JOURNAL OF COLLOID SCIENCE 15, 340--356 (1960)

M E A S U R E M E N T OF T H E PARTICLE SIZE D I S T R I B U T I O N AND

C O N C E N T R A T I O N OF CIGARETTE S M O K E BY
T H E "CONIFUGE ''1

C. H. Keith and J. C. Derrick

Liggett and Myers Tobacco Company, Research Department,
Durham, North Carolina
Received April 18/1960

ABSTRACT

The particle size d i s t r i b u t i o n and c o n c e n t r a t i o n of cigarette smoke has been meas-

ured with a centrifugal aerosol collector of the type developed b y Sawyer a n d Walton.
A conifuge designed to collect particles in the 0.05 to 10 micron d i a m e t e r size range
was c o n s t r u c t e d and used to s t u d y diluted tobacco smoke. F r o m these m e a s u r e m e n t s
it was found t h a t tobacco smoke as it comes from a cigarette is a n extremely con-
c e n t r a t e d aerosol w i t h a relatively stable d i s t r i b u t i o n of sizes ranging from 0.1 to 1.0
micron, peaked between 0.2 a n d 0.25 micron. The coagulation rate was found to agree
reasonably well w i t h theoretical predictions, and the initial c o n c e n t r a t i o n of nor-
m a l l y o b t a i n e d smoke was found to be 3 X 109 particles per cubic centimeter. The
usefulness of this e q u i p m e n t in the analysis of this difficult aerosol suggests a wider
application in other aerosol problems.

I. INTRODUCTION

The smoke produced by a burning cigarette is a commonly encountered

aerosol of considerable physiological significance. It also is an uncommonly
concentrated aerosol which presents many of the various complications
that make measurement of the particulate concentration and size distribu-
tion of aerosols quite difficult. These considerations have led a number of
workers, particularly Sano et al. (1), Langer and Fisher (2), and Holmes
et al. (3), to use it as a test aerosol with varying degrees of success. In this
paper, a centrifugal sedimentation technique of general applicability to
aerosols within a 0.05 to 10 micron size range is applied to tobacco smoke.
A centrifugal sedimentation technique is particularly suited to an aerosol
such as tobacco smoke which combines a very high particulate concentra-
tion, a small and somewhat nonuniform particle size, and volatile and
chemically unstable components (4, 5). The chief advantage arises from the
indirect size estimation possible with such a technique, thereby avoiding

1 P r e s e n t e d in p a r t at the S o u t h e a s t e r n Regional Meeting of the American Chemi-

cal Society, D u r h a m , N o r t h Carolina, N o v e m b e r 14, 1957.
340
 DISTRIBUTION AND CONCENTRATION OF CIGARETTE SMOKE 3z~l

the difficulties encountered with the use of microscopic, light-scattering,

or electrostatic weighing techniques for size measurement.
Of the centrifugal techniques, impactors have been most commonly
utilized for tobacco smoke (2-4). Some difficulty 'has been encountered,
particularly with multiple-stage impactors, in a considerable loss of ma-
terial through re-entrainment of previously deposited smoke particles. A
centrifugal instrument which avoids this problem and provides a con-
tinuously graded spectrum of particle sizes is the "Conifuge" developed
by Sawyer and Walton (6). Their demonstration of the ability of this
machine to collect and size grade a variety of liquid and solid aerosols
made it apparent that this approach had distinct possibilities for the
analysis of tobacco smoke.
The instrument developed by Sawyer and Walton collected particles
larger than 0.5 micron in diameter. To accommodate tobacco smoke, a
new instrument based on the same principles was constructed. The design
was such that particles with diameters ranging from 0.05 to 10 microns
could be sampled.
This paper describes this instrument and its application to the particle
size analysis of cigarette smoke. The limited data included serve to demon-
strate the usefulness of the equipment for this and other aerosol problems,
and allow the estimation of some of the factors of importance in altering
the concentration of smoke issuing from a cigarette.

II. EQUIPMENT
A. The "Conifuge"
A cross-sectional view of the centrifuge head and the surrounding air
chamber is given in Fig. 1. When rotating at high speeds, the head acts as
a centrifugal air pump. The centrifugal force imparted to the air in the
inner chamber between cones (E) and (0) causes it to flow from the axially
located entrance tube (P) to and out of the off-axis outlet jets (G). Since
the rotating head is enclosed in an airtight chamber (Q and S), this flow is
recirculated through the outer chamber back to the entrance tube, as
indicated by the arrows of Fig. 1. If a small portion of this circulating air
is removed through the outlet tube (T), an equivalent volume of smoke is
drawn in at (A). This incoming smoke is directed at the apex of the spinning
head and spreads out in a thin film on the surface of the inner cone. As
this film and the overlaying clean, circulating air are carried downward
and outward through the conical chamber an increasing centrifugal force
is applied to the individual particles owing to their excursion away from
the axis of rotation. In conjunction with the increasing centrifugal force,
the forward velocity of the air stream in which the particles are embedded
is decreasing as the annular area of the conical chamber increases. This
342 KEITH AND DERRICK

