United States
Environmental Protection
Agency
Environmental Sciences Research EPA-600/2-79-105
Laboratory May 1979
Research Triangle Park NC 27711
Research and Development
Aerosol
Measurements in the
Submicron Size
Range
Studies With an
Aerosol Centrifuge, a
New Diffusion
Battery, a Low
Pressure Impactor
and an Advanced
Condensation Nuclei
Counter
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U S Environmental
Protection Agency, have been grouped into nine series These nine broad cate-
gories were established to facilitate further development and application of en-
vironmental technology Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields
The nine series are
1 Environmental Health Effects Research
2 Environmental Protection Technology
3 Ecological Research
4 Environmental Monitoring
5 Socioeconomic Environmental Studies
6 Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8 "Special" Reports
9 Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECH-
NOLOGY series This series describes research performed to develop and dem-
onstrate instrumentation, equipment, and methodology to repair or prevent en-
vironmental degradation from point and non-point sources of pollution. This work
provides the new or improved technology required ior the control and treatment
of pollution sources to meet environmental quality standards.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161
-------
EPA-600/2-79-105
May 1979
AEROSOL MEASUREMENTS IN THE SUBMICRON SIZE RANGE
Studies with an Aerosol Centrifuge, a New
Diffusion Battery, a Low Pressure Impactor
and an Advanced Condensation Nuclei Counter
by
Othmar Preining and Axel Berner
Institute for Experimental Physics
University of Vienna
Austria
Research Grant No. 801983
Project Officer
Jack Wagman
Director, Emissions Measurement and Characterization Division
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U. S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
This report has been reviewed by the Environmental Sciences Research
Laboratory, U. S. Environmental Protection Agency, and approved for publica-
tion. Approval does not signify that the contents necessarily reflect the
views and policies of the U. S. Environmental Protection Agency, nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
-------
PREFACE
Polluting the atmosphere is one of man's most traditional ways to dis-
pose of unwanted products from cultural and industrial activities. But today's
large scale pollution forces society to regard its economical activities with
careful respect of man's and other beings' health. Related research programs
are still a demand in order to meet the needs for sufficient information to
make the necessary political decisions weighing calculable risks versus econ-
omic benefits.
The Institut fur Experimentalphysik, formerly I. Physikalisches Institut
of the University of Vienna has been and still is involved in a number of
research programs on the atmospheric environment, particularly it has a long
historic record in aerosol research dating back as far as the first decade of
this century. Present research is mainly concerned with visibility and respective
aerosol photometers, cascade impactors, condensation nuclei counters, their
fundamentals and their applications to atmospheric aerosols.
Prof. Dr. P. Weinzierl
Head
Institut fur Experimentalphysik
University of Vienna
m
-------
BACKGROUND
The longer research continues the more the importance of aerosols
becomes evident for all aspects of air pollution. The problems of recent
interest are related to industrial and civic hygiene in local environments
as well as to the global pollution of the earth's atmosphere. At all levels
the aerosol problems are of great, and mostly not completely understood
complexity. Hence the analysis of aerosols requires profound knowledge of
their physical and chemical properties.
The dynamics of aerosols, i.e. the development of the physical and
chemical properties in time, is a focus of todays'interest, and refined
measuring techniques are required in order to produce basic data such as
aerosol size distributions and the chemical composition of the different
particle sizes. Difficulties arise, because aerosol classifiers are generally
not able to cover the whole size range of interest. Hence different instruments
are to be combined into classifier systems in order to yield complete infor-
mation. New problems arise. The system components evaluate aerosol properties
within limits only, and moreover they may distort the information - problems
which can be recognized and overcome only by comparisons of different system.
In the past, this system approach led to the development of the
University of Minnesota Aerosol Classifying System, which led to profound
insights into the structure of aerosol size distributions. But this system
has its limitations^too ;and it is unsuitable for collecting chemically
analyzable samples. New techniques are required in order to meet such needs.
-------
ABSTRACT
The report summarizes the investigations of four aerosol classifiers
which cover finite, but overlapping ranges of the aerosol particle size
spectrum.
The first part is concerned with a cylindrical aerosol centrifuge,
which measures aerodynamic equivalent diameters precisely. This instrument
has been used as a reference instrument in diffusion battery experiments
reported in the second part. The diffusion battery has been investigated
for fairly large particle sizes (0.3 ym to 0.5 ym) to determine the influ-
ence of sedimentation, interception and impaction on the transmission of
the diffusion battery. These experiments have been performed with highly
monodispersed NaCl aerosols.
In the third part a five stage low pressure impactor is described,
which covers the size range from 0.1 ym to 25 ym diameter. It has been
developed specifically for the determination of the deposited particulate
mass. First data on mass-size distributions of atmospheric aerosols are
reported. The final chapter summarizes the development of a special con-
densation nuclei counter which measures number-size distributions in the
size range from 0.002 ym to 0.1 ym KELVIN-equivalent diameter. The
applicability to urban atmospheric aerosols is demonstrated.
-------
CONTENTS
Preface i i i
Background iv
Abstract v
Fi gures i x
Tables xiii
Acknowledgement xi v
Conclusions 1
Recommendations 3
The ROSL Aerosol spectrometer
Introducti on 4
Rotors of the ROSL Spectrometer 6
Particle Size Analysis 9
Operational Features of the ROSL Spectrometer 15
Discussion 26
The Diffusion Battery
Introducti on 28
The NaCl Generator 31
The NaCl Aerosol s 36
The Diffusion Battery 45
Deposition Mechanisms in a Diffusion Battery 48
Results and Discussion 54
VII
-------
The Five Stage Low Pressure Impactor
Introducti on 60
Particle Size Analysis 63
The Aerodynamic Equivalent Size 66
Deposits in the Impactor Stages 68
Note on Back Up Fi 1 ters 74
The Condensation Nuclei Counter
Introducti on 77
The Expansion Cloud Chamber and its Data Acquisition
System 77
The Analysis of the Recorded Data 81
Results on Atmospheric Aerosols 84
References 87
vm
-------
FIGURES
1 Cross section of rotor I. B, base plate; MC, measuring chamber;
CC, cleaning chamber; S, aerosol slit; H, ports to the cleaning
chamber; K, shaft 5
2 Cross section of rotor II. B, base plate; MC, measuring chamber;
S, aerosol slit; K, shaft; AFLO, aerosol flow limiting orifice;
TFLO, total flow limiting orifice 7
3 Head of rotor I with gasket. N,nylong spring disc; S,teflon sheet 8
4 Latex aerosol deposits on collection foil. S, projection of the
slit. 1), single spheres, 2) to 5), aggregates of 2,3,4,5 spheres 8
5 Number size distribution of latex aerosol. 1), single spheres;
2) to 5), aggregates of two to five spheres 11
6 The reduced deposition length and its relationship to the equiva-
lent particle size, calculated from DOW quotations. Bars indicate
the average and the range of the experimental data, the solid
line represents the theoretical expectation 17
7 Variance of number size distributions of single latex particles
and its relationship to the cross section of the AFLO orifice.... 18
8 Number size distributions of single latex spheres at different
AFLO cross secti ons 2o
IX
-------
9 Calibration curves of the aerosol centrifuge 22
lo Calibration curves of the aerosol centrifuge 22
11 Ambivalent deposition lengths 23
12 Range of speed fluctuations for the system rotor II/regulated
power supply 23
13 Number size distributions of single latex particles for cold
and warmed rotor 1 25
13a Calibration curves for rotor II 27
14 Furnace for the generation of monodispersed NaCl aerosols 29
15 Cross secti on of the aerosol generator 3o
16 Calibration curves for the gas temperatures in the center tube... 33
17 Gas temperatures along the axis of the center tube 34
18 Gas temperatures along the axis of the center tube,2ero flow rate.. 35
19 The sizes of the monodispersed NaCl aerosols in dependence of the
temperatures at zone I and zone II 37
2o Number size distribution of heavily coagulated NaCl aerosols 39
21 Number size distribution of weakly coagulated NaCl aerosols 4o
22 Number size distribution of diluted NaCl aerosols 42
23 Number size distribution of diluted NaCl aerosols 43
-------
24 Comparison of electron microscopical diameters of NaCl part-
icles to their deposition lengths in the centrifuge 44
25 Cross section of the diffusion battery 46
26 Tubing diagram of the diffusion battery 47
27 Theoretical impaction and interception efficiency of CHS plates .. 53
28 Experimental transmission data compared to theoretical
expectati ons 57
29 Cross section of the fice stage low pressure impactor 61
3o Wall losses in the five stage impactor 7o
31 Variations of deposited mass among the spots 72
32 Mass size distribution of a vaseline-dye-aerosol 73
33 Mass size distribution of an urban aerosol at Vienna 73
34 Threehourly variations of urban aerosols at Vienna 75
35 Diagram of the data acquisition system of the condensation nuclei
counter 78
36 Aerosol and gas flow diagram of the condensation nuclei counter .. 80
37 Scattered light intensity of the growing aerosol 82
38 Scattered light intensity calculated from the MIE theory 83
39 Cumulative number size distribution of KELVIN euqivalent
diameter 85
xi
-------
4o. Differential number size distribution of KELVIN equivalent
diameter for an urban atmospheric aerosol at Vienna
xn
-------
TABLES
Number page
1 Analysis of Latex particle deposits 16
2 Experimental and theoretical transmissions of
the diffusion battery 55
3 Deviations of theoretical Transmissions from
Experimental values 58
4 Data of the impactor stages 62
xm
-------
ACKNOWLEDGEMENTS
The research program has been sponsored by the Environmental Protection
Agency of the United States of America and by several Austrian Governmental
and Municipial Organizations including the Bundesministerium fur Wissenschaft
und Forschung, The Fonds zur Fbrderung der wissenschaftlichen Forschung and
the Hochschuljubilaumsstiftung der Stadt Wien. The support by these organi-
zations is gratefully acknowledged.
The authors wish to express their personal gratitude to the staff of
the Institute. Dr. Gerhard Kasper conducted successfully the research on
the diffusion battery, Dr. Paul Wagner, assisted by Mr. Franz Pohl, is respons-
ible for the study of the condensation nuclei counter. Mr. Christian Liirzer
performed the measurements of the urban mass size distributions. The authors
are greatly indepted to Mrs. Hannelore Kranner who performed the work of
typing and assembling the report with extraordinary patience.
xiv
-------
CONCLUSIONS
The results of the work reported here demonstrate:
1) The aerosol centrifuge is an instrument for measuring aerodynamic
diameters precisely and independent of external calibrations such as cal-
ibrations by latex particles. So far the instrument has been investigated
for particle sizes from o,2 _um to 3,ojjm diameter, but there are indications
that the instrument is capable to classify particles as small as o,o5jum,
at suitable operation conditions.
2) The diffusion battery can be operated up to sizes of o,5 jjm diameter,
if great care is taken in calibrating the instrument. For such particles
sizes impaction becomes an essential deposition mechanism.
3) Highly monodispersed NaCl aerosols with spherical particles of bulk
density can be produced by a generator of the SINCLAIR-LA MER type. The
aerosol properties are controlled by the cooling rate of the condensing
vapours.
4) Cascade impactors have been designed as to permit the direct and
accurate measurement of mass size distributions of atmospheric aerosols.
The first impactor under investigation has a measuring range from
o,l jjm to 25 jim, permitting the gravimetric evaluation of the accumulation
mode of atmospheric aerosols. There are indications that an impactor can
be designed for the collection sizes as small as o,o5 ym diameter or
less.
-------
5) The condensation nuclei counter technique can be used to determine
number size distributions of atmospheric aerosols in the size range from
o,oo2 urn to o,l jjm of KELVIN equivalent diameter.
6) The combination of a cascade impactor and a condensation nuclei
counter of the kind described here would constitute a system for the
evaluation of size distributions in the range from o,oo2 ym to 25 |im
diameter, with a sufficient overlap at sizes around o,ljim diameter. Moreover,
the data of the system components would be fairly independent of each other,
because the nucleation mode would represent almost completely the total
particle concentration by number, but does not contribute considerably to
the total particle concentration by mass.
-------
RECOMMENDATIONS
Since an aerosol sizing system of a cascade impactor and a condensation
nuclei counter as described in this report permits to evaluate the size
distribution of atmospheric aerosols in the same size range as the University
of Minnesota Aerosol Sizing System, a direct comparison of both systems
could provide so far unobtained insights in the structure of atmospheric
aerosols as well as in the operation characteristics of the classifier
systems.
-------
THE ROSL AEROSOL SPECTROMETER
Introduction
Aerosol centrifuges are well known instruments for the classification
of aerosols. Since their introduction by SAWYER and WALTON (1) in 195o they
underwent!arge improvements which culminated in the development of several
highly resolving spectrometers. The work of GDTZ et al. (2,3,4), KEITH and
DERRICK (5), KAST (6), STOBER and coworkers (7), (8), (9), (lo), (11), (12),
HAUCK and SCHEDLING (13), HOCHRAINER (14), (15), BERNER and REICHELT (16),
BURSON et al (17), and MATTESON et al (18) may be quoted here.
The main idea of these instruments is the separation of the aerosol par-
ticles from a steady, laminar flow by means of the centrifugal force. For this
purpose the aerosol is introduced into a rotating chamber. Under the influence
of the force the particles move to the outer wall where they are deposited on
a removable.foil. Classification occurs because the particle velocities produced
by the centrifugal force depend on the particle sizes: the larger the particles
are, the faster they drift to the wall. Complete classification is achieved
when the aerosol is introduced into the chamber by a narrow slit which super-
imposes the aerosol on a layer of particle free gas. For a thin aerosol layer
the particles fall to the outer wall from almost identical positions, and their
deposit location with respect to the entrance slit, i.e. the deposition length,
is a direct measure of their size because a given location corresponds to only
one particle size.
