Full text of DOI:10.1080/02786820119393

This article was downloaded by: [University of Southern Queensland]
On: 10 October 2014, At: 00:30
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK
Aerosol Science and Technology
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/uast20
A Transition from Heterogeneous to Homogeneous
Nucleation in the Turbulent Mixing CNC
Rashid Mavliev , Philip K. Hopke , Hwa-Chi Wang & Doh-Won Lee
Published online: 30 Nov 2010.
To cite this article: Rashid Mavliev , Philip K. Hopke , Hwa-Chi Wang & Doh-Won Lee (2001) A Transition from
Heterogeneous to Homogeneous Nucleation in the Turbulent Mixing CNC, Aerosol Science and Technology, 35:1,
586-595
To link to this article: http://dx.doi.org/10.1080/02786820119393
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”)
contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors
make no representations or warranties whatsoever as to the accuracy, completeness, or suitability
for any purpose of the Content. Any opinions and views expressed in this publication are the opinions
and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of
the Content should not be relied upon and should be independently verified with primary sources of
information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands,
costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or
indirectly in connection with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or
systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution
in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at
http://www.tandfonline.com/page/terms-and-conditions

–
(
)
Aerosol Science and Technology 35: 586 595 2001
c
° 2001 American Association for Aerosol Research
Published by Taylor and Francis
=
=
0278-6826 01 $12.00 C .00
A Transition from Heterogeneous to Homogeneous
Nucleation in the Turbulent Mixing CNC
Rashid Mavliev,1 Philip K. Hopke,2 Hwa-Chi Wang,1;3 and Doh-Won Lee2
1Illinois Institute of Technology, Chicago, Illinois
2Clarkson University, Potsdam, New York
3Air Liquide, Chicago Research Center, Countryside, Illinois
have been developed for more accurately characterizing heavy
ions and particles in the nanometer-size range.
Anew method for changing the supersaturationinthe Turbulent
Mixing CNC has been developed andused toexamine the transition
from heterogeneous nucleation of test particles to homogenous nu-
(
)
The Condensation Nuclei Counter CNC , which grows pri-
(
)
mary particles nuclei up to a more easily detectable size, is
one of the most widely used devices for studying particles below
0.1 ¹m. A general description of a CNC is given in many books
(
)
cleation of working  uid: dibutylphthlate DBP . Supersaturation
was controlled by changing the DBP vapor pressure in the nozzle
 ow by saturating only a predetermined part of the  ow, while the
total  ow and temperature remain constant. This approach allows
for the changing of the initial DBP vapor pressure, while keep-
ing the  ow structure and temperature Ž eld unchanged. The DBP
concentration in the outlet of the vapor generator was measured
experimentally for different ratios of saturated and bypass  ows
and found to be close to estimated values. Experimental results for
transitions from heterogeneous nucleation to homogeneous nucle-
ation are presented for NaCl and WOx particles at various DBP
vapor pressures. With an increasing of the DBP vapor pressure,
the concentration of enlarged particles increases until it reaches
a plateau. At higher initial values of DBP pressure, homogeneous
nucleation prevails, and the number concentration of particles fol-
lows a curve typical for homogeneous nucleation recorded in the
absence of nuclei. Nuclei with different mobility diameters were ac-
tivated at different values of vapor pressure. There are signiŽ cant
differences in the slopes of particle activation curves for NaCl and
WOx particles. The reasons for such differences are a subject for
continuing research.
(
(
))
and reviews for example, Willeke and Baron 1993 . Several
types of CNCs are used in aerosol research. The main difference
among these CNC designs is the way they produce supersatura-
tion that leads to particle growth up to a predetermined size for
subsequent detection. In an expansion-type CNC, supersatura-
tion is generated by adiabatic cooling during pressure reduction.
It is a batch instrument and has been used in atmospheric aerosol
(
research for many years. A continuous CNC Agarwal and Sem
)
1980 is widely used today. Supersaturation is formed by cool-
ing the laminar aerosol  ow that was saturated with working
 uid vapor. The third type of CNC is based on turbulent mixing
of the  ows with particles and working  uid vapor. This type of
instrument has not yet been commercialized. It was described
(
)
Ž rst by Kogan and Burnashova 1960 and was further devel-
(
)
oped in different versions by Okuyama et al. 1984 , Ankilov
(
)
(
)
(
)
et al. 1991 , Kousaka 1993 , and Mavliev and Wang 2000 .
(
)
The major advantage of the turbulent mixing CNC TMCNC
is the  exibility of generating supersaturation by simply mix-
ing the aerosol  ow and a  ow saturated with the working  uid
vapor.
INTRODUCTION
Recently, in the Ž eld of nanometer-sized particles, attention
has been paid to homogeneous nucleation, binary nucleation,
ion-induced nucleation, and simultaneous heterogeneous and
homogeneous nucleation because of the importance of these
The CNC is one of the most sensitive methods for detect-
–
ing nanometer particles. The detection limit can reach 2 3 nm
(
Stolzenburg and McMurry 1991; McDermont et al. 1991;
(
)
)
processes in atmospheric particle formation Kim et al. 1998 .
In addition, discrepancies between theory and experiment have
stimulated new research in this Ž eld. Several new techniques
Okuyama et al. 1984; Mavliev and Wang 2000 . The minimum
detection efŽ ciency of CNCs is very sensitive to the size of par-
(
)
ticles Makela et al. 1996 .
Although a CNC is primarily devoted to measuring the num-
ber concentration of particles, in recent years it has been shown
that a CNC can be used to measure the size distribution of
nanometer particles. The size distribution of nuclei can be mea-
Received 29 November 1999; accepted 11 May 2000.