A.SMOKE INLET TUBE K, CLAMPING RING

B.TEFLON GUIDE TUBE L, TEFLON CUSHION FOR BASE OF SUDE
C°APEX CAP NUT M, DOVETAILED DARK FfELD SLIDE
N,DOVETAILED NYLON SLIDE H O L D E R 0 I 2
D, MOUNTING NUT i , , , L i

E, ~NNER CONE O. OUTER CONE INCHES

F, FLOW EQUILIZING CHAMBERS P, AIR ENTRANCE TUBE
G, OUTLET JET QoCHAMBER WALL
H, C.S,I. ROTOR HUG R, CLAMP
I , "0" RING SHAFT SEAL S, CHAMBER TOP
J ~ COPPER CHAMBER SEAL T, OUTLET TUBE

FIG. 1. Cross sectional diagram of the Conifuge.

combination of centrifugal force and stream velocity determines the tra-

jectory of the individual particles. With a given set of dimensions and
rotational velocity and assuming no significant slippage between the ro-
taring head and incoming air stream, the trajectory of each particle is
uniquely determined by its Stokes-Cunningham settling velocity. Since
the trajectory of each particle determines its point of deposition on the
outer wall of the chamber, a deposit is formed there which is continuously
graded according to particle settling velocity. For spherical droplets of
uniform density, such as cigarette smoke, the gradation may be expressed
as a function of particle diameter. A representative sample of this graded
size distribution suitable for counting numbers of particles is obtained
from two dark-field microscope slides (M) embedded in the outer wall.
The conical shape of the chamber and the changing balance between
stream velocity and centrifugal force are such that an extensive range of
sizes are adequately represented on a relatively short collecting slide. The
easily centrifuged and relatively less numerous large particles are deposited
 DISTRIBUTION AND CONCENTRATION OF CIGARETTE SMOKE 343

near the apex of the cone or the head of the slide. The high stream velocity
in this region fully sorts these particles, so that their distribution is opened
out along the slide. However, an area deposit sufficient for counting pur-
poses is maintained since the ring area in this region is relatively small.
The more numerous smaller particles pass further down the conical chamber
into a region of slower stream velocity and higher centrifugal force. This
results in a compression of the range of sizes collected on a given length of
slide, but the larger ring area in this region counterbalances this effect,
so that the area deposit is not excessive.
By equating the centrifugal driving force and Stokes' law resistive forces
acting on a particle, Sawyer and Walton (6) formulated an equation for
the trajectory of individual particles. This equation is given in their paper
and is not reproduced here. The derivation contains a number of assump-
tions, so that their equation is at best only approximate, particularly for
large particles. Because the integrated equation is not analytical, the deposi-
tion points for particles of given settling velocities are best obtained by
superimposing these trajectories on an outline of the chamber dimensions.
Such plots were utilized to design a chamber of sufficient dimensions to
capture the range of sizes thought to be present in cigarette smoke.
One of the chief requirements for the satisfactory operation of the coni-
fuge is a steady, nonturbulent air flow in the space between the cones.
The design was made with this in mind, and sufficient flexibility was in-
corporated in the equipment so that turbulence might be avoided by an
experimental choice of operating conditions. Extensive precautions were
also taken in the machining of the head to eliminate any places which
might cause turbulent mixing of the smoke and clean air streams. A set of
operating conditions were experimentally chosen on the basis of excellence
of size gradation, this being taken as an indication of lack of turbulence.
These were a flout through the conical chamber of 54 c.c./sec, and a rota-
tional velocity of 8000 r.p.m. Other chosen or fixed parameters were a hori-
zontal cone separation of 1 cm., a semivertical angle of 45 °, a 12.5-cm.
slide length, and a smoke sampling velocity of 5 c.c./see.
The design and construction of the equipment was considerably simplified
and the cost reduced by adapting a commercial centrifuge to provide the
driving mechanism. The machine used was a C.S.I. angle centrifuge
(Custom Scientific Instruments, Arlington, New Jersey). The modifications
chiefly consisted of replacing the manufacturer's safety shield with the
airtight chamber and constructing the special centrifuge head, both of
which are shown in Fig. 1.

