebulized , dried, and sampled into the
OPC. They are available in sizes from about 0.1 ym to about 3 ym, and
have a refractive index of 1.6. They may be considered ideal because of
their spherical, smooth shape, i.e. they fit the assumptions for the
theoretical response calculations outlined above.
An alternate and more elaborate technique utilizes either the spin-
ning disc (Sierra Instruments, P.O. Bx. 909, Village Sq.; Carmel Valley, Ca.
93924) or the vibrating orifice monodisperse aerosol generator (Thermo-
Systems Inc., 2500 Cleveland Avenue, St. Paul, Minnesota 55113). The
705
-------
latter-^ allows the calculation of particle diameter to an accuracy of
about 2 percent using the experimental values of signal frequency, liquid
flow rate and concentration of aerosol material in solution with a vola-
tile substance. Aerosols generated with these instruments approximately
cover the entire size range of conventional OPC's. The effect of par-
ticle refractive index may be studied by generating ideal particles of
di-octyl phthalate (OOP, m = 1.49) and Cargille index-of-refraction
liquids with m = 1.4, 1.5, and 1.6 (R. P. Cargille Laboratories Inc.,
Cedar Grove, New Jersey). The Cargille 1.7 refractive index liquid was
found^ to be unstable and sufficiently volatile at room temperature to
cause significant uncertainties in the calculated particle diameter. An
example^ of an experimental calibration with DOP aerosols is shown in
Figure 3. The instrument's theoretical response-^ is shown in Figure 2.
Experimental calibration curves for other commercial optical particle
counters are shown in Figures 4 through 6, and will be discussed in the
next chapter. The effect of particle shape and light absorption may
be studied by generating particles such as methylene blue, sodium chlo-
ride and india ink.
10
0)
o>
a
I
£• o.
3
O
0.01
BAUSCH a LOMB
PC 40-1
O.I
10
50
Particle Diameter.
Figure 3 Experimental calibration curve for the Bausch & Lomb 40-1A
particle counter.
706
-------
o
CE
LU
I-
8
0.3
0.2
0.1
0.05
0.03
0.02
] I I T I j I I I | I I I I |
ORIFICE DIAMETER OOP AEROSOL
5 MICRONS
10 MICRONS
20 MICRONS
i I i i i i
ROYCO PC 215
i I i
0.3 0.5 I 235
PARTICLE DIAMETER,
10
Figure 4 Vibrating orifice generated calibration curve for the Royco
215 particle counter. The optical system of the Royco PC
218 is similar to the older Royco PC 215.
707
-------
Vl
o
TT
I I I
m
Ci
m
in
CJ
UJ
h-
UJ
o
UJ
o
cr
2
0) VI
> a)
Vi 4J
3 d
o 3
o
c o
o
•H >
•H O
Vi p<5
0)
a cu
X X
3
60
•H
1*4
in
- 8
snoA 'indino
§
o E
UJ
<
o
UJ
_l
o
0) Vi
> J-»
•H Vl
iH cd
cfl a,
u
o
1-1 r-J
Cfl CN|
4J
c o
OJ o
e s*.
•H O
Vi PS
0)
a, QJ
x x:
Pd 4-1
0)
Vl
60
708
-------
The refractive index and shape of airborne particle found in envi-
ronments such as mines and metallurgical processing may be quite diffef"-
ent from these artificial aerosols. The main difficulty in calibrating
OPC's for such dusts is the preparation of monodisperse calibrating aero-
sols. Conventional aerodynamic dispersion of powders such as coal does
not give sufficiently monodisperse and de-agglomerated aerosols. It has
been found that a fluidized bed3>35,36 serves well in the de-agglomera-
tion of powders. The airborne particle may then be classified to a known
size by means of a differential mobility analyzer. However, classifi-
cation by electrical mobility is limited to submicron particles with the
upper limit at about 2 to 3 um. Unclassified larger particles may be
sized by optical microscopy after passage through the OPC. However, the
counter's response is spread over a range of channels for a particle of
a given size so that the original size distribution of the dust must be
deduced by numerical inversion.3 This broadening effect limits the re-
solving power of the instrument as discussed in a later chapter.
Special attention should be paid to the uniform illumination of
the view volume. The light source of which an image is formed in the
view volume, should produce the most homogeneous radiation density pos-
sible. For instance, band-like tungsten filaments were found to be
superior over spiral tungsten filaments^'.
A sheath-air inlet further improves the quality of the view volume.
In a typical sheath-air inlet , shown in Figure 7, about 90 percent of
the incoming aerosol flow is diverted through an external filter which
separates the particles from the air stream. The filtered air is then
reintroduced around the remaining 10 percent of the original aerosol
flow which passes unimpeded through a straight, inner tube. The combin-
ed stream is then introduced into the optical view volume where the par-
ticles are counted. The flow is laminar and the aerosol stream flows as
a central core within the clean-air sheath into the view volume.
A sheath air inlet increases the resolution of the counter by fo-
cusing the aerosol sample stream into a more uniformly illuminated por-
tion of the view volume, and permits a higher aerosol concentration to
be counted because of the decrease in size of that view volume. This
also decreases the response time of the counter arid prevents recircula-
tion of aerosols into the view volume. The signal-to-noise ratio may be
improved if the aerosol sample stream is focused into a portion of the
view volume which has a higher mean intensity than the original view
volume.
The measured aerosol concentration will be lower than the true aero-
sol concentration, if particles are lost in the OPC inlet. The ratio of
709
-------
CLEAN AIR AEROSOL
FROM FILTER
1.0"
1
1
. 1
'•° T*- » 1
I '7f 56?5" 437'5. -L;
i | 1 noV 25 ,
I
,\x
\^
v\
\V i
s\XS
v\\
NV
.72" DIA.
1.375" DIA.
-.091 DIA.,6 HOLES
EQUALLY SPACED, l" DIA.
.4375" DIA.
FILTER
-.a5"o.D., .l875"l.D.
TJ
094"
Figure 7 Sheath-air inlet for the Royco 220 particle counter.
these two concentrations defines the sampling efficiency. If particles
of sizes larger than about 5 urn in diameter are to be sampled, the in-
let tube should preferably be short without any bends. An example is
given in Figure 8 which shows a conical inlet design^ for the Royco 245
high-flow-rate counter. With this inlet the original efficiency for
sampling from a calm air environment was improved to about 80% for 10
ym particles and about 43% for 30 ym particles as shown in Figure 9.
One may conclude from Figure 9, that measured concentrations should
always be modified by the sampling efficiency when particles larger than
5 to 10 ym in diameter are sampled.
710
-------
3 i" HOLES.
EQUALLY
SPACED
O3f
TWO INSERT PIECES
TO FORM
RECTANGULAR
CHANNEL
Figure 8 Conical inlet for the Royco 245 particle counter.
PARTICLE REFRACTIVE INDEX AND SHAPE
The chemical composition of natural aerosols may vary from particle
to particle. The particle index of refraction, m, is therefore unknown.
Many ambient aerosols contain a substantial quantity of water because of
their hygroscopicity. The refractive index may therefore range from m =
1.33 (water) to, say, about m = 1.9. For practical purposes, OPC's are
usually calibrated with ideal particles of refractive index 1.5 to 1.6.
The size measured by an OPC is then^ an "Equivalent latex sphere dia-
meter" or an "Equivalent DOP sphere diameter" depending on the calibra-
ting aerosol used. The theoretical response curve for the example
given in Figure 2 (Bausch & Lomb 40-1A Counter) shows that the counter
response is a strong function of refractive index, particularly for
711
-------
10
o
UJ
o
0-5
2
_J
ROYCO PC 245
10 15 20 25
PARTICLE DIAMETER, /im
30
35
40
Figure 9 Sampling efficiency as a function of particle diameter for
the modified conical inlet (Figure 8) of the Royco 245 high-
flow-rate particle counter.
Dp < 2 ym. This dependence varies considerably from one type of counter
to another because of the difference in optical configuration.
An experimental calibration3 of the Bausch & Lomb 40-1A counter for
DOP particles of refractive index m = 1.49 is shown in Figure 3. The
counter output voltage is seen to monotonically increase with particle
size. When calibrated with DOP particles, a similar monotonic increase
was found for the Royco 220 particle counter (Figure 5),whereas the cali-
bration curves for the Royco 215 (Figure 4) and the Royco 245 (Figure 6)
show a dip for particle sizes of about 1 ym in diameter. No useful
size distribution information can be obtained in the size range where
counters display such a dip in the calibration curve. A counter operat-
ing with monochromatic laser light may have a multivalued curve over a
large range of particle sizes.
Agreement between theoretical prediction and experimental calibra-
tion is generally very good^. After an experimental calibration curve
712
-------
has been established for a given refractive index, the theoretical re.s-
ponse for this refractive index and for other refractive indices may be
drawn onto the experimental plot. However, the reference size through
which the theoretical response curve is dimensionalized to counter out-
put voltage, should preferably be from a monotonic portion of the res-
ponse curve where the response is least affected by variations in re-
fractive index, i.e. Dp > 2 \im for the example in Figure 2.
Special aerosols may have refractive indices which greatly differ
from those generally found in the ambient air. Figure 2 shows the rela-
tive theoretical counter response to a particle of high refractive
index, 2.5, and one to an absorbing particle with m = 1.54 - 0.5i. A
typical light absorber is coal dust for which an experimental calibra-
tion curve-' is shown in Figure 3. As seen, the output signal amplitude
from the OPC ranges from a factor of 2 to 25 lower for coal than for
ideal transparent DOP particles of the same size. One may define a
size shift factor^ as the ratio of the coal particle size to that of a
DOP particle which generates the same output voltage from the OPC. This
size shift increases monotonically from about 3 for small particles to
about 5 for large particles in Figure 3. Submicron particles may have
unity size shift; e.g., the curves for m = 1.54 and for m = 1.54 - 0.5i
merge at a particle diameter of about 0.4 ym in the example of Figure 2.
The coal dust for the calibration curve (Fig. 3) was produced in
a ball mill so that the particles were nodular in shape. However, coal
dust particles produced by a cutting tool were found to have very ir-
regular shapes^Sj such as spears or platelets. This difference in shape
does not necessarily change the calibration curve but will decrease the
resolution of the counter because the scattered light received at a fix-
ed angle from the optical axis will depend on the orientation of the
particle in the view volume and will therefore have a larger range of
intensities than one would expect from a spherical particle.
The effect of particle shape is expected to be less for a counter
with a light collection aperture axis coaxial with rather than at some
angle to the illumination axis, because the scattered light is integra-
ted through 360° around the optical axis. A broad range of viewing an-
gles (3 - a ) for a coaxial system will further smooth out geometrical
variations.
713
-------
COINCIDENCE
r • -i 37» 39, 40 c . ,
Coincidence rerers to the simultaneous presence of more
than one particle in the optical view volume. This determines the maxi-
mum particle concentration, which an OPC may measure accurately. In the
case of coincidence, the indicated particle number concentration, NI, is
lower than the true concentration, Nt, and the shape of the size distri-
bution is weighted towards larger particles because the pulse height ana-
lyzer registers two or more particles in the view volume as one particle
of a larger size.
The statistical probability P, that no particle (p = o), one par-
ticle (p = 1) or more than one particle (p = 2, 3, 4, . . .) is present
at the same time in the view volume v is given by the Poisson distribu-
tion
P! (2)
where \i = Ntv
is the average number of detectable particles in v. The probability of
finding no particles in the view volume is then
T) M «».V
PO = e = e t (4)
and the probability of finding one, two or more particles in v, i.e. the
indicated number of particles in v is
(1 - PO) = 1 - e~NtV = N±v (5)
so that the ratio of indicated to true particle concentration becomes
714
-------
N.
i
N_
1
1 - e
-Nv
(6)
This derivation, however, considers only the static situation of find-
ing particles in the view volume^S. One should also consider the dynamic
situation of particles traversing the view volume. A particle will trig-
ger a pulse only, if the view volume is not already occupied by particles
in which case the indicated count is^7» ^9-51
N.
= e
-N,.v
(7)
Equations (6) and (7) are plotted in Figure 10. For small values of
N^v, equation (6) reduces to 1 - Ntv/2, and equation (7) to 1-N^v, i.e.
twice the coincidence of equation (6). Since the view volume can usually
not be defined accurately and the scattered light intensity of the par-
ticle entering the view volume may not immediately exceed the threshold
value, either equation (6) or (7) can be used to predict the limiting num-
ber concentration for which coincidence may become significant. Equation
(7) should be used for a conservative estimate.
1.0
0.9
0.8
N,
0.7
0.6
0.5
0.03 0.05
OJ 0.3
fj. - Ntv or N,Q(r
05
1.0
Figure 10 Ratio of indicated to true particle concentration as a func-
tion of average particle number concentration Ntv in view
volume v, or as a function of average particle count rate
during stretched pulse duration T.
715
-------
When several subcountable^lparticles are present in the view vol-
ume at the same time, the sum of the scattering intensities may result
in a detectable light pulse so that an excess in counts will be register-
ed by the multi-channel analyzer. This problem is usually alleviated by
dilution of the aerosol concentration and by setting the discriminator
voltage of the MCA to a sufficiently high value.
COUNT ACCURACY
The lowest particle concentration which an OPC may accurately meas-
ure is given by statistical considerations. The statistical probability
of correct counting is given by the Poisson distribution. An inherent
property of this distribution is the equality between the variance and
the mean, i.e. the standard deviation is equal to the square root of the
mean which equals the actual count to first approximation. The relative
standard deviation, ar, defined as the ratio of standard deviation, a, to
particle count, n, is then:
~ " (8)
The particle count, is given by
n = NtQst (9)
where Qs is the volumetric sample flow rate and t is the time of sampl-
ing.
Equation (8) is plotted in Figure 11. For instance, a count of 100
particles has a relative standard deviation of 10%, and a count of 1000
particles has a relative standard deviation of about 3%. The counter is
716
-------
i 1—i—r i r i i i i r i n
Particle Count ,n
Figure 11 Particle count accuracy of OPC's.
thus limited by particle coincidence at high number concentrations and
by particle count accuracy at low number concentrations. For normal
sampling times, this limits most commercial counters to a number concen-
tration range of about 3 to 4 decades. The number concentration of atmos-
pheric aerosols is about 3 to 4 decades lower for particles 10 ym in dia-
meter than for 0.5 ym particles, which corresponds to the nominal limits
of conventional OPC's.
For ambient atmospheres, the number concentration for a given size
range may vary within time as much as two decades, when sampled at sur-
face locations, and three decades when vertical concentration profiles
are measured^-^ e.g., by use of an aircraft. One therefore needs to
carefully observe the limits of operation of an OPC when sampling atmos-
pheres with time-varying particle concentrations. A large size range is
best sampled by two OPC's, one for sizes from, say,0.5 ym to 5 ym, and
the other for sizes from^say, 5 ym to 50 ym, with eacVi OPC optimized for
its respective size range. ^
The count accuracy for a given counter may be improved by increas-
ed time of sampling, which results in a loss of time resolution. It may
also be improved by registering a larger size subrange per channel of the
multi-channel analyzer (MCA). This will, however, result in a decreased
size resolution.
717
-------
RESOLUTION
An MCA can, in principle, resolve very small size intervals and
give detailed information on the structure of the size distribution
curve. However, every optical counter has a limit of resolution. The
term resolution^'3,7 refers to the ability of the instrument to distin-
guish two monodisperse aerosols of different mean particle size. This
is also referred to as channel cross-sensitivity.
The counter's response to monodisperse particles depends on the uni-
formity of the light intensity in the illuminated optical view volume,
the uniformity of sensitivity of the photodetector surface, and on the
baseline electronic and optical noise. The MCA therefore responds to
.monodisperse particles with pulses distributed over a range of channels.
o
o
1500
1000
500
MIT I 1 I I
.1. .1.
j. .1 i
20
30 40
CHANNEL
a)
50
4O 60
CHANNEL
b)
Figure 12 Optical particle counter output under good (a) and poor (b)
dispersion conditions.
For example, Figure 12 shows some typical pulse height spectra from the
Bausch & Lomb 40-1 counter when exposed to aerosols from the vibrating-
orifice generator operated under good and poor dispersion conditions.
718
-------
Good dispersion (Fig.l2a) is indicated by a very pronounced peak in the
displayed spectrum and by the near absence of pulses of a larger ampli-
tude above the peak. Under poor dispersion conditions (Fig. 12b) addition-
al peaks appear in the spectrum. These secondary peaks were caused by
droplets formed from the collision and coalescence of the primary droplets
during the dispersion process.
The shape of the counter signal, Fig 12a, is approximately Gaussian
and may be characterized by the overall variance OQ (cro = standard devia-
tion). The variance in signal from non-uniformity in illumination and
detection, a2,, and the variance in baseline electronic and optical noise,
a§, are essentially independent of each other, so that
02 = 02 + CT2 (10)
The relative standard deviation of the output pulse distribution for
monodisperse aerosols is then^
(ID
where Vo is the pulse height at the peak of the distribution for mono-
disperse aerosols of particle diameter Dp.
The electronic noise is approximately constant and is equal to the
counter output when no particles are passing through the optical view
volume. For example, it is approximately equal to 3 mV for the Bausch
and Lomb 40-1 counter . Since the counter voltage output increases with
particle size, approximately by one to two decades per decade of par-
ticle size, either electronic noise or Rayleigh scattering from the air
molecules sets the lower limit of particle size which an OPC can detect,
generally 0.3 ym toO.Sym in diameter. A discriminator is set at a volt-
age level exceeding the baseline noise signals.
