You are on page 1of 14

Aerosol Science 33 (2002) 609 – 622

www.elsevier.com/locate/jaerosci

Analytical formulae connecting the “real” and the “apparent”


nucleation rate and the nuclei number concentration for
atmospheric nucleation events
Veli-Matti Kerminena; ∗ , Markku Kulmalab
a
Finnish Meteorological Institute, Air Quality Research, Sahaajankatu 20E, FIN-00810 Helsinki, Finland
b
Department of Physics, University of Helsinki, P.O. Box 64, FIN-00014 Helsinki, Finland

Received 17 May 2001; accepted 18 October 2001

Abstract

A simple yet relatively accurate analytical formula has been derived between the “real” atmospheric nucle-
ation rate and the rate at which the resulting clusters, or nuclei, appear at some larger size (the “apparent”
nucleation rate) as a result of their growth by condensation. In addition, another analytical formula was de-
rived that connects the nucleation rate and the total nuclei concentration in a desired size range. The derived
formulae are applicable to situations in which there are no major 4uctuations in the pre-existing particle size
distribution or in the concentration of vapours responsible for the nuclei growth, and in which the total nuclei
number concentration remains su6ciently low (¡ 105 –106 nuclei cm−3 ) to prevent e8ective self-coagulation
between the nuclei. With the help of these formulae, an explicit nucleation scheme can be included into an
atmospheric model without the requirement that the modelled particle diameter range must be extended down
to one nanometer. In ;eld measurements only the “apparent” nucleation rate can currently be determined. The
derived formulae provide a means to convert this rate to a “real” nucleation rate, making it possible to test
more rigorously the viability of di8erent nucleation theories under atmospheric conditions. ? 2002 Elsevier
Science Ltd. All rights reserved.

1. Introduction

Formation and growth of new aerosol particles are key processes in atmospheric aerosol dynam-
ics, as demonstrated in several ;eld experiments (Weber et al., 1997; Kavouras, Mihalopoulos, &
Stephanou, 1998; Clarke et al., 1999; Birmili & Wiedensohler, 2000; Harrison et al., 2000; MCakelCa


Corresponding author. Tel.: +358-9-1929 5501; fax: +358-9-1929 5403.
E-mail address: veli-matti.kerminen@fmi.; (V.-M. Kerminen).

0021-8502/02/$ - see front matter ? 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 2 1 - 8 5 0 2 ( 0 1 ) 0 0 1 9 4 - X
610 V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622

et al., 2000a; Kulmala et al., 2001a; O’Dowd, 2001; Woo, Chen, Pui, & McMurry, 2001). Model
calculations based on either binary (e.g. water–sulphuric acid; Kulmala, Laaksonen, & Pirjola, 1998)
or ternary (e.g. water–sulphuric acid–ammonia, or water–hydrochloric acid–ammonia; Korhonen
et al., 1999; Arstila, Korhonen, & Kulmala, 1999) nucleation mechanisms suggest that nucleated
particles are initially of the order of 1 nm in diameter. This size is too small to be reached by
the existing measurement techniques, or to be covered by global or even regional-scale atmospheric
aerosol models. Because of this, it would be extremely useful if one could relate the formation rate
of critical clusters (the “real” nucleation rate) and the rate at which they appear at larger sizes (the
“apparent” nucleation rate) as a result of their growth by condensation in the atmosphere.
Recently, Kulmala et al. (2001b) have shown how particle size spectra, obtained from temporally
successive measurements at one site, can be used to calculate the particle growth rate, the conden-
sation and coagulation sink, the source rate and concentration of vapours responsible for the particle
growth, and the real nucleation rate. In this paper we will deepen this analysis by deriving simple
yet relatively accurate analytical formulae connecting the “real” and the “apparent” nucleation rate,
and the total nuclei concentration in a desired size range. The validity of these formulae under dif-
ferent atmospheric conditions is discussed, and the magnitude of associated errors is quanti;ed. The
expressions can be used both for modelling purposes and for analysing experimental data.