Address correspondence to Philip K. Hopke, Department of Chem-
istry, Clarkson University, Box 5705, Potsdam, NY 13699-5705.
E-mail: hopkepk@clarkson.edu
(
sured by means of changing the CNC’s sensitivity McDermont
586

587
TRANSITION FROM HETEROGENEOUS TO HOMOGENEOUS NUCLEATION
Table 1
Effect of particle charge on estimated Kelvin diameter
dkelv with charge
1.50 1.60 1.70 1.80 1.90 2.00 2.20 2.40 2.60 3.00
dkelv without charge 2.08 2.08 2.10 2.15 2.20 2.27 2.41 2.58 2.75 3.11
)
et al. 1991 and by means of measuring the size of grown parti-
2
(
)
=
4M
¾
q 1 ¡ 1 "
ln S D
¡
;
[2]
(
)
cles Ahn and Liu 1990; Rebours et al. 1996; Saros et al. 1996 .
The last approach is based on the fact that the growth of smaller
particles is delayed because of the Kelvin effect that results in
the Ž nal size of particles being dependent on initial nucleus size.
However, size resolution isnotsatisfactory and operationalrange
2¼2 dk4elv
"
0
½Rt dkelv
where q is the electrostatic charge and " and "0 are dielectric
constants of the droplet and vacuum, respectively Scheibel and
(
¨
)
Porstendorfer 1986 . The charge effect is shown in Table 1,
where estimated Kelvin diameters are compared for particles
with charge and without charge. Charge effects on nucleation
are negligible for particles above 3 nm. However, for smaller
particles, it could be substantial and should be taken into
account.
–
is limited to a range of 3 10 nm. In addition, the growth time for
particles of the same size depends strongly on spatial uniformity
of supersaturation. For most continuous  ow CNC’s, the spatial
distribution of supersaturation is not uniform because of the use
(
)
of diffusive cooling in the laminar  ow Stolzenburg 1988 .
In the present article, a new approach for changing the
TMCNC sensitivity is described. The approach is based on
changing the vapor pressure of working  uid by changing the
ratio of saturated and by-pass  ows in the vapor generator. This
approach was used to investigate the transition from hetero-
geneous to homogeneous nucleation for two types of initial
nuclei.
Effects of Nuclei Origin
The interaction between the nuclei and vapor molecules has a
strong in uence on the heterogeneous nucleation process. Equa-
( )
( )
tions 1 and 2 that relate the Kelvin diameter to the supersat-
uration ratio are simpliŽ ed cases of heterogeneous nucleation.
In reality, the interaction between the vapor and particles is not
well deŽ ned, especially for atmospheric aerosols of unknown
origin.
If the nucleating particle is not perfectly wettable by the
working  uid, the situation becomes more complicated. The
formation energy of the working  uid embryo will involve the
interaction of solid-liquid and liquid-gas interfaces. According
to Fletcher 1969 , one of the most important parameters in this
case is the contact angle between the particle surface and the
condensed phase. However, the contact angle on nanometer-size
particles is not known and the contact angle of bulk material may
not be valid. Experimental investigation of the effects of the nu-
clei substance on nucleation activation becomes extremely im-
portant. Nevertheless, only a few publications are available in
the literature.
APPROACH
Basics of Heterogeneous Nucleation
It is well known that particles start to grow when subjected to
sufŽ ciently supersaturated vapor conditions. This phenomenon
is known as heterogeneous nucleation and is the working prin-
ciple for the widely used CNC. In a CNC, particles are al-
lowed to grow to sufŽ cient size for further detection. The re-
lationship between vapor pressure and particle diameter when
the process of condensation starts to prevail over evapora-
tion is
(
)
4M
¾
(
)
Helsper and Neissner 1985 investigated the effect of the
vapor substance on the behavior of an expansion-type CNC for
two working  uids, water and butanol. They found that parti-
ln S D
;
[1]
½RT dkelv
(
)
cle substance Ag and NaCl did not affect the Kelvin diameter
when butanol was used as the working  uid. However, the size of
NaCl particles was overestimated by 2.5 times when water was
where dkelv is the Kelvin diameter, ¾ is surface tension, M is
molecular weight, ½ is density of liquid, R is the universal
( )
( )
T
=
gas constant, and the supersaturation S D P Psat T . Psat
(
)
is the saturation vapor pressure at temperature T . At a given used. Similar results were found by Porstendorfer et al. 1985 .
(
)
S value, particles greater than the Kelvin diameter are acti- Kesten et al. 1991 found different detection efŽ ciency of the
TSI CNC 3025 with butanol as a working  uid for NaCl and
vated and followed by condensational growth. Particles smaller
than the Kelvin diameter are not activated and eventually
evaporate.
(
)
Ag particles. Kousaka et al. 1984 found less profound but still
different results for NaCl and Ag particles. The brief description
(
)
If particles carry an electric charge, the evaporation/ of results by Madelaine and Metayer 1980 showed an essen-
condensation equilibrium condition is affected. The supersat- tial difference of CNC sensitivity for NaCl, V2O5, and H2SO4
uration required for droplet growth in this case can be expressed particles. These initial reports suggest the in uence of nucleus
origin and working  uid vapor. However, there is insufŽ cient
as

588
R. MAVLIEV ET AL.
data for establishing a complete theoretical description of the
phenomena.
Experimental Setup
The experimental system is shown schematically in Figure 1.
The basic components are the particle generators, the Differ-
(
)
ential Mobility Analyzer DMA , and the CNCs. The TMCNC
Formation of the Supersaturation by Turbulent Mixing
employed was essentially the same as that described previously
Mavliev and Wang 2000 , except for the vapor generator. A
(
)
In the TMCNC, the aerosol  ow “cold”  ow, Qa is mixed
(
)
with the gas  ow saturated with the vapor of a working  uid
more detailed description will be given in the following sec-
tions. A subset of this set up was used in previous experiments
on the detection efŽ ciency evaluation of a turbulent mixing CNC
at high  ow rates.