The variance in signal o| is found from measurements with large
particles (large V0) for which the ratio (ob/V0) tends toward zero, and
consequently (o0/Vo) -»• (as/V0) . The ratio (as/Vo) approaches 0.08 for
large DOP test particles when measured with the Bausch and Lomb 40-1
counter^. These measurements also show that os increased to a higher
value than predicted by equation (10) when the particle size was
reduced. A probable explanation of this discrepancy is that the smaller
particles experience a greater lack of uniformity in illumination and
detection than the large particles do, and that as is actually a variable
dependent upon particle size, rather than a constant, is as implied in
719
-------
equation (10). Also, small particles may scatter a finite number of pho-
tons, thus increasing the fluctuations in output signfL.
The resolution of an OPC depends therefore on the ability of the
counter to produce uniform pulses upon exposure to monodisperse aerosols.
It also depends on the slope of the calibration curve, a steep one being
the most desirable. Optical counters of the same model and produced by
the same manufacturer may not have the same resolution due to normal
manufacturing tolerances and differences in alignment of the optics.
When calibrating OPC's with non-ideal particles, such as monodis-
perse coal dust, the MCA may show a pulse height distribution with a
spread which is considerably wider than the intrinsic properties of the
instrument would dictate. The difference is attributed to the irregu-
lar shape of the monodisperse particles, and to a lesser extent, to re-
fractive index variations of the particles^. This spreading effect
further decreases the ability of the instrument to resolve small par-
ticle size differences. In general, the smallest variance is obtained .
with ideal aerosols, the worst with irregularly shaped particles, be-
cause the scattering intensity received in the collecting aperture from
irregularly shaped particles, depends upon the orientation of the par-
ticle in the view volume. Absorbing particles may further increase the
variance because the calibration curve of output voltage vs. particle
size, e.g. Figure 3, has, in general, a lower slope than the one for non-
absorbing particles.
PULSE PROCESSING
The photodetector in an optical counter, usually a photo-multiplier,
receives a flux of photons when particles pass through the optical view
volume. The photon flux is converted to a voltage signal and recorded
by a multichannel analyzer which stores the signal in a channel assigned
to a range of voltages and counts the number of pulses received in each
channel. However, the output signal shape from an OPC is generally not
compatible^-*»46 with the input pulse shape requirement of an MCA.
In general, an MCA requires a constant height pulse of about 3 to
5ysec duration. A pulse converter is therefore used as an interface to
convert the OPC signal to a compatible signal for the MCA. Some older
MCA's had a low input resistance, which was incompatible with the output
resistance of the OPC. Most of the newer MCA units, however, have an
720
-------
input impedance in excess of one megohm, which will not distort the out-
put signal from the OPC^6.
The performance of the pul.se converter will be illustrated by re-
ference to a specific device, the University of Minnesota Model 170-1
peak detector, which was developed from the Royco 170-1 pulse converter.
This device first removes low-frequency signals and d-c offset from the
incoming signal. The incoming signal is generally Gaussian in shape and
has a pulse width of a few to several hundred microseconds, depending on
the OPC used. The signal is then linearly or logarithmically amplified.
Logarithmic compression of the signal has the advantage of MCA signal
storage in logarithmic voltage intervals which approximately corresponds
to logarithmic particle size intervals. A baseline voltage offset is
necessary to prevent the converter from being triggered by noise in the
incoming signal.
When the amplified signal reaches its peak, the system stretches
the pulse at that voltage level for a preset period of time at the end of
which a "gating" pulse commands the MCA to read the signal amplitude.
The signal voltage is held as a charge on a capacitor which is quickly
discharged and reset to its baseline value after the MCA has read its
amplitude.
When the pulse stretching technique is used, the pulse has to be
stretched beyond the duration of the signal which introduces a dead time
during which no additional signals are processed. The calculation of
particle count loss is similar to the one for particle coincidence loss in
the view volume. Referring to the Poission distribution, Equation (2),
we have now
y = ntT = NtQsT
(12)
where n^ is the true particle count rate (Equation 9) and T is the time
duration of the stretched pulse, i.e. y is equal to the average number
of counts made during pulse duration T. The expression for the indicated
particle count n^ is then similar to Equation (5)
which is plotted in Figure 10. As seen from Figure 10, a less than 5%
loss in particle count approximately requires that
721
-------
N.Q T < 0.05
L S
Similarly, a less than 5% particle coincidence loss in the view volume
requires
Ntv < 0.05 (15)
The loss of signal count may be reduced by dilution of the aerosol
concentration. However, dilution techniques may incur a loss of large
particles.
An alternate technique of peak detection makes use of pulse differ-
entation. With this technique, a comparator looks at the slope of the
signal and gives a "gating" signal for amplitude reading by the MCA only
when the differentiator output changes from positive to negative at the
peak of the signal. However, fluctuations on the signal due to noise may
produce local maxima so that the MCA may record more than one output
pulse from a single input pulse.
A technique, reported by the University of North Carolina at
Chapel Hill , eliminates the problem of dead time. It is similar to the
stretching technique, but uses the comparator, which recognizes the
peak of the pulse, to give a gating signal to the MCA at the beginning
of the pulse stretching. The capacitor, which stretches the pulse at
peak amplitude, is then discharged when the original signal has disap-
peared, rather than after a preselected time period.
722
-------
CONCLUSIONS
Measurements by single particle optical counters are superior over
manual sampling and optical microscope counting techniques when the par-
ticle size distribution and concentration of air environments is to be
determined, because OPC's measure particles automatically and in situ.
Detailed information may thus be provided on the time variations of
these quantities, which cannot be achieved by microscope techniques with
any degree of statistical counting accuracy without great expenditure in
time. In addition, human error and bias is minimized, and possible
errors resulting from sample collection onto a substrate, and from trans-
fer of the sample to a proper measurement location are excluded.
Most OPC's can accurately count particles over a range of at least
four decades of particle concentration. The limits are set by coinci-
dence resulting from electronic dead time of simultaneous occurance of
more than one particle in the view volume, when sampling high concentra-
tions, and by particle count accuracy, when sampling low concentrations.
However, any range of particle concentration can usually be measured by
proper choice of particle counter(s), air dilution and sampling time.
The lower limit of particle size detected by conventional OPC's
with white light illumination is approximately 0.3 ym in diameter. The
upper limit is set by the sampling efficiency of the counter inlet tube.
When two decades of particle size are to be measured, say from 0.5 urn
to 50 urn, two OPC's should be operated, each one optimized for its res-
pective size range.
The size indicated by the counter is equivalent to the size of the
calibration aerosol, mostly an ideal (spherical and transparent) aerosol,
such as polystyrene latex. The shape and refractive index of the measur-
ed particle may be different from that of the calibration aerosol so that
its actual size may be larger or smaller than the one indicated by the
OPC. However, this problem is not limited to optical counters alone.
Practically all aerosol sizing instruments measure some equivalent sphere
diameter";' The inertial impactor, the aerosol centrifuge, the elutria-
tor and the cyclone measure an equivalent sphere diameter based on Stokes
drag and particle mass - the so-called aerodynamic diameter. The elec-
trical aerosol analyzer measures an equivalent sphere diameter based on
electrical charge and mobility. The diffusion battery measures an
723
-------
equivalent sphere diameter based on the diffusion coefficient of the par-
ticles. This indicates that all particle size data need to be inter-
preted with caution and with due consideration for the method used for
the me as ur emen t.
724
-------
REFERENCES
1. Whitby, K. T., and R. A. Vomela. Response of Single Particle Opti-
cal Counters to Nonideal Particles. Environ. Sci. Technol. 1: 801-
814, October 1967.
2. Liu, B. Y. H., R. N. Berglund, and J. K. Agarwal. Experimental
Studies of Optical Particle Counters. Atmospheric Environment 8:
717-732, 1974.
3. Liu, B. Y. H., V. A. Marple, K. T. Whitby, and N. J. Barsic. Size
Distribution Measurement of Airborne Coal Dust by Optical Particle
Counters. Amer. Ind. Hyg. Assoc. J. 8:443-451, August, 1974.
4. Gucker, F. T., and D. G. Rose. A Photoelectronic Instrument for
Counting and Sizing Aerosol Particles. Brit. J. Applied Physics:
Supplement No. 3, 138-143, 1954.
5. Lieberman, A., and R. J. Allen. Theoretical and Experimental Light
Scattering Data for a Near Forward System. (Presented at the
Amer. Assoc. for Contamination Control), May 19-22, 1969.
6. Martens, A. W., and K. H. Fuss. An Optical Counter for Dust Partic-
les. Staub-Reinhalt. Luft (English Transl.) 28:14-18, June 1968.
7. Whitby, K. T., B. Y. H. Liu, R. B. Husar, and N. J. Barsic. The
Minnesota Aerosol-Analyzing System Used in the Los Angeles Smog Pro-
ject. J. Colloid Interface Sci, 39:136-164, April 1972.
8. Willeke, K., K. T. Whitby, W. E. Clark, and V. A. Marple. Size Dis-
tributions of Denver Aerosols - A Comparison of Two Sites. Atmos-
pheric Environment, 8:609-633, 1974.
9. Willeke, K. and K. T. Whitby. Atmospheric Aerosols: Size Distri-
bution Interpretation. J. Air. Poll Control Assoc., 25:529-534,
May, 1975.
10. Quenzel, H. Influence of Refractive Index on the Accuracy of
Size Determination of Aerosol Particles with Light-Scattering
Aerosol Counters. Applied Optics 8:165-169, January, 1969.
11. Gebhardt, J., J. Bol, W. Heinze, and W. Lebschert. A Particle-
Size Spectrometer for Aerosols Utilizing Low-Angle Scattering
of the Particles in a Laser Beam. Staub-Reinhalt. Luft
(English Translation), 30:5-14, June 1970.
725
-------
12. Schuster, B. G., and R. Knollenberg. Detection and Sizing of
Small Particles in an Open Cavity Gas Laser. Applied Optics, 11:
1515-1520 July, 1972.
13. Knollenberg, R. G. Active Scattering Aerosol Spectrometry. In
Aerosol Measurements, W. A. Cassatt and R. S. Maddock (eds), NBS
Special Publ. No. 412, National Bureau of Standards (U.S.) October
1974, p. 57-64.
14. Cooke, D. D., and M. Kerker. Response Calculations for Light
Scattering Aerosol Particle Counters. Applied Optics 14:734-
739, March, 1975.
15. Martens, A. E0, and D. D. Doonan. Comments on: Influence of
Refractive Index on the Accuracy of Size Determination of Aerosol
Particles with Light-Scattering Aerosol Counters. Applied Optics
9:1930-1931, August, 1970.
16. Hodkinson, J. R., The Optical Measurement of Aerosols. In: Aerosol
Science, Davies, C. N. Academic Press, New York 1966, p.
287-357.
17. Hodkinson, J. R. and J. R. Greenfield. Response Calculations for
Light Scattering Aerosol Counters. Applied Optics 4:1463-1474,
November, 1965.
18. Hodkinson, J. R., and I. Greenleaves. Computations of Light-
Scattering and Extinction by Spheres According to Defraction and
Geometrical Optics, and some Comparisons with the Mie Theory.
J. Optical Society of America 53:577-588, May 1963.
19. Gucker, F. T., and D. G. Ross. The Response Curves of Aerosol Par-
ticle Counters. (Presented at the Third National Air Pollution Sym-
posium, Pasadena, Ca., April 18-20, 1955.)
20. Gucker, F. T. , R. L. Rowell, and G. Chui. A Study of Intensity
Functions for Dielectric Spheres Calculated from the Mie Theory
(Presented at the 1st Nat. Conf. on Aerosols, Liblice,
Czechoslovakia, October 8-13, 1962).
21. Gucker, F. T. and J. Tuma. Influence of the Collecting Lens
Aperture on the Light-Scattering Diagrams from Single Aerosol Par-
ticles. J. Colloid Interface Sci. 27:402-411, July 1968.
22. van der Hulst, H. C. Light Scattering by Small Particles. John
Wiley, New York, 1957.
726
-------
23. Mie, G. Beitraege zur Optik trueber Medien. Ann. Physik 25:377-445,
1908.
24. Kerker, M. The Scattering of Light and Other Electromagnetic
Radiation, Academic Press, New York, 1969.
25. Cassatt, W. A., and R. S. Haddock (eds.) Aerosol Measurements
(includes several articles on optical measurements.), NBS Special
Publ. No. 412, National Bureau of Standards (U.S.), 1974.
26. Pinnick, R. G., and D. J. Hoffmann.Efficiency of Light-Scattering
Aerosol Particle Counters. Applied Optics 12:2593-2597, November
1973.
27. Oeseburg, F. The Influence of the Aperture on the Optical System
of Aerosol Particle Counters on the Response Curve. Aerosol
Science 3:307-311, 1972.
28. Lundgren, D. A. and D. W. Cooper. Effect of Humidity on Light-
Scattering Methods of Measuring Particle Concentrations, J. Air
Poll. Control Assoc. 19:243-247, April 1969.
29. Rimberg D., and J. W. Thomas. Response of an Optical Counter to
Monodisperse Aerosols. Atmospheric Environment 4:681-688, 1970.
30. Boden, A., and W. Coenen. Darstellung von Teilchengroessenvert-
lungen mittels eines einfachen Spektrometers als Zustaz zu
Streulichtmessgeraeten. Staub-Reinhalt. Luft 30:483-487, Novem-
ber 1970.
31. Cooke, D. D., and M. Kerker. Particle Size Distribution of Col-
loidal Suspensions by Light Scattering Based upon Single Particle
Counts - Polystyrene Latex. J. Colloid Interface Sci. 42:150-155,
January, 1973.
32. Bakhanova, R. A., and L. V. Ivanchenko. The Calibration Curve of
Photoelectric Counters and Computation of Particle Size Distribu-
tion when the Relationship between Particle Size and Electrical
Pulse Amplitude is Ambiguous. Aerosol Science 4:485-490, 1973.
33. Whitby, K. T. and B. Y. H. Liu. Polystyrene Aerosols - Electrical
Charge and Residue Size Distribution. Atmospheric Environment 2:
103-116, 1968.
34. Berglund, R. N., and B. Y. H. Liu. Generation of Monodisperse
Aerosol Standards. Environ. Sci. Technol. 7: 147-153, 1972.
727
-------
35. Guichard, J. C. and J. L. Magne. Application du lit fluidise en
milieu gaseau a la generation des aerosols. Inst. Nat. Rech.
Chim. Appl., Vert-le-Petit, France. Note interieure No. 51, 1967.
36. Willeke, K., S. K. Lo, and K. T. Whitby. Dispersion Characteristics
of a Fluidized Bed. Aerosol Science 5:449-455, 1974.
37. Jaenicke, R., The Optical Particle Counter: Cross- Sensitivity
and Coincidence. Aerosol Science 3:95-*lll, 1972.
38. Marple, V. A. Development, Calibration, and Application of Size
Distribution Instruments at the University of Minnesota. In:
Aerosol Measurements, Cassatt, W. A., and R. S. Maddock, (eds.)
NBS, Special Publication No. 412, National Bureau of Standards
(U.S.), 1974, p. 176.
39. Bader, H., H. R. Gordon, and 0. B. Brown. Theory of Coincidence
Counts and Simple Practical Methods of Coincidence Count Cor-
rection for Optical and Resistive Pulse Particle Counters. Rev.
Sci. Instr. 43:1407-1412, 1972.
40. Nikiforova, N. K., and Y. S. Sedunov. An Investigation of Errors
Arising in Photoelectric Counters - Aerosol Particle Analyzer.
Aerosol Science. 3:441-453, 1972.
41. Whitby, K. T., and B. Y. H. Liu. Generation of Countable Pulses
by High Concentrations of Subcountable Sized Particles in the
Sensing Volume of Optical Counters. J. Colloid Interface Sci.
25:537-546, 1967.
42. Junge, C. E.. Air Chemistry and Radioactivity, Academic Press,
New York, 1963.
43. Whitby, K. T., Dept. of Mechanical Eng., Univ. of Minnesota,
private communication, 1975.
44. Whitby, K. T., W. E. Clark, V. A. Marple, G. M. Sverdrup, G. J.
Sem, K. Willeke, B. Y. H. Liu, and D. Y. H. Pui. Characterization
of California Aerosols. I. Size Distributions of Freeway Aerosol.
Atmospheric Environment 9:463-482, May, 1975.
45. Graedel, T. E.. Channel Width Determination and Electronic Pulse
Processing Losses in Optical Particle Counters. Aerosol Science
5:125-131, 1974.
728
-------
46. Memenway, D. R.. Pulse Processing from Particle Detectors. Ph.D.
Thesis, Dept. Env. Sci. and Eng., Univ. of North Carolina at
Chapel Hill, 1974.
47. Liu, B. Y. II., R. N. Berglund, and J. K. Agarwal. Author's Reply.
Atmospheric Environment 8:1344, December 1974.
48. Wales, M., and J. N. Wilson. Theory of Coincidence in Coulter
Particle Counters. RSI 32:1132-1136, 1961.
49. Princen, L. H., and W. F. Kwolek. Coincidence Corrections for Par-
ticle Size Determinations with the Coulter Counter. RSI 36:646-
653, 1965.
50. Pisani, J.F., and G. H. Thomson. Coincidence Errors in Automatic
Particle Counters. J. Phys. E. 4:359-361, 1971.
51. Mercer, T.T. Aerosol Technology in Hazard Evaluation, Academic
Press, New York, 1973.