B. Auxiliary Equipment
Several pieces of equipment were necessary for reprodueibly obtaining
and immediately diluting cigarette smoke for use in the conifuge. The
latter was necessary because coagulation of the extremely concentrated
344 KEITH AND DERRICK

raw smoke stream would significantly change its particle size distribution
very rapidly. A several hundred fold dilution was made by surrounding
the smoke stream as it issued from the cigarette with a high-velocity, clean
air stream. Both streams were directly introduced into a 12-1. flask from
which smoke samples could be drawn through a small sampling chamber
into the conifuge. The sampling chamber was fitted at either end with
plunger valves so designed that a small volume of smoke could first be
drawn into the chamber and subsequently into the conifuge without in-
terrupting the normal flow of clean air into the instrument.
Since immediate dilution of the smoke stream was necessary, it was
not possible to smoke the cigarettes in the orthodox manner by applying
a controlled vacuum pulse to the end of the cigarette. It was, instead,
necessary to puff on the cigarette by forcing air through the burning cone.
This was accomplished by placing a bell over the previously lighted cigarette
just before a pulse of slightly compressed air was delivered from a 4-1.
storage tank3 The duration and intensity of the puff were controlled by a
clock-operated solenoid valve and an adjustable pressure drop in the supply
line to the bell. The larger diluting air stream was supplied from the same
tank and was similarly controlled so that the ratio of diluting air to smoke
remained constant during the puff. The timing was such that the diluting
air was flowing slightly before and after the puffing air stream. Save for
the use of compressed air in place of vacuum, the mechanism employs
the sample principles as the smoking machine described by Keith and
Newsome (7), and was adjusted to take a similar, reproducible 44-c.c.,
2-see. puff, every ball minute.
A Bausch and L o m b research microscope equipped with a cardioid
dark-field condenser, 20-power apochromatic objective, and a 25-power
compensated eyepiece fitted with a standard Whipple disk was used for
counting the number of particles collected on the dark-field slides. The
light source was a B and L Model 48 carbon are microprojector. The
slides require an excellent flame-polished optical surface for they must be
examined with a dry upper surface to avoid possible shifting of the position
of the captured smoke particles. Suitable slides were obtained from the
Baltimore Instrument Company of Baltimore, Maryland. The slides were
prepared for use by cleaning in nitric acid and rinsing with water and
redistilled alcohol. Any residual dust was then removed by coating the
slide with collodion and subsequently removing the collodion film. The
final step consisted of rubbing on a thin film of silicone oil, which prevented
the captured smoke particles from spreading.
In some early work, a puffing mechanism was used which utilized the suction in-
duced by the high velocity diluting air stream to draw on the cigarette. This was dis-
carded because of the extreme difficulty in controlling the volume and other char-
acteristics of the puff,
 DISTRIBUTION AND CONCENTRATION OF CIGARETTE SMOKE 345

III. EXPERIMENTAL :¥[ETttODS

A. Calibration of the Conifuge

In order to establish the relationship between particle size and place of
deposition (as measured by distance from the head of the collecting slide),
the conifuge was calibrated with several test aerosols, which were physically
similar to tobacco smoke. These were atomized molten paraffin wax,
homogeneous dioetyl phthalate aerosols obtained from a LaMer-Sinclair
generator (8), and the homogeneous aerosols formed by atomizing greatly
diluted samples of Dow monodisperse polystyrene latex by the method
described by O'Konski and Doyle (9).
For the paraffin aerosol, the place of deposition of 0.5-micron or larger
particles was correlated with particle size, as measured by ordinary micro-
scopic techniques. For the homogeneous test aerosols the position of maxi-
mum deposition was correlated with the previously determined particle
size. These sizes were determined by the red band structure of the D.O.P.
aerosol and from the given particle size of the polystyrene dispersion,
the measurement of which has been described by Bradford and Vanderhoff
(10).
Figure 2 illustrates the distribution of pal%icle counts obtained from
0.814-micron polystyrene particles. The ordinate of this curve is the cal-
culated number of particles present in a conical segment of the outer cone
of 1 mm. slant height, the calculation being based on the average slide
count in that region. The abscissa is the distance from the top of the
sampling slide in millimeters. The extreme sharpness of the maximum
deposition peak is an indication of the quality of the size gradation obtained
in this machine. The slight inflection on the left side of the curve and the
somewhat slower rise on this side are thought to be caused by the presence
of double and larger aggregate polystyrene particles which were formed
either during the atomization process or by subsequent coagulation. The
rapid fall of the curve on the right side, where the choice is between one
or no polystyrene particles per atomizer droplet, is more representative of
the deposition of a truly monodisperse aerosol.
The calibration curve for the instrument is given in Fig. 3. The separate
calibrations form a continuous curve for particles between 5 and 0.08
microns, indicating an acceptable size grading action for liquid and solid
aerosols over this range of particle size. The deposition curve predicted by
Sawyer and Walton's equation (6) is given by the dashed curve of Fig. 3.
The depai%ure of the two curves in the region of large particles is not un-
expected in light of the assumptions made in their development. In the
small particle region the two curves again depart. This occurs because
the equation predicts the maximum distance at which deposition of a particle
of a certain size would occur. The experimentM points are, however, based
346 KEITH AND DERRICK

'06
x
I--
Z

E
_%
re"
ILl
0.
ffl
W

I...-
rr
n,
LL
0 2 --

ILl
an
;E

0
o
.~
Io
ji L 20 :.'.'50
I
40
DISTANCE ALONG SLIDE (rnm.)
50
| & I
60

F I e . 2. D e p o s i t i o n p a t t e r n of 0.814 m i c r o n p o l y s t y r e n e aerosol in the Conifuge.

on the position of maximum deposition of a given particle size. From the

distribution of deposition distances of the homogeneous aerosols, such as
that illustrated in Fig. 2, it is found that the points of furthest deposition
agree well with the theoretical curve.