(
)
“hot”  ow, Qv . The mixing of “cold” and “hot”  ows causes
the supersaturation of the working  uid vapor because of the
immediate temperature decrease of the mixed  ow. As a result,
particles greater than the equivalent Kelvin diameter will start
to grow. In our system, the initial “hot”  ow can be considered
as a turbulent jet in a conŽ ned space. According to Abramovich
(
)
1963 , the spatial structure of a jet in a conŽ ned space is similar
Particle Generation
to that of a free jet. The temperature and pressure Ž elds are then
described by
Two different types of particles were used in the experiments.
Particles of NaCl were generated by evaporation in a high tem-
perature furnace and subsequently cooled off the vapor by mix-
ing it with a cold  ow. This method allows particle generation
(
)
T D T1 C T2 ¡ T1 n;
[3]
–
in the size range of 2 20 nm. Test particles of WOx were gener-
(
)
P D P1 C P2 ¡ P1 n;
ated by a hot wire method similar to that described by Reischl
(
)
(
)
where the local mixing ratio n is a function of the jet coordi-
nates and indices 1 and 2 correspond to the “cold” and “hot”
components, respectively. This simpliŽ ed approach is valid for
low vapor concentration, and the thermodynamic parameters of
the mixture of these two  ows are practically the same as those
in the absence of vapor. The mixing ratio changes from 0 at the
jet boundary to 1 at the center of the jet, immediately down-
stream of the nozzle. The  ow farther away from the nozzle can
be considered uniform. The transition from a jet to a uniform
 ow occurs at the distance from the nozzle where the axial ve-
locity is close to the average velocity of the  ow. The Ž nal value
of the mixing ratio is
et al. 1997 . Tungsten wire Aldrich was heated by electric cur-
rent inside a glass tube with a Ž ltered air ow of 3 lpm. Particle
size and number concentration were controlled by varying the
heating power. The  ow pattern was arranged to allow a quick
change from NaCl toWOx particleswithout disturbing the DMA
and CNC operating parameters. This  exibility allowed for the
investigation of heterogeneous nucleation for both substances
under very similar conditions.
The nucleation activity of the particles depends strongly on
their size. The monodisperse fraction of particles was obtained
from the original polydisperse distribution by the Vienna-type
(
)
DMA Qsh D 28 lpm, Qs D 3 lpm and directed to the Faraday
(
) (
)
Cup Electrometer FCE Q1 D 2:5 lpm and to the TMCNC
Qv
(
)
Q2 D 0:5 lpm . At a short distance from the  ow split, the
n
D
;
[4]
Ž n
(
)
Qv C Qa
CNC  ow was diluted with Ž ltered air of 1.5 lpm providing the
total sample  ow of 2 lpm. The dilution air ow was pushed
through a heat exchanger cooled by water from the circulating
bath at a temperature of 20±C. Diffusion losses of particles in
the connection line between the DMA outlet and the CNC inlet
where n is equal to parameter Rh used for estimations by
Ž n
(
)
Okuyama et al. 1984 . Instead of numerically solving the
 uid dynamic equations, a simple approach can be employed by
treating the mixing ratio as an independent variable in order
to examine the impact of various designs and operating
parameters. For a given mixing ratio, the local values of
(
)
total length 0.55 m were estimated using well-known formu-
las for particle penetration through cylindrical tubes Willeke
and Baron 1993 . Estimated diffusion losses can reach 45% for
2 nm particles; for bigger particles, diffusion losses become
lower with 20% for 4 nm and 10% for 10 nm particles. Addi-
tional particle losses that could occur in the entrance of the CNC
system are negligible according to our estimations.
(
)
( )
=
supersaturation S D P Psat T , the corresponding Kelvin
diameter, and the homogeneous nucleation rate can be determ-
ined. Calculations were performed using the Excel spreadsheet
program. The Calculating procedure consists of the following:
–
²
(
)
generating a set of n values in the range of 0.02 200,
The Vienna-type DMA HAUKE, Vienna, Austria is a suit-
²
²
²
²
calculating local temperature and pressure values for
able instrument for particle separation in a size range below
(
)
( )
50 nm Fissan et al. 1996 . Recent research on selection efŽ -
ciency shows that proper adjustment of the sample/sheath  ow
ratio allows an attainment of 5% resolution at the high end and
each value of n using Equation 3 ,
calculating saturated pressure and surface tension val-
ues at the temperature corresponding to value n,
calculating supersaturation and Kelvin diameter val-
ues, and
calculating the homogeneous nucleation rate using pre-
viously determined parameters.
(
)
10% at the low end for sizes»1 nm Rossel-Lompart et al. 1996 .