2. Derivation of the formulae

Let us assume an atmospheric nucleation event, during which new nuclei of diameter dnuc; ini
are produced at a rate J (t) (nuclei cm−3 s−1 ). When the nuclei grow in size by condensation and
possibly by self-coagulation, their number concentration is decreased because of various removal
mechanisms. As a result, the 4ux at which the new nuclei appear at some larger particle diameter
dp , the so-called “apparent” nucleation rate Japp (dp ), is always smaller than the “real” nucleation
rate J . In order to derive an analytical formula between J and Japp , we will assume a nucleation
burst occurring over a small time interval Mt, during which a monodisperse population of new
nuclei is produced. In the following we will investigate how the particle number concentration of
this population N (nuclei cm−3 ) changes as the nuclei grow to larger sizes.

2.1. Assumptions and approximations

An analytical formula for the time evolution of the nuclei number concentration N during the
growth from their initial size dnuc; ini to some larger size dp cannot be derived without several
simplifying assumptions and approximations. The assumptions are: (1) the only important sink for
the nuclei is their coagulation to larger pre-existing particles, (2) the nuclei grow by condensation
at a constant rate, and (3) the pre-existing population of larger particles remains unchanged during
the nuclei growth.
Assumption (1) can be written mathematically as
dN (dnuc ) 
= −N (dnuc ) K(dnuc ; dp; j )Nj ; (1)
dt j
V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622 611

where dnuc ∈ [dnuc; ini ; dp ] is the time-dependent diameter of the nuclei, K(dp; 1 ; dp; 2 ) is the Brownian
coagulation coe6cient between particles of diameter dp; 1 and dp; 2 , the index j goes over the size
classes of the pre-existing particle population, and Nj is the number concentration of particles in a
size class j. Assumption (2) takes the form
ddnuc
= GR = constant; (2)
dt
where GR(m s−1 ) denotes the nuclei growth rate. From assumption (3) it follows directly that the
4ux of non-volatile vapours to the pre-existing particle population remains constant during the nuclei
growth. Mathematically, this can be written as
1
dp; j m (Knj ; )Nj = CS = constant; (3)
j
2
where m is the transitional correction for the condensational mass 4ux given by (Fuchs & Sutugin,
1971)
1 + Knj
m (Knj ; )= : (4)
1 + 0:377Knj + 1:33Knj (1 + Knj )=
Here is the mass accommodation coe6cient of the condensing vapour, and Knj = 2=dp; j is the
Knudsen number with  representing the vapour mean free path in the air. The parameter CS (m−2 )
in Eq. (3) is called the condensation sink (e.g. Kulmala et al., 2001b).
The Brownian coagulation coe6cient between nanometer-size nuclei and larger pre-existing parti-
cles is proportional to d− 
nuc , where  is somewhere between 1.5 and 2. We will approximate  = 2,
which leads to the following relation:
K(dnuc ; dp; j ) d2nuc = K(d0 ; dp; j ) d20 ; (5)
where d0 is equal to 1 nm. Coagulation of very small nuclei to larger particles can be thought
as if large vapour molecules were condensing to the same particles with a unity accommodation
coe6cient. The rate at which a 1-nm particle coagulates with the pre-existing particle population is
thus expected to be proportional to the rate at which a non-volatile vapour molecule is condensing
to the same particle population:

K(d0 ; dp; j )Nj =  × 4Di CS =1 =  CS =1 : (6)
j

Here Di is the vapour di8usion coe6cient. Note that the proportionality factor  (m2 s−1 ) is constant
with time when there is no change in the pre-existing particle population (CS is constant) or in
the ambient temperature and pressure. The validity and resulting errors associated with assumptions
(1)–(3), as well as approximations (5) and (6), will be discussed in more detail in Sections 3.1
and 3.2.

2.2. Connection between the “real” and the “apparent” nucleation rate

As a ;rst step in deriving an equation for the time behaviour of the nuclei number concentrations
N (dnuc ), we note that the left-hand side of Eq. (1) can be written as
dN (dnuc ) dN (dnuc ) ddnuc
= : (7)
dt ddnuc dt
612 V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622

By combining Eqs. (1), (2) and (7), we obtain


dN (dnuc ) N (dnuc ) 
=− K(dnuc ; dp; j )Nj : (8)
ddnuc GR j

Using then approximations (5) and (6) to the right-hand side of Eq. (8), we obtain further
dN (dnuc ) N (dnuc ) d20  
=−  CS =1 : (9)
ddnuc GR d2nuc
Eq. (9) can be rearranged to a form
dN (dnuc ) ddnuc
= − 2 ; (10)
N (dnuc ) dnuc
where the factor  is given by
d20  CS =1  CS =1
= = : (11)
GR GR
By integrating Eq. (10) from dnuc; ini to some larger size dp , and noting that  is constant according
to assumptions (2) and (3), we ;nally obtain
 