B. Size Distribution of Cigarette Smoke

Uniformly blended cigarettes of known weight and resistance to draw
were conditioned to an equilibrium moisture content by storage in an
atmosphere of constant humidity and temperature. In the smoking opera-
tion, the cigarette was lighted by the operator, and the routine was con-
tinued in the previously described equipment. After sampling and collec-
tion of the smoke, the slides were removed and the numbers of particles
 DISTRIBUTION AND CONCENTRATION OF CIGARETTE SMOKE 347

I0
8
6
•
%% \ ® PARAFFIN SPHERES
• DI'OCTYL PHTHALATE
5 '%~, \ • POLYSTYRENE LATEX
4 "*'\ ~ - - ~ EXPERIMENTAL CURVE
~', ~ --THEORETICAL CURVE

¢:
ti,,I
I
I--
l~1.8
=E
_--..6 2-,
Q.5
I,iJ
--I.4
r,..3
I-
¢:.3

.2

.I
.08
,06
.05
.04

DISTANCE ALONG SLIDE (ram)

FIG. 3. Theoretical and experimental calibration curves for the Conifuge.

c o u n t e d 3 a t 5 - m m . i n t e r v a l s a l o n g t h e slides. I n g e n e r a l for e a c h slide, 5

W h i p p l e d i s k squares, e a c h r e p r e s e n t i n g a field 0.033 b y 0.033 r a m . a n d
c o n t a i n i n g f r o m 1 t o 30 p a r t i c l e s , were c o u n t e d . T h e a v e r a g e of t h e s e
c o u n t s f r o m b o t h slides was m u l t i p l i e d b y t h e a p p r o p r i a t e f a c t o r to o b t a i n
a c o u n t of t h e n u m b e r of p a r t i c l e s in t h e f r u s t r u m of a cone of 0.033 r a m .
s l a n t h e i g h t a n d w i t h a r a d i u s e q u i v a l e n t to t h a t of t h e o u t e r cone of t h e

3 Although there was a generally low and somewhat variable background count for
clean slides, no correction was applied to the observed counts. During the counting
operation particles and slide imperfections which obviously appeared to be foreign
to the sample were not included in the count.
348 KEITtt AND DERRICK

conifuge at the counting position. Plotting these counts against a particle

diameter as obtained from the calibration curve 4 formed a size distribution
curve. A logarithmic scale was used for the diameter axis to open out the
smaller particle end of the scale.
Graphical integration of the size distribution curves was used to compute
the total number of particles collected in the conifuge. This figure divided
by the known volume of the sample gave the concentration of diluted
smoke, which in turn was converted to a concentration of raw smoke by
multiplication with the dilution ratio. This ratio was generally 295 volumes
of diluted smoke to 1 volume of raw smoke.
As was pointed out by Sawyer and Walton (6), the size distribution curves
can be readily transformed into mass distribution curves, and from these
the total mass of the sample m a y be obtained by integration. Computations
of these quantities were not extensively used in this work, since the results
are generally imprecise. This arises from the relatively few counts for
large particles, which contribute heavily to the mass distribution and total
mass of the smoke deposit. Calculations of the total mass of smoke were,
however, found to be in reasonable agreement with estimates of this quan-
tity obtained by direct weighing of the total particulate m a t t e r collected
in a suitable trap.
Measurements of the size distribution of side stream smoke were obtained
by allowing a cigarette, selected as above, to freely burn without puffing
in the 12-1. collection flask for 1 rain. The resulting sample was collected
and counted in the same manner as main stream smoke samples.
IV. RESULTS AND DISCUSSION
A. Reproducibility of Collection of Cigarette Smoke
I n order to test the reproducibility of the particle size distribution ob-
rained from selected cigarettes, four separate cigarettes were smoked and
the smoke from the fourth puff was diluted, aged for 30 sec., and collected
in the conifuge. The results of these measurements are illustrated in Fig. 4,
where the circles represent the measured counts and the smooth curve
represents the average size distribution. Although there is considerable
scatter of the points, the figure shows that it is possible to reproduce the
particle size distribution from one cigarette to the next. A subsequent
4 The diameters, as read from the calibration curve, are those for unit density par-
ticles, and may be called a "settling diameter" for particles of other densities. The
"settling diameter" is defined as the product of the actual particle diameter and the
square root of the density of the particle. These diameters were used throughout this
work for lack of exact data on the density of cigarette smoke particles. Sam (4) meas-
ured a value of 0.98 g./c.c, and Sano (1) obtained values ranging from 0.62 to 1.00
g./e.c. Using Sam's value or Sano's highest values causes no sensible change in con-
vetting from settling to actual diameters. Using Sano's lower value would increase the
listed diameter values by a factor of 1.27.
 DISTRIBUTION AND CONCENTRATION OF CIGARETTE SMOKE 349