At sample and sheath  ow ratios used in the experiments, the
width of the transfer function should be approximately 12%,
but because of the effect of diffusion broadening in the small

589
TRANSITION FROM HETEROGENEOUS TO HOMOGENEOUS NUCLEATION
(
Figure 1. Schematic diagram of the experimental set up. Vienna-type DMA, FCE, TMCNC, Laser Aerosol Spectrometer PMS
)
(
)
(
)
LAS-X , Optical Particle Counter OPC-501 PMS, OPC , TSI Condensation Nuclei Counter TSI 3760 , and Aerosol Charger
(
)
Kr-85 TSI .
particle range, the real width can reach 20% Stolzenburg 1988; up to 150±C. Temperature is controlled by a circuit board with
)
(
Reischl et al. 1997 .
an accuracy of § 0.1±C. Two chambers of equal size, which can
3
–
be Ž lled with a working  uid of 15 20 cm by volume, are also
CNC Operation Principle
machined. Flows from these two chambers are connected and
mixed before passage to the growth tube. One of the chambers
The scheme of the TMCNC is shown in Figure 2. In
(
)
(
)
“bottom” is partially Ž lled with the working  uid dibutylph-
TMCNC, the aerosol  ow “cold”  ow mixed with the gas
(
)
(
)
(
)
thalate DBP , while the other chamber “top” is kept dry. The
degree of supersaturation is adjusted by varying the ratio of  ows
passing through these two chambers, which are equilibrated at
the same temperature.
 ow saturated by working  uid vapor “hot”  ow . The mixing
of “cold” and “hot”  ows causes supersaturation because the
saturation pressure corresponding to the system temperature af-
ter mixing is lower than the vapor pressure in the mixed  ow.
(
)
The saturated vapor  ow is directed to a mixing chamber
througha 1 mm diameternozzle. The aerosol  ow isdirected into
the mixing chamber through a circular opening which is formed
Flows at least one should be turbulent to provide quick mixing.
The common technique to vary the degree of supersaturation in a
TMCNC is to adjust the temperature of the vapor generator and
therefore the equivalent Kelvin diameter. A disadvantage of this
method is that the temperature of the mixing zone depends on
the temperature of the vapor generator and therefore the effect of
supersaturation is convoluted with that of the temperature Ž eld
(
)
by the nozzle and a plastic insert see Figure 2 . The main pur-
(
)
pose of the insert minimal opening 4 mm, total length 40 mm
is to reduce the “dead” space and the recirculation of the  ow.
These two coaxial  ows are mixed and directed to a condensa-
tion tube downstream. A condensation tube is made of a stainless
steel tube with an internal diameter of 20 mm and a length of
200 mm. The residence time of particlesinthe condensation tube
is about 2 s, assuming a fully developed  ow. The actual resi-
dence time may be somewhat shorter because of the turbulent
jet  ow structure at the entrance of the growth tube. However,
it is sufŽ cient for particle growth if the saturator temperature is
(
)
in the mixing zone Adachi et al. 1992 . In the present research,
we used another approach as described below to control super-
saturation without changing the saturator temperature and  ow
rate and thus preserving the temperature and  ow Ž eld.
The vapor generator is machined from a block of aluminum
(
)
alloy that is coated with aluminum oxide thickness »100¹m .
The  ow channels are machined by drilling through the alu-
minum block. Additional holes are drilled to install the heating
elements and a solid-state temperature sensor. This approach
allows uniform temperature distribution within the block. The
vapor generator is placed in an outer Te on shell for insulation
±
(
)
set at 110 C or above Mavliev and Wang 2000 . The growth
(
)
tube is cooled by a heat exchanger copper tube with water  ow
from a circulating bath at a temperature of 20±C to prevent the
condensation of DBP vapor on optical elements downstream.

590
R. MAVLIEV ET AL.
smaller particles and lowering the detection limit. As supersatu-
ration increases toa certain level, homogeneous nucleation starts
to compete with heterogeneous nucleation. Therefore homoge-
neous nucleation is the terminal limiting factor for this method
since the operating conditions have to avoid the initiation of ho-
mogeneous nucleation. Important questions are how far from
this point should conditions be set and what criteria can be used
to choose the operating parameters.
These criteria can be determined by scanning the supersat-
uration over the range of values up to the onset of homoge-
neous nucleation. Scaling parameters can be applied to obtain
proper working conditions. Homogeneous nucleation causes a
very characteristic exponential growth in particle number and
can be easily identiŽ ed. The onset point of homogeneous nu-
cleation can be used to determine the supersaturation value and
the corresponding Kelvin diameter under the speciŽ c working
conditions.
Supersaturation scanning in the TMCNC is achieved by
changing the initial vapor pressure of DBP in the “hot”  ow.
(
)
The carrier gas to the vapor generator 0.8 lpm is split into two
 ows at the entrance. One fraction of  ow is directed to the
( )
Figure 2. Scheme of TMCNC. Nozzle 1 ,  ow compressor
(
)
vapor generator chamber containing DBP “bottom” and satu-
rated with DBP vapor. Another part of the vapor generator  ow
( ) ( ) ( )
2 , boundary of jet 3 , “cold”  ow with particles 4 , growth
( ) ( ) ( )
tube 5 , outlet of grown particles 6 ,  ow control system 7 ,
( ) ( )
(
)
is directed to the chamber without DBP “top” but equilibrated
at the same temperature. These two  ows are recombined inside
the vapor generator before being directed to the nozzle. This
approach permits changing the initial DBP vapor pressure in
the vapor generator  ow by changing the ratio of  ows through
“top” and “bottom” chambers while keeping the  ow structure
and the temperature Ž eld unchanged.
critical oriŽ ce 8 , aerosol Ž lter 9 , growth tube temperature
(
)
(
)
control 10 , “top” volume of the vapor generator 11 , “bottom”
(
)
(
)
volume of the vapor generator 12 ,  ow mixer 13 , and main
(
)
electronic board MEB . Solid lines arethegas ow connections;
dotted lines are electrical connections.
The DBP concentration in the outlet of the vapor gener-
ator was measured gravimetrically. Porous metal cups were
placed in the air stream at the nozzle outlet. The weight of
the cups was measured before and after exposure. The DBP
accumulation is directly proportional to the exposure time, in-
dicating that the measurement procedure is correct. To measure
the penetration of vapor through the porous cups, two cups are
placed sequentially. The DBP amount on the second cup is neg-
ligible, indicating complete condensation of vapor on the Ž rst
cup.