-------
Fig.l: Cross section of rotor I. B,base plate; MC, measuring chamber;
CC, cleaning chamber ; S, aerosol slit ; H, ports to the cleaning chamber;
K, shaft.
-------
Rotors of the ROSL spectrometer
An adequate slit system has to meet one essential condition: the
superposition of the aerosol flow on the particle free gas flow should be
as undisturbed as possible. This is achieved when all parts of the slit
participate fully in rotation (19), (2o). The technical solutions of this
problem are very limited in number. The particular solution realized in
the ROSL spectrometer is a radial slit, or a cut through the inner parts
of the rotor (see fig. 1). This concept, of the slit leads almost forcibly
to a concentric arrangement of all other rotor elements. The choice has
been given to concentric cylinders, but not to cones.
In operation the rotor is driven by a high speed motor via the shaft K
and the flow is maintained by the selfpumping action of the rotor. The aero-
sol flow enters the slit S by an orifice in the center of rotation. This
socalled AFLO orifice restricts the aerosol flow to the desired magnitude.
Partially particle free gas is produced by the rotor itself. Some aerosol is
passed through several holes, H, into a cleaning chamber, CC, where large par-
ticles are removed from the gas. The remaining particles are usually too small
as to be detected by light microscopy. Particles of this background aerosol
will be deposited on the foil, but they do not appear optically. This cleaning
procedure bears some advantage. At first, the gas composition remains un-
changed and therefore the superposition of the aerosol flow and the particle
free flow is not influenced by density gradients. Secondly, external filtering
devices are unnecessary as long as the remaining particles do not interfere
with the observations. The particle free flow enters the measuring chamber,
MC, through a port in the top of the rotor. Another rotor of slightly different
design is shown in fig. 2. In this case the particle free gas is not produced
in a cleaning chamber, but it is drawn from the flow boundary layer at the
outside of the rotor. The gas is particle free to the same degree as described
above. Size classification is bound to laminar flow in the measuring chamber,
therefore the total amount of flow has to be limited. Adequate restriction
is achieved by a set of six orifices, i.e. the TFLO or total flow limiting
orifices, in the bottom of each rotor. In order to change the collection
foils the rotors are disassembled by lifting the outer parts comprising the
6
-------
Fig.2 : Cross section of rotor II. B, base plate; MC, measuring chamber;
S, aerosol slit; K,shaft; AFLO, aerosol flow limiting orifice;
TFLO, total flow limiting orifice.
-------
Y////A
FIS3
FIG. 4
Fig.3: Head of rotor I with gasket. N, nylon spring disc. S, teflon sheet.
Fig.4: Latex aerosol deposits on collection foil. S, projection of the slit.
1), single spheres.2) to 5),aggregates ot two, three, four and five spheres,
8
-------
outer wall of the measuring chamber, the upper wall of the aerosol slit
and the cleaning system^rom the base plate, B, which is fixed to the
inner parts of the rotors. In the first rotor (see fig. 1) the removable
piece is bolted to the base plate by six bolts and it is positioned by a
very narrowly spaced cylindrical guide, whereas in the second rotor the
parts are connected by a large thread and held in position by a conical
guide.
For measuring the rotational speed, the first rotor is equipped with
ten small permanent magnets in the top of the rotor. The magnets pass
beneath an electromagnetic pfck up in the housing of the centrifuge. The
top of the second rotor bears a disc with twelve blank fields on a dark
background, for switching a photo transistor. Primarily, the rotors have
not been designed for direct measurements of flow rates. This inability
became a serious draw back during a certain period of the work, and there-
fore the first rotor has been equipped with a seal in order to enable
the measurement of the total flow rate. This seal, which is fixed to the
housing of the centrifuge, consists essentially of a nylon spring disc
and of a teflon sheet which is pressed to the top of the rotor (see fig.3).
The seal is tight, up to rotational speeds of 6000 rpm. The flow rates
are measured by means of a soap film moving in a calibrated glass tube
mounted on top of the housing. (It should be mentioned, that the teflon
sheet undergoes heavy wear by the rotor, and therefore must be replaced
frequently).
Particle size analysis
The aerosol slit has a width of o,l mm which is small compared to
the chamber width of 3 mm, i.e. the distance between the inner wall and
the outer wall of the chamber. The aerosol particles therefore start their
movement to the outer wall from fairly equivalent positions at the end
of the slit. Consequently the; particles of an extremely monodispersed
aerosol will occur at similarily equivalent deposition locations, which
after unrolling the foil appear as a straight line parallel to the projec-
tion of the slit. The deposits of a latex aerosol, which contains a number
9
-------
of different, aerodynamical ly separated classes of particles (15), (19),
(21), thus appear as a set of separated lines (see fig. 4), and after
counting as a set of separated number size distributions (see fig. 5).
The distance between these lines and the projection of the slit is intro-
duced as the deposition length, L , of the particles.
From the point of an observer rotating with the chamber, the particle
velocity has essentially two components, i.e. the radial velocity, v , in
the radial direction, and the velocity, v , in the axial direction. The
azimuthal velocities of the particle vanish, if solid body rotation is
assumed for the gas and the particles. The influence of the centrifugal
force on the movement of the particle follows from STOKES1 law, which
may be accepted in the form
(,r/6).DjJ.pp.u2.R = 3.,.ng.Dp.v;/Bpg (1)
for spherical particles of diameter D , and density, p. The slip correction
factor, B , is given by
Bpg = 1 + (Ag/Dp).(l,36 + o,7.exp(-3,68(xg/Dp)) (2)
The gas has the viscosity,n , and the mean free path, A , of the molecules.
The velocity, v' , is the relative velocity of the particles in the
gas, in the radial direction. But because of the cylindrical design of the
rotor, the gas does not possess radial velocities within the measuring
chamber, and consequently the radial velocity, v , of the particle and its
radial velocity, vj,, relative to the gas^are identical. This leads to the
equation
vr =
for the radial velocity of the particle.
10
-------
Zoo _
loo _
DEPOSITION LENGTH (
Fig.5: Number size distribution of latex aerosol. 1), single spheres;
2) to 5), aggregates of two to five spheres.
11
-------
2
The acceleration to .R needs some consideration. The radius, R, is the
radial distance of the particle from the axis of rotation. Thus the
acceleration increases as the particle moves to the outer wall. But the
variation of this radial distance, R, is AR = 3 mm from the inner to
the outer wall, which is small compared to the mean radius
R = (R + R.)/2 = (4o + 37)/2 = 38,5 mm of the chamber. The acceleration
is transmitted to the particles by the gas, and therefore the angular
velocity, m, is the angular velocity of the gas at the position of the
particles. In general, there will be an ancular motion of the gas relative
to the rotor, and consequently the angular velocities of the gas differ
from the angular velocity,u0» of the rotor.
Fortunately, the solution of these problems is fairly simple, if
the assumption is justified that the gas within the centrifuge rotates
like a solid body with angular velocity u)Q. In this case the radial
velocity of the particles is determined by
2 _
with a constant acceleration term u .R for all particles.
It should be mentioned here, that particles with the same
* 1/2
D = Dp-(Bpq-pp) are obviously not discriminated by the centrifuge. The
equation
is therefore introduced as definition for the aerodynamic equivalent
size of a particle, and it is this parameter, which is directly related
to the radial velocity of the particle.
During their way frcm the inner wall to the outer one, the particles
are transported downstream together with the gas. The assumption is justified
that the axial velocities of the gas and the particles are identical and
therefore the particle velocities will show the same velocity profile as
1-2
-------
the gas velocities. But ST'OBER et al. (22) have proven that the deposition
length, L , of a particle is independent of the true path of the particle;
rather the deposition length is determined by the average axial velocity,
V
2
z .(R - R)) = Q/(2W.R.AR) (6)
v =
Regarding the path of a particle, one finds that within the same period
of time the particle has travelled the distance, AR, radially, and the
deposition length, L , axially, and consequently the relation
Lp/AR = vz / vr (7)
holds, where v is the average radial velocity of the particle on the path
to the outer wall. Replacing of v£ by equation (6) and v by equation (4)
yields the equation (8) for the deposition length, L ,
L = o,8959.1o~*.(Q/D*2.N2) (8)
2 22
where w has been replaced by 4.77 .N and n by its numerical value at
standard temperatures. (It should be mentioned, that the exact calculation
of the deposition length would change the constant from k = o,8959 to
k = o,8963 (19)).
However, equation (8) is based on the assumption that the gas rotates
like a solid body with angular velocity to . A direct proof by measuring
the angular velocities of the gas is difficult, but the obser-
vations with latex particles of well known size indicate that this assump-
tion is justified. First, the calibration curves, i.e. the empirical
relations between the sizes of the latex particles and their deposition
lengths follow equations like
13
-------
where the constants, A., depend on the flow rate, Q, and on the rotational
speed, N. These equations coincide perfectly with equ. (8) as far as
particle size and deposition length is concerned. By comparing (8) and
(9), the relation
A1 = o,8959.1o"6.(Q/N2) = k.(Q/N2) (lo)
follows, which immediately leads to the equation
L* = (D*)"2 = (l/k).(Lp.N2/Q) (11)
where the inverse square of the equivalent size has been introduced as
Jy
the reduced deposition length, L . By means of (11), the reduced deposition
length and therefore the equivalent size of the latex particles is com-
pletely determined by the data of the centrifuge, i.e. the deposition
length, the flow rate and the rotational speed.Other values for the
equivalent sizes of the same particles are calculated from their known
diameter using equ. (5). Both methods should be in agreement if the
assumption of solid body rotation is correct for the particle motion.
The results of confirming experiments are listed in table I. Four
different latex particle sizes have been used, i.e. D = I,3o5 urn,
D = o,79 urn, D = o,557 ym and D_ = o,357 ym. A density of I,o5 has been
assumed for these particles. The deposition lengths, L , have been
measured for different rotational speeds and different flow rates. From
these data, the reduced deposition lengths, L*, have been calculated
according to equ. (11). For a given latex size these values are in fair
agreement, and the average reduced deposition length,T*, has been used for
calculating the equivalent sizes, D , of the latex particles. These sizes
are in agreement with the equivalent sizes, D*, calculated from the latex
diameters as quoted by the manufacturer. The relatively large deviations
for latex particles of o,79o ym diameter indicate, however, that the
quoted size might be wrong.
14
-------
The data of table I represent nire different calibration curves. These
curves should coincide to a single curve, when the reduced deposition
lengths are introduced according to equation (11). Its logarithmic form
is given by
In L* = 2.In D* (12)
P P
where L stands for the reduced deposition lengths calculated from the
centrifuge data, and D for the equivalent size calculated from the
diameters quoted by the manufacturer. The graph of this curve, which
is represented in fig. 6 together with the experimental data, indicates
very clearly the consistency of theory and experiment forttiree of the
latex sizes.
Operational features of the ROSL spectrometer
The aerosol flow limiting orifice restricts the flow into the aerosol
slit. Depending on the cross section of this orifice, the aerosol flow is
low for small orifices and vice versa. The influence of the AFLO orifice
on the resolving power of the centrifuge is of more importance. With
smaller orifices the aerosol layer entering the measuring chamber is
narrowed and consequently the initial positions as well as the deposition
locations of the particles are better defined. This effect is demonstrated
by the variance of the deposit locations of single latex spheres, which
have been collected with different AFLO orifices. As demonstrated in fig. 7,
the variance decreases with diminishing orifice cross sections. A further,
though minor decrease could be expected for still smaller orifices, but
for such experiments long collection periods are needed, which increase
the errors due to variations of the rotational speed and other long term
instabilities.
Some experiments have been conducted to determine the resolving power
of the aerosol spectrometer, at low aerosol flow conditions. The number
size distributions of the latex aerosols are assumed to be Gaussian. Then
the variance, Vm, of the observed size distribution is the sum of the
15
-------
•
00
t—
t— 1
c/)
o
a.
UJ
Q
UJ
i
— J
C_>
I""H
i—
^^
Q.
X
UJ
\—
LO
LD
t
CM
f*.^
fx^,
f.
«— 1
cn
LO
CM
ft
i— i
r*^
00
«t
o
LO
OO
LO
O
1 —
ro
«*
0
oo
ft
cn
i— i
o
o
o
ro
co
CO
oo
rt
LO
CM
0
•t
ID
00
cn
r<
CM
CM
00
ft
CM
00
i— 1
CM
ft
i— 1
LO
rsN«
oo
n
i— 1
ro
CM
LO
rt
O
cn
LO
o
LO
rt
1— 1
OO
0
o
0
oo
i
i
i
i
oo
cn
LO
ft
CM
0
ft
*±
CM
i— 4
CM
rt
i— 1
LO
i— 1
A
CM
CM
i— «
LO
O
cn
0
cn
n
O
LO
rt
cn
^j-
0
0
0
ro
Lfj
CM
CM
f
LO
CM
i— <
rt
OO
«^-
LO
LO
rt
CM
00
CM
f.
i— 1
f-x
oo
CM
n
I— 1
CM
rt
O
|
|
\
t
CO
ft
LO
CO
O
0
00
oo
CM
ft
LO
r— 1
•51-
f*
.f^.