729
-------
COMPARISON OF IMPACTION, CENTRIFUGAL SEPARATION AND
ELECTRON MICROSCOPY FOR SIZING CIGARETTE SMOKE1
R. F. Phalen, Department of Community and Environmental
Medicine, College of Medicine, University of California,
Irvine, California 92664
W. C. Cannon, Biology Department, Battelle Northwest
Laboratories, Richland, Washington 99352
D. Esparza, Aerosol Physics Department, Inhalation Toxicology
Research Institute, Albuquerque, New Mexico 87108
ABSTRACT
Determination of the size distribution of aerosols composed of
small liquid droplets involves special problems, especially when the
particles are numerous per unit volume of carrier gas. Electron micro-
scopy, useful in examining most solid aerosols, is not generally appli-
cable to liquids. Using unfiltered smoke from research cigarettes as a
test aerosol, three techniques for obtaining a size distribution were
compared. The aerosol, having a mass median aerodynamic diameter of
about 0.7 micrometers, was examined with a seven-stage cascade impactor,
a centrifugal separator, and by rapid fixation of the aerosol with
methyl 2-cyanoacrylate vapor followed by electrostatic precipitation and
electron microscopy. The resulting size distributions were compared and
found to be in fairly good agreement. Rapid fixation in the airborne
state appears to be a useful technique for preparing certain liquid aer-
osols for electron microscopy.
731
Preceding page blank
-------
COMPARISON OF IMPACTION, CENTRIFUGAL SEPARATION AND
ELECTRON MICROSCOPY FOR SIZING CIGARETTE SMOKE1
R. F. Phalen, Department of Community and Environmental
Medicine, College of Medicine, University of California,
Irvine, California 92664
W, C. Cannon, Biology Department, Battelle Northwest
Laboratories, Richland, Washington 99352
D. Esparza, Aerosol Physics Department, Inhalation Toxicology
Research Institute, Albuquerque, New Mexico 87108
INTRODUCTION
The objective of this study was to compare three methods of obtain-
ing size distributions of main-stream cigarette smoke. No attempt was
made to reproduce cigarette smoke as it exists in the human smoking
situation. Instead, laboratory methods of puffing, diluting, mixing and
sampling were used so that the three sizing methods could be compared
and so that other laboratories could repeat the studies. The methods
for sizing were: a) use of a cascade impactor, b) use of a centrifugal,
spiral aerosol separator, and c) coating of the smoke with methyl 2-
cyanoacrylate, collecting with an electrostatic .precipitator and sizing
from electron micrographs. The latter method of sizing is not considered
a standard laboratory technique.
Fresh cigarette smoke has several properties that make it diffi-
cult to characterize; it is chemically complex, initially highly concen-
trated and not in equilibrium with its surroundings with respect to
moisture content, temperature, or vapor pressures of volatile components.
In its fresh state it consists primarily of aqueous, spherical droplets
with dissolved and suspended solid materials. Unless it is diluted,
coagulation proceeds rapidly, and depending on the factors of dilution,
the median size of a cloud may increase or decrease with time.
A great deal of research has been and is being done with respect to
the health effects of cigarette smoke and there is a need for standardi-
zation of sizing methods. The methods compared in this study are all
relatively straightforward and one or more probably could be made avail-
able to the majority of those investigating cigarette smoke.
732
-------
MATERIALS AND METHODS
CIGARETTES
The Kentucky Reference IRI cigarette, stored at 60% RH and 20-23°C
was used throughout.
SMOKING PARAMETERS
The smoking machine (Figure 1) was recommended by C.H. Keith
(Celanese Fibers Company, Charlotte, N.C.). It consisted of a short
glass cigarette holder connected to a 3-necked 500 ml glass flask fitted
with a metal stirring paddle and a critical orifice through which a
timed flow of air could be drawn. The cigarette holder was removed to
permit sampling.
filter
valve
vacuum
500 ml flask
Figure 1. Machine used to generate cigarette smoke for
comparison of three methods of sizing
sub-micron droplets.
The smoking regime began with lighting the cigarette during a
35 * .5 ml puff of 0.94 seconds duration. Two subsequent puffs of the
same size were taken at 1 min. intervals. After the third puff the
smoke was purged from the flask. The fourth puff, 1 min. after the
previous puff, was sampled for 15 seconds, 5 seconds after it was drawn
into the flask.
SAMPLING DEVICES AND METHODS
A seven stage cascade impactor of the type designed by T. T.
Mercer2 was used. When operated at 500 cc/min. the effective cut-off
diameters of the impactor used were:
733
-------
Stage
1
2
3
4
5
6
7
6.
4.
3.
2.
1.
1.
0.
7 ym
3 ym
0 ym
3 ym
6 ym
0 ym
46ym
filter 0.0 ym
Impactor stages were covered with glass cover slips coated with a
silicone-oil spray. After sampling, the deposit on each impactor stage
was dissolved in a fixed volume of isopropyl alcohol and aliquots were
assayed with a spectrophotometer at a wavelength of 350 nm. The rela-
tive distribution of material on each stage was fitted to a log-normal
distribution and the mass median diameter and geometric standard devia-
tion determined. Two cigarettes were used for each impactor run.
Smoke was drawn into a Lovelace Aerosol Particle Separator (LAPS)
(P. Kotrappa and M.E. Light3) for 15 seconds at 310 cc/min. The device
had been calibrated with monodisperse polystyrene spheres and found to
collect particles from 0.54 ym to 5.1 ym along a 40 cm-long metal foil
strip. After sampling, the strip was cut into 22 segments and assayed
photometrically by the same method used with the impactor. The exit fil-
ter was also assayed. Six cigarettes were used for each run.
Samples for electron microscopy were collected using a point-to-
plane electrostatic precipitator operated at 50 cc/min. for 15 seconds
(P.E. Morrow and T. T. Mercer4). Before these sampling runs 3 drops of
methyl-2-cyanocrylate (Polysciences, Inc., Warrington, Pa.) were placed
in the 500 ml smoke chamber and the flask was heated to 70°C (this
method is similar to that described by G.P. Morie, C.H. Sloan and V.G.
Peck"1). Samples were collected on carbon-coated diffraction grating
replicas with line spacings of 0.833 ym. Sizing was performed using a
Zeiss TGZ3 Analyzer (Carl Zeiss, Inc., N.Y., N.Y.) from photomicrographs
with a magnification factor of about 30,000. An estimate of mass median
aerodynamic diameter was made by assuming a density of unity and
that the aerodynamic diameter of agglomerates containing n primary par-
ticles was equal to n/3times the average primary particle diameter.
Impactor and spiral-centrifuge samples were taken at two locations,
the Inhalation Toxicology Research Institute (ITRI), Albuquerque, New
Mexico, and at Battelle Northwest Laboratories in Richland,Washington(BNW"
using essentially identical smoking machines and sampling methods. The
precipitator samples were taken only at ITRI.
734
-------
RESULTS AND CONCLUSIONS
The agreement between all sampling methods and, in fact, all sample
runs was within the experimental errors of a given technique. The table
gives the results of each sampling run. In each case, the size distri-
bution was approximately log-normal and the mass median and geometric
standard deviation (og) were used to describe the results.
run
1,2
3-7
8
9-11
12
method
Impactor
Impactor
LAPS
LAPS
EM
average
MMAD
0.74 ym
0.71 ym
0.79 ym
0.72 ym
0.57 ym
££
1.4
1.4
1.3
1.4
1.5
location
ITRI
BNW
ITRI
BNW
ITRI
The precipitator sample contained spherical particles, many of which
were in cluster or chainlike agglomerates (Figure 2). Primary parti-
cles, sized from electron micrographs (n=191),were log-normally distrib-
uted with a count median diameter of 0.29 ym and a og of 1.4. A back-
ground (no cigarette) precipitator sample yielded some spherical parti-
cles, negligible in quantity when compared to samples of cigarette smoke.
Since results from each of the three methods were in close agree-
ment, it is concluded that any of the three techniques could be used for
describing the size distributions of cigarette smoke except when extreme
accuracy is required. Furthermore, the fixation by coating with methyl-
2-cyanoacrylate may be of use in electron-microscopic studies of other
aqueous aerosol particles.
ACKNOWLEDGEMENTS
This study was suggested by Dr. Charles H. Keith and performed with
the invaluable assistance of Robert Yarwood.
735
-------
Figure 2. Electron micrograph of cigarette smoke
particles fixed by coating with methyl-2-
cyanoacrylate vapor. Diffraction-grating-
replica spacing is 0.833 pm.
6
a
•H
•o
-------
REFERENCES
1. Research conducted at the Inhalation Toxicology Research Institute
of the Lovelace Foundation and at Battelle Northwest Laboratories.
2. Mercer, T.T., M.I. Tillery, and G.J. Newton. A Multi-stage, Low
Flow Rate Cascade Impactor. Aerosol Sci. 1:9-15, 1970.
3. Kotrappa, P. and M.E. Light. Design and Performance of the Lovelace
Aerosol Particle Separator. Rev. Sci. Instrmts. 43:1106-1112, 1972.
4. Morrow, P.E., and T. T. Mercer. A Point-to-Plane Electrostatic Pre-
cipitator for Particle Size Sampling. Am. Ind. Hyg. Assn. J. 25:
8-14, 1964.
5. Morie, G.P., C.H. Sloan and V.G. Peck. Study of Cigarette Smoke by
Means of the Scanning Electron Microscope. Beitrage Z. Tabakfor-
schung. 7:99-104, 1973.
737
-------
EXTENDED ELECTRIC MOBILITY METHOD
FOR MEASURING AEROSOL PARTICLE SIZE
AND CONCENTRATION
Earl 0. Knutson
IIT Research Institute
10 W. 35th Street
Chicago, Illinois 60616
ABSTRACT
An accurate method for determining the concentration and size distribu-
tion of aerosols has been developed, based on bipolar charging followed
by differential electric mobility analysis. The size resolution is
10% or better throughout the size range 0.005 to 1.0 ym. The particles
are kept airborne throughout the analysis, and are never subjected to
heat or high vacuum. Thus, the particle size obtained by the method
describes the particles in their natural state. Moderately volatile
particles can be analyzed in the same way as solid particles.
The method depends critically on accurate expressions for particle
mobility and charge distribution. The expressions used have been well
established in the literature, at least for spherical particles above
0.01 ym diameter.
The method is applicable to both atmospheric and laboratory aerosols,
provided that the number of particles larger than 1 ym is small com-
pared to the number smaller. The lower limit of concentration depends
on the type of aerosol sensor used in conjunction with the apparatus.
739
Preceding page blank
-------
EXTENDED ELECTRIC MOBILITY METHOD
FOR MEASURING AEROSOL PARTICLE SIZE
AND CONCENTRATION
Earl 0. Knutson
IIT Research Institute
10 W. 35th Street
Chicago, Illinois 60616
INTRODUCTION
The feasibility of measuring aerosol particle size by controlled elec-
tric charging and subsequent electric mobility analysis was first
demonstrated by Rohmann1 in 1923. Several variants of the method have
since been described in the literature. The first convenient, practi-
cal, and accurate device was that of Whitby and Clark2.
In a parallel but independent line of development, devices for the
measurement of electric mobility spectra of atmospheric ions have been
in use since about 1905. This led to the discovery and characteriza-
tion of the small, intermediate, and large atmospheric ions. Junge3
appears to have been the first to attempt deduction of the size and
concentration of atmospheric particles from measured ion mobility
spectra.
A common limitation of electric mobility-based methods for aerosol
particle size measurement is poor size resolution in the range
0.1 to 1.0 ym diameter. Junge3, whose method made use of the natural
charge on atmospheric particles, restricted the application to sizes
less than 0.1 Urn to avoid multiple charged particles. Whitby and
Clark's2 method, which uses unipolar diffusion charging, is usable up
to 1.0 urn, but has limited size resolution above 0.1 ym due to the
relatively flat particle mobility versus size curve.
This paper describes an extended electric mobility method for aerosol
size and concentration measurement. The method provides high resolu-
tion in the entire range 0.005 to 1.0 ym.
The method described here is similar to that suggested by Kudo and
Takahashi"* in that bipolar charging is used prior to the electric
mobility analysis. The bipolar charger is a passive device employing a
radioactive source. The mobility analyzer used in this work is a
740
-------
two-inlet, two-outlet flow coaxial cylinder design similar to that
described by Hewitt . The present method may be compared to the method
of Whitby and Clark , which uses unipolar charging and a two-inlet,
one-outlet mobility analyzer.
To date, the method described here has been used for approximately 150
size distribution and concentration measurements on laboratory and
atmospheric aerosols. For laboratory aerosols, 10% size resolution can
be obtained throughout the size range 0.005-1.0 urn. For atmospheric
aerosols, the size range coverage is 0.005-0.1 ym, since the concentra-
tion above 0.1 ym is usually too low to be accurately measured with the
present sensors. A programmable desk calculator with plotter attachment is
used to reduce the data and plot the resulting size distribution.
For those laboratory aerosols which are nearly monodisperse, the
present apparatus may also 1
by Knutson6 to obtain highl;
deviation of particle size.
present apparatus may also be operated in the mode described previously
by Knutson6 to obtain highly accurate values of the mean and standard
APPARATUS
An overall sketch of the apparatus required for this method of particle
size analysis is given in Figure 1. The main components are the
aerosol charger, the mobility analyzer, and the aerosol sensor.
BIPOLAR CHARGER
The aerosol bipolar charger is shown in Figure 2. The aerosol is
drawn through a 1.27 x 1.27 x 5.0 cm long channel. One wall of the
channel is an aluminum-coated Mylar alpha-particle window. A foil con-
taining 1.9mCi2l+1Am is cemented to the floor of a 0.48 cm deep recess
behind the window. The radioactive area is 1.0 x 5.0 cm. Alpha
particles penetrate the window and cross the channel, penetrating the
aerosol flow at right angles. The ion concentration and production
rate were determined by measuring the current between the A]-coated
window and an electrode on the opposite channel wall, as a function of
applied voltage. At low valtage, the resistance of the air gap was
3,170 M^. The saturation current (at 500 volts) was 63 nano-amps.
These values imply an ion concentration of 1.74 x 10 cc and a pro-
duction rate of 3.9 x 1011 ion-pairs sec ~l. The measured ion
concentration checks well with the value 2.24 x 10 cc~ estimated as
per Cooper and Reist .
The physics of aerosol charging by exposure to bipolar ions is known
from extensive studies reported in the literature. The basic
741
-------
AEROSOL IN
241
Am
N
AR
ER,
iCi
II
ABSOL
FILT
AEROSOL
UTE
ER
AIR
CLEAN
FLOW, qc
24.6 Inm
FLOW, qa
3.1 1pm
FLOW
CONTROL
VALVE
CAPILLARY
FLOWMETER
AEROSOL
CLASSIFICATION
PRECISION
ADJUSTABLE
HIGH VOLTAGE
POWER SUPPLY,
10 kV
FILTER/ELECTROMETER
AEROSOL SENSOR
VACUUM SAMPLING
PUMP, FLOWRATE
50 cm Hg q = 2.8 1pm
2-INLET,
2-OUTLET
HEWITT-TYPE
ELECTRIC
MOBILITY
ANALYZER
I
MAIN
OUTLET
FLOW, qm
24.9 1pm
CRITICAL
ORIFICES
(2)
Figure 1: AEROSOL SIZE MEASURING SYSTEM
742
-------
AEROSOL INLET
MATERIAL:
HIGH DENSITY
POLYETHYLENE
II
II
FLAT PLATE
ELECTRODE
AEROSOL.
OUTLET
3.81 cm
1.9mCi 241
ON 1.27 x 3.81 cm
FOIL CEMENTED TO
FLOOR OF RECESS
ALUMINIZED MYLAR
GEIGER COUNTER
WINDOW MATERIAL
STRETCHED OVER RECESS
Figure 2
AEROSOL BIPOLAR CHARGER
743
-------
meter x which carry V units of charge
theoretical equation was derived in various ways by Lissowski <
Pluvinage9, Gunn10, Luchak1 1 , and Keefe, Nolan, and Rich12. This
theory shows that, after sufficient exposure to bipolar ions, the
particle charge reaches a stationary state which can be described to
good approximation by
* (x) = (2TVa)~1/2exp[-V2/2a ] (1)
where
(x) = the fractional number of particles of dia-
meter x which carry V
(v = 0, + 1, + 2,...)
a = xkT/2e2, a dimensionless particle size
k = 1.38054 x 10~16 ergs/°K, Boltzmann's constant
T = the absolute temperature, K
e = 4.803 x 10" 10 statcoulomb
Experimental studies of bipolar charging of 0.4 to 4.0 urn diameter
particles of various materials were conducted by Lissowski8, Woessner
and Gunn13, and Gunn and Woessner14. Luchak11 analyzed data by
Kunkel and by Gillespie. These studies confirmed Equation 1 for
slowly equilibrated aerosols, regardless of initial charge. For rapid-
ly equilibrated aerosols, however, Gunn and Woessner found that the
charge distribution becomes assymetric, as described by the equation
$v(x) = (2TTa)"1/2exp[-(v-aa)2/2a J (2)
where
o = ln(A+/A_)
A , A_ = the air conductivity due to positive and
negative ions, respectively
This correction to Equation 1 has been found necessary for the charger
in Figure 2. The best value of a for the present charger is -0.1. The
correction affects mainly particles above 0.5 ym diameter.
Takahasi and Kudo15 have shown that Equation 1 is valid for 0.1 ym Pb
and 0.14 ym Zn aerosols conditioned by exposure to a 2 mCi 2'*1Am alpha
source similar to the one shown in Figure 2.
744
-------
An extensive study of bipolar charging for 0.01 to 0.07 ym diameter
hot nichrome wire particles was conducted by Pollak and Metnieks16.