The concentration of particles after condensation growth is
(
)
measured using three particle counters: a CNC TSI 3760 , an
(
)
optical particle counter PMS OPC-501 , and a laser aerosol
(
)
(
)
spectrometer PMS LAS-X . The optical counter OPC-501
has a detection range of 0.5 ¹ and 5 ¹ and is designed for a  ow
(
)
rate of 2.8 lpm. The laser aerosol spectrometer LAS-X has
four measuring ranges with detection limits of 1.5 ¹m, 0.3 ¹m,
0.17 ¹m, and 0.12 ¹m, respectively. Because the size distribu-
tion of grown particles usually did not Ž t in a single measuring
range, the scanning mode of the LAS-X was used. This oper-
ational mode increases the time necessary to measure one size
distribution to 40 s, but provides detailed size distribution of
grown particles. The third instrument used to measure number
concentration of grown particles, TSI CNC model 3760, has a
detection limit of 14 nm and was used at a standard  ow rate
Experiments were performed for vapor generator  ows of 0.8
(
)
and 1.6 lpm no “cold”  ow was used and for temperatures of
115, 120, and 130±C. The measurement results for a vapor gen-
erator  ow rate of 0.8 lpm and saturator temperatures of 120±C
are presented in Figure 3 for different ratios of vapor  ow to total
 ow. The dotted lines represent estimations based on saturated
vapor pressure, and the symbols are experimental values. The
mass of accumulated vapor substance m was estimated using the
gas law:
(
)
of 1.5 lpm. The detection limit 50% detection efŽ ciency of a
prototype TMCNC is approximately 2.7 nm for  ow rates from
(
)
1 to 2.8 lpm Ankilov et al. 1994; Mavliev and Wang 2000 .
PVM
Supersaturation Scanning
m D
;
[5]
RT
As was shown above, the activation of particle growth and
( )
CNC detection limits depend on supersaturation. The increase where Psat T is saturated vapor pressure, V is volume of air
( )
(
)
of supersaturation allows starting heterogeneous nucleation for stream L , M is molecular weight of DBP 278.35g/mol , R

591
TRANSITION FROM HETEROGENEOUS TO HOMOGENEOUS NUCLEATION
Figure 3. Dependence of the DBP vapor concentration in the
outlet ofthesaturator chamber onthe ow saturationratiofor sat-
urator temperatures of 120±C. Dotted lines represent estimations
based on saturated vapor pressure, and symbols are experimental
values.
Figure 4. Variation of particle concentration measured by TSI
(
)
(
)
CNC 3760 Ntsi and by PMS OPC Nopc in the supersaturation
scanning experiment. Supersaturation was changed by means of
(
)
DBP saturation  ow Qbottom in the presence of 1.7 nm particles
(
)
t < 105 min and in the absence of particles. The presence of
¤
¤
)
(
is gas constant 0.0821 atm L/mol K , and T is temperature of
( )
(
)
.
particles is indicated by FCE current Ifce
the saturator. Saturated vapor pressures Psat T were estimated
(
)
with parameters as used by Okuyama et al. 1984 and were
18 and 24.7 Pa for saturation temperatures of 115 and 120±C,
respectively. As can be seen in Figure 3, the dependence of DBP
vapor concentration in the outlet of the saturator chamber is
in good agreement with the calculated values. Measured vapor
concentration values for other  ow rates and temperatures are
also in agreement with estimations.
tial nuclei indicating homogeneous nucleation. When particles
are present, there are two “waves” of concentration increase
related to the start of heterogeneous and homogeneous nucle-
ation. Similar behavior is observed on increasing and decreas-
ing the supersaturation, indicating the monotonic relation be-
tween the  ow ratio in the vapor generator and the resulting
supersaturation.
Experiment and Data Reduction
The number concentration of particles detected by the dif-
(
(
)
All of the instruments used in the experiment were controlled ferent sensors TSI CNC 3760 Ntsi; d > 14 nm , PMS OPC
–
(
)
(
))
by a single personal computer that allowed the synchronization
N
opc; d > 0:5 ¹m , and LAS-X Nlas 1 2 at different DBP
of different devices and simplifying data collection of multiple vapor pressures is presented in Figure 5. For the LAS-X data,
variables in a dynamic situation. Typical experimental proce- two size ranges are presented with detection limits of 0.3 ¹m
(
)
dures consist of setting a working temperature for the TMCNC, and 0.17 ¹m, respectively. The FCE current Ifce represents the
setting parameters for particle generators, setting the desired nu- initial concentration of nuclei that remains constant during the
clei size and composition, and scanning TMCNC supersatura- experiment. There are signiŽ cant differences in the data from
tion by changing the ratio of saturated/bypass  ows in the vapor sensors at vapor pressure values below 0.93 Pa. The OPC and
generator. During supersaturation scanning, all of the experi- LAS-1 data are signiŽ cantly lower than the TSI points. The
mental variables were recorded. Figure 4 represents the typical grown particle size decreases as the vapor pressure decreases
variation of these variables during a supersaturation scanning and causes these differences because the particles then become
(
)
experiment in the presence of 1.7 nm particles t < 105 min
smaller than the detection limit of OPC. At the end of the pres-
(
)
and in the absence of particles t > 105 min . Variations in sure scanning range, the LAS-2 points become lower than the
(
)
particle concentration were measured by TSI CNC 3760 Ntsi
TSI value, indicating the shift of grown particle size below the
(
)
and by PMS OPC Nopc and are presented along with the vari- 170 nm range. The good agreement between the TSI CNC and
(
)
(
)
.