35

o
~o 30 o o

~z5 ) o

w
~20
°
~ o
i
~e~ o

,,,a:°'15
O.
) o

II1 o

I I I , ; ~ n, I I I~
.06 ,08 .I .2 .3 .4 .5 ,6 .8 i 2 3 4
P A R T I C L E D I A M E T E R (Microns}
FIG. 4. Reproducibility of the particle size distribution of cigarette smoke.

series of ten measurements, each on a separate cigarette, gave essentially

the same results for diluted smoke aged 60 sec., the most frequently oc-
curring diameter being 0.21 micron, the geometrical mean diameter being
0.23 micron, and the standard deviation being 0.14 micron, the latter being
obtained in the manner outlined in Table I.

B. Aging Studies on Cigarette Smoke

Although the previously described puffing mechanism provides a large
sample of diluted tobacco smoke, this smoke is still relatively concentrated
in comparison to other aerosols and therefore would be expected to coagu-
late at an appreciable rate. In order to study this process, the particle size
distribution of cigarette smoke was determined at varying times after the
puff. These distributions are presented in Fig. 5, and as expected for coagu-
lation of small particles into larger aggregates there is a progressive de-
crease in the numbers of particles and an increase in mean particle size
with aging. In Fig. 6 it is apparent that the particulate volume or reciprocal
of the particulate concentration is a linear function of aging time, as has
been observed for many coagulating aerosols (11). From this plot, the
initial number of particles in the sample, No, was estimated to be 1.8 X 107
particles per cubic centimeter of the diluted smoke sample, or 5.3 billion
particles per cubic centimeter in the original smoke. The slope of the
curve, or coagulation constant, K, was found to be 2.4 X 10-1° c.c./sec.
350 KEITH AND DERRICK

:35

,d- 3 0
~30 SEC.
'0

~25
Ld
t3
UJ
//
°rt-1 5
LIJ
O_
r~

0 I I I;~+~ l I I L I , , . , ,, !
.06 .08 .I .2 .3 .4 .5 .6 ,8 I 2
PARTICLE DIAMETER {Microns)

l~IG. ~. The effect of aging on diluted cigarette smoke.

12
O

10

8
O
x .9

I I I I i i t I I I .... I i , ]
40 80 120 160 200 240 280
TIME (Seconds)
FIG. 6. Relation between particulate volume and aging time.
 D I S T R I B U T I O N A N D C O N C E N T R A T I O N OF C I G A R E T T E S M O K E 351

F u r t h e r m e a s u r e m e n t s of these q u a n t i t i e s y i e l d e d a n a v e r a g e c o a g u l a t i o n
c o n s t a n t of 3.4 X 10 -1° e.c./sec, a n d 3 billion p a r t i c l e s p e r cubic c e n t i m e t e r
for t h e f o u r t h puff o n a r e g u l a r l e n g t h , b l e n d e d c i g a r e t t e of a v e r a g e w e i g h t
a n d m o i s t u r e c o n t e n t . T h i s c o a g u l a t i o n c o n s t a n t is s o m e w h a t less t h a n
t h e v a l u e s o b t a i n e d b y o t h e r s for a v a r i e t y of c o n s i d e r a b l y m o r e d i l u t e
aerosols. U s i n g W h y t l a w - G r a y ' s (12) m o d i f i c a t i o n of t h e S m o l u c h o w s k i
coagulation equation, a slightly curved relation between particulate volume
a n d t i m e was c a l c u l a t e d w i t h slopes r a n g i n g f r o m 4.2 to 4.5 X 10 -1° c.c./sec.,
w h i c h a r e in r e a s o n a b l e a g r e e m e n t w i t h t h e e x p e r i m e n t a l l y d e t e r m i n e d
slope of 3.4 X 10 -1° c.c./see., e s p e c i a l l y since t h e t h e o r e t i c a l slope is for
a n i d e a l m o n o d i s p e r s e d aerosol, w h e r e a s t h e e x p e r i m e n t a l v a l u e was ob-
rained with a concentrated heterogeneous smoke.