 
N (dp ) = N (dnuc; ini )exp − : (12)
dp dnuc; ini
Eq. (12) can also be written in terms of the nucleation rate. Since the “real” nucleation rate J is
equal to N (dnuc; ini )=Mt, the “apparent” nucleation rate Japp can be written as
 
 
Japp (dp ; t  ) = J (t) exp − : (13)
dp dnuc; ini
Here t  − t = (dp − dnuc; ini )=GR is the time di8erence between nuclei formation and their growth to
a size dp .
Eq. (13) is useful for modellers who want to predict the production of new particles using some
nucleation theory, but are unwilling (or unable) to extend the modelled particle size range down to
one nanometer. In this case the applied nucleation theory gives both J (t) and dnuc; ini , and the factor
 can be calculated from other model variables such as the pre-existing particle size distribution and
condensable vapour concentrations (for details, see Section 3.1).
Conversely, to estimate the “real” nucleation rate from atmospheric observations, one needs to
measure the time evolution of the particle number size distribution in the size range [dp ; dmax ],
where dp should be as small as possible (preferably a few nanometers) and dmax as large as possible
(¿ 500 nm). This information allows GR and CS , and thereby the factor , to be calculated. The rate
of change of the total particle number concentration in the size range [dp ; dmax ] can be interpreted
as Japp (dp ), after which the “real” nucleation rate becomes equal to
 
  
J (t) = Japp (dp ; t )exp − : (14)
dnuc; ini dp
The only unknown in Eq. (14) is dnuc; ini , the value of which must be estimated from a nucleation
theory. According to nucleation theories and also laboratory experiments, a good estimate for this
quantity is 1 nm.
V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622 613

2.3. Connection between the nucleation rate and the nuclei number concentration

An analytical formula for a total nuclei number concentration cannot be derived unless the nucle-
ation rate is constant with time, J (t) = J . For a constant nucleation rate, the concentration of nuclei
in any size range [dp; 1 ; dp; 2 ] can be calculated from the following equation:
 dp;2
1
N[dp; 1 ;dp; 2 ] = Japp (dp ) ddp : (15)
GR dp;1
By substituting Eq. (13) into (15) we obtain
J
N[dp; 1 ;dp; 2 ] = exp(−=dnuc; ini )F(; dp; 1 ; dp; 2 ); (16)
GR
where the function F, after denoting x = =dp , is given by
 =dp;2
F= −(=x2 ) exp(x) d x: (17)
=dp;1

The integral in Eq. (17) can be solved analytically, and the result is

 k+1 (d− k −k
p; 1 − dp; 2 )
F = dp; 2 exp(=dp; 2 ) − dp; 1 exp(=dp; 1 ) +  ln(dp; 2 =dp; 1 ) + : (18)
k k!
k=1

The number of terms to be taken into account in calculating the in;nite sum in Eq. (18) depends
on the magnitude of . If  ¡ 1 nm, only the ;rst ;ve terms are needed to receive a 1% accuracy in
the value of F. If  approaches 10 nm, a similar accuracy requires the ;rst 20 terms to be included.
Eq. (16) relates the nuclei number concentration in any size range to the “real” nucleation rate. A
similar relation between the nuclei number concentration and the “apparent” nucleation rate can be
obtained by connecting Eqs. (14) and (16). A special case of this relation is that between Japp (dp )
and the concentration of nuclei ¡ dp :
Japp (dp )
N[dnuc; ini ;dp ] = exp(−=dp )F(; dnuc; ini ; dp ): (19)
GR
Eq. (19) is valuable if one is able to measure the concentration of nuclei larger than some ;xed
diameter dp , but is interested in the concentration of nuclei smaller than this size.