ation of DBP saturation  ow Qbottom and FCE current Ifce
LAS-X data at higher values of vapor pressure indicates that the
The presence of particles is indicated by FCE current, which size distribution of grown particles is stable and the measured
(
)
changes as DMA voltage is lowered to zero t D 105 min . As data are valid.
can be seen in Figure 4, both counters give comparable values
The sizedistributionsof grown particles areshown inFigure 6
of particle concentration at higher values of Qbottom, at lower for both heterogeneousnucleation and homogeneousnucleation.
values of Qbottom the Ntsi and Nopc data differ due to the differ- Figure 6a shows the size distributions of heterogeneous nucle-
ent detection limits. The particle number concentration changes ation of DBP, at three initial vapor pressures, on NaCl particles
sharply with changes in supersaturation in the absence of ini- with a mobility equivalent size of 1.7 nm. As vapor pressure

592
R. MAVLIEV ET AL.
Figure 5. Number concentration of particles detected by dif-
(
)
(
)
ferent sensors: TSI CNC 3760 Ntsi , PMS OPC Nopc , and
–
(
)
LAS-X Nlas 1 2 at different DBP vapor pressures. For LAS-X
data, two size ranges are presented with detection limits of 0.3
(
)
mm and 0.17 mm, respectively. The FCE current Ifce represents
the initial concentration of nuclei.
increases, the modal size of grown particles shifts to the right.
At a pressure of 0.80 Pa, the lognormal size distribution Ž ts well
with the experimental results. At high pressure, however, a small
tail can be seen at the lower size end. This tail becomes predom-
inant for even higher pressure. Apparently, homogeneous nu-
cleation of DBP at this value of vapor pressure overtakes hete-
rogeneous nucleation. This observation is conŽ rmed by data
presented in Figure 6b, where size distributions are presented
for homogeneous and for heterogeneous nucleation at the same
initial DBP vapor pressure of 1.73 Pa. The Ž lled symbols are
data for 1.7 nm NaCl nuclei, while the open symbols are data
obtained from homogeneous nucleation of DBP vapor in the ab-
sence of NaCl nuclei. The solid line is the superposition of two
lognormal size distributions: one corresponds to heterogeneous
( )
Figure 6. a Size distribution of grown particles measured
by LAS-X at different DBP vapor pressures. NaCl particles of
1.7 nm were used as nuclei. b Size distribution of grown par-
ticles for homogeneous and heterogeneous nucleation at DBP
vapor pressure of 1.73 Pa. Points inboth Ž guresare experimental
data; lines are lognormal approximation.
( )
nucleation of 1.7 nm nuclei and the other corresponds to homo- before starting to increase again. The second increase of parti-
geneous nucleation of DBP vapor. The solid line Ž ts well the cle concentration is caused by homogeneous nucleation of DBP
data for NaCl nuclei, indicating the simultaneous occurrence of vapor, as indicated by the curve for homogeneous nucleation
homogeneous nucleation and heterogeneous nucleation.
in Figure 7. For particles of 1.4 nm, the concentration curve
The data presented in Figure 5 shows a typical example of the changes slope as the DBP vapor concentration reaches the ho-
transition from heterogeneous to homogeneous nucleation. The mogeneous nucleation range.
effect of the initial nuclei sizes on this transition is presented in
Simultaneous measurementsof FCE current and particle con-
Figure 7 and shows the dependence of the particle concentra- centration detected by CNC allow for comparison between these
tion on DBP vapor pressure for different initial nucleus size and two concentrations in order to determine the detection efŽ ciency
for homogeneous nucleation. The FCE current for each particle of the CNC. The FCE current is recalculated to particle concen-
size is also presented. In the case of homogeneous nucleation, tration, assuming each particle carries a single negative charge
(
the DMA voltage was turned to zero and FCE current was at
the fraction of multiple charged particles for a particle size be-
(
)
)
background levels see Figure 5, t > 105 min . Both of the het- low 20 nm is negligible . The concentration is reduced by the
(
)
erogeneous nucleation curves in Figure 7 have similar patterns. dilution factor in the outlet of the DMA system see Figure 2
As the vapor concentration increases, the number of detected and in the CNC. Diffusion losses in the tube connecting the
particles starts to increase. For particles of initial DMA sizes of DMA system and CNC inlet are estimated and taken into ac-
1.4 and 2 nm, the increasing concentration reaches a “plateau” count. CNC/FCE ratios measured for different particle sizes at

593
TRANSITION FROM HETEROGENEOUS TO HOMOGENEOUS NUCLEATION
Figure 9. The dependence of particle concentration on DBP
vapor pressure for different initial nuclei sizes and compositions
and for homogeneous nucleation. The results for 1.7 nm and for
Figure 7. The dependence of particle concentration on DBP
vapor pressure for different initial nucleus size and for homoge-
neous nucleation. The FCE current for each particle size is also
presented.
(
)
homogeneous nucleation Homog. consist of two sets of data
obtained by decreasing and increasing the vapor pressure.
different DBP vapor pressures are presented in Figure 8. Chang-
ing the DBP vapor pressure within the operational limits of the
system does not in uence the detection of particles >2.6 nm to
any signiŽ cant degree. As shown in the previous Ž gures, homo-
geneous nucleation prevails over heterogeneous nucleation at
DBP pressures above 1.73 Pa, giving unreasonable values of the
CNC/FCE ratio thatexceed 1. For smaller particles, thedetection
efŽ ciency shows a strong dependence on DBP vapor pressure.
(
)
results for 1.7 nm and for homogeneous nucleation Homog.
consist of two sets of data obtained by decreasing and increas-
ing the DBP vapor pressure. The transition from heterogeneous
nucleation to homogeneous nucleation is very similar for both
types of nuclei. Particle concentration increases with increasing
vapor pressure, reaches a “plateau,” and converges to the curve
for homogeneous nucleation.