00
cn
LO
r-
CM
0
O
rt
CM
^-
ro
CM
rt
r—i
LO
cn
rt
O
o
CM
LO
O
0
*^~
*t
0
o
n
LO
LO
O
0
00
00
o
•t
LO
LO
oo
«
LO
LO
00
«^-
rt
CM
r— 1
LO
ft
CM
ID
00
CM
n
i — 1
LO
oo
ft
I— 1
LO
00
^-
n
O
1— 1
LO
ft
O
0
A
LO
r---
0
o
CO
CM
tvj
LO
ft
LO
LO
OO
rt
CM
i — 1
LO
LO
*>
CM
cn
o
ft
1—1
CM
CXj
CM
rt
r—t
CM
LO
ft
O
1
1
1
1
LO
ft
f-N
^J-
o
o
o
o
CM
LO
ft
LO
0
LO
CO
r-t
LO
^~
•*
CM
r^»
LO
ft
i-H
^<
vo
CM
rt
1 — 1
i— 1
00
ft
o
LO
1— 1
LO
rt
O
oo
OO
ft
0
LO
rt
I— 1
r^
o
o
o
to
CO
00
LO
rt
LO
LO
LO
rt
LO
O
LO
LO
rt
CM
0
LO
rt
CM
CT>
^~
CM
•*
»— 1
CM
CM
r.
r— 1
CM
O
LO
rt
0
CTt
^f-
ft
o
o
ft
CTi
O
i— 1
O
O
O
LO
E E
3- 3.
oo r*^ o
«d- r-l CM
i^. ^- ^~
ft ft rt
LO O O
II II II
X CL X CL X CL
1 — 1 IQ Q
£ E
3- 3.
o i-«- cn
<3- CM CM
LO LO LO
rt rt ft
CM O O
II II II
X Q. X CL X CL
1— 1 IQ Q
E E
3. 3.
CM r--. o
^f Cn Is*".
CM 00 00
*\ n A
i-l O O
II II II
X CL X CL X CL
I_J IQ Q
E E
3- 3.
CO LO t— 1
i— i cn o
LO OO «d"
n *» •»
O 1— 1 r— 1
II II II
X Q. X Q- X Q.
I—I 1 Q Q
1
£
5 E
J=^ °
o> _,Q-
0) J^
'^
p- ^^
o ^r
4-> z
'I/j
0 o
O ni
QJ /rt
"° --C.
1 3
^
"O
a;
$*- X. Q.
1 i
«t
X CL E
l_l O
i.
'f. M-
CL
Q -0
CU
Et %
-1—*
O fl3
S_ i—
M- 3
0
QJ (O
-(-> 0
ra
"3 cu
U N
(O CO
o
* c
ai cu
^J ^—
•r- ea
1/1 .^
4-* 13
c cr
cu cu
CO **
> X CL
••- IQ
"^
cr •«
CU CU
CD
ft ^
X CL<+-
Q ••-
• « 4-J
i- C
CU CU
4-> U
E cu
fO ^"~
"O
>^
-a J3
cu
§ 1
O" 3
*» fO
Q_ QJ
Q E
16
-------
I ~ '
XCL"> 5 Q
a *- 3- —
T
in
_ in
cu
Nl
O)
u
-p
c
cu
s_
cu
Q.
X
cu
cu
O; -P
0)
p—
fO
>
•r—
rs
cr
cu
cu
-P
o
-P
Q.
•r—
.C
(/)
c
o
1J3
fO
r—
0)
(/,
•p
-t~
"X3
C
tO
-C
-P
en
c
cu
p—
c
o
•r-
•p
'[/)
o
Q.
CU
•o
-a
cu
o
^
•a
cu
s.
cu
1
1
^XD
•
cn
•r-
U_
CU
cn
c
to
^.
JC
-p
T3
to
cu
^:
-p
cu
IO
o
•r"
•o
c
•r-
(/I
i-
tO
ca
•
10
c
o
• r-
-P
•P
0
3
cr
31
o
(~)
E
O
S-
M-
T3
CU
tO
3
O
i—
IO
o
•
0
•r™"
to
-P
a
cu
a.
X
cu
1
to
o
•r—
4_>
CU
o
cu
^
cu
-p
I/)
1 ^
C"
cu
in
cu
c
a.
cu
i.
cu
c
• p—
r—
-a
•P-
P^
o
cu
-C
-p
*
-p
-a
P—
to
-P
17
-------
^"*"^OL^
^^•^^
'V.
\
\
\
\
\
\
I
\
h
\
\
\
fa
Y
1'
_
i LJ
( ^** i- QT ) A iD^WIiWA
i
_ E:
o
q 4->
— °
CNI +J
^ "=
^
_q ^ o
^ Ti
•r—
5 £
c re
o CL
~*^ Sit ^
« ^-)
8 re
WJ |—
p
^^ ^L
— • «R ^
01
c
cv« w
0
t/)
c
o
-fj
—"^ 3
•i —
S-
1 1
CO
O)
N
'Lo
S-
OJ
3
C
O
O)
o
c
rO
•i—
s-
(O
O)
u
• r— i
4-
'^
O
o
LJ_
+->
o
;-
O
Q.
18
-------
variance, V , of the original size distribution (quoted by the manufacturer),
and of the variance, V^ , added by the centrifuge:
Vm = VQ + Vi (13)
The contribution of the instrument could then be estimated from equ.(14)
Vi = Vm - Vo <14>
The data show, however, that this method is not consistent because of
inaccuracies of the original variances, V . Two examples may demonstrate
the difficulties. The latex particles of o,557 ym diameter have an
-4 2
original variance of V = l,1664.1o ym , whereas the measured variance
-42
has a value of Vm = o,7225.1o ym . Consequently, the contribution of
-42
the instrument would be negative, V- = -o,4439.1o ym . The latex particles
-4 2
of o,79o ym diameter have anoriginal variance of V = o,1936.1o ym ,
-42
whereas a value of V = o,5373.1o ym has been found by the centrifuge.
-42
In this case, the contribution of the instrument would be V- = o,3436.1o ym ,
The occurancesof negative variances demonstrate the infeasability of the
procedure, and it is therefore not adequate to draw quantitative conclusions
even on the positive data. However, the statement holds, that the variance
contributed by the centrifuge is of the same order of magnitude or less
than the variance of the latex particles, for high resolving conditions
The highest aerosol flow rate and the lowest resolving power is
obtained when the rotor is operated with no AFLO orifice in the center of
the slit. In this case, the number size distributions of single latex
spheres exhibit a rectangular shape. In addition, the size distribution
of the different components of the latex aerosol, i.e. the singlet,
doublet and the other multiple particles, may overlap considerably. As
shown by the number size distribution in fig. 8, the rectangular shape
develops from larger to smaller deposition lengths. Therefore the narrow
distributions and the wide ones coincide at the descent to smaller particles.
In order to achieve classification, the flow has to be laminar and
19
-------
CO
c
o
•r"
O
CU
CO
CO
co
O
c
CU
Ol
M-
-o
•4-)
(X3
CO
CU
s_
CU
c~
Q.
CO
X
CD
-(_>
(O
r—
CU
'01
c
co
M-
O
CO
c~
O
•r—
";3
-Q
'll
CO
•i —
"O
O)
N
CO
s_
CD
o
E
3
, ,
00
CD
., —
u_
CO
c
o
o
CU
CO
CO
CO
0
S-
CJ
CU
CD
to
1
'
,
CO
CU
c
^
T3
CU
.c
CO
CO
T3
f\
CO
c~
O
•r-
4_>
O
CU
CO
CO
co
O
i-
o
^Z
-------
vortex-free within the chamber. Such conditions are set by the TFLO in the
base plate of the rotor. Good flow conditions are indicated by straight and
narrow deposition lines, if latex aerosols are used. At unsuitably high
flow rates, however, the latex deposits exhibit a regular, sinusoidal
structure, and in extreme cases a regular array of overlapping arcs. The
TFLO are used to set the measuring range of the centrifuge. Because of the
definite length of the chamber, the deposition lengths, L , are restricted
to values between L = o cm and L = 11,o cm. However, the useful 1 range
for data analysis is smaller. Near the slit at L = 0, the particle sizes
crowd together, and the sensitivity, AL /AD , is almost zero. Therefore,
deposition lengths below o,3 cm or o,5 cm do not bear much information on
the difference of particle sizes. With increasing deposition lengths, the
sensitivity grows continuously, according to equ. (9 )
ALp/AD* = -C.(Lp)1/2 (15)
where C is a constant parameter. But on the other hand the concentrations
of the deposits decrease reciprocally, with the consequence of higher
statistical errors. Moreover, the flow pattern at the end of the chamber
may be disturbed by the outgoing flow. For these reasons it is not advisable
to extend the range of evaluated deposition lengths beyond L = 8 cm or
L = 9 cm. This range of deposition lengths comprises about half an order
of magnitude in particle size, according to equ. (9). This is a fairly
narrow measuring range, and should therefore be adjusted carefully to the
particle sizes of main interest. This is achieved by the appropriate choice
of the TFLO and of the rotational speed. As a rule, the particles of main
interest, e.g. the peak of a monodispersed aerosol, should be deposited at
deposition lengths between L = 2 cm and L = 5 cm (see fig. 9, lo).
For a given rotational speed, the total flow rate increases when the
total orifice cross section, i.e. the sum of the TFLO cross sections, is
enlarged. Therefore a certain calibration of the centrifuge is finely
adjusted by using TFLO of different sizes. It should be mentioned, however,
that the flow rate may be ambivalent for certain TFLO orifice combinations.
This fact is illustrated in fig. 11. According to the graph the flow rate
21
-------
FIG.9 (ABOVE)
FIG.IO(BELOW)
2
Fig.9: Calibration curves. Rotational speed 48oo rpm; TFLOs 2,36 mm (lower
2 2
curve), 4,71 mm (middle curve ), 11,78 mm ( upper curve).
2
Fig.lo: Calibration curves. Rotational speed 6000 rpm. TFLOs 2,36 mm (lower
2 2
curve), 4,71 mm ( middle curve), 18,85 mm ( upper curve ).
22
-------
7o
60 -
5o _
3o
2o
0 INDIFFERENT TO START CONDITIONS
C SLOW START
® FAST START
g
UJ
ea
REGION OF >
BISTABILITY i
AFLO CROSS SECTION (
5.
FIG.11
+ „-
i 3_
+ 2 -
t 1
tu
a.
o
t-
<
o
z
o
FIG.12
ROTATIONAL SPEED ( •
6.1o3 12,J
Fig. 11: Ambivalent deposition lengths. 48oo rpm, o,557jim latex size.
Fig.12: Range of speed fluctuations for the system rotor II/regulated
power supply.
23
-------
is ambivalent in a limited domain of total orifice cross sections, and the
actual value of the flow rate depends on the mode of accelerating the rotor:
a "fast" start will yield the higher flow rate, whereas a "slow" start prodices
the lower flow rate. This behaviour is effected by the TFLO orifices and their
arrangement in the rotor base. Ambivalence occurs for certain orifice sets
with three orifices of one size and the other three of another size, but only
if these orifices are arranged alternately, i.e. each orifice being neigh-
boured by orifices of the other size (19).
During operation, short term and long term fluctuations of the rotor
speed must be taken into account. The motor is a high speed collector motor,
and it is operated from the power line of the laboratory. The motor speed is
usually set by the voltage of an variable autotransformer. In long term
operation, the warming up of the rotor is a further source for speed
fluctuations. In order to overcome these deficiencies, a speed controller
has been developed, which keeps the rotational speed within the limits of a
few rotations per minute. Typical short term fluctuations are less than
+ 2 rpm, but ;as shown by fig. 12vthere are a few points of lower stability,
where the fluctuations are as high as +10rpm. The long term stability
measured for hours of continuous operation is better than o,5 %, i.e. the
measured speed does not deviate more than lo rpm from any preset value
between 2ooo rpm and 12ooo rpm.
Another effect of long period operations is the warming up of the rotor
by the heat of the bearings. Temperature differences of lo°C and 2o°C above
room temperature have been observed at the outer side of the rotor.Consequently
the inner parts should be warmer, and temperature gradients across the
measuring chamber are to be expected. Fortunately, the inner wall of the
chamber is warmer than the outer wall, and therefore the temperature gradients
is in favour of a stable stratification of the flow. However, the gas inside
the chamber is altered in density and viscosity, and consequently changes of
the flow rate could occur. The experiments show that such changes in
calibration are negligible. Latex aerosols have been collected on the same
foil for a cool and'a warm rotor. As fig. 13 demonstrates, there is no
significant difference in flow rates between a cool and a warm rotor.
24
-------
2oo,
loo
2o 22 2M 26 28
DEPOSITION LENGTH, ( MM )
24 26 28
DEPOSITION LENGTH ( m )
Fig. 13: Number size distributions of single latex particles for cold (solid
o
line) and warmed (dashed line) rotor I. 1), 6000 rpm, AFLO 3,14 mm, TFLO
y ? 2
18,8 mm . 2),5ooo rpm, AFLO o,o9 mm TFLO 18,8 mm . Particle size o,557jjm
diameter.
25
-------
Discussion
As indicated by the calibration curves of the ROSL spectrometer the
angular velocities of the particles do not differ significantly from the
angular velocity of the rotor. This property opens a way to precision
measurements of aerodynamic diameters. The precision depends on the errors
in measuring the deposition length, the total flow rate, and the angular
velocity. The errors in the deposition length are about 0.1 mm at the
large particle end of the deposition foil and about 0.3 mm near the
small particle end. These errors could still be lowered by elaborate mea-
suring techniques, however, care must be taken that the foil is smooth
and perfectly contacting the wall during operation in order to prevent
flow disturbations. The precision of the angular velocity measurements
is sufficient, the velocity is easily measured and well controlled by the
regulated power supply. Precision measurements of the total flow rate are
a serious problem. In order to use external flow meters the rotor is to
be connected leakageless to the housing. Solutions as represented for
rotor I are not satisfactorily because they do not work at higher rotati-
onal speeds and they do not allow continuous functional control during
operation. Probing the flow with particles of well known sizes poses the
problem of the precision of size measurements. As demonstrated by fig. 13a
the scatter of the particle size data is unacceptable for precisely determ-
ining the calibration curves, and consequently the total flow rates would
be submitted to the same uncertainty.