C. Effect of Various Smoking and Cigarette Variables

T h e d a t a r e p o r t e d in earlier sections were o b t a i n e d f r o m t h e f o u r t h , 44-c.c.
puff on a n u n f i l t e r e d b l e n d e d c i g a r e t t e of n o r m a l w e i g h t a n d m o i s t u r e
c o n t e n t . S o m e d a t a o b t a i n e d f r o m c i g a r e t t e s in w h i c h one o r m o r e of t h e s e
v a r i a b l e s were a l t e r e d are p r e s e n t e d in T a b l e I.

TABLE I
Effect of Smoking and Cigarette Variables
Most freq. No. of
occurring Meanb ¢ Standardc parficlesb
diameterb diameter deviation per c.c.
Cigarette conditions Smoking conditions a (~) (~) @) X lO-g

Blended, 1.08 g., 11% H~O 4th, 44-c.e. puff 0.21 0.23 0.14 3.01
Blended, 1.08 g., 11% H20 4th, 35-c.c. puff 0.23 0.23 0.14 2.97
Blended, 1.08 g., 11% H~O 4th, 55-c.c. puff 0.21 0.21 0.14 2.89
Blended, 1.08 g., 11% It20 2nd, 44-c.c. puff 0.22 0.21 0.14 2.24
Blended, 1.08 g., 11% H20 5th, 44-c.c. puff 0.23 0.23 0.14 4.13
Blended, 1.08 g., 11% tt~O 7th, 44-c.c. puff 0.23 0.21 0.14 4.66
Blended, 1.08 g., 11% H~.O 10th, 44-e.c. puff 0.22 0.21 0.14 4.31
Blended, 1.00 g., 11% H20 4th, 44-e.c. puff 0.23 0.22 0.14 3.57
Blended, t.I8 g., 11% H20 4th, 44-c.c. puff 0.20 0.21 0.13 2.86
Blended, 1.08 g.,~ 3.6% H20 4th, 44-c.c. puff 0.22 0.22 0.14 4.66
Blended, 1.08 g.,~ 6.2% H20 4th, 44-c.e. puff 0.23 0.21 0.14 4.54
Blended, 1.08 g.,e 20.4% H20 4th, 44-c.c. puff 0.20 0.22 0.14 1.65
Bright, 1.09 g. 4th, 44-c.c. puff 0.20 0.21 0.14 3.25
Burley, 0.90 g. 4th, 44-c.c. puff 0.20 0.20 0.14 3.01
Turkish, 1.15 g. 4th, 44-c.c. puff 0.20 0.20 0.14 3.07
All puffs of 2 seconds duration taken at ~ minute intervals.
b Extrapolated values at zero aging time.
c Estimated from a logarithmic-probability plot of particle size against cumulative
frequency, the geometric mean being the size at which 50% of the particles are
greater and less than that size. The standard deviation is taken as the ratio of the
84% size to the 50% size adjusted to the size range under consideration (13).
d Weight of the cigarettes at 11% moisture.
352 KEITH AND DERRICK

In Table I it is apparent that the particle size distribution is a relatively

stable property of cigarette smoke over a considerable range of variations
of cigarette and smoking conditions. It is also apparent in Table I and the
previous distributions that cigarette smoke is a relatively homogeneous
aerosol with regard to particle size. This stable and narrow distribution of
sizes suggests that the distribution is formed by a combination of removal
processes common to all cigarettes. At the small particle end of the spec-
trum, it is probable that processes such as coagulation, diffusional capture
by the tobacco strands, and growth by condensation of vaporized materials
would tend to eliminate particles less than 0.1 micron diameter. With re-
gard to the coagulation process, in Fig. 5 it was shown that the peak di-
ameter of diluted smoke would double within approximately 180 sec. In
raw smoke, with its three hundred fold greater particulate concentration,
this should occur in 0.6 see., which is of the same order of magnitude
as the 0.1 sec. residence time of the smoke within the cigarette. Thus it
is apparent that this process would significantly contribute to the re-
moval of small particles.
Another process which probably controls the particle size distribution
of the exit smoke stream is removal of the larger particles through filtration
by the tobacco strands in the cigarette butt. It is known from other work
(14) that appreciable amounts of smoke are collected by this portion of the
cigarette. In a later section this process will be considered in more detail.
Although the particle size distributions in Table I are relatively stable,
there is a considerable variation in the total number of particles per cubic
centimeter of smoke gasses for these cigarettes. In general these variations
are in line with the effect of these parameters on the weight of smoke (14).
Increments in puff volume have a nearly linear effect on the weight of
smoke, which is consistent with a constant particulate density, the in-
crement coming from the increased total volume. Later puffs on a cigarette
have been found to produce more smoke, chiefly because less is filtered out
by the stub, and such an increment is detected in Table I. With the smoking
method employed herein the cigarette tends to become extremely moist
after the seventh or eighth puff. This is thought to cause the observed de-
crease in numbers of particles between the 7th and 10th puff. The variables
of cigarette weight and moisture content show the expected increase in
numbers of particles with decreasing weight and moisture content. Bright,
burley, and Turkish tobaccos appear to give slightly anomalous results
as they produce nearly the same number of particles, although these
tobaccos give somewhat different weights of smoke.
D. Side Stream Measurements
The smoke rising from the burning cone of a cigarette during the interval
between puffs should have a different size distribution than the smoke
drawn from the cigarette during a puff since it is not subjected to the same
 DISTRIBUTION AND CONCENTRATION OF CIGARETTE SMOKE 353