3. Application to atmospheric nucleation events

3.1. Information needed to apply the derived formulae

In the previous section, analytical formulae (Eqs. (13), (14), (16) and (19)) connecting the “real”
and the “apparent” nucleation rate, as well as the total nuclei number concentration in a desired
size range, were derived. Before these formulae can be applied to atmospheric nucleation events,
the value of the parameter  given by Eq. (11) must be calculated. This requires determining the
following three terms: CS =1 ; GR, and .
The value of CS =1 is obtained directly from Eqs. (3) and (4) once the pre-existing particle number
size distribution is known. When calculating CS =1 for use in Eq. (11), two things must be kept in
614 V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622

mind: the parameter is set equal to unity, and the Knudsen number is calculated by replacing 
with the air mean free path air . This procedure does not restrict the generality or applicability of our
results, it simply makes the proportionality factor  independent of the physico-chemical properties
of condensing vapours.
The nuclei growth rate GR can be extracted from atmospheric measurement or, when this in-
formation is not available, it can be calculated from condensable vapour concentrations using the
following equation:
4  Di m (Knnuc ; i )
GR = [Ci − Ci; eq ]: (20)
nuc i dnuc
Here nuc is the nuclei density, m is given by Eq. (4), the index i goes over condensable vapours,
Di is the vapour di8usion coe6cient, and Ci and Ci; eq are the vapour concentrations in the gas-phase
and over the particle surface, respectively. Noting that Knnuc 1 and assuming that the condensable
vapours responsible for nuclei growth are non-volatile (Ci; eq Ci ), a following approximation to GR
(in nm h−1 ) can be obtained:
3:0 × 10−9 
GR ≈ cVi Mi Ci : (21)
nuc i

Here Ci and nuc are given in molecules cm−3 and in kg m−3 , respectively, cVi (m s−1 ) is the molec-
ular speed of the condensing vapour, and Mi (g mol−1 ) is its molecular weight.
The proportionality factor  is approximately constant, even though it is expected to vary weakly
with the shape of the pre-existing particle size distribution, the initial size and density of the nuclei,
and the ambient temperature and pressure. In order to quantify the magnitude and the variability
of the factor , we constructed a numerical model that can accurately simulate the time evolution
of a growing nuclei population in the presence of larger pre-existing particles. Since the Brownian
coagulation kernels between di8erent-size particles are well de;ned, this kind of model is relatively
simple to implement as long as the vapours responsible for the nuclei growth are assumed to be
non-volatile. By making a large set of simulations with di8erent model parameters (the initial and
;nal size of the nuclei, the nuclei density, the mean diameter of the pre-existing particle population,
the ambient temperature), and comparing the results to the analytical predictions (Eq. (13)), the
following semi-empirical formula for  could be obtained:
       −0:33  −0:75
dnuc; ini 0:2 dp 0:075 dmean 0:048 nuc T
 = 0 : (22)
1 nm 3 nm 150 nm 1000 kg m−3 293 K
Here 0 is equal to 0:23 nm2 m2 h−1 , dmean is the number mean diameter of the pre-existing particle
population, and T is the ambient temperature. The accuracy of Eq. (22) will be discussed separately
in Section 3.2.2.
After demonstrating how the parameter  can be determined in practice, it is worth giving some
idea on its variability in the atmosphere. The main factors a8ecting the value  are the total concen-
tration of pre-existing particles, their mean size, and the nuclei growth rate (Table 1). An estimate
for the magnitude of  in the lower troposphere can be calculated using measured particle number
size distributions (Seinfeld & Pandis, 1998; Heintzenberg, Covert, & Van Dingenen 2000; MCakelCa,
Koponen, Aalto, & Kulmala, 2000b; Raes et al., 2000) and nuclei growth rates (Weber et al., 1997,
1998; Birmili & Wiedensohler, 2000; MCakelCa et al., 2000a). As a result, typical values of  were
V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622 615

Table 1
The in4uence of a pre-existing particle population and the nuclei growth rate on the factor . A uni-modal particle size
distribution has been assumed, with Ntot representing the total particle number concentration, dmean the number mean
diameter, and  the geometric standard deviation. Also given are the condensation sink in forms CS and CS = 4Di CS