The capacity of the centrifuge to work at speeds up to 12ooo rpm or
more is still unused. Experiments have been performed successfully in
collecting small sized Latex aerosols near the slit. Classification
turned out to be perfect justifying the conclusion, that particles with
sizes around 0.05 jjm could be sized.
26
-------
4 5
lo
2o 3o 4o 5o 7o
DEPOSITION LENGTH (MM)
Fig.lSa : Calibration curves for rotor II. 5ooo rpm and TFLO lo,6 mm (upper
2
curve) , 5ooo rpm and TFLO 4,7 mm ( middle curve ), 9ooo rpm and TFLO
o
4,7 mm ( lower curves, dashed and solid), o sizes by DOW, O sizes by (.£3),
C> sizes by
27
-------
THE DIFFUSION BATTERY
Introduction
Aerosol particles undergo BROWNian movement, which causes them to
migrate through the gas. If they meet a surface they will stick to it and
never rebound. The chance for a particle to be deposited on walls depends
on particle size and on the dimensions of the container, once the gas, its
temperature and its pressure is given. The smaller the particles are and
the narrower the container, the higher are the chances.for deposition by
diffusion.
TOWNSEND (23) in 19oo was the first to derive an analytical equation
which applies to diffusion deposition in cylindrical tubes. Later on,
GORMLEY and KENNEDY (24) and many others worked out similar solutions of
this problem. Progress with respect to certain corrections has been made
by the use of computers (TAN (25), TAN and HUE (26)). But diffusion is not
always the sole mechanism for deposition in tubes. Sedimentation within the
tubes and impaction and interception at the entrance must be taken into
account, especially for particles of more than o,l pro in diameter.
Many investigations have been performed in order to solve these
problems theoretically, but the results are to some extent in disagreement.
Therefore it has been the goal of this work, to study the effects of
impaction and interception experimentally, by applying a diffusion battery
and the ROSL aerosol spectrometer to monodisperse NaCl aerosols.
28
-------
s_
o
o o
<4-
c
CD
O
10
-*:
o
£
CO
"o c
CO O
O •!-
O) (O
CO r—
3
(C -i-
fO
O)
-o
OJ
s-
(U
Q- -M
-o
O
o
o
CD
-C
i- CM
(U
O)
a)
.c
-U
s_
o
d)
fO
C
OJ
Q.
o
o
OJ
CO
"o
CO
o
CD
CO
CD
J=
+o
M-
O
O
to
o
0)
fO
C
O
O T-
3 T3
-o
O 0)
S- -C
f-i 4J
CD
-------
Fig.15: Cross section of the aerosol generator at the position of a heater
element. A, fire bricks.B, ceramic tube for sheltering the heating wires.
C, ceramic tube for carrying the heating wires. D, center tube for aerosol
production. E, thermo couple. F, heating wire lead.
30
-------
The earlier classifiers by diffusion, i.e. the diffusion batteries,
were made from bundles of long and narrow tubes or from packages of sturdy
plates with narrow spacings. The large draw back of such batteries was their
bulkiness and weight. Recently SINCLAIR (27) described a handy diffusion
battery which he has used in field work. For this battery thin plates with
a large number of extremely fine, collimated holes (collimated hole
structures = CHS) have been used. MATTESON et al. (28) specified another
diffusion battery with CHS plates, where the plates had about 3.1o holes
of 15 ym diameter. A similar battery of six, nominally identical CHS plates
has been used for this work (29) (3o).
A special generator for the NaCl aerosol has been developed. The
generator can produce fairly monodispersed aerosols in the desired particle
size range (3o), (31).
The NaCl aerosol generator
The generator is a tube furnace with a center tube of approximately
1,5 m length. Different tubes have been used, with 3 cm o.d. and 2,7 cm i.d.,
or 2,5 cm o.d. and 2,o cm i.d.. The center tubes are made from pure
aluminium oxide which is highly resistant to attack by NaCl vapors. The
center tube is surrounded by other ceramic tubes which carry three seperated
heaters. These heaters which could be operated up to looo°C have lengths of
2o cmj they are arranged serially with distances of 2o cm between them.
They are electrically independent thus forming three independent heating
zones. The heaters are protected by a third ceramic tube. Finally, the
whole tube system is imbedded in thick fire bricks for thermal insulation
(see fig. 14 and fig. 15). The bricks are resting on iron rods, and the
furnace is hold together by two springloaded end plates which carry special
flanges for the aerosol and the gas ports. The aerosol port is equipped
with an ejector for rapid dilution of the aerosol. The seal between the
center tube and the flanges is made by a silicone elastomer, which pos-
sesses some flexibility after hardening. The center tube undergoes
extension and bending by heating and therefore cannot be connected rigidly
to the ports.
31
-------
The electrical power of each heater is controlled by thermostats ,specifi-
cally PLASTOMATIC SCR by PHILIPS. The reference temperatures are measured
by three thermocouples just above the heaters in the middle of each zone.
These temperatures are lower than the gas temperatures in the center tube,
and a calibration is needed to relate the gas temperatures to the reference
temperatures of the thermostats. Such calibration curves are shown in
fig. 16 for two different tubes. In these calibrations, the highest gas
temperatures of the heating zones are indicated. Temperature profiles
along the center tube are shown in the next figures. The temperature
profiles are slightly different for a stagnant and for a streaming gas,
at a flow rate of 1 1/min. A downstream shift is observed, and a slight
decrease of the highest temperatures, but these changes are negligible
(see fig. 17). The temperature profiles are time dependent. Observations
demonstrate that the highest temperatures are reached fairly quickly and
no significant increase is found after a time lapse of one hour from the
start of heating. In the valleys between the heating zones, however, changes
are observed up to 18 hours after heating commences (see fig. 18).
For aerosol production, ceramic boats are placed in the center tube.
The zone next to the gas port, i.e. zone I, contains a boat filled with
NaF for nuclei production. Zone II contains one or two boats with NaCl
for aerosol production. Zone III at the end of the generator is left free.
Nitrogen, which is carefully dried in some applications, has been used for
carrier gas.
A drawback of the furnace is the frequent rupture of the center tube .
Thermal stresses due to the high temperature gradients along the tube are
a main source, but the influence of the NaCl and NaF va'pours should not be
neglected. The risk of breakage increases when cold boats are placed in
the furnace. The problem is still far from being solved satisfactorily.
Thin walled aluminum oxide tubes seem to stand longer than thick walled
ones; steel tubes and quartz tubes will disintegrate within days.
32
-------
8oo
7oo
600
600 7oo
THERMOSTAT TEMPERATURE, °C
Fig.16: Calibration curves relating the temperatures in the center tube to
the temperatures of the thermostats.
33
-------
O>
4->
fO
o
S3
0)
OJ
M-
M-
•o
ta
i-
d)
4J
C
OJ
o
0)
o
(/I
X
(0
QJ
fO
>
S-
O
^0 3
s- o
OJ >—
Q. M-
d) O
•4-> S-
OJ
(/) N
cr> s-
•i- 3
U- O
34
-------
-o
•o
c
to
CO
01
to
o
s-
O)
N
03
*t
5
3
S-
(1)
4J
C
0)
o
O)
4-
O
X
fl3
CD
01
c
o
fO
CO
3
rO
O)
O.
O)
OJ
1-
O)
rO 03
CD CO
CD
O)
CD >
4-> 3
O
CD
CO
Q. S-
03 O)
(O
I
OJ
CD
S-
3
O
35
-------
The Nad aerosols
The aerosols produced by the generator are fairly monodisperse. When
illuminated with a beam of white light they exhibit a sequence of definite
colours in the scattered light, from the red to the yellow, the green and
the blue. Not only does this phenomenon, i.e. the higher order TYNDALL
spectra, indicate good monodispersity, but it permits the continuous
monitoring of the particle size by measuring the angular position of a
given colour, e.g. the red (29). This size measurement represents an
optical equivalent diameter of the aerosol.
The aercsol production depends on three parameters, i.e. the temperature,
Tl, of the NaF source, the temperature, T2, of the NaCl source, and the
flow rate, Q. The temperature } Tl; influences the size and the monodispersity
of the aerosol. For high temperatures the nuclei contribute considerably
to the mass of the aerosol particles and therefore the size increases when
the temperature is raised beyond a certain level. For lower temperatures,
the nuclei do not contribute to mass, and therefore the particle size becomes
independent of the temperature at the NaF boat. The limit between these two
regimes is about 550°C (see fig. 19). The monodispersity, which decreases
for high and for low temperature, is best at temperatures of about 6oo°C.
The main purpose of heating the NaF source is the production of condensation
nuclei. The tube itself produces some nuclei, but these are insufficient in
number or perhaps inadequat in kind to effect the condensation of mono-
dispersed aerosols. With the NaF source, nuclei concentrations of the order
of lo to lo per cc are obtained. These nuclei concentrations are almost
insensitive to changes in temperature, Tl, at the NaF boat. It should be
mentioned that the NaF boat becomes unnecessary after an interval of operation,
after which it does not alter the efficiency of the nuclei production.
Nuclei are presumably produced from contaminations at the walls. The
temperature, T2, of zone II determines by evaporation the amount of NaCl
which condenses on the nuclei, and therefore this temperature dominates
the size of the aerosol. The relation between size and temperature, T2, is
fairly reproducible, if certain rules are obeyed. First, the procedure of
exchanging the NaCl boats should be performed very carefully to minimize
36
-------
1,2
Lo
o,8
6 72o°C
65o°C
Moo
Soo
600
7oo
Fig. 19: The sizes of the monodispersed NaCI aerosols in dependence of the
temperatures at zone I and zone II.
37
-------
contamination introduced by the boats. Secondly, after a change of the
thermostat temperature, the furnace is out of equilibrium for one hour,
during which time the average size of the aerosol is unstable. Finally,
the furnace should be conditioned ahead of longer periodsof stand by
under power, i.e. the NaCl boats are to be removed from the furnace, and
the tube is to be flushed with clean gas for a couple of hours to evaporate
salt remnants from the heating zones.
Zone III, which does not contain a material source, influences the
degree of monodispersity. The mechanisms, however, are not very clear.
Monodispersity is best, when the temperature of zone III is kept below
the temperature of zone II. Therefore, zone III does not act like the
classical LaMER generator's reheater, which revaporizesintermediate
condensation products, but instead controls the temperature drop at the
end of zone II and thereby the cooling rate of the growing aerosol.
The influence of zone III on the structure of the particles will be dis-
cussed later.
The flow rate determines the concentration of the NaCl vapours, and
therefore influences the size of the aerosol. More important are the effects
on the degree of monodispersity which is best at flow rates from 0,8 1/min
to l,o 1/min. Outside this fairly narrow range the monodispersity decays
rapidly. This result is another indication that the degree of monodispersity
is controlled by the cooling rate of the growing aerosol.
As already mentioned the number concentrations of the aerosols are
very high, and therefore the coagulation processes effect rapid changes
of the size distributions. Some very broad size distributions, measured
by the ROSL spectrometer after a coagulation time of 2o seconds, are
represented in fig. 2o and fig. 21. These size distributions posses a
definite structure. Besides the main peak of the single particles several
other peaks occur which belong to particles formed by two^ three and more
NaCl particles. The equivalent sizes of these peaks correspond very well
to the equivalent sizes of the doublets, triplets and other aggregated
particles in the latex aerosols (21). After dilution at the end of the
38
-------
CM CD
S-
O)
o
to
o
i-
O)
to
fO
-a
a>
tO
to
o
o
to
CD
o
S-
•p
c
0)
ra
3
CT
(U
O
IO
c
3 T3
cr> o
(1)
(O
0)
•4->
4-
O
to
4- O)
O i—
B
(U
•i- O
13 CU
Q.
to
ai
N
to
(D
o
CM
C7>
-------
o s-
« QJ
S-
O)
.a
(O
•I—
-o
i— 3
O 0"
3
JD
O)
S_
O)
cu
(O
C
-o"
o
0)
(0
CO
O
CD
fO
O)
Q.
O
OJ
O. C
CO
CD
n3
Ol
en
o>
OJ
-C
Q.
CO
dl
C
CO
40
-------
center tube, the size distributions are fairly narrow, but still in most
of them there occurs a small amount of double particles. The most "pure"
of these distributions are logarithmic normal distributions, as shown by
the almost straight lines in the log normal plot of the data (fig. 22,23).
Deviations at the upper end are due to the double particles.
The density of the particles is a very essential parameter of these
aerosols. Deviations from the bulk density which have been reported in
the literature (32), would severely restrict the further use of these
aerosols, especially with regard to their later application in the
diffusion battery experiments, where the centrifuge is used as the
reference classifier. The diffusion battery measures a diffusion equivalent
size, whereas the size by the centrifuge is a aerodynamic equivalent one.
Both sizes are correlated by the density of the particles.
The density of the NaCl particles has been determined by a method
described by MATTESON et al (32). The aerodynamic equivalent sizes of
the particles are measured by the aerosol centrifuge. The same deposits
are analysed by electron microscopy to determine the actual particle
diameters at respective deposit locations. Since the aerodynamic size,
*
D , and the diameter, D , are well known, the density is deduced fromequ.5
Pp = (D*p / Dp)-Bpg (16)
The density derived by this method has been found to be in good
agreement with the bulk density of NaCl, and therefore the diameters of
the particles can be determined from the aerodynamic equivalent size,
D , by using the bulk density. Diameters measured by the electron microscope
and by the centrifuge are in excellent agreement, as shown by fig. 24. (33).