35

,:- 30
s
o

25
b.I

0
bJ

u) 20
E
E
m')
m')
o 15
L,J
r,

,,m,
nm
m0
:E
Z
5

0 I ; I & I I I I
.06 .08 .I .2 .3 .4 .5 .6 .8 I 2 3 4
PARTICLE DIAMETER {Microns}
FIG. 7. Particle size d i s t r i b u t i o n of side s t r e a m cigarette smoke.

size-determining processes as the latter. Figure 7 shows the distribution

obtained for this smoke, which was collected by the previously described
method. Larger numbers of small particles, down to 0.08 micron, were
observed and the most frequently occurring size was 0.15 micron as com-
pared to 0.20-0.23 micron for main stream smoke. I t was estimated from
this curve that free burning produces particles at the rate of 6.3 billion per
second. Since the sample was obtained over a period of 60 see., it is likely
that the original smoke contains greater numbers of still smaller particles.
E. Filtration of Tobacco Smoke
To investigate the role of coagulation and filtration in the removal of
larger tobacco smoke particles, the effect of drawing the smoke stream
through additional cigarettes, open tubes, and efficient cigarette filters
was studied. To isolate the effects of coagulation and filtration two as-
semblies were prepared by taping together three 70-ram. cigarettes and by
attaching a 140 ram. long, 8-ram. glass tube to a single cigarette. The
assemblies are thus equivalent cigarettes with an open 140-ram. mouth-
piece and with a 140 ram. tobacco filter. The distributions obtained from
the fourth puff on these cigarettes are illustrated in Fig. 8.
In Fig. 8, it is readily apparent that the tobacco filter removed con-
siderably greater numbers of particles than the equivalent mouthpiece and
also that this removal was selective for larger particles since the distribu-
tions are slightly displaced. The raw smoke coming from the mouthpiece
was found to contain 2.1 billion particles per cubic centimeter, whereas the
~54 KEITIt AND DERRICK

15

~-- --'0--- 140 ram. T U B ~ '

Z
~:IO ACCO
~9
uJ

Q
~5
n

W
II1

'7

O I I I I~ I I t I I~'~-; : " ' ' ~ I I I

.06 D8 .I 2 .3 .4 . 5 . 6 .8 I 2 3 4
PARTICLE D I A M E T E R (Micronsl
Fla. 8. Particle size distributions of tobacco smoke after coagulation and filtra-
tion in a 140-ram. tube and tobacco column.

tobacco filter delivered only 0.5 billion. F r o m previous data it was esti-
mated that 3 billion particles entered both assemblies, so that 0.9 billion
particles were removed by the mouthpiece and 2.5 billion by the filter.
F r o m the residence time of the smoke in the mouthpiece and the coagula-
tion curve, it was calculated t h a t 0.5 billion particles were removed by
coagulation, and that 0.4 billion remained in the tube at the end of the
puff or were lost at the entrance or walls of the tube.
For the tobacco filter equivalent losses would be expected from coagula-
tion and smoke remaining in the filter, so that the net removM of particles
through filtration is 1.6 billion per cubic centimer, amounting to a better
t h a n 50 % removal by this mechanism.
F r o m the distributions of Fig. 8, a figure for the weight filtration effi-
ciency of the 140-mm. tobacco filter m a y be obtained and compared with
an independent estimate of this quantity. Converting the number distribu-
tions of Figs. 4 and 8 to mass distributions and integrating in the region
0.1-1.0 micron gave a mass of raw smoke coming from the cigarette of
32 #g. per cubic centimeter, while that issuing from the tube and tobacco
filter was 30 and 5 gg. per cubic centimer, respectively. These yield an
overall filtration efficiency for the tobacco filter of 84 %. F r o m measure-
ments on shorter tobacco filters, an efficiency of 82 % is calculated for a
140-mm. tobacco column; this agrees well with the efficiency calculated
from the particle size distributions.
Figure 9 illustrates the effect of a relatively efficient 17-mm. cellulose
acetate cigarette filter on the particle size distribution of cigarette smoke.
The distributions illustrated were the average for diluted smoke aged for
 D I S T R I B U T I O N AND C O N C E N T R A T I O N O F C I G A R E T T E SMOKE 355

,~o30
35f
E2o
E

QI5 ~ WITHOUT FILTER

WITH FILTER

lIE
z
5

0 i i 1 I ' ' ~ n, I I
.06 .08 .I .2 .5 ,4 .5 .6 .8 I 2. :5 4
PARTICLE DIAMETER ( M i c r o n s )
FIG. 9. Particle size distributions of the smoke from filtered and unfiltered ciga-
rettes.