Ntot dmean  GR CS CS 


(cm−3 ) (nm) (nm h−1 ) (m−2 ) (s−1 ) (nm)
100 150 1.5 5 5 5:8 × 10−4 0.22
500 150 1.5 5 23 2:9 × 10−3 1.10
2000 150 1.5 5 92 1:2 × 10−2 4.40
5000 150 1.5 5 230 2:9 × 10−2 11.01
500 100 1.5 5 12 1:6 × 10−3 0.58
500 500 1.5 5 110 1:4 × 10−2 5.70
500 150 1.3 5 21 2:6 × 10−3 1.00
500 150 2.0 5 29 3:6 × 10−3 1.38
500 150 1.5 2 23 2:9 × 10−3 2.75
500 150 1.5 10 23 2:9 × 10−3 0.55

found to be in the range 0.1–1 nm for remote marine and continental areas, in the range 1–5 nm
for rural areas, and close to or above 10 nm for urban and other heavily polluted environments. For
aerosols in the free troposphere, practically no information on nuclei growth rates is available. As-
suming the nuclei growth rate to be of the order 1–5 nm h−1 and using observed particle number size
distributions (Weber & McMurry, 1996; Raes, Van Dingenen, Cuevas, Van Velthoven, & Prospero,
1997; Nyeki et al., 1998), the values of  in the background free troposphere can be estimated to
be of the order 0.1–1 nm. Larger values are expected to found in the presence dust or anthropogenic
pollution, or when the nuclei growth rates are low ¡ 1 nm h−1 .

3.2. Uncertainties associated with the derived formulae

Since several simplifying assumptions and approximations had to be made in deriving the ana-
lytical formulae (13), (14), (16) and (19), some error in predictions given by these formulae is to
be expected. In the following we aim to give some idea on the magnitude of these errors under
di8erent atmospheric conditions. The analysis is based on detailed comparisons between the analyt-
ical predictions and accurate numerical results. The latter are obtained from simulations using the
detailed numerical model mentioned in Section 3.1.

3.2.1. Validity of our assumptions


Our ;rst assumption in deriving the analytical formulae was that the only important sink for nuclei
is their coagulation with larger pre-existing particles. In the absence of rain, two other potential sinks
are the dry deposition to various surfaces and self-coagulation. The nuclei dry deposition velocities
are unlikely to reach values much ¿ 0:1 m s−1 (Zufall, Bergin, & Davidson, 1998; Zhang, Gong,
Padro, & Barrie, 2001), making the nuclei lifetime against dry deposition of the order of several
hours or more in the atmospheric boundary layer. This is usually much longer than the time scales
over which nuclei either grow to a few nanometers or are scavenged by the pre-existing particle
population. We may conclude that the nuclei dry deposition has practically no in4uence on the
validity of assumption (1).
616 V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622

Fig. 1. The fraction of nuclei left after their growth from 1 to 5 nm as a function of the parameter . Self-coagulation
has been excluded in the set simulations marked with a solid line, and included in the three set simulations marked with
a dashed line. The initial nuclei concentration has been set equal to 104 , 105 or 106 cm−3 (the lowest, middle, and upper
dashed line, respectively).

Self-coagulation a8ects the nuclei population in two ways: it reduces the nuclei number concen-
tration and increases their mean size. To quantify when self-coagulation starts to a8ect the system,
two sets of simulations using our numerical model were performed: those including and those ex-
cluding self-coagulation. An illustrative subset of the simulation results is shown in Fig. 1. In these
simulations, the growth of a nuclei population from 1 to 5 nm was followed. During their growth,
the nuclei were allowed to coagulate with both pre-existing larger particles (coagulational scaveng-
ing) and other nuclei present (self-coagulation). The concentration of condensable vapours was ;xed
such that the nuclei growth rate due to condensation alone was equal to 5 nm h−1 . The initial nuclei
number concentration was varied between 104 and 106 cm−3 . The shape of the pre-existing larger
particle population was kept ;xed, but the concentration of these particles was varied from simula-
tion to simulation by varying the parameter  given by Eq. (11). From Fig. 1 we can see that in
a very clean environment ( ≈ 0:1), only a small fraction of the nuclei are scavenged away by the
pre-existing larger particles. The in4uence of self coagulation becomes important when the initial
nuclei number concentration exceeds about 105 cm−3 . In polluted environments ( ¿ 1), most of the
nuclei coagulate with larger pre-existing particles and the role of self-coagulation is diminished.
When summarizing the above ;ndings, we may conclude that assumption (1) is valid until self
coagulation starts to in4uence the time evolution of the total nuclei number concentration. In clean
environments and at low nuclei growth rates, this may occur at nuclei concentrations of the order
105 cm−3 . In polluted environments and at higher nuclei growth rates, a nuclei concentration in
excess of 106 cm−3 is required to make self-coagulation important.
Our second assumption was that the nuclei grow at a constant rate GR by condensation. The
validity of this assumption can be investigated by looking at the time evolution of the di8erent
terms in Eq. (20). During the nuclei growth, the magnitude of GR can be changed by changes in
(i) the ratio m =dnuc , (ii) the nuclei density, or (iii) the concentrations Ci and Ci; eq . If the mass
V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622 617