It should be mentioned, that the errors, indicated by the bars, and the
systematic deviations from the exact relation arise from deficiences
in the microscope technique. Moreover the particles are spherical, and
the data confirm the assumption that they are perfectly solid and not
composed of aggregates of finer particles. The third heating zone seems
to favor these properties, because other authors who have produced NaCl
41
-------
Fig.22 : Number size distribution of diluted NaCl aerosols. Cumulative size
distribution ( open dots and solid line ) and differential size distribution
( insert).
42
-------
7o
30.
DO
-------
13
11
Fig.24: Comparison of electron microscopical diameters of NaCl particles to
their deposition lengths in the centrifuge. Experimental data ( open dots)
and theoretical relationship ( solid line ) assuming bulk density for the
particles.
44
-------
particles from single heating zone furnaces report surface roughnesses
and lower densities ((32), (34), (35)).
The Diffusion Battery
A six stage miniature DB was built using ^identical" CHS plates. Its
performance was investigated w[th monodispersed spherical sodium chloride
aerosols. Relative standard deviations during the experiments averaged 7 %.
The particle concentrations at the exit of the DB were measured by collecting
the aerosols on Nuclepore filters and measuring the deposited mass with
a flame photometer. The expected mass collection efficiencies were
calculated from the DB geometry and the number size distributions of
the aerosols and compared with the experimentally determined collection
efficiencies. The particle size distributions were measured by the aerosol
centrifuge concurrently with each DB experiment.
The experiments were carried out at room temperature and at temperatures
between -6° and -75°C. In a previous investigation (28) considerable
deviations between experimental and theoretical collection efficiencies
at low temperatures were reported. The results described below, obtained
with monodispersed spherical aerosols in a water-free, inert carrier gas,
do not substantiate the deviations encountered earlier.
The DB (Fig. 25) consists of one prestage (2) and six identical stages
(3) (only one is drawn) which are held together by four bolts (5) and are
sealed by teflon gaskets. Each stage contains one CHS plate (3b), pressed
into position by a steel ring (3a) and sealed by a 0.02 mm gold foil. The
CHS plates have diameters of 15 mm and a large number (ca. 4oo ooo) of not
quite circular pores with diameters of about 15 ym and lengths of about
5oo pm; porosity is 44 %. The porosity is a very essential parameter for
the calculations and has therefore been determined very carefully by four
different methods, i.e. (i) from the difference between geometric and true
volume by size and weight measurement and known density of the material,
(ii) from the difference of weight in air and water, (iii) from determination
of open and total surface area by planimetry (from photographs), and
45
-------
Fig.25: Cross section of the diffusion battery. 1), port into the battery;
2), prestage; 3), diffusion stage with steel ring, 3a), CHS plate,3b), and
exit port,3c);4), end plate; 5), clamping bolts.
46
-------
Fig.26: Tubing diagram of the diffusion battery. Aerosol port, 1), and cool-
ind coil, 2), in the thermostated bath, 4) ; 3), battery with samplin mani-
fold,5) ; 6) exchangeable filter holder and back up filter, 7) ; 8), flow
meter with regulator, 9), and vacuum pump, lo) ; 12), clean gas port with
filter, 11).
47
-------
(iv) from measured pressure drop versus flow rate across the CHS plates,
assuming Hagen-Poiseuille flow. The parameters of different CHS plates
vary in a range of about 5 % from the mean. The aerosol flow (fig. 26)
enters the DB through a cooling coil (2) (even when no cooling is
effectuated to ensure identical conditions for all experiments). At the
inlet of the DB the temperature is monitored with a thermocouple.
Each stage of the DB has an exit tube of 3 mm inner diameter with
a stainless steel valve. By switching the appropriate valves the aerosol
passes a desired number of stages (3) and is then led through the exit
tube into a sampling manifold (5). The manifold is immediately followed
by a nuclepore filter (6), which serves as the concentration measuring
device. Though a specially designed filterholder (35) permits quick
exchange of filters, this procedure still is slower than reading e.g.
a condensation nucleus counter. Nevertheless the filtration method
was chosen because of the difficulties in compensating for the pulsations
in aerosol flow introduced by most condensation nucleus counters. Tubing
can be held very short and particle losses are minimized; in addition
care is taken to avoid high particle velocities or sharp bends and
eddies with subsequent impaction in the flow system. The particle concen-
trations are measured alternately at the prestage and behind the selected
number of stages of the DB. Between measurements the manifold is always
flushed with filtered clean gas through an additional inlet. For experiments
at low temperatures the DB is immersed in a cold bath, consisting either
of a mixture of KaCl and ice (to - 2o°C) or of dry ice - acetone (to -75°C).
All gas streams entering the DB are dried with Mg (ClO^^and P205. (If
the aerosol is not dried sufficiently, water vapour condenses on the CHS
plates and freezes in their pores). Glass tubing is used exclusively.
Deposition mechanisms in a diffusion battery
The theory of deposition in flow systems has been investigated
extensively; only the relevant expressions will therefore be reviewed briefly.
Particle deposition takes place by impaction and interception on the
face of a CHS plate and by diffusion and sedimentation in its channels.
48
-------
An estimation of the effects of electrical deposition by image forces
and of sedimentation in the presence of diffusion (37), (38), (39), shows
that both may be neglected. So only diffusion, impaction and interception
will be considered.
Diffusion
The semiempirical expression given by THOMAS (4o) was chosen as most
convenient for diffusional deposition in circular channels at Poiseuille
flow because it covers the whole studied range of the diffusion parameter
which is given by
D.l.n /17x
y = •„ Q • (17)
D ..... diffusion coefficient
1 ..... channel length (plate thickness)
n ..... number of parallel channels
Q ..... volume flow rate
THOMAS gives the following expression for the diffusional deposition
efficiency:
ED = 1 -{0.819 exp(3.657v>) + 0.097 exp (-22. 3y) +
+ 0.032 exp( - 57u) + 0.027 exp (- 123y) +
+ 0.025 exp( - 75oy) (18)
Results obtained from the analogue expressions ofTWOMEY (41) differ only
by less than 0.1 % from the values of THOMAS (4o) in the range of^ /row,
0.001 to 0.05.
Impaction
A simple model of particle flow to the surface of filters was studied
49
-------
by PICH (42). He gives the following expression for the impaction
efficiency E.:
r _ 2 Z
~
6 =
1 - /P
z = 2 St. /G + 2 St2.G(exp{l/St./G}-l) (19)
with
P porosity
St Stokes number
SMITH and PHILLIPS (43) found values for inertial collection by solving
the differential equations of flow with a computer. Their model is probably
closer to reality and also accounts for finite particle size, but the lack
of analytical expressions makes its application difficult, so it is not used.
Interception
A formula governing interception efficiency ED is given by SPURNY (44).
It can be derived from simple geometric considerations and may be applied
to CHS plates treating them like Nuclepore filters in first approximation
ER = NR (2 - NR) (2o)
where NR is the ratio of particle to pore diameter.
SMUTEK and PICH (45) studied a more complicated model of flow which leads
to rather complex expressions, giving a combined impaction and interception
efficiency as well as an impaction efficiency alone; in the following they
will be refered to as E.D' and E.'.
IK 1
50
-------
Combinations of Deposition Mechanisms
Deposition by diffusion, impaction and interception occurs simultaneously,
so expressions have to be derived for the combination of these processes.
One possible way is to assume their independence. Total penetration through
the system (e.g. a filter, a CHS plate) is then found by multiplication of
the partial penetrations. Hence the combined penetration for diffusion and
impaction, T~., is
TDi = VTi <21)
and for additional interception it is
TD1R=TD.T..TR ; (22)
the respective collection efficiencies are
EDi - ' - TD1 ' ED * Ei - ED'Ei <23>
and
EDiR=1-TDiR> (20
the use of either T or E is a question of convenience.
Equ. (21) is accepted by several authors for filter deposition (39),
(44), (46), (47), generally in the form of equ. (23). It has also been
used to correct DB data for impaction (28). The combined deposition
efficiency of all three mechanisms, however, does not seem to be given
correctly by equ. (22). Semiempirical formula have been used instead,
which are derived from equ. (24) by introducing weight factors for
interception (46), (47). These factors are determined from experiments
and reduce the relative influence of interception considerably. A
theoretical explanation has not yet been given and the fact that no
filter material is perfectly regular introduces additional problems.
51
-------
The expression which was used here is
EDIR-V^I) (ED + ERO^-NR)) (25)
with the same numerical constants as used by SPURNY et al. (47).
Another possible formula for combined penetration is derived from equ.
(22) by using the SMUTEK-PICH expression for impaction and interception,
E-R'» instead of equ. (19) and (2o). Combined penetration then reads
TD1R = TD.(1-E1R') (26)
It is interesting to note that the values computed for impaction
efficiencies from the SMUTEK-PICH expression are negligibly small compared
to those from equ. (19) in the range of particle sizes and velocities
studied. The values from the SMUTEK-PICH expression represent practically
only interception as can be seen from fig. 27 showing computed curves
for combined impaction and interception efficiences. The flow rate is
chosen as parameter and increases from 0.2 to 1 1/min in steps of 0.2 1/min
which represents the typical range of experimental values. For curves
1 to S^equ. (25) for EDiR was used, setting ED = 0. Efficiency increases
markedly with flow rate by increased impaction while interception remains
constant. For the same variation of flow rates the efficiency curves from
the SMUTEK-PICH formula, E-D', coincide practically (curve 6 for o.2 1/min
* K
and curve 7 for 1 1/min) because the contribution from impaction is very
small.
The above discussion shows the difficulties in selecting appropriate
expressions for deposition from impaction and interception purely from
theoretical considerations - in contrast to diffusional deposition where
numerical results calculated from several authors are nearly identical.
Values derived from different impaction and interception formulas and
their combinations were therefore directly compared to experimental data.
52
-------
uu
»-h*
u
0,2 0,5 ( pM ) 1,0
Fig.27:Theoretical impaction and interception efficiency of CHS plates at
different flow rates. Curve 1 to curve 5, efficiencies increasing from o,2
1/nrin to 1 1/min. Curves 6 and 7, theoretical expression with underestimation
of impaction.
53
-------
Results and Discussion
The main goal of this work is the comparison of theoretical and
experimental values for diffusion coefficients at various temperatures.
For this purpose the diffusion deposition efficiencies must be derived
with high accuracy from the total deposition in the DB. Contributions
of different particle deposition mechanisms to total deposition were
therefore investigated. Sedimentation proved to be negligible under the
experimental conditions given.so, in addition to diffusion,only impaction
and interception were considered. The results of 13 experiments are
presented in Table 2. Four parameters are given in columns 1 to 4
characterising each experiment: mean geometric particle diameter and
relative standard deviation measured with an aerosol centrifuge, aerosol
flow rate and temperature. Columns 5 and 6 contain the measured penetration,
T , per stage averaged over all 6 stages of the DB, and its relative
exp
standard deviation o,. The calculated penetration values are listed in
columns 7 to 12 for the following set of mechanisms and their combinations:
Tn, T., RR for diffusion, impaction and interception,
respectively, from formulae (18), (19) and (2o).
T combination of diffusion and impaction by formula (21)
'DI
Tn-p combination of diffusion, impaction and interception
U1K by formula (25)
Tn.T. ' combination of diffusion, impaction and interception
u 1K by formula (26)
Values are always calculated for a diffusion length of one stage, with
parameters averaged over all 6 stages. Other possible combinations with
TR, as TD.TR or T0-T-j-Tp» nave been omitted because their values were
too far off from T according to our measurements.
C A U
For better correspondence between computed penetration and experimental
values, T , the penetrating mass for each fraction of the aerosol number
54
-------
at.
•r—
O
Oil
Q
C£.
co
•z.
o
I-H
co
Lu
U_
i—i
Q
U-l
u:
i—
u_
o
co
o
i—i
CO
co
t—i
co
CU
^°^ ^—
-4_> r—
+ 1 tO
leviation
cperimentc
experimer
cu cu
•o .c
i. « +J
to CL
-o x cn
c: cu c
to i— -r-
4-> to
LO
eu >,
•p o -o
•r- E
i- +J -i-
tO -r-
Q. « 1—
Q.
Q. CU CM
Q 4-3 «— 1
1
r-r ^ rv.
-------
distribution was calculated independently. Total penetration was then
obtained by adding the single fractions and dividing by the whole incoming
mass of aerosol.
In Table 3 the relative differences in percent between T and
theoretically predicted penetrations are listed using the same set of
combinations as in Table 1 and additionally Trs.T~ and T~.T..T~. The
mean values given below each column can be interpreted as systematic
deviations and thus indicate whether the respective theoretical expressions
tend to give higher of lower values than total experimental deposition.
Diffusion alone (column 1) obviously leads to a value of T which is
markedly higher than T (+ 4 %); diffusion together with impaction
cXp
(column 2) gives the closest fit to the experiment (+ 1,6 %) when equ.(21)
is used for impe.ction. Column 2 with an average difference of 3 % between
theory and experiment shows clearly, that interception is overestimated
by equ. (2o), as already mentioned above. The values in columns 4 to 6
comprise all three mechanisms. Simple combination of these mechanisms by
multiplication (column 4) gives penetration values that are markedly lower
than the experimental ones (- 5 %). The probable reason again is an
overestimation of interception. The average deviation for the semi empirical
equ. (25) from SPURNY (47) in column 5 is about the same as for the combined
impaction-interception equ. (26) from SMUTEK-PICH (column 6). It is somewhat
higher than that for T.,. in column 2, but still lies within the error
margins of about 3 % for our DB experiments. In Fig. 28, finally, the
above results are displayed graphically for one selected DB experiment.