There are also differences between these two types of nuclei.
These differences are more pronounced in Figure 10. In this
Results and Discussion
(
Ž gure, the dependence of the detection efŽ ciency CNC/FCE ra-
Experiments were performed with two types of nuclei. WOx
andNaCl particles weregeneratedsimultaneously and wereused
in equal experimental conditions. The dependence of detected
particle concentration on DBP vapor pressure for different ini-
tial nuclei sizes and compositions is presented in Figure 9. The
)
tio on vapor pressure for different sizes and composition of nu-
clei is presented. The impact of homogeneous nucleation caused
by the increase in particle concentration at DBP pressures above
(
)
1.60 Pa see Figure 9 has been subtracted. The major difference
in data for WOx and NaCl is in the slope of the rising part of
(
(
Figure 8. Dependence of detection efŽ ciency CNC/FCE con-
Figure 10. Dependence of detection efŽ ciency CNC/FCE ra-
)
)
centration ratio on DBP vapor pressure for different sizes of
tio on vapor pressure for different sizes and compositions of
(
)
tungsten oxide WOx nuclei.
nuclei.

594
R. MAVLIEV ET AL.
The transition from heterogeneous nucleation to homoge-
neous nucleation wasinvestigated using thescanning CNC. With
an increase of DBP vapor pressure, the concentration of enlarged
particles increased and reached a plateau. At higher values of
DBP pressure and correspondingly higher supersaturation val-
ues, homogeneous nucleation prevails and the number concen-
tration of particles follows the curve typical for homogeneous
nucleation recorded in the absence of nuclei. Nuclei with dif-
ferent mobility diameters were activated at different values of
vapor pressure. The differences between homogeneous and het-
erogeneous nucleation were correlated with nucleus mobility
sizes.
Factorsaffecting heterogeneousnucleation for different types
of nuclei were investigated using the scanning CNC. Experimen-
tal results are presented for NaCl and WOx particles at various
DBP vapor pressures. There are signiŽ cant differences in the
slopes of particle activation curves for NaCl and WOx particles.
The reasons for such differences are a subject for continuing
research.
(
Figure 11. The detection efŽ ciency of TMCNC CNC/FCE
)
concentration ratio for different nuclei sizes and compositions.
Recent data for NaCl and WOx is compared with NaCl data at a
 ow rate of 1 l/min previous data , which were obtained during
the Vienna Workshop of CNC’s Intercomparison Ankilov et al.
(
)
(
)
1994 .
ACKNOWLEDGMENT
This workwas supported by the US EnvironmentalProtection
Agency under grant R826654.
the detection efŽ ciency curves. The curves for different particle
sizes of the same substance are essentially parallel. When curves
are reaching a “plateau,” the detection efŽ ciency also varies with
particle size and substance. The dependence of detection efŽ -
ciency on particle size for different compositions of nuclei is
REFERENCES
(
)
Abramovich, G. N. 1963 . The Theory of Turbulent Jets, MIT Press, Cambrige,
MA, p. 633.
(
presented in Figure 11 data presented is for Ž xed value of va-
(
)
Adachi, M., Okuyama, K., and Seinfeld, J. H., 1992 . Experimental studies of
)
por pressureof 1.33 Pa . The data pointsfor thesample ow rates
–
ion-induced nucleation. J. Aerosol Sci. 23, 4:327 337.
(
)
Agarwal, J. K., and Sem, G. J. 1980 . Continuous Flow, Single-Particle Count-
of 1 l/min were obtained during the International Workshop on
–
ing Condensation Nucleus Counter, J. Aerosol Sci. 11:343 357.
(
comparison of CNC instruments at Vienna University Ankilov
(
)
Ahn, K. H., and Liu, B. 1990 . Particle Activation and Droplet Growth Process
)
et al. 1994 and are presented here for comparison with recent
in Condensation Nucleus Counter-I. Theoretical Background, J. Aerosol Sci.
measurements at a sample  ow rate of 2 l/min. The NaCl data
points for different  ow rates converge to a single curve. At the
same time, the CNC detection efŽ ciency for WOx particles is
substantially lower than that for NaCl particles for DBP vapor
pressure values of 1.33 Pa. As observed in Figure 10, the situa-
tion reverses at lower DBP pressures where WOx particles have
higher detection efŽ ciency. This fact cannot be explained with
available theories. Further experimental and theoretical research
on this subject is required.
–
21:263 276.
Ankilov, A., Baklanov, A., Colhoun, M., Enderle, K.-H., Filipovicova, D., Gras,
J., Julanov, Y., Lindner, A., Lushnikov, A., Majerowicz, A., Mavliev, R.,
McGovern, F., Mirme, A., O’Connor, T., Podzimek, J., Preining, O., Reischl,
G., Rudolf, R., Sem, G., Szymanski, W., Tamm, E., Wagner, P., Winklmayr,
W.,and Zagainov, V. 1994 . Sensitivityof Various AerosolCounting Methods
for Different Particle Size and Compositions: Results of Vienna Workshop
1993. Proc. of Fourth International Aerosol Conference, edited by R. Flagan.
Los Angeles, CA, pp. 375 376.
Ankilov, A., Baklanov, A., Eremenko, S., Mavliev, R., Reischl, G., and
Majerowicz, A. 1991 . Comparison of the Novosibirsk Diffusion Battery
(
)
–
(
)
–
with the Vienna Electromobility Spectrometer, J. Aerosol Sci. 22:s325 328.
Fissan, H. D., Stratmann, F., Buscher, P., Neumann, S., Pui, D., and Chen, D.