They found that Equation 1 is accurate in the size range 0.027 to
0.07 ym, but underestimates the number of charged particles for
particle size below 0.027 ym. Their measurements corroborated earlier
results by Nolan and Kennan17. The theoretical reasons for the
discrepancy were identified by Fuchs18, who presented improved calcu-
lations. Data and theory are shown in Figure 3.
The most comprehensive theory of small particle charging at present is
that of Gentry and Brock19, as amended by Brock20 to include the image
effect. The corresponding curve in Figure 3 was calculated by the
simplified method of Gentry21, assuming conducting particles and free
molecular regime ion transport. Of the three theories shown, the
Gentry curve best fits the data below 0.03 ym. In this range, Gentry's
curve may be approximated by:
*v(x) = 2 + (0.1/x/'5444if V = -+1 (3)
( 0 if v > 1
where x is in ym.
Based on the above considerations, the following expression for
$ (x) was adopted in this work:
For x > 0.03 ym, Equation 2 with a = -0.1 is used;
For x < 0.03 ym, Equation 3 is used.
Further data on bipolar charging have appeared recently. Liu and Pui? "*
present data on NaCl and methylene blue particles with diameter from
0.02 to 0.2 ym and on dioctyl phthalate droplets 0.53 to 1.17 ym in
diameter. The bipolar charger used was a 2 mCi 85Kr beta-emitter.
Several of their data points are shown in Figure 3.
Liu and Pui's data agreed well with Equation 1 down to the smallest
particle size considered. It should not be inferred, however, that
Equation 1 is valid for particle diameters less than 0.02 ym. For the
present, Equation 3 is preferred over Equation 1 for particle size less
than 0.03 ym due to its better theoretical basis and the supporting
data of Pollak and Metnieks16. Equation 3 differs from Liu and Pui's
data by at most 16%.
Liu and Pui21* found no difference as to charging between NaCl and
methylene blue.
745
-------
0,30
£ 0,20
cc
o
^-s LJJ
v_x (J3
r-l Z
I |-~*
O CO
a: LL
o o
^-s >-
0,10
•e
CQ
<
CQ
O
0,05
0,03
0,02
DATA:
A POLLAK & METNIEKS
'O NOLAN & KENNEN
OLIU & PUI
FUCHS
THEORY.
KEEFE, NOLAN,
AND RICH THEORY
GENTRY THEORY
I
0,005 0,01 0,02 0,03 0,05
x, PARTICLE DIAMETER, pm
0,1
Figure 3
BIPOLAR CHARGING THEORY
AND DATA
746
-------
K02ima and Sekikawa present data on bipolar charging of urban
aerosols upon exposure to 210Po alpha particles. Their results (not
shown) lie somewhat above the theoretical curve of Fuchs18.
MOBILITY ANALYZER
The differential electric mobility analyzer used in the present work
is shown in Figure 4. It is a two-inlet flow, two-outlet flow coaxial
cylinder design similar to those described by Hewitt5, Knutson5, and
Lui and Piu22.
The aerosol mechanics of the electric mobility analyzer have been
discussed in some detail by Knutson6. The main result was an expression
for the transfer function, fi. This function is defined as the ratio of
the number of particles carried out via the sampling outlet flow to the
number entering via the aerosol flow. The function fi is most easily
presented in the form of a graph, as in Figure 5. It is seen that fi
is a function of: the particle mobility, K; the applied voltage, V;
the mobility analyzer length, L; the radius ratio, ra/ri and the
four flow rates qc, qa, qm, and qs. For a given voltage, V, the
half width, AK, of the mobility fraction extracted from the aerosol is
AK = 1/2 (q + q )/(2irAV) (4)
3 S
The centroid, K*, of this mobility band is
K* = l/2(q + q )/(2nAV) (5)
c m
The resolution is therefore
AK/K* = (q + qD)/(qn + q ) (6)
3 S C Tu
AEROSOL SENSOR
The aerosol sensor (Figure 1) is used to determine the amount of aerosol
emerging from the mobility analyzer sampling outlet at each mobility
setting. Several devices can be used. The simplest is the collecting
filter/electrometer arrangement indicated in Figure 1. This measures
the electric current due to the stream of charged particles. Alterna-
tives which have been used include a single particle optical counter
(set to count all emerging particles within its size capability), a
condensation nuclei counter, and an aerosol mass monitor. The latter
sensor leads in a direct way to the aerosol size distribution by mass.
., 747
<4
-------
Clean Air Galley, Fed
By Two Radial Pipes
(not shown)
Aerosol Galley, Fed
By Two Radial Pipes
(not shown)
2345
Scale in cm
2r2 -7.62 cm Dia
2 r( -5.08 cm Oio
Twenty-four 0079 cm Dio
Radial Holes for Aerosol
Sample
Twelve 0 32 cm Dia
Radial Holes for
Main Outlet Air
L-7l.6cm
Twelve 0 32 cm Dio
Radial Clean Air
Entry Holes
58 p-m Mesh Nylon
Screen in This
Plane
Twenty-four 0.079 cm
Dio. Axial Aerosol
Entry Holes
Flow
Direction
1
Materials
Alum mum
Delrin Plastic
Mom Outlet
Sample Outlet —
Hole for H. V.
Lead to Center
Electrode
Figure 4: DETAIL OF MOBILITY ANALYZER
748
".*,
-------
qg/qa)
(q +q
Mc Hm
Figure 5
MOBILITY ANALYZER TRANSFER FUNCTION, ft
= aerosol flow rate qr = clean air flow rate
q = sampling flow rate
s
V = center rod voltage
L = distance between inlet and exit gaps
r , r9 = inner and outer radius of annulus
q = main outlet flow rate
m
A = L/ln(r2/r1)
749
-------
OPERATION
The aerosol to be analyzed is drawn through the apparatus as shown
in Figure 1. The bipolar charger establishes the charge distribution
given by Equations 2 and 3. The mobility analyzer extracts mobility
fractions selected by the applied voltage in accordance with the
transfer function, Figure 5. The electric current carried by the
extracted mobility fraction is measured by the aerosol sensor. Thus,
the raw data consists of a table of current vs. voltage.
The mobility analysis process may be visualized with the aid of
Figure 6. The sloping lines in figure 6 show the relationship between
particle electric mobility and particle diameter for various particle
electric charges. The horizontal lines labeled K^, K2, and Ko,
represent a sequence of narrow mobility fractions. For reasons which
will become clear, adjacent mobility fractions are chosen in the
ratio 1:1.429. The vertical lines in Figure 6 represent the sequence
particle sizes defined by the intersection of the mobility lines with
the curve for mobility of singly charged particles.
Each mobility fraction consists of several discrete particle sizes. For
example, the mobility fraction K^ (2.042 x 10""* cm /volt-sec.) in
Figure 6 consists of particles of diameter 0.120, 0.186, 0.240, and
0.292 ym, carrying 1, 2, 3, and 4 units of charge, respectively. The
first of these sizes is that labeled x^ in Figure 6. The remaining
three particle sizes nearly coincide with the sizes labeled x^, x^, and
x5 (0.183, 0.233, and 0.297 ym, respectively). This relationship is
true also of the other mobility fractions: the mobility fraction K.
consists of particles whose sizes are given to close approximation by
xi> xi+2> xi+v anc^ xi+4- This approximation, which is made possible
by the choice of spacing for the mobility fractions, is used in the
analysis which follows.
Particles of electric charge greater than four units are neglected
in this treatment, since they are usually much less numerous than those
considered above. An alternate procedure is available when greater
accuracy is needed (see Appendix).
With the above approximations, the response, R^, of the aerosol sensor
at the i voltage setting may be written as the sum of four terms:
,A,,N,w^ (7)
V
v=l
; 750
i"
-------
u
uu
in
i
o
c;
LU
E
u
m
o
o
cc
h-
u
LU
_i
LU
LLl
Od
<
CL.
0,15 0,2 0,3 0,5 0.7
X, PARTICLE DIAMETER, ym
Figure 6
ILLUSTRATION OF THE SERIES OF
MOBILITY FRACTIONS AND PARTICLE SIZES
751
-------
Where qa is the aerosol flow rate through the charger, N-; is the number
concentration of aerosol particles in the size range (x, x + dx)
centered at x-^, W j is a weight factor giving the response of the sensor
to a single particle of charge V and size xj, w ^ is a weight factor
related to the finite width of the mobility band, and j = i + 0,
i 4- 2, i + 3, and i + 4 for V = 1, 2, 3, and 4, respectively. A rigor-
ous derivation of Equation 7, giving the weight factors, is given in
the Appendix.
The aerosol analysis procedure consists of obtaining aerosol output
readings RI, R2, R3,... for the descending sequence of mobility frac-
tions KI, K2, K-},... A standard sequence of 25 voltage settings has
been adopted for the operation, as shown in Table 1. The flow rates
may be adjusted to suit the aerosol, subject to the restriction of
high resolution, that is: (qa + qs)/(qc +<3m) <<: !• The sequence
K^, K2» £3,... is computed from the voltages using Equation 5. The
corresponding particle sizes are computed from the Stokes-Cunningham
mobility formula (Equation 25 of Wahi and Liu23). Particle size
sequences at various flowrates are given in Table 1.
The voltage sequence may be entered at any point, but must be continued
upward until the aerosol output drops to essentially zero, say at the
ntn step. The zero output indicates that the particle concentration
for size xn and above is negligibly small. Data taking is then stopped
and Equation 7 is applied to the data to solve for the concentrations
N(X1), N(x2),...N(xn_1).
The solution of Equation 7 is accomplished by means of a programmable
desk calculator. A plotting accessory automatically plots out the
particle size distributions.
Aerosol losses in the analyzer and its piping due to diffusion and im-
paction have been estimated theoretically. Impaction losses appear to
be inconsequential. Diffusion losses for the operating conditions
shown in Figure 1 are estimated at 35% for 0.004 ym, 10% for 0.01 ym,
and less for larger particles.
EXAMPLE MEASUREMENT RESULTS
Figure 7 shows three example aerosol size distributions obtained by the
method just described. The plot format is dN/d(ln x) on a linear scale
as a function of x on a logarithmic scale. The curves are not normal-
ized. Thus, the area under the curve in each case is the total number
concentration of the aerosol.
752
-------
Table 1
COURSE VOLTAGE SEQUENCE AND CORRESPONDING PARTICLE SIZE
FOR IITRI ELECTRIC MOBILITY AEROSOL ANALYZER
Voltage
Setting
1.902
2.718
3.884
5,551
7.932
11.33
16.20
23.14
33.07
47.26
67.54
96.51
137.9
197.1
281.6
402.4
575.1
821.8
1,174.
1,678.
2,398.
3,427.
4,897.
6,998.
10,000.
Particle Size (vim) at Various
q - 10
0.0052
0.0062
0.0075
0.0089
0.0107
0.0128
0.0154
0.0.83
0.0221
0.0271
0.032
0.039
0.047
0.057
0.069
0.085
0.103
0.129
0.159
0.200
0.25
0.33
0.43
0.56
0.76
15.0
0.0043
0.0052
0.0061
0.0073
0.0087
0.0105
0.0125
0.0149
0.0179
0.0216
0.026
0.031
0.038
0.046
0.055
0.067
0.082
0.100
0.124
0.156
0.19
0.24
0.32
0.41
0.54
20.0
0.0038
0.0045
0.0053
0.0065
0.0075
0.0090
0.0108
0.0129
0.0155
0.0186
0.022
0.027
0.032
0.039
0.048
0.057
0.070
0.085
0.105
0.129
0.16
0.20
0.26
0.33
0.43
Flow Rates, q (1pm)
25.0
0.0034
0.0039
0.0048
0.0057
0.0067
0.0080
0.0097
0.0116
0.0138
0.0166
0.020
0.025
0.029
0.035
0.042
0.051
0.062
0.075
0.092
0.113
0.14
0.18
0.22
0.28
0.37
30.0
0.0030
0.0037
0.0044
0.0051
0.0061
0.0074
0.0088
0.0105
0.0128
0.0151
0.018
0.022
0.026
0.032
0.038
0.046
0.056
0.068
0.087
0.102
0.13
0.16
0.20
0.25
0.32
q = mobility analyzer clean airflow
q = mobility analyzer main outlet flow
4
753
-------
10x10
8
5
o
u
X
c
2
0
PSL RESIDUE
AEROSOL
ATMOSPHERIC
AEROSOL
O
o
STEARIC ACID
CONDENSATION
AEROSOL (x50)|
- o
x50
3 5 10 20 50 100 200 500 1000
x, PARTICLE DIAMETER,
Figure 7
EXAMPLE AEROSOL SIZE DISTRIBUTIONS
754
-------
The left curve in Figure 7 is an atmospheric aerosol sample taken in
Chicago on February 12, 1974. The analyzer flow rates were as shown
in Figure 1. The collecting filter /electrometer current measuring
device indicated in Figure 1 was used as the aerosol sensor. The size
distribution shows a weak mode at - 0.01 ym. The curve falls rapidly
to the right of the mode, but continues leftward out of the limits of
the analyzer. The area under the curve represents 1.59 x 106
particles/cc. This was among the highest concentration observed in
68 atmospheric aerosol samples in Chicago.
The total concentration for the atmospheric aerosol shown in Figure 7
was checked with a condensation nucleus counter (Gardner Associates,
Schenectady, N.Y., Small Particle Counter, Type CN) . The result was
3.4 x 10 particles/cc, or about 1/5 of the total number concentration
indicated by the mobility analyzer. The reason for this difference has
not been determined.
The center curve in Figure 7 was obtained when nebulizing an aqueous
suspension of 0.557 ym polystyrene latex particles. The aerosol
represented in Figure 7 consists of surfactant particles left when
droplets not containing a latex particle evaporate. The latex particles
themselves were too few in number to be detected accurately. The flow
rates in this test were qa = 1.4 1pm, q = 9.6 1pm, qg = 1 1pm, and
qm = 10 1pm. The current sensor was used.
The number concentration of the surfactant residue particle aerosol, as
determined by the area under the curve in Figure 7, was 1.17 x 10 cc"1.
The geometric mean diameter and geometric standard deviation were
0.054 ym and 1.6, respectively. For comparison, the concentration of
latex particles were measured separately and found to be 1,200 cc"1.
The right curve in Figure 7 depicts a stearic acid aerosol generated by
a high-volume condensation aerosol generator developed at IITRI. The
generator output is 34 1pm, which was diluted to 7.1 x lO4 Iptn for the
test shown in Figure 7. The analyzer flow rates for this test were
qa = 0.38 1pm, qc = 7.07 1pm, qs = 0.31 1pm, and qm = 7.0 1pm. A.n
optical single-particle counter (Royco, Inc., Menlo Park, California,
Model 245) was used as the aerosol sensor in this test. The counter
was set to count all particles above 0.3 ym diameter. The sampling
flow rate (0.31 1pm) was diluted with filtered air to make up the
28.3 1pm flow required by the counter.
The number concentration computed from the stearic acid aerosol curve
in Figure 7 is 4,600 cc"1. The geometric mean diameter is 0.73 ym.
The geometric standard deviation of the stearic acid aerosol was too
small to be accurately measured by the present technique. However, the
755
-------
analysis method described by Knutson6, applicable to near-monodispef9e
aerosols, was applied to the data and yielded geometric mean diamether =
0.73 ym and geometric standard deviation = 1.12.
The number concentration of the stearic acid aerosol was checked by a
microscope count of a glass slide sample taken with an electrostatic
sampler (Thermo-Systems, Inc., St. Paul, Minnesota, Model 3100). The
result was 2.03 x 10 particles/cc, 4.4 times the concentration deter-
mined by the mobility analyzer. The reason for this difference has not
been identified.
SUMMARY
It has been demonstrated that the electric mobility method for measur-
ing aerosol particle size and concentration can be extended to provide
10% size resolution throughout the range 0.005 to 1.0 ym. This is
accomplished by using bipolar charging, a differential mobility analy-
zer, and a specially designed measurement procedure.
The extended electric mobility method has been applied to about 150
atmospheric or laboratory aerosols. The method was found to be conveni-
ent and moderately fast (20-30 min. for data collection). The size
resolution has been adequately demonstrated with near-monodisperse
aerosols. Further work is required, however, to determine the accuracy
with which the device measures concentration.
ACKNOWLEDGEMENT
This work was supported by the U.S. Atomic Energy Commission through
its Contract No. AT(ll-l)-578.
REFERENCES
1. Rohmann, H. Method of Size Measurement for Suspended Particles.
Z. Phys. 17: 253-265, 1923.
2. Whitby, K. T., and W. E. Clark. Electric Aerosol Particle
Counting and Size Distribution Measuring System for the 0.015 to 1 y
Size Range. Tellus XV111.2: 573-586, 1966.
756
-------
3. Junge, C. E., The Size Distribution and Aging of Natural Aerosols
as Determined from Electrical and Optical Data on the Atmosphere.
J. Meteor. 12: 13-25, February, 1955.
4. Kudo, A., and K. Takahashi. A Method of Determining Aerosol
Particle Size Distribution Applying Boltzmann's Law. Atmos. Envir.
6: 543-549, August, 1972.
5. Hewitt, G. W. The Charging of Small Particles for Electrostatic
Precipitation. Trans. Amer. Instit. Elec. Engrs. 76: 300-306, 1957.
6. Knutson, E. 0., The Distribution of Electric Charge among the
Particles of an Artificially Charged Aerosol. PhD thesis at
University of Minnesota, 1971. 210 p.
see also Knutson, E. 0., and K. T. Whitby. Aerosol Classification
by Electric Mobility: Apparatus, Theory and Applications. To appear
in J. Aerosol Sci. 6(6), 1975.