Values of T for increasing numbers of DB stages (dashed line) are drawn
C AvL)
together with the error limits of the measured penetration. The other curves
correspond to the above set of computed values for T. It is interesting to
compare the values of diffusion coefficients D and particle diameters D
from DB measurements to those from the aerosol centrifuge. Considering
that the relative deviation of T from T_j, averaged over all data,
was only 1,6 % (table 3) with constant tendency in the same direction
(such an error is insignificant, for a DB measurement !), it was at first
surprising to find that experimental and theoretical values of D and D
differ by about lo %. It may be recalled here, that this is a purely
56
-------
PARTICLE SIZE: o/'fi fn
FLOW RATE : o/67 L/MIN
TEMPERATURE : - 69 °C
DIFFUSION IENGTH, STAGE NR.
5
Fig.28: experimental transmission data compared to theoretical expectations,
The shaded field illustrates the region of experimental confidence.
57
-------
TABLE 3. DEVIATIONS OF THEORETICAL TRANSMISSIONS FROM EXPERIMENTAL
VALUES. T(experim) = loo %
1)
TD
2,9
2,5
2,2
1,2
7,9
3,5
4,8
3,7
3,2
4,2
6,2
4,8
5,8
2)
TDi
1,9
1,8
1,7
o,3
1,7
1,4
2,3
2,2
1,6
o,2
2,2
1,3
2,1
3)
TDR
-4,2
-2,4
-1,3
-3,5
-0,9
-3,8
-2,8
-2,4
-3,o
-6,4
-1,2
-4,o
-2,6
4)
ViTR
-5,o
-1,8
-1,7
-4,3
-6,7
-5,8
-5,1
-3,8
-4,4
-lo,o
- 5,2
- 7,2
- 4,9
5)
TDiR
-3,2
-3,1
-0,8
-3,o
-4,o
-3,7
-2,9
-2,1
-2,8
-4,4
-2,8
-4,5
-3,4
6)
VIR-
-0,5
-1,5
-2,7
-3,2
-2,8
-3,o
-1,5
-1,5
-0,8
-4,1
-1,9
-1,7
-1,7
4,o
1,6
-2,9
- 5,o
-3,1
-2,1
-------
numerical consequence of the evaluation method: The value of T is at
\Z f\lJ
first corrected for impaction by dividing by T-; then the transcendental
equ. (18) is solved for y from which D and D are finally obtained. As T
depends exponentially on y , errors in D cannot be simply calculated from
the relative error in T, they also depend on the absolute value of T.
For this reason T was used to study deposition in our DB and not D or D .
As already mentioned above, the DB was also tested at lower than
ambient temperatures, between -6° and -75°C, with spherical sodium chloride
particles in a stream of carefully dried nitrogen. Results do not differ
significantly from the measurements at room temperature listed in Tables 2
and 3. It should nevertheless be mentioned here, that deviations were
observed in a few early experiments; these were in the same direction
and order of magnitude as earlier reported data (28). It seems, though,
after studying possible correlations to other effects that most probably
the cold bath surrounding the DB was not sufficiently homogeneous in
temperature and so thermal diffusion could have taken place. The additional
deposition effect was about 5.5 %. One suche experiment at -74,5°C, e.g.,
gave a penetration T which was about 7 % below the theoretical value
GXp
Tn. listed in column lo of table 1. This led to a diffusion coefficient
fi -fi
of 1.9 x lo cgs units instead of o.63 x lo" - the difference is 2oo %.
In subsequent experiments care was taken to avoid this source of error
and good agreement was obtained afterwards.
59
-------
THE FIVE STAGE LOW PRESSURE IMPACTOR
Introduction
Cascade impactors are used for the determination of size distributions
of airborne particulate matter. In the impactor the aerosol is precipitated
on a series of stages, where each stage collects a certain size category
of the aerosol. Cascade impactors have been developed in many different
designs, mainly for application to special purposes. A stimulating goal
for further developments of cascade impactors is the determination of the
mass size distributions of urban aerosols in correlation with their chemical
composition.
A five stage cascade impactor for mass analysis is represented in
fig. 29. The aerosol enters the instrument, by a ring slit between the top
cover and the cylindrical housing. This slit and the inner side of the
cover constitute the entrance stage, i.e. stage 5, which has a cut off
point of 25 ym. Particles of larger sizes are deposited on a thick layer
of vaseline in order to prevent blow off and bouncing. The rest of the
aerosol passes to the next stage which collects all particles larger than
6,4 ym. This stage, i.e. stage 4, consists of a circular orifice plate
with a set of identical holes with sharp, rectangular edges, and of an
impaction plate with a large center hole for the flow.Distance between
these plates is made by a ring at the periphery of the stage. This
arrangement is repeated in each of the following stages. The cut off
points of these stages are 1,6 ym for stage 3; o,4 ym for stage 2;
and o,l ym for stage 1. The volume flow rate is 80 1/min at standard
60
-------
Fig.29: Cross section of the five stage low pressure impactor. A, base plate
with critical orifice.F, and vacuum pump adapter, G; parts of the housing:
B, cover; C, mantle ; D, nut ; E, impaction stages.
61
-------
TABLE 4. DATA OF THE IMPACTOR STAGES
1) 2) 3) 4)
D* W S L N p./p0 Pa/pQ
4 6,25 o,55 l,2o l,lo 11 l,oo l,oo
3 1,58 o,17 0,60 o,5o 24 l,oo 0,996
2 o,38 o,o7 o,2o o,2o 33 o,996 o,859
1 o,lo o,o5 o,15 o,15 2o o,859 o,335
1) stage number
2) cut off diameter (
3) W, orifice diameter (cm); S orifice-to-plate distance; (cm);
L, orifice length (cm); N, orifice number
4) p , atmospheric pressure; p^, pressure ahead of the orifices;
p. pressure below the orifices.
a
62
-------
temperature and pressure. The flow is controlled by a critical orifice
in the base plate of the impactor. Other data on this impactor are listed
in table 4.
Each orifice plate has a number of identical orifices which are
placed on a circle, concentric to the axis of the stage. This arrangement
effects an equal flow partition, and consequently the aerosol deposits
are equally partitioned during sampling. For sufficiently large distances
between the orifices, the deposits are localized in a number of separated
spots. Investigations have shown that these spots carry equal amounts of
the total deposit.
The impaction plate is covered by a foil. This feature is advantageous
for weight analysis because the foils can be cut from very thin material
to keep their weight at a low level and because the selection of foil
materials can vary according to the needs of the analysis. Aluminum foils
of loxtm thickness have been used successfully for measuring mass size
distributions of solid and semi-liquid particles, and foils of filter
membranes such as mi Hi pore cellulose membranes are very convenient for
the collection of liquid materials like oil.
A certain amount of the material is deposited on other surfaces of
the stages. These wall deposits are of significance, because they limit
the practical use of the impactor. In many cases there is no possibility
or justification to determine these deposits separately in order to include
them into the results. Therefore they should preferably be negligibly small.
The impactor described here has wall deposits the order of a few percent
of the total deposits of a stage, an amount which may be neglected in
practice. The wall deposits, however, increase when the total load of a
stage is raised beyond certain limits.
Particle size analysis
An impactor stage is composed of an orifice plate and of an impaction
plate, which are separated by a distance S. The orifice plate has circular
63
-------
orifices of diameter W. The orifices have the length L.
The aerosol is drawn through the orifices by means of a vacuum pump.
Aerosol jets are produced, which are deflected by the impaction plate.
For sufficiently large orifice distances, S > W, the flow has three sections,
i.e. the free jet flow ahead of the impaction plate, the: stagnation flow
at the impaction plate near the stagnation point, and the free wall jet
at the impaction plate outside the region of stagnation. For sufficiently
long orifices, L > 2W, and for moderate pressure drops along the orifice,
the free jet has the diameter, W, and its velocity is identical to the
gas velocity, u , in the orifice. The wall jet has similar characteristics.
The velocity is again u , if friction losses are neglected, and the jet
flows in a layer of thickness, H = (1/8).W, approximately. (48).
In the region of stagnation the gas velocities undergo changes with
respect to amount and direction, and relative velocities between the
particles and the gas occur. The particles change their velocities
according to the STOKES law, which is given by
J«-(3.ir.ng.Dp/yBpg).tf -U) (29)
where ^ is the velocity of the particle, il the velocity of the gas at the
particle's position, and m the mass of the particle. The slip correction
factor, B , depends on the KNUDSEN number, K = 2.A-/D , according to
r y y M
equ . (3o)
Bpg = (1 + K.(1,257 + o,4 exp (-1,1.K"1)) (3o)
where x is the mean free path of the gas molecules which equates to
Xg = lo,8 . ( ng/P) • (T/M)1/2 (31)
where the gas has the temperature, T, the pressure, P, and the molecular
weight, M. It should be noted that T and P are the values of temperature
and pressure at the location of the particle.
64
-------
By introducing the relaxation time, T , of the particle
,p- (l/l8).(D2p.pp.Bpg).(l/ng) (32)
the equation of motion (29) becomes
) (33)
which shows very clearly the significance of the relaxation time. Changes
of the particle velocities are inversely proportional to the relaxation time,
with regard to comparable relative velocities. Consequently, in the region
of stagnation the trajectories of particles with large relaxation times are
almost undisturbed continuations of the trajectories in the free jet. These
trajectories will end on the impaction plate. Sufficiently small particles
follow almost perfectly the streamlines of the gas, and therefore their
trajectories do not end on the impaction plate.
The separation curve T(T ) is a very convenient means to describe the
efficiency of the stage. T(t ) is zero for sufficiently small particles,
and its numerical value is one for large particles. In an intermediate
range of relaxation times, AT , the separation curve increases from
T = 0 to T = 1 monotonously, at ideal conditions. In the range ATQ the
separation curve assumes the value T(T ) = o,5 for a certain relaxation
time, T , i.e. the critical relaxation time. This time is determined by
calculation or experimentally. The approximate numerical value for the
critical relaxation time is estimated by considering the particle flow in
the region of stagnation. All of the deposited particles, at least those
with relaxation times near the critical time, enter the region of stagnation
with the velocity, u , of the jet. Their path from the periphery of the
region of stagnation down to the plate, vertically, is of the order of the
wall jet thickness, H . The smallest particles that contact the plate
do so with almost zero velocity (vertically to the plate) and arrive after
a time of the order
T1 = Ho/uQ = (l/8).(W/u0) (34)
65
-------
This time is approximately the critical relaxation time, T , for round jets.
According to equ. (34), the critical relaxation time is independent of the
physical constants of the particles and of the gas, but depends on the jet
velocity and the jet diameter, and therefore characterizes the stage
properties.
Deposition of a particle occurs when its relaxation time is larger than
the critical relaxation time, and particles are not deposited when their
relaxation time is smaller. Thus, the relative relaxation time, , of a
particle is decisive for deposition,
* = Tp/To = (4/9)-(£pp-VV-(Uo/W) (35)
In accordance with this equation the collection efficiency curve is often
represented by T(). Other equivalent parameters, which differ from )
assumes the valueT^) = o,5 at = 0,88. In the intermediate range, which
is for most impactors less than A = o,3.<|> , the collection efficiency curve
increases monotonically, and may be influenced by the surface structure of
the impact!on plate or by flow turublences. However, the exact shape of the
curve in this region is not very significant for the performance of a
cascade impactor.
The aerodynamic equivalent size
In the stagnation flow different particles can exhibit trajectories
which are identical point for point in laminar flow, or which coincide on
an average curve in turbulent flow. Regarding the points or short elements
of such a trajectory, one may state that it is correlated to a definite
relaxation time. Consequently, all particles on such a path have identical
66
-------
relaxation times, and can not be discriminated aerodynamically, i.e. by
observation of different paths.
The particle category, which belongs to a definite relaxation time,
T , does not only contain spherical particles, but also particles of many
other regular and irregular shapes. The spheres within this category possess
2
the relaxation time T = (DD-PD-BDa)/(18.'j ), ancl among them is the sphere
with the density p = p = 1 and with the diameter, D ,
De = ((V18.ng)/(Pe.Beg))1/2 (36)
This diameter is the aerodynamic equivalent size for the particle category
under consideration.
For a certain relaxation time, the equivalent size can not be deduced from
equ. (36), because the slip correction factor is unknown. But this equations
leads immediately to
De •Beg= .T - o,88.T into equ. (36)
DQ = ((o,88.18.ng.W)/(8.u0.pe.Beg))1/2 (39)
and is calculated by using equ. (38) for the KNUDSEN number KQ.
67
-------
Impactors are often tested with spherical particles with densities
different from p = 1. For identical relaxation times the relation
2 _ 2 •
equivalent size is defined by
partic1es and their
De =
In this formula the brackets accompanying the slip correction factors
indicate their dependence on the mean free path and of the particle diameters,
In consequence, the equivalent size of a given particle with diameter,D,
and density, p , will vary for different mean free paths, x . However,
the variations are small for common aerosols.
Deposits in the impactor stages
The main deposits, which occur immediately beneath the orifices, are
round spots with a diameter of 1,2 times to 1,7 times the orifice diameter.
These spots are clearly isolated when the interorifice distances are large
enough, and for this reason values of four times to seven times the orifice
diameter have been chosen. Beside the spcts other deposits occur. Halos
surrounding the spots are found which at low loads are clearly separated
from the spots by an almost particle - free zone. The mechanisms of halo
formation are still controversial, but it has been proposed that turbulence
in its very becinningsis responsible for this deposition (49).Deposits occur
between the spots along a narrow straight line, in radial direction with
respect to the axis of the stage. This is a very interesting phenomenon,
because the line appears at the confluence of two wall jets from adjacent
orifices. In this region the wall jets detach from the impaction plate
forming a new, upward flowing jet. The existence of this jet is indicated
by straight deposit lines on the upper plate in between the orifices.