60 sec. from 4 cigarettes with, and 4 cigarettes without, filters. As in the

case of the tobacco filter, a preferential reduction in the numbers of large
particles is apparent in the shift of the filter distribution towards smaller
sizes in addition to an overall lowering of the curve.
The total number of particles was found to be 2.3 and 3.6 billion per
cubic centimeter of raw smoke for the filtered and unfiltered smoke, re-
spectively. From these figures it is calculated t h a t the filter removes 36 %
of the numbers of particles from the smoke stream. Converting the curves
to mass distributions for particles less than 1.0 micron in diameter and
again computing the filtration efficiency yields a figure of 39 %, the in-
crease stemming from the greater contribution of the larger particles to
the total mass. Considering the fact that separate samplings of cigarettes
were necessary for each determination, these figures agree well with the
measured filtration efficiency of 37 %.
These experiments with tobacco filters and relatively efficient cigarette
filters indicate that the filtration mechanism is effective in removing larger
particles. This in conjunction with the coagulation and other growth
processes produce an essentially stable and narrow distribution of particle
sizes in cigarette smoke as it comes from the cigarette,
V. SUMMARY
The particle size distribution and particulate concentration of cigarette
smoke has been measured using a centrifugal collection instrument called
356 KEITIt AND DERRICK

the "Conifuge." The instrument avoids the difficulty of direct size measure-
ment by continuously grading the particles according to settling velocity
or size. It is capable of collecting particles ranging in diameter between
0.05 and 10 microns. For these reasons the instrument is well suited for
measurement of the size distribution of cigarette smoke and other aerosols.
The size distribution and particulate concentration of smoke were found
to be reproducible quantities for similar cigarettes, but were found to vary
according to the age of the smoke sample, even after considerable dilution.
The rate of decrease of the number concentration of smoke particles was
found t o agree reasonably well with the rate predicted by the modified
Smoluchowski coagulation equation.
Alteration of a number of smoking and cigarette variables was found to
have no detectable effect on the particle size distribution, but changed the
concentration of particles in a manner similar to the effect of these vari-
ables on the weight of smoke.
A preferential removal of larger particles was observed for high-efficiency
tobacco and cigarette filters, in addition to a considerable decrease in
particulate concentration.
The essentially constant size distribution appears to arise through the
removal of small partlcles through coagulation and larger particles through
filtration by the tobacco strands.

ACKNOWLEDGMENT
The authors are indebted to Mr. Henry Pierce for his considerable assistance in the
design and construction of the conifuge and accessory equipment.

REFERENCES
1. SANO.,K., FUJIYA,Y., AND SAKATA,S., J. Chem. Soc. Japan 74,664 (1954).
2. LANGER,G., ANDFISHER, M. A., Am. Med. Assoc. Arch. Ind. Health 13,373 (1956).
3. HOLMES, J. C . , HARDCASTLE, J. ]~., AND MITCHELL, R. I., Tobacco Sci. 3, 148
(1959).
4. SAM, A., "A Study of Particle Size Distribution and Dilution of Cig-
arette Smoke." Thesis, Duke University, Durham, North Carolina, 1956.
5. KAItLER,H., AND LLOYD, B. J., JR., J. Natl. Cancer Inst. 18,217 (1957).
6. SAWYER,K. F., ANDWALTON,W. H., or. Sci. Instr. 27,272 (1950).
7. KEITH,C. H., ANDNEWSOME,J. 1~., Tobacco Sci. 1, 51 (1957).
8. SINCLAIR, D., AND LAMER, V. K., Chem. Revs. 44,245 (1949).
9. O'KoNsKI, C. W., ANDDOYLE, G. J., Anal. Chem. 27,694 (1955).
10. BRADFORD,E. B., AND VANDERHOFF,J. W., or. Appl. Phys. 26,863 (1955).
11. WHYTLAW-GRAY,R., AND PATTERSON, H. S., "Smoke: A Study of Aerial Dis-
perse Systems," Chapter 5. E. Arnold & Co., London, 1932.
12. WHYTLAw-GRAY,R., AND PATTERSON, I-I. S., "Smoke: A Study of Aerial Dis-
perse Systems," Chapter 6. E. Arnold & Co., London, 1932.
13. DRINKER, P., AND HATCH, T., "Industrial Dust," pp. 144-149. McGraw-Hill,
New York, 1936.
14. NEWSO~fE,J. R., ANDKEITR, C. H., Tobacco Sci. 1, 58 (1957).