7. Cooper, D. W., and P. C. Reist. Neutralizing Charged Aerosols with
Radioactive Sources. J. Colloid Interface Sci. 45: 17-26, 1973.
8. Lissowski, P. Das Laden von Aerosolteichen im ciner Bipolaren
lonenatmosphare. Act Phys. Chem. URSR. 13: 157-192, 1940.
9. Pluvinage, P. Etude Theorique et Experimentale de la Conductibilite
Electrique dans les Nauges Non Orageux. Annales de Geophys. 2:
31-54, 160-178, 1946.
10. Gunn, R. The Statistical Electrification of Aerosols by Ionic
Diffusion. J. Colloid Sci. 10: 107-119, 1955.
11. Luchak, G. The Theory of the Electric Charge Distribution of
Monodispersed Lightly Charged Aerosols of Spherical Particles
Coagulating in a Bipolar Ionized Atmosphere. J. Colloid Sci. 12:
144-160, 1957.
12. Keefe, D., P. J. Nolan, and T. A. Rich. Charge Equilibrium in
Aerosols According to the Boltzmann Law. Proc. Roy. Irish Acad.
60 (Sect. A): 6-45, 1959.
13. Woessner, R. H., and R. Gunn. Measurements Related to the Funda-
mental Processes of Aerosol Electrification. J. Colloid Sci. 11:
69-75, 1956.
14. Gunn, R., and R. H. Woessner. Measurements of Systematic Electrifi-
cation of Aerosols. J. Colloid Sci. 11: 254-259, 1956.
757
-------
15. Takahashi, K., and A. Kudo. Electrical Charging of Aerosol Parti-
cles by Bipolar Ions in Flow Type Charging Vessels. J. Aerosol
Sci. 4: 209-216, 1973.
16. Pollak, L. W., and A. L. Metnieks. On the Validity of Boltzmann's
Distribution Law for the Charges of Aerosol Particles in Electrical
Equilibrium. Geofis. Pura Appl. 53: 111-132, 1962.
17. Nolan, P. J., and E. L. Kennen. Condensation Nuclei from Hot
Platinum: Size, Coagulation Coefficient and Charge Distribution.
Proc. Roy. Irish Acad. (Dublin) 52A: 171-190, 1949.
18. Fuchs, N. A. On the Stationary Charge Distribution on Aerosol
Particles in a Bipolar Ionic Atmosphere. Geofis. Pura Appl. 56:
185-193, 1963.
19. Gentry, J., and J. R. Brock. Unipolar Diffusion Charging of
Small Aerosol Particles. Chem. Phys. 47: 64-69, 1967.
20. Brock, J. R. Aerosol Charging: The Role of the Image Force.
Appl. Phys. 41: 843-844, 1970.
21. Gentry, J. W. Charging of Aerosol by Unipolar Diffusion of Ions.
Aerosol Sci. 3: 65-76, 1972.
22. Liu, B. Y. H., and D. Y. H. Pui. A Submicron Aerosol Standard and
the Primary, Absolute Calibration of the Condensation Nucleus
Counter. Colloid Interface Sci. 47: 155-171, 1974.
23. Wahi, B. N., and B. Y. H. Liu. The Mobility of Polystyrene Latex
Particles in the Transition and the Free Molecular Regimes.
Colloid Interface Sci. 37: 374-381, 1971.
24. Liu, B. Y. H., and D. Y. H. Pui. Equilibrium Bipolar Charge Dis-
tribution of Aerosols. Colloid Interface Sci. 49: 305-312, 1974.
25. Kojima, H., and T. Sekikawa. An Attempt for Obtaining the
Aerosol Size Distribution. Meteor. Soc. Japan. 51: 287-293, 1973.
APPENDIX
MATHEMATICAL DETAILS
In this appendix, a formal derivation is given for Equation 7 of the
758
-------
text, and methods of solving the equation are discussed.
The number concentration of particles with size in the range (x, x + dx)
and charge V after bipolar charging is given by the product the area under the curve in Figure
5 of the text. This area is qs/2ir. With these steps, Equation 8
becomes:
R(V) = q
759
-------
00
2q q \
= (-jr~ / Wy(x*)$ (x*)N(x*)K(x*) (dx/dK)* (Jl)
c m /
v=l
The last form is obtained using Equation 10.
In practice, the sensor response, R, is read at a discrete set of
voltage settings, V-^, V2» V3,.., V^, . . . . Thus x* depends on two
indices, V and i. A great simplification could be realized if x*
could be made to depend on one index only. As indicated in Section 3
of this text, this can be accomplished by a proper choice of voltage
settings.
Note that by Equation 10, x* depends only upon the product to V and V.
Consider the voltage sequence defined by V^+-^ = 1.429V^. This sequence
has the property that
V.+2 - 2.042V. = 2V.
V.=3 = 2.918V. = 3V. (]2)
V.=4 = 4.170V. = 4VjL
Equations 10 and 12 show that the doubly charged particles at the itn
voltage have nearly the same size, x*, as singly charged particles at
the (i + 2)tn voltage. Similar statements apply for the particles
bearing 3 and 4 units of charge. This was demonstrated graphically in
Figure 6. To good approximation, the reduction of x* to a single-
index qualtity has been accomplished, at least for particle charges
1, 2, 3, and 4.
In many cases, Equation 11 can be truncated after four terms.
Introducing index j,
fi if V = 1
i = 1+2 lf V = 2 (13)
3 )i + 3 if v = 3 UJ;
/ i + 4 if v = 4
Equation 11 becomes
760
-------
where R-j^ = R(Vi), N. = N(xp, etc., and x* denotes the root of Equation
10 for V = 1 and V = V j . J
Equation 14 is the formal version of the intuitive Equation 7. The
weight factor wv< in Equation 7 is seen to have the value
It is seen that w^ is independent of the index v.
The weight factor W^j depends on the aerosol sensor used. If an optical
single-particle counter is used, each particle contributes 1 count so
that Wv.: = 1. If a current sensor is used, as in Figure 1, Wv^ = the
particle charge, Ve. if a mass flow rate sensor is used Wv^ = the mass
of a particle of size Xi .
Equation 14 may be treated as a fifth-order, inhomogeneous , linear
difference equation. The coefficients, although not constant, are
calculable from the sequence of particle sizes. Solution may be
accomplished by applying Equation 14 recursively for Nn_j , Nn_2>-..,
NI (in that order) . The required starting values are Nn+^ = Nn+^ =
Nn+2 = ^n+l = Nn = 0. Alternatively, Equation 14 may be regarded as
defining a linear system of equations for Nj_, N2»..., Nn_|. The set
of equations may be solved by standard methods, or by special methods
which exploit the fact that the coefficient matrix has zeros below the
main diagonal.
If the aerosol being analyzed contains appreciable numbers of particles
larger than ~ 0.5 ym, a significant number of particles carries more
than 4 units of charge. In this case, truncation of Equation 11 after
4 terms causes serious error in interpreting the data. This error is
most serious when the current measurement is used to sense the output
aerosol, because the weight factor Wv(x) then increases with particle
charge. An alternate procedure has been developed for use when appre-
ciable numbers of particles exceed 0.5 ym diameter.
The alternate procedure makes use of the voltage sequence v±^i = 1. 1053V -^,
For this sequence,
Vi+6 = 2. 015V ±
Vi+10 = 3.008Vi
Vi+13 = 4.062Vt
Vi+15 = 4.962V±
vi+17 = 6-062vi
*' 761
-------
Hence, this voltage sequence permits incusion of the terms fof
particle charges 1 through 6 in Equation 11. This voltage sequence
was used to obtain the stearic. acid size distribution at the right
of Figure 7.
762
-------
RAPID MEASUREMENT OF PARTICULATE SIZE
DISTRIBUTION IN THE ATMOSPHERE
R. L. Chuan
Celesco Industries
Costa Mesa, Calif.
ABSTRACT
A new device has been constructed and tested, which allows the rapid
determination (in terms of minutes) of the distribution of particulate
mass by size in the range 0,05 to 50 micrometers. The device is essen-
tially an active cascade each stage of which consists of a quartz cry-
stal microbalance. The mechanical and fluid-mechanical design of the
system is entirely equivalent to a conventional cascade, except that the
filter or impactor is replaced by a piezo-electric crystal whose reso-
nant frequency decreases in proportion to accumulated mass, brought to
the adhesive-coated crystal surface by aerodynamic impaction. The type
of crystals and electronics employed offers a basic sensitivity of 7 x
108 Hz/gm and a frequency stability of 10~2 Hz/sec, which makes it pos-
sible to obtain sensible frequency changes in each of the 6-to 10-stage
cascade in sampling time of 2 to 6 minutes at a sampling flow rate of
100 ml/min whei the total particulate mass concentration is between 50
and 100 micrograms/m . Tests with the instrument have been conducted
in both indoor and outdoor atmospheres, with total particulate mass con-
centration ranging from 25 to 200 >>g/m . Some outdoor measurements dur-
ing the passage of photo-chemical smog over the test site show the tran-
sition from a normal mono-modal to a bi-modal distribution.
763
-------
RAPID MEASUREMENT OF PARTICULATE SIZE
DISTRIBUTION IN THE ATMOSPHERE
R. L. Chuan, Celesco Industries Inc.
INTRODUCTION
The determination of the distribution of particulate mass accord-
ing to size is of importance to the study of the dynamics of particu-
late generation and transport. The study of dynamical features neces-
sarily requires the rapid measurement of size distribution. It is also
desirable that a single means of measurement be applied over the com-
plete size range of interest, for evident reasons of consistency in
data interpretation. The technique of inertial separation of particu-
lates by size is well developed and relatively free of theoretical and
experimental uncertainties or ambiguities, and should lend itself well
to particulate size determination if the means of particulate mass
measurement can be made rapid. The conventional cascade impactors usu-
ally require the order of hours to collect sufficient samples for ac-
curate weighing when sampling ambient atmosphere, with the development
of the quartz crystal microbalance active impactor and its ability to
measure in real time the mass of impacted particulates, as reported by
Chuan^-'2, it is possible to achieve rapid determination of particulate
size distribution by using a cascade of these active impactors. This
paper describes the design and operation of such an active cascade sy-
stem, and discusses some typical results obtained by sampling ambient
air.
764
-------
DESCRIPTION OF APPARATUS
The cascade is designed in the conventional way of pumping air
through a series of chambers with increasing velocity as the air
passes through successive stages. The flow rate to be sampled is
determined by considerations of the area available for impacting par-
•~\
ticulates on the crystal (of the order of 0.2 cm ), the sensitivity
of the crystal (about 7 x 10 Hz/gm), the stability of the electronics
and the concentration of particulates per size interval to be sampled.
The basic response of the system expressed in terms of a frequency
shift rate is
~ = io"12cvk (i)
dt
where C is the mass concentration in microgram/cubic meter (/.g/irf ) ,,
V the flow rate in rnilliliter/minute (ml/min) and k the senslhivicy
in Hertz/gram (Hz/gm).
The electronics in each of the cascade stages consist of two
oscillators and a mixer, all built into a hybrid chip. With the typs
of A-T cut matched crystals and the hybrid chip used the overall
accuracy of the microbalance unit is typically about 0.02 Hz/Tin. A
frequency shift of 1 Hz/min would then constitute a signal to noi =e
ratio of 50; and is considered a criterion for designing the <".-• scale
p
system. With a sensitivity of 7 x 10 Hz/gm and a concentrai.i,on of
10 /•« g/m , one then arrives at a flow rate of 140 ml/min as that neces-
sary to produce an adequate response.
In order to obtain a 50% cut-off for mass density 2 particulcl.es
of 0.05 micron diameter, the impaction jet velocity is to be set at
about 3 x 104 cm/sec with a jet diameter of about 2 x 10" CTI. nil>e
fabrication technique involved in making very small precision j 3ts
allows one to make an array of 37 jets within a circle of about 0.4
cm. This is within the working area of the crystal, and the total
flow rate through the array of jets is about 100 ml/min, which is near
the amount needed for adequate system response.
Once the final stage configuration is established the preceding
stages can be configured to provide any size intervals one desires.
The data to be shown later in this paper were produced by two cascade
systems, one a 6-stage and the other a 10-stage. The 50% cut-off
points for each cascade, calculated from actual, measured flows and
pressure drops, are listed below:
765
-------
;;' x--stage cascade
s' -jgi.' 5(1'a cut-off
1 5 3 /(ra
2 25
5 6.6
4 1.4
!;. 0.32
6 0.06
Ton-stage Cascade
3 77
2 36
:; 17
4 7.9
3.7
6 1.7
/ 0.87
8 0.40
9 0.17
LO 0.06
The signal from ^ach of the cascade stages is in 'the form of a
beat frequency (between th<= sensing and reference Crystals) in the
kilohertz range. It is amplified and clipped to a square wave in the
base of the cascade as- shown In i:ig. 1. The base also contains the
power-supply for the stager, nnd the corresponding amplifiers, as well
as the pump and flow-net •> for the cascade. Fig. ? shows the internal
details of a stage. :\ r'onpa.ij on control unit, also seen in Fig. 1, con-
tains a programmer and n pri.rrcrr to process the signals froir. the cascade.
The beat-frequency Croin ey drift in the electronics and partly by slow
temperature changes which affect both the crystals and the electronics
(but in no event greater: than about 0.2 Hz/min) , the instrument is usu-
ally operated continuo'isly lor several hours at a time and a filter is
-------
Reproduced from
best available copy.
0)
-d
rt
O
w
a)
u
o
0)
oo
cfl
j-i
01
X
•H
00
•H
767
-------
0)
oo
ca
o-
-------
Used to obtain a tare correction before and after a short sampling tun.
It has been found from tests that at a total concentration of about
SO^g/m^, five successive 2-minute scans are sufficient to produce
good accuracy/ with an 8- to 10-minute tare run with the filter before
and after the sampling run if higher accuracy is desired. At total
concentrations higher than 100 /"g/m a single 2-minute scan is suf-
ficient.
769
-------
RESULTS AND DISCUSSION
Some results of outdoor ambient air measurements are shown in
Figs. 3 and 4. The measurements were made in Costa Mesa, California,
which is located southeast of Los Angeles and about 7 miles from the
coast. The location is generally free of the more severe smoggy con-
ditions to the east of downtown Los Angeles, except during short per-
iods (lasting a few days) when desert winds push the smog over and past
the location out to sea, to return later when the wind shifts to an
on-shore direction. The area is mainly light industrial and agricul-
tural, with a major freeway and residential areas upwind.
Fig. 3 shows four distributions measured (with the 6-stage cascade)
in the morning and afternoon of a day without any smog transported from
Los Angeles. The time of day, brief weather description, the total
particulate mass concentration and the mass mean diameter are shown on
the figure.
The first distribution, taken in the early morning, appears to be
log-normal, probably representing fairly clean condition (109 ,/ug/m )
left from the previous night, with a fairly small mass mean diameter of
0.19yum. As the day warmed up and a light wind developed, the distri-
bution showed a shift to the right, with the trend continuing into the
early afternoon, as seen in the distributions taken at 1120 and 1345.
The 1345 distribution showed a very large mass mean diameter of 1.5 >'m,
almost an order of magnitude larger than that of the early morning dis-
tribution. As a cooling trend developed in the late afternoon, accom-
panied probably by both locally generated and transported photochemical
smog, a distribution suggestive of a bi-modal character developed, with
a mass mean diameter of O.SO^m.
Fig. 4 shows two distributions taken three hours apart on a day
that saw a very visible and narrow band of smog that had been out at sea
sweep over the measuring station around noon. The change from a bi-
modal distribution with equal peaks in mid-morning to a bi-modal one
with a much higher sub-micron peak is quite dramatic.
In measurement of indoor air, with presumably a large population
of sub-micron particulates, a cascade with a stage between the last two
(at 0.32 and 0.06/im) of the 6-stage cascade was used. This was a 10-
stage system with one stage at 0.17>«m. The result of sampling the air
inside an instrumented van is shown in Fig. 5. The van had filtered
and air-conditioned atmosphere; but smoking was allowed. The result
was a very narrow distribution with a mass mean diameter of 0.15 -um and
a total concentration of 282 >tg/m3. The filter system apparently allow-
ed the very small aerosols from cigarette smoke to remain in the air.
770
-------
o
<\
«o
6
a
o
S
< 2
cs
c>
UJ
00
o
0£
01
X
a.
O
(A
§
J-l
4-1
OJ
O
O
T3
4-1
O
c
•1-J
(3
O
•H
-U
CO
•H
O
at
N
•H
CO
3
U
•H
4J
(-1
C9
to
•H
771
-------
w z
o o
i §
2 oo
LL «/>
(U
Cu
co
I
i
CO
C
O
4-J
CO
•H
O
0)
N
•H
0)
4J
tfl
O
•H
-------
I
I
(/)
z
-^- s
Q.
O
-o
q
o
q
o
S
ft
05
i
.u
M
"d
0)
C
•H
a
o
•H
w
•H
(U
M
•H
d
a
•H
4J
oo
-H
773
-------
CONCLUDING REMARKS
The particulates impacted on the crystal remain there and even-
tually build up to such an extent (with many tens of layers of parti-
cles) that the response of the crystal to further accretion of mass
becomes non-linear. The limits of linear response are established
experimentally. In the case of the cascade it has been found that the
final stage (with 0.05/im cut-off) can accommodate a frequency change
of 200 Hz before going non-linear. Other stages can accommodate more
than 200 Hz. In the examples shown above the total frequency shift
in the final stage per sampling run lasting long enough to yield ac-
curate data was about 4 Hz. Thus, with a 200 Hz limit on total mass
accumulation, 50 sampling runs can be made before it is necessary to
change the sensing crystals. It is therefore possible, with one
change of crystals (which requires about 10 minutes for a 10-stage
cascade), to obtain four distributions per hour over a 24-hour period.