The deflection of the wall jets requires a stagnation flow in the region
of confluence, and consequently the deposit line marks the stagnation line.
Moreover, turbulence along this line is indicated, because particles would
not be deposited otherwise. The amount of material in the halos and in the
68
-------
stagnation lines is in most cases a few percent of the mass in the spots.
Therefore these deposits do not affect the results seriously when they
are integrated into the measurement, i.e. by weighing the samples.
Other deposits occur on the inner wall of the ring spacers where the
wall jets hit, and at the underside of the orifice plate as already mentioned
above, and predominantly around the mouth of the orifices on top of the
orifice plate. These wall deposits limit the practical usefulness of the
impactor because they are usually not easily included in the measurements,
and therefore they should preferably be small. The wall deposits of the
impactor have been determined by using vaseline aerosols, produced by
spraying from a heated nebulizer and containing a soluble dye, i.e. sudan
red, as a tracer. This mixture disolves readily in toluene, and consequently
the amount of dye as well as the deposited aerosol mass can be determined
by absorption photometry. In addition, this aerosol permits the control
of the weight analysis by photometry.
As shown by the graphs in fig. 3o, stage 2 and stage 3 have wall losses
of 3 % to 4 % of the total mass deposited in the stages for moderate loads
of a few milligrams per stage. For higher loads the amount of wall deposit
increases. One possible explanation for this behaviour is based on the
assumption that the wall deposits of vaseline aerosols build up a loose,
dendritic structure which is permeable to the flow. In such a case additional
amounts of particles would be removed by filtration. It is another consequence
of this assumption that the deeper layers of the wall deposits should become
less permeable in the course of time, and therefore a "saturation" in the
amount of wall deposit should be observed for high loads. Such a trend
is probably indicated in the graphs for stage 2 and stage 3. The wall
deposits of stage 4 are substantially larger than those of stage 3 and stage 2,
and in addition they scatter broadly between lo % and 3o %, as indicated
in fig. 3o. It is not clear if these high amounts are intrinsic to stage 4,
or if they are related to the special mass size distribution which has a
low and varying amount of mass in the size range of stage 4. From the
observations one may conclude that the amount of wall deposit is higher
when there are fewer particles above the size cut point of the stage.
69
-------
3oX
2oZ
loX
D
O
0
>°»
&•
-------
The wall deposits of stage 1 have not been determined reliably enough for
presentation of data at this time. However, the measurements indicate wall
deposits in amounts of 1 % to 3 %.
Sample partitioning has been investigated by means of the vaseline-
dye aerosol. After sampling the aerosol, the aluminum collection foils are
cut into pieces, each one containing a spot. The mass is determined photo-
metrically. The first results were discouraging, because deviations of
5 % to lo % from the average spot mass occured (see fig. 31). However, these
numbers were not in agreement with the flow measurements for single orifices,
which indicated errors within I % to 2 % of the average flow. It turned
out, that the discrepancy is related to the procedure of filling the aerosol
chamber used in these experiments. If this chamber is filled inhomogeneously,
the impactor collects on one side particle-free air for a longer period of
time than on the other side. This has been proven twofold. First, homogenization
of the chamber aerosol by stirring the air during filling lowers the
deviations to acceptable limits of 2 % to 3 % of the average spot mass
(see fig. 31). Secondly, at non-uniform filling the deviations exhibit
systematic structures like maxima or minima, and these occur in the same
angular positions with respect to coordinates of the chamber, irrespective
of the angular position of the stages relative to the impactor (see fig. 31).
Moreover, the results indicate that the flow in the impactor does not mix
up even after having passed several stages.
In routine work the samples are weighed with a semi-micro balance
with a reproducibility of + lo yg. The weight of blank aluminum foils is
determinate with total errors of + 15 yg to + 2o yg depending on the skill
of the experimenter and on the atmospheric conditions. The errors introduced
by the routine sampling procedure are still higher. In this case the blank
aluminum foils are transferred from the balance directly to the stages, and
after closing and opening the impactor, are weighed again. By these manipu-
lations, the errors increase to a total of + 3o yg to + 4o yg. (The
manipulation errors would increase still more, if the foils were stored in
containers). Such error limits are quite satisfactory for most practical
purposes. The quality of the method is demonstrated by comparing the mass
71
-------
Q REFERENCE ANGLE
v\
-lo*
0 REFERENCE ANGLE
d REFERENCE ANGLE_ 2TL
-loZ
0 REFERENCE ANGLE
2TT
Fig.31: Variations of deposited mass among the spots of stages 2 and 3
of the impactor, at inhomogeneous (a) and homogeneous (b) sampling conditi
ons. Examples (c) and (d) illustrate the mass variations for subsequent
impactor stages.
72
-------
C/5
ir
3o
2o
lo
FIG .32
3 STAGE fe. 4
E
1
3o
2o
lo
.2
PARTICLE DIAN TER
.8
3.3
RG33
12.8
-------
size distribution by gravimetry to the mass size distribution by photometry
(see fig. 32).
First measurements of urban aerosol were performed in January 1978.
The sampling station is on an open balcony lo m above ground in the back
yard of the institute building. Sampling periods of three to four hours
proved to be sufficient for reliable measurements. The mass size distributions
show definitely the accumulation mode of the urban aerosols (see fig. 33).
The modal equivalent diameter is in the size range from o,4 ^m to 1,6 ym
for all size distributions during the sampling period. Coarse particles
are rare because of the height above ground and because of the remote position
of the sampling site with respect to the dust sources. A series of subsequent
measurements during several days has been conducted. These aerosols exhibit
uniform mass size distributions with respect to shape, but the particle
concentration vary strongly during the day, reflecting the variations of
the strength of the sources as well as the variations of the atmospheric
exchange conditions (see fig. 34).
Note on backup filters
Constant flow rates are necessary for operating impactcrs correctly.
Consequently the pressure drop across the impactor, or the pressure in the
final stage must be controlled. This is easily achieved by critical orifices
at the exit of the impactor. According to our experience the method is
practicable even for heavier deposits because these do not influence the
flow resistance of the stages.
Backup filters will introduce additional resistances which have different
initial values and are changing continuously due to the filter loading. These
problems may be overcome by carefully selecting the filter material, or by
automatic and manual control of the pressures. With respect to labour and
equipment, however, these measures are more expensive than critical orifices.
74
-------
fO CM
-
0)
>,
•—
S-
3
o
(/I
s_
O)
co
O) CT>
O) t—I
S-
O
4->
O
to
o.
OJ
co
S-
-Q
•r- co
75
-------
For a given aerosol the need of a backup filter depends on the pro-
perties of the impactor. The Viennese urban aerosols, e.g., exhibit an
almost log normal mass size distribution in the fine particle range, with a
mode in the range from o,5 ym to o,7 ym. The standard deviation is such
that the size range below o,l ym does not contribute considerably to the
total mass of the aerosol. In this case the low pressure impactor with its
final cut off size of o,l ym will deliver a fairly valid estimate of the
total aerosol mass-, commonly used impactors, however, would not because their
final stages have cut off sizes around o,5 ym or larger. These impactors are
to be equipped with backup filters, or additional stages, in order to complete
the data on the aerosol.
Nevertheless there might be interest in the ultrafine particles, espec-
ially when questions arise about their chemical nature. To meet such needs
the low pressure impactor should be equipped with another stage, rather than
with a backup filter. The stage would permit the application of critical ori-
fices for flow control and it would offer the advantage of concentrating the
particles on small areas, compared to the area of adequate backup filters.
76
-------
THE CONDENSATION NUCLEI COUNTER
Introduction
Condensation Nuclei Counters (CMC's) are frequently used to measure
the particle concentration in the atmosphere. However, it is not clear,
if all atmospheric particles are activated as condensation nuclei, grew
to visible size and can be detected. Investigations of the heterogeneous
nucleation in supersaturated vapor are required to determine detection
limit and sensitivity of CNC's. Furthermore, additional information
about the nuclei aerosol can be obtained from measurements at different
supersaturations.
The Expansion Cloud Chamber and its data acquisition system
Over a period of several years an expansion cloud chamber with digital
process control and on-line data acquisition has been developed (5o). The
experimental arrangement is shown in fig. 35. The necessary supersaturations
are generated in a pressure-defined subpressure expansion cloud chamber EXP.
For observation, the droplets, growing in the expansion cloud chamber, are
illuminated by a He-Ne laser beam. During the expansion process and the
subsequent droplet growth three parameters are monitored and recorded in
the memory of a digital storage oscilloscope (transient recorder TR):
1) Gas pressure in the expansion chamber, measured by a piezoelectric
pressure transducer PT.
2) Intensity of the light scattered by the growing droplets under
a selectable fixed scattering angle, measured by the photomultiplier PM.
77
-------
PROCESS CONTROL
VC
Fig.35: Diagram of the data acquisition system of the condensation nuclei
counter.
78
-------
3) Intensity of the light transmitted through the expansion chamber,
measured by the opto-element OPT.
Because of the high data rate during a single measuring run it
appeared to be necessary to interface the experimental apparatus to a
digital computer (Digital Equipment Corp., POP 15/3o). The data, obtained
during each measuring run are stored on magnetic tape. Temperatures of
the humidifier, the measuring chamber, and the ambient pressure are
recorded continuously. According to a selectable measuring program
different expansion ratios and scattering angles are chosen automatically.
In case a malfunction is detected, the system is shut off after saving
the already recorded experimental data.
The scattering intensity varies over more than two orders of magnitude,
as the scattering angle is changed. Therefore it was necessary to develop a
programmable amplifier PAMP for amplification of the light scattering
signal. Depending on the actual intensity, adequate amplification is
selected automatically.
Gas flow and pressure during each measuring run are controlled and
maintained by two pumps and five solenoid valves. The valves are operated
by an electronic valve control VC. The corresponding experimental arrangement
is shown in fig. 36. Constant flow in the humidifier HUM is provided by
the tubing pump TP in connection with VI and the bypass valve V4. By means
of the vacuum pump P in connection with the differential pressure transducer
DP and valves V3 and V5 a preset subpressure is obtained in the recipient
Rl. The expansion is initiated by opening of valve V2. The measuring volume
and the low pressure system are separated by a thin rubber membrane. The
degree of supersaturation obtained is reproducible within a range of less
than + o,5 %. The expansion time is about 5 msec.
A typical set of experimental data from a single measuring run is shown
in fig. 37. The upper curve shows the pressure as a function of time, the
middle curve the transmitted light intensity and the lower curve the scattering
79
-------
Fig.36: Aerosol and gas flow diagram of the condensation nuclei counter.
80
-------
intensity. It can be seen thet the expansion is completed before the growing
droplets become visibile. Therefore, one can conclude that the influence
of the growing droplets on the thermodynamic parameters in the measuring
chamber is negligible during the expansion. Accordingly, vapor depletion
and the release of latent heat of condensation will not change the initial
supersaturation obtained and the expension is dry-adiabatic.
The Analysis of the Recorded Data
The scattering intensity (fig. 37, lower curve), shows a series of
maxima and minima (51). Fig. 38 shows theoretical light, scattering curves
for a single droplet. The good agreement between these curves indicates
that the droplets are quite monodisperse. While for the experimental data
the abscissa is a time axis, for the theoretical curves the abszissa is a
size axis. Accordingly, after establishing a one-to-one correspondence
between the extremes in the experimental anc1 theoretical intensity curves,
size and concentration of the droplets can be determined at various times
during the growth process (52). From the position of the experimental
extremes relative to the time axis the droplet size can be determined.
On the other hand, the height of the experimental maxima is a quantitative
measure for the drcplet concentration. It should be stressed that this
procedure provides an absolute method for independent measurement of
droplet size and concentration. At high drcplet concentrations the light
extinction in the measuring chamber cannot be disregarded. Therefore the
measured height of the experimental light scattering maxima must be corrected.
Dividing the light scattering signal by the light intensity, transmitted
through the expansion chamber, will eliminate the extinction effect.
It turned out that the experimental droplet growth curves are very
sensitive to temperature changes in the expansion chamber and the humidifier.
Therefore a two stage thermostat!ng system was used which reduced temperature
variations to + o,oo5 K. For an accurate determination of the initial
supersaturation in the measuring chamber a precise knowledge of the expansion
ratio is essential. Measurement of the gas pressure using the piezoelectric
pressure transducer (PT, fig. 35) with a response time as small as 5 psec
81
-------
THETA=15°
S=128
300 400 500 ms
Fig.37: Scattered light intensity of the growing aerosol, recorded at a
scattering angle of 15° .
82
-------
on
en
CO
II
(C
in
cn
•z.
UJ
* A
t~ CD
a)
c
fO
O>
s
0)
03
O
«/)
03
S-
o
o
cu
O)
o
S_
-o
0)
(0
O
••- q-
u. o
83
-------
allows a determination of the supersaturation with an error of less than
Measurements of droplet concentration have been performed at different
supersaturations between Io3 and 4oo %. Urban aerosol particles served as
condensation nuclei. At low supersaturations only the larger perticles
will start to grow. As the supersaturation is increased during several
measuring runs, more and more particles will be activated. According to
the Gibbs-Kelvin equation the lower size limits for particles activated
at given degrees of supersaturation, can be calculated. In this way a
cumulative "Kel vin-equivalent"-size distribution of the condensation nuclei
can be obtained. This equivalent size of a particle can be described as the
size of a water drcp growing at the same supersaturation as the particle.