CONCLUSIONS
(
)
A novel CNC was developed that is capable of scanning
a range of supersaturation values via turbulent mixing of two
 ows, one with nuclei and the other with DBP vapor. The vapor
pressure of DBP was controlled by splitting the vapor generator
 ow and saturating only a predetermined part of the  ow while
the total  ow and temperature remained constant. This approach
allows for the changing of the initial DBP vapor pressure while
keeping the  ow structure and the temperature Ž eld unchanged.
The DBP concentration in the outlet of the vapor generator was
measured experimentally for different ratios of saturated and
bypass  ows and found to be close to the estimated values.
1996 . Experimental Comparison of Four Differential Mobility Analyzers
–
for Nanometer Aerosol Measurement, Aerosol Sci. Technol. 24:1 13.
Fletcher, N. 1969 . The Physics of Rainclouds, Cambrige University Press,
London.
Helsper, C., and Niessner, R. 1985 . On the In uence of the Vapour Substance
on the Behavior of an Expansion-Type Condensation Nucleus Counter, J.
Aerosol Sci. 16:457 461.
Kesten, J., Reineking, A., and Porstendorfer, J. 1991 . Calibration of a TSI
Model 3025 UltraŽ ne Condensational Particle Counter, Aerosol Sci. Technol.
15:107 111.
Kim, T. O., Ishida, T., Adachi, M., Okuyama, K., and Seinfeld, J. H. 1998 .
Nanometer-Sized Particle Formation from NH3/SO2/H2O/Air Mixtures by
(
)
(
)
–
¨
(
)
–
(
)
–
Ionizing Irradiation, Aerosol Sci. Technol. 29:112 125.

595
TRANSITION FROM HETEROGENEOUS TO HOMOGENEOUS NUCLEATION
(
)
¨
Porstendorfer, J., Scheibel, H., Pohl, F., Preining, O., Reischl, G., and Wagner, P.
Kogan, Y., and Burnashova, Z. 1960 . Size MagniŽ cation and Measurement of
–
(
)
1985 . Heterogeneous Nucleation of Water Vapor on Monodispersed Ag and
Condensation Nuclei in the Continuous Flow, J. Phys. Chem. 34:2630 2639
(
)
in Russian .
NaCl Particles with Diameters between 6 and 16 nm, Aerosol Sci. Technol.
–
(
)
Kousaka, Y. 1993 . Mixing Type CNC and its Applications. Report on Inter-
national Conference on Aerosol Science and Technology. Taichung, Taiwan,
R.O.C.
4:65 79.
(
)
Rebours, A., Bouland, D., and Renoux, A. 1996 . Recent Advances in Nanopar-
ticle Size Measurement, J. Aerosol Sci. 27:1227 1242.
–
(
)
(
)
Kousaka, Y., Okuyama, K., and Niida, T. 1984 . Size-Supersaturation Rela-
tionship in Activation of UltraŽ ne Aerosol Particles. In Aerosols, edited by
B. Liu, D. Pui, and H. Fissan. Elsevier, New York, p. 51.
Reischl, G. P., Makela, J. M., and Necid, J. 1997 . Performance of Vienna Type
–
Differential Mobility Analyzer at 1.2 20 Nanometer, Aerosol Sci. Technol.
–
27:651 672.
(
)
Madeline, G., and Metayer, Y. 1980 . Technical Note, J. Aerosol Sci.
Rossel-Liompart, F., Losceratales, I., Bingham, D., and Fernandez de la Mora,
(
)
11:358.
J. 1996 . Sizing Nanoparticles and Ions with a Short Differential Mobility
–
(
)
Makela, J. M., Mavliev, R., and Jokinen, V. 1996 . On the Detection EfŽ ciency
Analyzer, J. Aerosol Sci. 27:695 720.
(
)
(
)
of UltraŽ ne Condensation Particle Counter UCPC, TSI Model 3025 Below
Saros, M. T., Weber, R. J., Marti, J. J., and McMurry, P. H. 1996 . UltraŽ ne
3 nm. In Nucleation and Atmospheric Aerosols, edited by M. Kulmala and P.
Aerosol Measurement Using a Condensation Nucleus Counter with Pulse
–
–
Wagner. Elsiver Publishing, pp. 643 646.
Height Analysis, Aerosol Sci. Technol. 25:200 213.
(
)
¨
(
)
Mavliev, R., and Wang, H.-C. 2000 . Design and Performance Characteris-
tics of a Turbulent Mixing Condensation Nuclei Counter, J. Aerosol Sci.,
in press.
Scheibel, H., and Porstendorfer, J. 1986 . Counting EfŽ ciency and Detection
Limit of Condensation Nuclei Counters for Submicrometer Aerosols, J. Col-
loid Interface Sci. 109:261 274.
–
(
)
(
)
McDermont, W. T., Ockovic, R. C., and Stolzenburg, M. R. 1991 . Counting
Stolzenburg, M. 1988 . An UltraŽ ne Size Distribution Measuring System, Ph.D.
EfŽ ciency of an Improved 30-A Condensation Nucleus Counter, Aerosol Sci.
thesis, University of Minnesota, Minneapolis, MN.
–
(
)
Technol. 14:278 287.
Stolzenburg, M. R., and McMurry, P. H. 1991 . An UltraŽ ne Aerosol Conden-
–
(
)
Okuyama, K., Kousaka, Y., and Motouchi, T. 1984 . Condensation Growth of
sation Nucleus Counter, Aerosol Sci. Technol. 14:48 65.
(
)
UltraŽ ne Aerosol Particles in a New Particle Size MagniŽ er, Aerosol Sci.
Willeke, K., and Baron, P., eds. 1993 . Aerosol Measurement: Principles, Tech-
niques and Applications, Van Nostrand Reinhold, New York.
–
Technol. 3:353 366.