Preparations are under way to install a 10-stage cascade in an
aircraft for plume measurements. With a plume concentration of se-
veral hundred >ig/m it will be possible to obtain a good sample in
about 30 seconds, which is about the time required to traverse a plume
at about one mile from the source. Other applications include the
characterization of particulates in critical areas of a ship (such as
the engine room) to develope fire warning criteria, and the monitoring
of particulates at 1-hour intervals in a submarine to develope air
quality data under prolonged cruise conditions.
774
-------
REFERENCES
Chuan, R. L. An Instrument For the Direct Measurement of Particulate
Mass. J. Aerosol Sci 1, No. 2, May 1970.
Chuan, R. L. Application of An Oscillating Quartz Crystal to Measure
the Mass of Suspended Particulate Matter.
In: Analytical Methods Applied to Air Pollution Measurements, Stevens,
R. K., and Herget, W. F. (ed.) Ann Arbor, Ann Arbor Science Publish-
ers, Inc., 1974. p. 163-189.
775
-------
IDENTIFICATION AND MEASUREMENT OF PARTICULATE TRANSPORT PROPERTIES
Donald L. Fenton
IIT Research Institute
10 West 35th Street
Chicago, Illinois 60616
ABSTRACT
The continuum approach to particulate flow in the wake region behind a
circular cylinder is utilized to identify and experimentally evaluate
the appropriate transport properties. The criteria of a dilute sus-
pension is applied which allows separation of the gaseous phase equa-
tion of mean motion from the particulate equation of mean motion. This
type of unrestricted turbulent flow is also found in jets and within
boundary layers on the free side. Whereas the gaseous wake flow in the
fully developed region can be characterized by a similar solution, the
particulate flow cannot because of the nonlinearity of the gaseous-
particulate transfer term. Neglect of this term in the governing equa-
tion of motion for particulate flow permits a solution analogous to the
gaseous flow — Gauss' function.
Measurements within the particulate wake include particulate velocity
and particulate density (mass concentration) where the wake centerline
defects persisted far downstream before fully developed conditions were
established. Distributions of the particulate variables transverse to
the wake became Gaussian at the downstream distance where the gaseous
flow is fully developed. The experimental system consisted of a closed-
loop air flow facility and the particulate material was composed of
glass beads (count mean diameter = 11 microns). Knowledge of the par-
ticulate transport properties, especially their relation to the gaseous
transport properties, is useful in calculations applied to particulate
control devices.
777
Preceding page blank
-------
NOMENCLATURE
b half width of wake
C constant
Du hydraulic diameter of duct
H
d diameter of circular cylinder
E y-component of the electric field
F time constant for momentum transfer between the particle and gas
(N ) electroviscous number based on d
eVd
(N ) gas-particle momentum number based on d
md
(N ) Reynolds number based on d
Rd
(-*•) charge-to-mass ratio
tn
P
u x-component of the mean gaseous velocity
u.. maximum deviation in gaseous velocity from u^ within gaseous
max wake
u maximum deviation in particulate velocity from u within
1 particulate wake p
max ^
v y-component of gaseous velocity
x free stream flow direction
x geometrical origin of similarity of the wake
y transverse flow coordinate
Greek
c eddy diffusion coefficient for the particulate density
e eddy diffusion coefficient for the gaseous momentum
778
-------
e eddy diffusion coefficient for the particulate phase momentum
tup
H similarity parameter
£ non-dimensional distance downstream from circular cylinder
p gaseous density
p particulate density
lp density of the particulate material
p maximum deviation in particulate density from p m within
1 particulate wake
max r
p* ratio of particulate density to gaseous density at free stream
conditions
T shearing stress of the gaseous phase
T shear stress due to particulate collisions with the wall
P
$ volume fraction occupied by the particles
Subscripts
1 deviation from free stream conditions
<» free stream condition
g gaseous phase
p particulate phase
779
-------
IDENTIFICATION AND MEASUREMENT OF PARTICULATE TRANSPORT PROPERTIES
Donald L. Fenton
IIT Research Institute
10 West 35th Street
Chicago, Illinois 60616
INTRODUCTION
Many studies have been made concerning the behavior of dilute particu-
late suspensions in gaseous media. One motivation for this interest is
the multitude of particulate control and instrumentation devices in use
at present. In spite of this, however, little work has been expanded
on the determination of the appropriate eddy diffusion transport coef-
ficients. Determination of these transport coefficients for various
flow systems and particulate materials would greatly improve the pre-
diction of overall flow behavior. Also, the length of conduit required
for the establishment of fully developed conditions could be more easily
determined.
The continuum approach to particulate flow is utilized here where the
ordinary plane two-dimensional wake is studied. This approach accounts
for the separation of particulate and gaseous streamlines, relative slip
between the particulate and gaseous phases, electrical field and charg-
ing effects, and interactions between particles and gas, mutual parti-
cles, and particles and gas with a solid boundary. For dilute suspen-
sions, this continuum approach has been demonstrated to be valid1'2 .
Taking p as the particulate cloud density and "p as the density of the
particulate material, the volume fraction $ is given as PD/T^D. Soo
indicates that 0 must be less than 0.05 for the existence of a dilute
suspension (interparticle spacing > two particle diameters). Recently,
successful application of the continuum approach has been made concern-
ing conduit flows and other specialized geometries3'1*^.
In the study reported here, the ordinary two-dimensional wake (generated
by a circular cylinder) is investigated with emphasis on the transport
coefficients. The particulate momentum and density eddy diffusion
transport coefficients are compared to the gaseous momentum transport
coefficient. The actual determination of the transport coefficients is
based on the measured wake profile for the appropriate variable.
THEORETICAL BACKGROUND
The coordinate system used in conjunction with the conservation equa-
tions is given in Figure 1. The downstream distance along the wake axis
780
-------
u
b
Figure 1. COORDINATE SYSTEM FOR THE PLANE
TWO-DIMENSIONAL WAKE
781
-------
is the x-coordinate and the upward vertical direction is the y-coordin-
ate. The center of the circular cylinder is the origin of the coordin-
ate system.
In light of the experimental conditions and following the continuum ap-
proach to particulate flow as given by Soo1, the steady state continuity
relations for the gaseous and particulate phases are
9u 9v
(1)
p p
-
3y
(2)
under the constraint of no phase transformations and no internal gen-
eration. The momentum equations for the gaseous and particulate phases
in the x-direction are after simplification
pu H. + pv " = _KFp (u-u ) +
3x 9y p p 3y
(3)
3u 3u
p u -r-2- + p v -^- = Fp (u-u ) + p E (-3-) -f
p p 3x p p 3y p p p x m 3y
(4)
The velocities u and v are to be interpreted as mean velocities for
turbulent flow. The constitutive equation for T has not been specified
and therefore, the gaseous momentum equation is not restricted to lam-
inar flow. Because the main flow velocity is large relative to the
transverse velocity in wake flow, the y-component of fluid momentum is
neglected. The pressure distribution in the main flow and throughout
the wake was essentially uniform. The effectiveness parameter, K, in
equation (A) incorporates the possibility of irreversible momentum
transfer from the particulate phase to the gaseous phase.
With the assumption of a dilute suspension, equation (2) in combination
with Pick's Law becomes
3P
P _.
-T -)-
3x
3P rs
P 9
- 1- =: _ -
3y 3y
m
V-
F
3y
cp 3y
(5)
where mass diffusion in the x-direction is taken to be much less than in
the y-direction. Equation (5) includes diffusion by an electric field
force and the transport coefficient, £CD> takes into account eddy dif-
fusion of particulate density.
782
-------
A similarity solution can be obtained for the gaseous phase and is re-
ported in the literature6. The interparticle spacing is sufficient to
prohibit the particulate phase from affecting the motion of the gaseous
phase. Therefore, the effectiveness parameter is very nearly zero and
gaseous motion is uncoupled from the particulate phase motion. Boundary
conditions for the gaseous phase are as follows
0— =0
ay
v=0
(6)
(7)
In seeking a similarity solution for the wake profile, a similarity
parameter is defined as
n =
and
(8)
where x is the distance from the origin of the coordinate system to
the geometrical origin of similarity and d the diameter of the circular
cylinder. An additional constraint on the gaseous motion is
DRAG
(9)
which also must be satisfied. In the process of identifying the trans
port coefficient, the following assumption is made
= PE -£
xy rag 9y
(10)
where £ is the coefficient of eddy viscosity for the gaseous phase.
The resulting solution is
max
u d
ndn
mg
(11)
where n equals y//d(x + x ), £ is taken as a sealer, and u is the
deviation between the velocity in the wake and the free stream velocity.
The integral can be evaluated if E is prescribed and for simplicity is
rno
taken as a constant. Carrying out the integration yields
783
-------
u d
= exp
4e
(12)
max
mg
which is the Gauss function.
An analogous situation exists — similarity at sufficient downstream
distance — with the particulate phase if the following postulation is
made7
)n)
f
(13)
where C and n are defined in like manner to the gaseous phase and C
is a constant. A condition of quasi-similarity for the particulate^
phase can then be formulated
(14)
With the above postulation, the solution for u proceeds as for the
gaseous phase. The result is
= exp
u d
4e
mp
(15)
max
where the integration is performed under the condition that e is a
constant sealer.
mp
The transport of particulate density in the wake region can also be
obtained. With the assumption that the transport of p is analogous to
the transport of gaseous momentum,
- vp
= C
cp 3y
(16)
and taking e as an unprescribed sealer, similar wake profiles of p
should exist. If e is taken as a constant, Gauss' function is again
obtained
784
-------
u d
= exp
max
4r
(17)
cp
where Pp is the mean deviation between the wake region and free stream
particulate density. Also, if the ratio c /£ is constant throughout
the wake region, then mP CP
max
e
mp
e
u
Pl
u
PI
max
cp
(18)
EXPERIMENTAL APPARATUS AND MEASUREMENT RESULTS
A recirculation-type wind tunnel 12 inches square was used to generate
the turbulent suspension. The upstream distance from the test section
was over 40 hydraulic diameters to insure fully developed gaseous flow
upon generation of the wake. The Reynolds number based on the hydrau-
lic diameter varied between 5.6 x 10 and 6.4 x 10 for the tests.
A circular cylinder placed transverse to the flow generated the two-
dimensional wake. Circular cylinders were used because this geometry
occurred most often in the literature to generate an ordinary symmetri-
cal wake. Comparison directly to the gaseous flow literature is
facilitated. To eliminate effects of periodic flow, the Reynolds num-
ber based on cylinder diameter was greater than 1,000.
In the determination of gaseous flow, a modified Prandtl pitot-static
tube was employed. The gaseous phase requires only the measurement of
velocity for characterization. Figure 2 gives typical results for the
gaseous wake at various distances downstream. Similarity is observed
to exist for downstream distances greater than x/d = 40. The curve
drawn on the profile is a result of selecting the appropriate e which
best satisfies the experimental data. The central portion of the gas-
eous wake is therefore seen to behave as predicted by the gradient-type
diffusion concept of transport. Agreement at the outer wake boundary
is not good and is due to the reduced coefficient of eddy diffusion
near the boundary. Also important is the coalescence of the wake pro-
files for all Reynolds numbers in the region of similarity which, of
course, indicates no peculiarities in the flow about the circular
cylinder.
785
-------
x*« 40
O x*- 80
D X*=I60
O x*=80, Cylinder In Vertical
Position
Figure 2. GASEOUS WAKE PROFILE AT (N ) = 3.42 x 10"
d
786
-------
Three quantities must be measured to fully characterize mean motion of
the particulate phase: p , Ppu , and (q/m) . The electrostatic term
was neglected in the solution or equation (^). Experimental justifica-
tion (measurement of charge-to-tnass ratio) is provided by the dimension-
less form of the equation of motion2 and typical values for the di-
mensionless groups are
(N ) % 10~5 and (ND) *> 103
evd Rd
In the dimensionless particulate diffusion equation, electrostatic ef-
fects can also be neglected because
(N )2 (N ) ^ 10'11 and (N) ^ 103
6Vd md Rd
7 R
With the measurement techniques utilized ' , the particulate velocity
was not measured directly but through combination of the particulate
mass flux and particulate density measurements. Essentially, the local
mass flux divided by the particulate density at the same location yields
the particulate velocity. Consequently, no information is obtained con-
cerning the velocity of different particle sizes.
The particulate material was Ballentine Blast Beads (specific gravity =
2.65) that were bimodally distributed with the modes at 3 and 35 ym.
Microscopic investigation determined the count mean diameter to be 11 ym
and showed the particles to be 95% spheres and 5% platelets. Figure 3
indicates the duct centerline particle size distribution for the experi-
mental conditions.
Figure 4 shows the particulate velocity defect for conditions corres-
ponding to the gaseous momentum data already given. In all the data ob-
tained, n serves as an effective similarity parameter. This is not
surprising because the gas imparts momentum to the particles through
drag effects and therefore the particulate momentum distribution should
be approximately the same shape. The curve drawn through the data is
the apparent coefficient of eddy diffusion for particulate momentum.
Also, note that for the dilute suspensions employed, the distribution of
particulate momentum is not significantly altered by the free stream
particulate density. The two curves drawn reveal the actual influence
of the free stream particulate density on the particulate momentum wake
profile.
Particulate density defects within the wake region were also distributed
according to Gauss' function and again, r\ served as an effective simi-
larity parameter. Figure 5 indicates data obtained corresponding to
787
-------
Q)
e
Q>
Q.
50
40 -
30
1 20
10
_L
co
20
40
2a
« 59.3 fps
= 88.6 fps
60
Figure 3. PARTICLE SIZE DISTRIBUTION AT DUCT CENTERLINE
80
0.4 -
0.2 -
A X =40
O X+= 80
n X*=I20
A X*= 20
A X*=40
* X*= 80
• X*=I20
/>• = 0.00657
/>* = 0.00657
/>* = 0.00657
/,» = O.OI55
/>£ = O.OI55
/»* = 0.0155
P* = 0.0155
0.8
1.0
Figure 4. PARTICULATE VELOCITY WAKE PROFILE AT (ND) = 3.42 x 10"
Rd
788
-------
0.00657
0.00657
Figure 5. PARTICULATE DENSITY WAKE PROFILE AT (N ) = 3.42 x 1(T
K A
789
-------
gaseous momentum in Figure 2. Also, the free stream particulate density
Slightly influences the distribution of thu particulate density defect.
The curves drawn indicate the magnitude of this influence on the partl-
culate density wake profile.
DISCUSSION OF RESULTS
Table 1 gives results for the eddy diffusion transport coefficients ob-
tained for all the experimental conditions. Free stream values of the
gaseous eddy diffusion coefficient, em, are normalized and constant over
the experimental conditions. The values for em were obtained from
measured duct velocity profiles. The eddy diffusion coefficient for
gaseous momentum is also normalized and varies with the free stream
velocity.
Table 1. EDDY TRANSPORT COEFFICIENTS
(Vd
2.40xl03
3.42xl03
3
3.42x10
6.93xl03
6.93xl03
5.12xl03
5.12xl03
5.12xl03
u
(m/s)
12.7
18.1
18.1
18.3
18.3
27.0
27.0
27.0
d
(cm)
0.318
0.318
0.318
0.635
0.635
0.318
0.318
0.318
0.
0.
0.
0.
0.
0.
0.
0.
00
0141
00657
0155
00657
0157
0106
0146
0232
1
1
1
1
1
1
1
1
2e
m
u Du
OO JJ
.41xlO~4
.41xlO~4
-4
.41x10
.41xlO~4
.41xlO~4
.41xlO~4
.41xlO~4
.41xlO~4
£
U
0.
0.
0.
0.
0.
0.
0.
0.
mg
d
OO
0481
0481
0481
0418
0418
0443
0443
0443
e
mp
e
mg
0.489
0.469
0.489
0.608
0.653
0.551
—
0.616
e
cp
e
mg
0.632
0.675
0.698
0.804
0.830
0.686
0.747
0.783
t
mp
Lcp
0.773
0.695
0.700
0.755
0.786
0.803
—
0.787
The two particulate transport coefficients are normalized by the gaseous
momentum eddy diffusion coefficient for wake flow. The particulate
momentum eddy diffusion coefficient is roughly half the eddy diffusion
coefficient for gaseous flow. Values are also given for the apparent
coefficient of eddy diffusion for particulate density in Table 1. The
diffusion of particulate density is seen to occur more readily than
particulate momentum for the same wake structure. Also, the ratio of
eddy diffusivities for the particulate momentum and density for the
various experimental conditions are calculated. Within experimental
error, the ratio e /£ is constant (about 0.76) which justifies
equation (18) determined from the integral forms of particulate momentum
790
-------
and density defect distributions.
Most of the symmetrical wake studies reported use circular cylinders for
wake generation because at the downstream distances where similarity is
established, object geometries are not significant. This turbulent wake
type of unrestricted shear flow can also be found in jets and within
boundary layers on the free side. Although the specific results of this
work are unique to the particulate suspension investigated, significant
information is yielded concerning the relative magnitude of the three
transport coefficients.
REFERENCES
1. Soo, S. L., Fluid Dynamics of Multiphase Systems, First Edition,
Blaisdell Pub., Mass., 1967.