Results on Atmospheric Aerosols
Fig. 39 shows a cumulative size distribution which can be derived from
the experimental data using the Kelvin-Gibbs-equation for the equivalent
diameter
d = 4-CT'Mv . I _
In S
Here a denotes the surface tension, MV the molecular weight of the vapor,
PL the density of the liquid, S the supersaturation and T the absolute
temperature. The particle sizes range from 0,002 to 0,11 ym diameter.
Differentiation of the cumulative size distribution yields a differential
particle number distribution (Fig. 4o). The mode of this size distribution
corresponds to the so called Nucleation Mode of the urban aerosol.
The indicated experimental data are mean values over about lo measuring
runs for each point. To obtain statistically significant data on the urban
aerosol, long measuring series are required. It is expected that these
measurements will give a clearer understanding of the urban aerosol in the
range below 0,1 ym.
84
-------
rO
.a
Z3
(O
S-
o
4-
0)
fC
"O
-!->
C
-!->
rO
3
E
O
cr>
CO
ro
0)
to
'o
CO
o
O)
rO
O
£
Ol
-C
Q.
rO
85
-------
-
N
!^- ^
s
o
o
o
0>
E
C
o
oo *^-
o LU
C^ I
LU
o 2
40 <
Q
o
in
\-
^.
5 LU
-J
<
o >
00 ID
a
o LU
(N
o
0
D
NOIlONHd 'diSIQ
c
to
rO
c
o
J3
'si
to
•r—
-a
0)
i—
rO
c
0)
CO
O
to
o
s_
OJ
o
-------
REFERENCES
1. Sawyer, K.F., and Walton, W.H. The Conifuge - A size-separating sampling
device for airborne particles, J. Sci. Instruments 27 (195o), p. 272
2. Goetz, A. An Instrument for the quantitative separation and size-classi-
fication of airborne particulate matter down to o.2 micron. Geofisica
Pura e Appl. Proc. II, 36, (1957), p. 49
3. Goetz, A. Stevenson, H.J.R. and Preining, 0. The Design and Performance
of the Aerosol Spectrometer. A.P.O.A. Journal lo (196o) p. 378
4. Goetz, A. and Preining, 0. Bestimmung der GrbBenverteilung eines Aerosols
mittels des Goetz'schen Aerosolspektrometers. Acta Physica Austriaca,
14 (1961) p. 292
5. Keith, C.H. and Derrick, J.C. Measurement of the particle size distribution
and concentration of cigarette smoke by the "conifuge". J. Coll. Science,
15 (196o), p. 34o
6. Kast, W. Neues StaubmeBgerat zur Schnellbestimmung der Staubkonzentration
und der Kornverteilung. Staub, 21 (1961), p. 215
7. Stbber, W. and Zessak U. Zur Theorie einer konischen Aerosolzentrifuge.
Staub, 24 (1964), p.295
8. Stbber, W. und Zessak U. Zur Messung von Aerosol-TeilchengrbBenspektren
mit Hilfe von Zentrifugalabscheidern. Zentralblatt fur Biologische Aero-
solforschung, 13 (1966), p. 1
87
-------
9. Stbber, W. and Flachsbart, H. Aerosol Size Spectrometry with a Ring
Slit Conifuge. Environmental Sci., 3 (1969) p. 641
lo. Stbber, W. and Flachsbart, H. High Resolution Aerodynamic Size Spectro-
metry of Quasi-Monodisperse Latex Spheres with a Spiral Centrifuge.
J. Aerosol Sci., 2 (1971), p. Io3
11. Stbber, W., 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 Precipitators. J. Coll. Sci.
39 (1972), p. Io9
12. Stbber, 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.
13. Hauck, J. und Schedling, J.A. Ober ein modifiziertes Modell einer Konifuge.
Staub 29 (1969), p. 18
14. Hochrainer, D. and Brown, P.M. Sizing of Aerosol Particles by Centrifugation.
Environmental Sci. and Technology, 3 (1969), p. 83o
15. Hochrainer, D. A New Centrifuge to Measure the Aerodynamic Diameter of
Aerosol Particles in the Submicron Range. J. Coll. Interface Sci., 36
(1971), p. 191
16. Berner, A. und Reichelt.H. Ober EinlaBspaltsysteme in Konifugen. Teil I:
Das ROSL-System. Staub 29, (1969), p. 92
17. Burson, J.H., Keng, E.Y.H., and Orr, C.,Jr. Particle Dynamics in Centri-
fugal Fields. Powder Technology 1 (1967) p. 3o5
18. Matteson, M.J., Boscoe, G.F. and Preining, 0. Design Theory and Calibra-
tion of a Field Type Aerosol Spectrometer. Aerosol Sci. 5 (1974), p. 71
-------
19. Abed-Navandi, M., Zum Auflbsungsvermbgen des Aerosolspektrometers,
Ph.D. Dissertation, Vienna 1973
2o. Abed-Navandi, M., Berner, A. and Preining, 0. The Cylindrical Aerosol
Centrifuge in:Fine Particles, B.Y.H. Liu Ed., Academic Press, London,
New York, 1976
21. Stober, W., Berner, A. and Blaschke, R. The Aerodynamic Diameter of
Aggregates of Uniform Spheres. J. Coll. Int. Sci. 29 (1969), p. 71o
22. Stober, W. and Boose, Ch. Developing Flow and Particle Deposition in
Horizontal Elutriators and Semi-Dispersive Aerosol Centrifuges.
Atmospheric Env. 7 (1973) p. 119
23. Townsend, J.S. The Diffusion of Ions into Gases. Proc. Roy. Soc. 65
(19oo) p. 192
24. Gormley, P.G. and Kennedy, M. Proc. Roy. Irish Acad. 52 A (1949), p.163
25. Tan, C.W. Diffusion of Disintegration Products of Inert Gases in Cy-
lindrical Tubes. Int. J. Heat Mass Transfer 12 (1969), p. 471
26. Tan, C.W. and Hsu, C.J. Diffusion of Aerosols in Laminar Flow in a
Cylindrical Tube. J. Aer. Sci. 2 (1971), p. 117
27. Sinclair, D. A Portable Diffusion Battery. Its Application to Measuring
Aerosol Size Characteristics. Amer. Ind. Hyg. Ass. J. 33 (1972), p. 729
28. Matteson. M.J., Sandlin, C.W. and Preining, 0. Diffusion of Aerosols
at Various Temperatures. J. Aer. Sci. 4 (1973) p. 3o7
29. Kasper, G. Untersuchungen der Durchlassigkeit einer kompakten Diffusions-
batterie im T<
Vienna, 1977.
batterie im Temperaturbereich von + 35°C bis -75°C. Ph.D. Dissertation,
89
-------
3o. Kasper, G., Preining, 0. Penetration of a Multistage Diffusion Bat-
tery at Various Temperatures. J. Aerosol Sci. 9 (1978) p. 331
31. Kasper, G., Berner, A. Ein Generator zur Erzeugung extrem monodisperser
Kochsalzaerosole. Staub-Reinh. Luft 38 (1978) p. 183
32. Matteson, M.J., Fox, J.J. and Preining, 0. Density Distribution of
Sodium Chloride Aerosols Formed by Condensation. Nature Phys. Sci. 238
(1972) p. 61
33. Kasper, G. On the Density of Sodium Chloride Aerosols Formed by Condens-
ation. J. Coll. Interface Sci. 62 (1977), p. 359
34. Espenscheid, W.F., Matijevic, E. and Kerker, M. Aerosol Studies by Light
Scattering: III. Preparation and Particles Size Analysis of Sodium Chlor-
ide Aerosols of Narrow Size Distribution. J. Phys. Chem. 68,(1964),p.2831
35. Armbruster, L., Stahlhofen, W. und Gebhart, J. Produktion von monodisper-
sen Feststoffaerosolen nach dem La Mer-Sinclair-Prinzip. in: Tagungs-
bericht Ges. f. Aerosolforschung, Bad Soden/BRD, 1976.
36. Kasper, G. Membrane Filter Mounts and a Quick-change Device for Aerosols
and Liquid Suspensions. J. Physics E: Sci. Instr. lo, (1977), p. 600
37. Hollander, W. and Schumann, G. On the Influence of Image Forces on the
Performance of Diffusion Batteries, in: Tagungsbericht Ges. f. Aerosol -
forschung, Bad Soden/BRD, 1976.
38. Davies, C.N. Diffusion and Sedimentation of Aerosol Particles from Poisseu-
ille Flow in Pipes. J. Aer. Sci. 4 (1973), p. 317
39. Taulbee, D.B. and Yu, C.P. Simultaneous Diffusion and Sedimentation of
Aerosols in Channel Flows. J. Aer. Sci. 6 (1975), p. 433
90
-------
4o. Thomas, J.W.
Particle Loss in Sampling Conduits
in: Assessment of Airborne Radioactivity,International Atomic Energy
Agency, Vienna, 1967.
41. Towney, S. Bull. Observ. Puy de Dome 173 (1963).
42. Rich, J. Impaction of Aerosol Particles in the Neighborhood of a Circ-
ular Hole. Coll. Czech. Chem. Commun. 29 (1964).
43. Smith, T.N. and Phillips, C.R. Inertial Collection of Aerosol Particles
at Circular Aperture. Env. Sci. Techn. 9 (1975), p. 564.
44. Spurny, K. Zentralbl. biol. Aerosolforschung, 13 (1966), p. 44
45. Smutek, M. and Pich, J. Impaction of Particles on the Surface of Membrane
Filters. J. Aer. Sci. 5 (1974), p. 17
46. Spurny, K. and Pich, J. Staub 24 (1964), p. 25o
47. Spurny, K. and Madelaine, G. Czech. Chem. Commun. 29 (1971), p. 2857
48, Berner, A. Zur Abscheideeigenschaft von Impaktorstufen bei nicht statio-
na'rer Strbmung. Sitzungsber. Osterr. Akademie der Wissenschaften,
Mathem.-Naturw. Klasse Abt. II, 185. Bd. (1976), p. 327
49. Berner, A. Zur Ursache sekundarer Partikelniederschlage bei Impaktoren
Staub-Reinh. Luft 38 (1978), p. 1
5o. Wagner, P.E. und Pohl, F.G., Eine proze|3gesteuerte Anlage zur Untersuchung
der Kinetik von Kondensationsvorgangen. Staub-Reinh. Luft 38 (1978),p.72
51. Wagner, P.E. Optical Determination of the Size of Fast-growing Water Drop-
lets in an Expansion Cloud Chamber. J.Coll.Interf.Sci.44 (1973),p.181
91
-------
52. Wagner, P.E. The Interdependence of Droplet Growth and Concentration.
I. Theory of Droplet Growth and Applications on Condensation Nuclei
Counters. J. Colloid Interface Sci. 53 (1975), p. 429
53. Porstendorfer, J., Heyder, J. Size Distributions of Latex Particles.
J. Aerosol Sci. 3 (1972), p. 141
54. Heard, M.J., Wells, A.C., Wiffen, R.D. A Re-determination of the Dia-
meters of DOW Polystyrene Latex Spheres. Atm. Environm. 4 (197o),p.l49
92
-------
TECHNICAL REPORT DATA
(Please read fnstrucnons c-i the in-ersc before c,mipU ting1
—_ : r
1. REPORT NO.
EPA-600/2-79-105
3 RECIPIENT'S ACCESSION NO
IN THE SUBMICRON SIZE RANGE
Studies with an Aerosol Centrifuge, a New Diffusion
Battery, a Low Pressure Impactor and an Advanced
Condensation Nuclei Counter
5. REPO'H DATE.
MayJL979
|6 PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Othmar Preining and Axel Berner
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Institute for Experimental Physics
The University of Vienna
A-1090 Wien, Strudlhofgasse 4
Austria
10. PROGRAM ELEMENT NO.
1AD712 BC-16 (FY-78)
11. CONTRACT/GRANT NO.
801983
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Sciences Research Laboratory—RTF, NC
Office of Research and Development
U. S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final 1973-1978
14. SPONSORING AGENCY CODE
EPA/600/09
15. SUPPLEMENTARY NOTES
16. ABSTRACT
The report summarizes the investigations of four aerosol classifiers which cover
finite, but overlapping ranges of the aerosol particle size spectrum. The first part
is concerned with a cylindrical aerosol centrifuge, which measures aerodynamic equiva-
lent diameters precisely. This instrument has been used as a reference instrument in
diffusion battery experiments reported in the second part. The diffusion battery has
been investigated for fairly large particle sizes (0.3 ym to 0.5 ym) to determine the
influence of sedimentation, interception and impaction on the transmission of the
diffusion battery. These experiments have been performed with highly monodispersed
NaCl aerosols. In the third part a five stage low pressure impactor is described,
which covers the size range from 0.1 ym to 25 ym diameter. It has been developed
specifically for the determination of the deposited particulate mass. First data on
mass-size distributions of atmospheric aerosols are reported. The final chapter
summarizes the development of a special condensation nuclei counter which measures
number-size distributions in the size range from 0.002 ym to 0.1 ym KELVIN-equivalent
diameter. The applicability to urban atmospheric aerosols is demonstrated.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
Air pollution
*Aerosols
*Particle size
*Centrifuges
Diffusion
Size separation
Condensation nuclei
COS AT I 1'ield/Group
13B
07D
14B
13H
11C
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
19. SECURITY CLASS (This Report!
CLASSIFIED
21 NO. OF PAGES
107
20 SECURITY CLASS (This page)
CLASSIFIED
22. PRICE
EPA Form 2220-1 (Rev. 4-77)
PREVCM-'S EC1 TION i s OBSOLETE
93
------- |