2. Soo, S. L., Dynamics of Charged Suspensions, International Reviews
in Aerosol Physics and Chemistry, V.2 (ed. Hidy, G. M., and
Brock, J.), Pergamon Press, Oxford, 1971.
3. Soo, S. L., and Tung, S. K., Appl. Sci. Res., 24:83-97, June 1971.
4. Stukel, J. J., and Soo, S. L., Powder Technol., 2:278-289, 1969.
5. Ramadan, 0. E., and Soo, S. L., Physics of Fluids, 12:1943-1945,
September 1969.
6. Hinze, J. 0., Turbulence, McGraw-Hill Book Company, New York, 1959.
7. Fenton, D. L., Ph.D. Thesis, Dept. of Mech. Engr., Univ. of 111.,
1974.
8. Soo, S. L., Stukel, J. J., and Hughes, J. M., Environmental
Science and Technology, 3:386-393, 1969.
791
-------
OPTICAL AEROSOL SIZE SPECTROMETRY BELOW AND ABOVE THE
WAVELENGTH OF LIGHT - A COMPARISON
J.GEBHART/ J.HEYDER/ C.ROTH/ W.STAHLHOFEN
Gesellschaft fiir Strahlen- und Umweltforschung m.b.H.
6 Frankfurt/Main, Paul-Ehrlich-Str.21, Fed. Rep. Germany
ABSTRACT
A critical review of the abundant literature on scattering
of light on small particles revealed that different light
scattering devices should be used in order to adjust their
response to the light scattering properties of the partic-
les. Below the wavelength of light laser light illumination
and a mean scattering angle in the nearer forward direction
are preferable. Such an instrument is the LASS (laser aero-
sol size spectrometer)with a sizing range between 0.05 and
0.7 p.m. Above the wavelength of light white light illumi-
nation and low angle scattering is more advantageous. The
LASI (low angle scattering instrument) which has a sizing
range between 0.7 and about 6 p.m serves for this purpose.
The performance of these instruments with respect to par-
ticle diameter, refractive index and shape has been checked
with monodisperse polystyrene spheres, condensation drop-
lets and solid Fe^-.-particles as well as agglomerates of
polystyrene spheres; It turned out that the response of
the LASI is a function of the projected area of the par-
ticles independent of the optical properties and shape of
the particles. The LASS measures directly the volume equi-
valent diameter of a particle regardless of its shape.
However, the response of the LASS depends on the refractive
index of the particle material. The response curves for
different refractive indices differ more or less by a con-
stant factor.
1. INTRODUCTION
Particle sizing by means of light scattering on single par-
ticles is known since more than 25 years. Meanwhile the sub-
ject has been steadily developed and since about 10 years
optical particle counters using white light illumination
are commercially available. After the invention of the laser
793 preceding page Wank
-------
principle several attempts have been made to replace the
white light illumination of scattering devices by coherent
and monochromatic laser light illumination. Although all
instruments measure particle size distributions of aerosols
they have been regarded as optical particle counters. Since
the sensitivity as well as the resolution power of light
scattering devices have been improved considerably by using
laser light illumination, recent developments utilizing
lasers have been regarded as opticle size spectrometers.
Their performances will be reviewed in this paper. It will
be also part of this paper to discuss the advantages and
disadvantages of both types of illumination for particle
size analysis.
To adjust light scattering devices to the light scattering
properties of small particles it has been proved advisable
to use two different optical arrangements because of the
different optical behaviour of particles smaller or larger
than the wavelength of light. For the size range below the
wavelength a laser aerosol size spectrometer (LASS) has
been developed in this laboratory. Above the wavelength
a low angle scattering instrument (LAST) with white light
illumination has been applied. The performance of both
instruments with respect to particle diameter, refractive
index and shape will be discussed based on experimental
results.
2. WHITE LIGHT OR LASER LIGHT ILLUMINATION?
The outstanding characteristics of laser light are the high
stability of the output, the degree of spatial and temporal
coherence and the high flux density.
The high stability renders unnecessary a reference light
source.
The spatial coherence or the low beam divergence leads to
a high intensity at the focal point of an optical system
and therefore the need of parabolic mirrors and compli-
cated lens systems is eliminated. In the case of focal
plane illumination the intensity of a typical 5 mW
He-Ne-laser radiation is 4 to 5 orders of magnitude
greater than the intensity from the most intense inco-
herent light source. Therefore the sizing range of usual
optical particle counters can be extended by utilizing
lasers. Since due to its spatial coherence a laser beam
can be focussed to a very tiny, diffraction limited
794
-------
Spot the dimensions of the sensing volume can be kept
very small. For,this.,reason aerosols of high concentra-
tions (up to 10 cm" ) can be analysed with negligible
coincidence losses.
The temporal coherence or monochromacy on the other hand
is advantageous for instruments analysing the angular
dependence of the scattered light, but is a drawback for
particle counters, since in the size range above the
wavelength the response curves of monochromatic counters
exhibit oscillations. These fluctuations can be smoothed
over by applying white light and a large collecting aper-
ture, however the smoothening of white light is more
effective. The application of only a large collecting
aperture is not sufficient since for monochromatic light,
even for an aperture of 4/T , oscillations occur as can
be seen from the extinction coefficient of transparent
spheres (Hodkinsonl, Quenzel^).
It has to be mentioned in this connection that the disad-
vantage of an oscillating response curve can be partly over-
come by a new evaluation method. By this method a differen-
tial size distribution spectrum of a polydisperse aerosol
of spherical particles and known refractive index is re-
corded with the instrument. The peaks in the spectrum are
then related to the turning points of the theoretical cali-
bration curve and act as internal size references (Sch6ck3 ,
Bakhanova and Ivanchenko^). But the resolution power of
this method is poor. If growing droplets are illuminated
with laser light and the scattered light is observed under
a fixed angle the response of the instrument as function
of time shows also characteristic fluctuations which can be
correlated with particle size. By this way the growth rate
of a droplet due to condensation can be determined
(WagnerS).
The influence of monochromatic and white light upon the
calibration curve of an optical counter is demonstrated
by the following graphes. Detailed calculations in this
field have been made by Hodkinson and Greenfield^, Quenzel^,
BroBmann'and OesebergS. in Fig. 1 and 2 the partial scatter-
ing cross-section as function of the particle size for mono-
chromatic and white light and a mean scattering angle of
45° is shown. For particle sizes larger than the wavelength
light scattering can be considered as a surface effect
and outside the diffraction lobe the mean scattered light
flux depends on the square of the particle diameter. For
795
-------
n i i i—i—SJM 11 i i—i 111111 i i—r
i i i i
rrr
u i r
E
>0
o
o
w
o
M-l *W
o
tn
en M
cu cu
> -P
d 3
u o
u
cu
tn cu
C rH
O O
en
cu
CO
H -P i
(0 X! m
ft tT"~~
CU
> rH
•H (C ^
-P U 0) CO
(0
II
•H -P
-P -H
C
(0
o
en
C
•H
T)
O
ffi
O
-P
• -H CU
T3 'H
5-1 *4-4
O C
o cu
o cu
cu ax: o us 5-4
a o 5 co *-' o
CM
S-l
o
E
I
CO
w M
0) 0)
> -P
n c
V 3
O O
u
w
W CU
C M
o u
(VH
CO ^>
CU M
00
CU
>
-H
-P
fO
D4
-H c
rH •-
U ^.
•H ro
-P r-
ffl •
I?
C
C
m
e
02
O
M
CQ
O
-P
C
•H
rM **~^ M
O CO O
o u
C II U
O O
-------
monochromatic light the curves shpw the typical oscilla-
tions. In the range below the wavelength a d6-dependence
is observed and the curves are straight lines even for
monochromatic light. Fig. 2 shows the smoothening effect
due to white light above the wavelength. In Fig. 3 the re-
lative response curve for a receiver aperture between 2.5°
and 5.5° is plotted against the particle diameter,d. For
monochromatic light and transparent spheres the response
curve exhibits periodical fluctuations with 5 maxima be-
tween 1 and 6 p.m, which can be smoothed over by white light.
(see chapter 6) .
CONCLUSIONS: It has been shown that for size analysis of
particles bigger than the wavelength white light illumi-
nation is advisable in order to avoid ambiguous response
of the instrument. For size analysis of particles smaller than
the wavelength either laser or white light illumination
can be used. However, laser light illumination is preferable
in order to achieve a higher sensitivity and resolution
power of the instrument.
3. OPTICAL INSTRUMENTS USING WHITE LIGHT
A critical discussion of the performance of optical instru-
ments using white light exceeds the limits of this paper
Therefore, the reader is refered to the following papers
describing the instruments: Gucker et al.9, ZinkylO,
Sinclair 11, Martens and Fuss12, Turpin13. Because most of
the instruments are nowadays commercially available their
performance has been thoroughly studied in several aerosol
laboratories: Whitby and Vomela14, Jaenicke15, Williams and
Hedleyl6, Liu et al. 17.
4. OPTICAL INSTRUMENTS USING LASER LIGHT ILLUMINATION
According to the special properties of laser light the op-
tical setups can be devided into three groups:
i. Instruments with the sensing volume inside the op-
tical cavity of the laser.
ii. Instruments detecting the angular intensity function
of scattered light.
iii. Instruments with a fixed mean scattering angle and
collecting aperture including the solid angle 4ft*.
ad i:
Schleusener1^ suggested recently a new approach to the de-
tection and sizing of aerosol particles. In this method,
797
-------
E
u
to1-
w6-
A= 0.6328 ^im
n =/.5
i *n =1.75-0.08751
t—*n =2.0-0.51
d.jum
T
5
-r
6
0
-r
2
Fig.3. Theoretical response curves of the LASI
solid line : transparent spheres
broken line: absorbing spheres
798
-------
each particle passing through the laser cavity,creates a
cavity disturbance that results in a pulse shaped decrease
in the laser output. Individual particles can be counted
up to 1O^ per second and classified according to their
light extinction. The light extinction is amplified by
about a factor of 100 inside the laser resonator. Lateron
Schehl19 tried to minimize the sensing volume by passing
the particles through the focus of one of the resonator
mirrors and to increase the sensitivity by measuring the
scattered light instead of the extinction. But neverthe-
less resolution and measurable size range is poor compared
to the other optical counters. Although it is not mentio-
ned the response of the device must be disturbed by par-
ticles passing not the sensing volume.
ad ii:
Instruments determining the whole scattering diagram of a
particle with a small collecting aperture (Phillips and
Wyatt20, Wyatt and Phillips21, Cooke and Kerker22, Gucker
et al.2 , Moser2^, Maier2-") produce more" information about
the properties of the particle. These devices are useful
for the exact determination of the refractive index or the
diameter of a particle with an absolute accuracy comparable
with that of electron microscopy. They are applicable for
spherical particles in the size range from about 0,3 to
10 |_im. The step by step detection of the angular dependence
of the scattered intensity, however, is time consuming.
Furthermore the immense amount of information has to be
processed by a computer to obtain the best fitting of the
experimental diagrams with the theoretical predictions.
The most skillful work on this subject has been done by
Gucker et al. . Their high speed photometer detects the
scattering diagram of single aerosol particles from 7°
to 353° in as little as 14.4 ms. Although this instrument
has a high speed, the concentration of the aerosol must be
lower than 10 cm~3 in order to prevent coincidences. All
other devices have a much slower recording system.
ad iii:
Optical particle counters with a fixed mean scattering
angle and collecting aperture can be further divided into
two types. The first type of instruments detects with a
moderate collecting aperture the light scattered in the
forward direction. The optical design is realized with dark
field illumination for the low angle scattering instruments
and with an asymmetrically arranged collecting aperture for
bigger mean scattering angles.
799
-------
Although these instruments have quite different designs,
they all attempt to minimize the sensing volume. For the
low angle instruments the coherent radiation is absolutely
necessary for the exact separation of illuminating and
scattered light. Useful measurements can be made at angles
as small as 1.5°.
In the second type of instruments a maximum quantity of
scattered light from all directions is collected in order
to extend the sizing range of the instruments to smaller
particles. If the collecting aperture is enlarged, however,
the background noise - due to light scattering on gas mole-
cules and on structural parts of the instruments - increa-
ses in the same way as the radiant flux through the aper-
ture due to light scattering on the particle. An enlarge-
ment of the collecting aperture, therefore, does not result
in a lower detection limit.
Low angle light scattering instruments with small sensing
volumes and fairly high resolution powers are described
by Kaye^G anci Gebhart et al. ^. The instrument of Gebhart
et al. can be used for particle size analysis between
0.15 and 1.2 jim. For larger particles the response curve
becomes ambiguous. The testing procedure of the instrument
of Kaye is still incomplete.
Another optical particle counter with dark field illumina-
tion having a collecting aperture from 7° to 12° has been
developed by Hill _ . This high resolving instrument is
designed for high concentrated aerosols, but limited to a
size range between 0.15 and 1 jam. Instruments detecting
the light scattered in the forward direction by means of an
asymmetrically arranged receiver aperture have been repor-
ted by Bol et al.29 ancj pinnick et al.30. The latter publi-
cation gives not much information about the actual per-
formance of the device since the proof of the light scatter-
ing theory was given preferance.
The counter described by Bol et al. 29 was designed for
highly concentrated aerosols in the size range below the
wavelength of light; but the lower detection limit of this
instrument is already met for 0.15 jam particles.
The most popular instrument which detects the light scat-
tered in nearly the solid angle 4 7f is the Jacobi-counter
(Jacobi et al.31) using a lucite cylinder as light collec-
ting device. A copy of this instrument is described by
800
-------
Walsh . These counters have the discussed disadvantages of
non-monotonous response curves for particles in the micron
size range. The smallest detectable particles are about
0,3 jam. Because the laser beam is not focussed the sensing
volume is relatively large and the measurable concentra-
tions are low, but nevertheless the resolving power of the
instrument is quite excellent. Lioy et al.^3 tried to in-
crease the sensitivity of their particle counter by collect-
ing the scattered light from all directions by an ellip-
soidal mirror and achieved a lower detection limit of about
0,2 |im.
5. RESPONSE TO PARTICLES OF DIFFERENT OPTICAL PROPERTIES
AND SHAPES
5.1. Definitions
To analyse the influence of the optical properties and the
shape of a particle upon the response of a light scatter-
ing device it is useful to introduce the definitions
(Heyder et al.34) given in Fig. 4. In these diagrams the
partial scattering cross-section, S, is drawn against the
geometrical diameter, d, of the calibration spheres. If the
aerosol to be investigated consists of spherical particles
with the same refractive index, nc, as the calibration
spheres the geometrical diameter, d, is measured. For a
spherical particle with a refractive index, n * nc, an
equivalent light scattering diameter, dsce, is determined.
For a non-spherical particle with refractive index, nc, the
light scattering diameter, dsc, is obtained. The influence
of the particle shape may be described by an optical shape
factor, w, also defined in Fig. 4.
5.2. Particles larger than the wavelength
For particles larger than the wavelength light scattering
becomes more and more a surface effect. The incident light
entering the particle is scattered due to refraction. This
component dominantes in the angular range up to about 100°
except within the forward scattering lobe of diffraction.
For particles of absorbing materials the part of light
entering the particle is absorbed. Instruments which mainly
collect the light scattered due to refraction, therefore,
show great differences between transparent and absorbing
materials. This can be seen from Fig. 1 and Fig. 2. An ex-
perimental verification of this behaviour using commer-
cially available particle counters has been given by Whitby
801
-------
isotropic calibration sphere
nc : refractive index
d ; geometric diameter
n 4 n.
d : geometric diameter
d . : equivalent volume diameter
dve dsc
dsce: equivalent light scattering
diameter
dsc : light scattering diameter
S(dscl
Sldvel
optical shape factor
Fig.4. Definitions used in optical particle size spectro-
metry
and Vomela"14 for particles of india ink and by Liu et al.3^
for monodisperse coal particles.
Within the forward scattering lobe of diffraction the scat-
tered light intensity is a function of the projected area
of the particle independant of its optical properties and
its shape. This can be seen theoretically in Fig. 3 and will
be confirmed experimentally (see chapter 6). The oscillati-
ons of the response curve for transparent spheres are caused
by interferences of light diffracted and refracted on the
particle. They can be smoothed over by white light. For
opaque materials the refracted component is absorbed inside
the particle and no interference occurs (see Fig. 3).
5.3. Particles smaller than the wavelength
For particles smaller than the wavelength light scattering
802
-------
becomes more and more a volume effect. Calibration curves
for different real parts of refractive index tend to run
parallel in a double logarithmic plot (Fig. 5). That means,
that these curves differ more or less only by a constant
factor which is a function of the real part of the refrac-
tive index. For absorbing materials with a complex refrac-
tive index, n = n-ik,additionally the influence of the
k-value is reduced considerably. As can be seen from Fig. 6
all curves of the same real part of refractive index cross
over for particle diameters below the wavelength.
As long as light scattering is a volume effect the scatter-
ing cross-section of a particle becomes mainly a function
of its volume and less of its shape. This can be shown
theoretically by applying the light scattering calculations
to spheroids which are smaller than the wavelength (van de
Hulst ). Consequently, a light-scattering device below
the wavelength measures the equivalent volume diameter,
dve» °f a non-spherical particle (see Fig. 4). An experi-
mental verification of this fact will be given in chapter 7.
6. MEASUREMENTS WITH THE LASI ABOVE THE WAVELENGTH
For particle size analysis in the size range above the
wavelength a low angle scattering instrument (LASI) using
the white light of a mercury lamp has been developed
(Gebhart and Roth37). The optical setup of the instrument
is shown in Fig. 7. The lens, LI, forms an image of the
mercury arc, Q, on the slit, D-\ , of cross-section
(0.04 x 0.20) mm2. D-] is imaged into the sensing volume by
the lense, L2; the magnification is 5 so that the beam
cross-section in the sensing volume is about (0.2 x 1 ) mm^.
By means of a hole in the mirror, M, the primary beam is
separated from the light scattered by the particles. The
primary beam vanishes in a light trap, T. The scattered
light is sampled by the lens, L3, which forms an image of
the particle onto the stop, 04. Behind this stop the scat-
tered light reaches the photomultiplier,PM,where the light
flashes are converted into electrical pulses. The electri-
cal pulses are linearly amplified and then classified in
a multi channel analyser, MCA. The primary beam has a semi-
angle of 1.5° and the aperture of the receiver is extended
from 2.5 to 5.5°. The whole primary beam passes a tank with
filtered air. By means of a constant pressure drop between
the tank and the aerosol inlet particles are sucked into
the sensing volume. The aerosol tube has an inner diameter
of0.500 mm. The sampling rate is about 100 cm-^ min~1 . By
803
-------
in
in 5-i
0)0)
5-1 C C
3 3 fd
O O
rj in
e
-1 in
oo C
CM O
n -H I
^D -P O
U PS
G
in a) M -P
0) xl
o o
Cn-P
•H
in -p m
d)
rH -H -P
O fd SM g
-HUGO
-P -H in 5-1
0) -p X! XI
fd
i-H
o
en u 0)
C -H -P O
C -P
•H
tn
M C
(0 -H
-
o o
QJ
x;
EH
u
o en o
M G
•woo
O M-I g
c u o
(C U O
(d X3
Cn
•H
U
GOO
in ri. in
in 5-i T! oo G
0) (D C CN O 'O
;> .p fd ro -H C
5-| C 1^ I I
-------
Q
filtered air
filtered air
Fig. 7. Set-up of the LAST
MCA
4<]
jj
1
Fig. 9. Set-up of the LASS
805
-------
use of a clean air jacket the aerosol stream is focused to
a filament of about 100 )-im. The pulse duration is a few us
and the effective sensing volume about 1 0~6 cm3 _ Thus con-
centrations of up to 10^ particles per cm^ are allowed
with counting rates of more than 10^ particles per second.
The experimental response curve of the instrument in the
size range between the lower detection limit of about 0.7|um
and the upper limit of about 6 urn is shown in Fig, 8. Three
kinds of monodisperse aerosols with different optical pro-
perties have been used for the calibration. To find the
geometrical diameters of the particles independent methods
have been applied. The size of polystyrene spheres was
checked by electron microscopy (Porstendorf er and Heyder-^")
For the aerosol of di-2-ethylhexyl-sebacate (DES) produced
in a Sinclair-La Her generator (Stahlhofen et al.39) size
measurements of the droplets were carried out in a sedi-
mentation cell (Stahlhofen et al.40). Fe2O3~particles
were produced by nebulizing a colloidal solution of
(Albert et al. ') by means of a spinning top(May^2). The
diameter of the dried solid particles was determined by
electron microscopy. In spite of the different optical
properties of these particles all measured points lie al-
most on one curve. No oscillations of the experimental ca-
libration curve are observed. The partial scattering cross-
section, S, depends on the third to fourth power of the
particle diameter, d. These results confirm the theoretical
predictions that light scattering within the forward lobe;
of diffraction is nearly independent of the optical con-
stants of the particle material.
The response of the instrument to non-spherical particles
has been investigated using aggregates of uniform polys-
tyrene spheres. Let dscj be the diameter of a sphere with
refractive index, nc , which has the same partial scattering
cross-section, S, as an agglomerate formed by j uniform
spheres with diameter, d-| , and refractive index, nc . Then
a relative light scattering diameter, F-; , of the aggregates
can be defined as (Heyder and Porstendorf er^^) :
Some relative light scattering diameters measured for
different diameters, d-|, of the single spheres are listed
in Table 1. As can be seen, F-j=jV2, so that dscj is identi-
cal with the diameter of the projected area of a non-
spherical particle. In other words, as long as a compact
806
-------
10
' ' ' I • ' • • I • I • I
' x polystyrene spheres n = l59
' O DES~ droplets n=M5
« ^e^ -particles n =25-ik
I I I I I I I
3 456
0.6 as ;
Fig.8. Experimental response curves of the LAST for par-
ticles of three different refractive indices
807
-------
d1 (urn)
1 .158
1 .83
average
values
1 r
F3 = ^
F2
1 .39
1 .41
1 .40
1 .41
F
1
1
1
1
3
.69
.75
.72
.73
Table 1: Relative light
scattering diameters, F-; ,
of aggregates of uniform
polystyrene spheres measur-
ed with the LASI.
particle larger than the wavelength is considered and the
scattered light is collected within the forward lobe of dif-
fraction the response is a function of the projected area
of the particle regardless of its shape. Outside the diffrac-
tion lobe, however, light scattering is effected by reflec-
tion and refraction at the particle surface. Then specular
reflections and internal reflexes can produce a light
scattering distribution which deviates considerably from
the scattering diagram of a sphere of equal projected area
(BroBmann7, Zerull44).
Finally it should be mentioned that the response of a pro-
jected area instrument depends on the orientation of the
particle in the sensing volume. It can be shown that even
for randomly oriented agglomerates of spheres those orien-
tations are more frequent which exhibit projected areas
being multiples of the cross-section of a single sphere.
Therefore separate peaks can be recorded for the different
agglomerates.
7. MEASUREMENTS WITH THE LASS BELOW THE WAVELENGTH
In the submicron size range (below the wavelength of light)
the laser aerosol size spectrometer (LASS) is used for par-
ticle size analysis (Heyder et al.45, Roth and Gebhart46).
The optical arrangement of the spectrometer is shown in
Fig. 9 as cross-section through the plane of observation.
The spectrometer is supplied with two light sources, Q, a
helium-neon laser ( \- 0.6328 (am) with an output of 7 mW
808
-------
and an argon ion laser (A= 0.5145 jam) with an output of
1.2 W. The laser light is focused by an astigmatic system
of lenses, L, into the sensing volume. The focus has an
elliptical cross-section of (0.02 x 0.12)
The light scattered by the particle is collected by micros-
cope objective, O, under a mean scattering angle of 40°
and is passed via a mirror, M, to a red sensitive photo-
multiplier, PM. By turning the mirror, M, the scattered
light is either reflected on the photomultiplier cathode
or into an eyepiece, E, for observation. The sensing volume
is imaged by the objective, 0, into a plane of exchangeable
stops, D. Behind the sensing volume the primary beam
vanishes in alight trap, T. The aerosol nozzle is directed
perpendicular to the laser beam and to the plane of ob-
servation. The width of the aerosol stream having an origi-
nal diameter of 0.2 mm is reduced by aerodynamic focusing
to a small filament of about 0.03 mm width. In this way
a sensing volume of only 1.5 • 1 0~8 Cm3 is obtained, so
that aerosols with concentrations up to 10 cm~3 can be
measured without coincidence losses. Under these conditions
a sampling rate of about 4 cm3min~1 is obtained. The output
pulses of the photomultiplier (~0.25 jj.s) are linearly am-
plified and then accepted by a multi-channel analyser, MCA.
Calibration curves, S (d) , of the instrument for two refrac-
tive indices are given in Fig. 10, where in a semi-logarith-
mic diagram S is plotted against the dimensionless size
parameter, \X = //"d/^_ . The corresponding diameters of the
particles for the two laser wavelengths are also drawn on
the abscissa. The diameters of the polystyrene spheres
with refractive index, nc=1.590, have been checked by elec-
tron microscopy (Porstendorfer and Heyder3^) . The experi-
mental values and the theoretical curve derived from light
scattering theory summarized by Kerker^ were matched for
d = 0.206 (am. It turned out that the experimental points
and the theoretical curve are almost in line. Where small
deviations occured the experimental results have been con-
sidered more reliable. Calibration curves for aerosols of
refractive indices, n jt nc , differ from the matched experi-
mental one by a factor A(n,nc,d) which can be derived from
light scattering theory. The curves in Fig. 10 increase
monotoneously up to particle diameters of 0.7 jim and differ
below 0.5 (am (
-------
d1 (ym)
0.1 OO
0.151
0.206
0.318
average
values
Fj=j1/3
1
1
1
1
1
1
F2
.254
.285
.233
.267
.259
.260
1
1.
1
1
1
1
F3
.446
.457
.417
.478
.449
.442
F4
1 .610
1 .616
1 .588
—
1 .604
1 .588
F5
1 .708
1 .702
1 .718
—
1 .709
1 .710
F6
--
1 .841
1 .830
—
1 .836
1 .818
Table 2: Relative light scattering diameters, F-, of
aggregates of j uniform polystyrene spheres
measured with the LASS.
diameter, d-, , have been used. Due to the high resolution
power of the instrument aggregates up to six spheres can
be distinguished. By means of the relative light scattering
diameter, F-;, defined in chapter 6 the results are summa-
rized in Table 2. As can be seen, Fj = J1/ , so that dscj
is identical with the equivalent volume diameter, dve,
of a particle regardless of its shape. In other words the
optical shape factor of a non-spherical particle smaller
than the wavelength of light is 1. These experimental find-
ings are valid within the experimental error of the method
which is less than 2 %.
8. HIGH RESOLUTION OPTICAL SIZE SPECTROMETRY
In order to evaluate the resolution power of the spectro-
meters LASI and LASS monodisperse polystyrene latex aero-
sols of known size distributions are used. The spectra
stored in the multichannel analyser are converted into
particle diameter distributions by means of the calibration
curves. Since the size distributions of the polystyrene
aerosols can be approximated by Gaussian distribution func-
tions, the modal diameter, 3, and the standard deviation,
& , are used to characterize the distributions;
810
-------
depends on the standard deviation, 6^ei» °f tne polysty-
latex aerosols (evaluated by electron microscopy) and,
additionally, on the variance of the spectrometers.
ren
y\
LASS O.O73
O.100
0.151
O.206
0.318
0.488
LASI 1 .1 58
1 .83
2.68
oel(ym)
O.OO50
O.0040
0.003O
O.OO40
0.0050
O.O070
0.021
0.028
—
aop(ym)
O.OO4O
O.OO35
0.0032
0.0035
0.0037
O.OO65
0.043
0.040
0.050
s\
a /d ,
op' el
0.055
O.035
0.021
0.017
0.012
0.013
0.037
0.022
0.018
Table 3: Resolution power of the LASS and the LASI
measured with monodisperse polystyrene latex
aerosols (3 = modal diameter of the polysty-
rene latex spheres measured with an electron
microscope).
The comparison of QT and & -. is shown in Table 3.
These values agree vePy well as far as the LASS is concerned.
Hence, the variance of this instrument is negligible. The
variance of the LASI is not negligible probably primarily
due to the short-time stability of the mercury arc lamp
which is less than the stability of the laser.
To achieve a high resolution power an optical size spectro-
meter has to be designed in such a way that the partial
scattering cross-section of a particle depends strongly on
particle size. Furthermore, the resolution can still be
improved by applying a light illumination of high short-
time stability and aerodynamic focusing of the aerosol stream
811
-------
The latter results in a constant aerosol flow and homoge-
neous illumination of the particles. However, a hiqh reso-
lution optical size spectrometer has a limited dynamic range
for instantaneous size detection. Therefore, the size distri-
butions of polydisperse aerosols have to be measured step-
wise by varying the amplification of the electrical pulses.
»-'.
10
10
10
10
10
0
05 1.0 15 2.0 25 3.0 35 40 45 5.0.
a_
005 Or 015 02 025 03 d.fun (A.=0.5K(tm}
02 03 04 0.5 06 07 08 09 10
=06328 urn)
10. Experimental response curves of the LASS for
transparent spheres of different refractive
index
812
-------
REFERENCES
1) Hodkinson,J.R.: Dust measurements by light scattering
and absorption, Ph.D.Thesis, University of London 1962
2) Quenzel,H.: Appl.Optics 8:165,1969
3) Schock,W.: Kolloquium Aerosolmefitechnik, Aachen 1975
4) Bakhanova,R.A. and Ivanchenko,L,V.: Aerosol Sci.4:485,
1973
5) Wagner,P.: Untersuchung des Tropfchenwachstums in einer
schnellen Expansionsnebelkammer, Ph.D.Thesis, Univer-
sitat Wien 1974
6) Hodkinson,J.R. and Greenfield,J.R.: Appl.Optics 4:1463,
1965
7) Bro6marm,R.: Die Lichtstreuung an kleinen Teilchen als
Grundlage einer TeilchengroSenbestimmung, Ph.D.Thesis,
Universitat Karlsruhe 1966
8) Oeseburg,F.: Aerosol Sci. 3:307,1972
9) Gucker,F.T. and Rose,M.A.: Brit. J. Appl. Physics, Supple-
ment No 3:138,1953
10) Zinky,W.R.: J. Air Pollution Control Assoc. 12:578,1962
11) Sinclair,D.:J. Air Pollution Control Assoc. 17:105,1967
12) Martens,A.E. and Fuss,K.H.: Staub-Reinhaltung Luft 28:
229,1968
13) Turpin,P.Y.: Revue d'Optique 46:309,1967
14) Whitby,K.T. and Vomela,R.A.: Env.Sci.Techn. 1:801,1967
15) Jaenicke,R.: Aerosol Sci. 3:95,1972
16) Williams,!, and Hedley,A.B.: Aerosol Sci. 3:363,1972
17) Liu,B.Y.H.;Berglund,R.N. and Argawal,J.K.: Atm. Env. 8:
717,1974
813
-------
18) Schleusener,S.A. and Read,A.A.: Rev.Sci.Instr. 38:1152
1967
19) Schehl,R.;Ergun,S. and Headrick,A.: Rev.Sci. Instr. 44:
1193,1973
2O) Phillips,D.T.;Wyatt,P.J. and Berkman,R.M.: J.Coll.
Interface Sci. 34:159,1970
21) Wyatt,P.T. and Phillips,D.T.: J.Coll. Interface Sci.
39:125,1972
22) Cooke,D.D. and Kerker,M.: J.Coll. Interface Sci. 42:
150,1973
23) Gucker,F.T.;Tuma,J.;Lin,H.-M.;Huang,C.-M.;Ems,S.C. and
Marshall,T.R.: Aerosol Sci. 4:389,1973
24) Moser,H.O.: Appl.Optics 13:173,1974
25) Meyer,U.: Bestimmung von TeilchengroBen mit einer Streu-
lichtmethode, M.Sc.Thesis, TH Darmstadt 1975
26) Kaye,W.: J.Coll.Interface Sci. 44:384,1973
27) Gebhart,J,;Bol,J.;Heinze,W. and Letschert,W.: Staub-
Reinhaltung Luft 30:238,1970
28) Hill,C.A.: Conference on Aerosol Physics, Southhampton
1973
29) Bol,J.;Gebhart,J.;Heinze,W.;Petersen,W.D. and Wurzbacher,
G.: Staub-Reinhaltung Luft 3O:475,1970
30) Pinnick,R.G.;Rosen,J.M. and Hofmann,D.J.: Appl.Optics
12:37,1973
31) Jacobi,W.;Eichler,J. and Stolterfoht,N.: Staub-Reinhal-
tung Luft 28:314,1968
32) Walsh,M.: private Mitteilung, 1-B-73
33) Lioy,P.J.;Rimberg,D. and Hanghey,F.J . : Aerosol Sci 6:
in press,1975
34) Heyder,J.;Gebhart,J. and Stahlhofen,W.: Water, Air and
Soil Pollution 3:567,1974
814
-------
35) Liu,B.Y.H.;Marple,V.A.;Whitby,K.T. and Barsic,N.J.: Am.
Ind. Hyg. Assoc. J. 74:443,1974
36) van de Hulst,H.C.: Light scattering by small particles,
John Wiley,New York,1957
37) Gebhart,J. and Roth,C.: 1st Annual Meeting of Gesell-
schaft fur Aerosolforschung (Bad Soden/Ts. Rossert-
str.11, Germany) 1973
38) Porstendorfer,J. and Heyder,J.: Aerosol Sci. 3:141,1972
39) Stahlhofen,W.;Gebhart,J.;Heyder,J. and Roth,C.: Aerosol
Sci. 6,in press,1975
4O) Stahlhofen,W.;Armbruster,L.jGebhart,J. and Grein,E.:
Atm. Env., in press,1975
41) Albert,R.E.;Petrow,H.G.;Salam,A.S. and Spiegelman, J .R. :
Helath Physics 1/,933,1964
42) May,K.R.: J.Appl. Physics 20:932,1949
43) Heyder,J. and Porstendorfer,J.: Aerosol Sci. 5:387,1974
44) Zerull,R.: Mikrowellenanalogieexperimente zur Lichtstreu-
ung an Staubpartikeln, Ph.D.Thesis,Univ.Bochum 1973
45) Heyder,J.;Roth,C. and Stahlhofen,W.: Aerosol Sci. 2:341
1971
46) Roth,C. and Gebhart,J.: 1st Annual Meeting of Gesell-
schaft fiir Aerosolforschung (Bad Soden/Ts. Rossert-
str.11,Germany) 1973
47) Kerker,M.: The scattering of light, Academic Press, '\
New York-London, 1969
48) Maguire,B.A.: Aerosol Sci. 2:417,1971
49) Lowan,A.N.: Nat.Bur.Standard, Appl. Math.Series 4,
Tables of scattering functions for spherical particles
50) 01af,J. and Robock,K.: Staub-Reinhaltung Luft 21:495,
1961
815
------- |