Fig. 2. Ratio between the “apparent” and the “real” nucleation rate. The parameter  has been assumed to be constant
(solid line), or to change by ±20% (dotted line) or ±50% (dash–dotted line) during the nuclei growth from 1 nm to the
;nal size of 3 nm. In all simulations, GR = 5 nm h−1 , dmean = 150 nm, nuc = 1 g cm−3 , and T = 293 K.

accommodation coe6cients of the condensable vapours do not change as a function of the nuclei
diameter, the ratio m =dnuc is approximately constant (within 2%) over the diameter range 1–10 nm.
Changes in the nuclei density might be somewhat larger (up to 20 –30%), but only if the ambient
relative humidity (RH) and=or the chemical composition of the nuclei varies considerably. Changes
in Ci; eq are probably of less signi;cance, since the nuclei growth is likely to be controlled by rela-
tively non-volatile vapours for which Ci; eq Ci (Kerminen, Virkkula, Hillamo, Wexler, & Kulmala,
2000). We conclude that the main variable controlling the variability of GR is the concentration
of condensible vapour(s) Ci which, depending on time of day, could vary by a factor two or even
more during the nuclei growth to 5 –10 nm.
Our third assumption was that the condensation sink CS is constant. In reality, CS may either
increase or decrease during the nuclei growth. An increase in CS results when secondary material
is transferred from the gas-phase to the pre-existing particle population, or when the ambient RH
increases. A decrease in CS results if the ambient RH decreases, or when the air mass is diluted,
for example, due to the growth of the boundary layer depth. However, according to recent studies on
daily variation of CS , it is seems to be relatively constant during particle formation events (Kulmala
et al., 2001b).
The above discussion demonstrates that both the nuclei growth rate GR and the condensation sink
CS could change signi;cantly (by factor two or even more) during the nuclei growth to several
nanometers in the atmosphere. Fig. 2 illustrates how a certain change in the parameter  during
the nuclei growth (illustrating potential changes in GR and=or CS ) a8ects the ;nal nuclei number
concentration, or the “apparent” nucleation rate Japp . The results are based on simulations with our
numerical model, and the parameter  is assumed to increase or decrease linearly with time. A couple
of things can be immediately seen from Fig. 2. First, the ratio between the “apparent” and the “real”
nucleation rate decreases strongly with an increasing value of . This highlights the important role
of pre-existing particles in scavenging fresh atmospheric nuclei, especially in polluted air masses.
Second, the uncertainty in Japp resulting from the potential variability of the parameter  during the
nuclei growth remains relatively small for  up to about 1 nm. When  is greater, a larger fraction
618 V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622

Fig. 3. Ratio between the “apparent” and the “real” nucleation rate as a function of the parameter . Calculations are
based on either accurate numerical simulations (solid line) or formula (13), with  given by Eq. (22) (dotted line) or 
put equal to 0 = 0:23 nm2 m2 h−1 (dash–dotted line). In the top picture, GR = 2 nm h−1 , dmean = 150 nm, dnuc; ini = 1:2 nm,
nuc = 1:5 g cm−3 , and T = 293 K. In the bottom ;gure, GR = 5 nm h−1 , dmean = 500 nm, dnuc; ini = 1 nm, nuc = 1 g cm−3 ,
and T = 293 K. Note the di8erent ;nal nuclei diameter between the two pictures.

of the nuclei will be scavenged away during their growth, and Japp becomes more sensitive to the
potential changes in either the nuclei growth rate or the pre-existing particle size distribution.

3.2.2. Accuracy of the analytical formulae


So far we have investigated the atmospheric validity of the various assumptions made in deriving
our analytical formulae. In this sections the accuracy of these formulae against detailed numerical
simulations is investigated. The analysis is performed under conditions where assumptions (1)–(3)
are valid, so it measures mainly those uncertainties that arise from approximations (5) and (6) or,
more explicitly, from the functional form of formula (22).
To start with, let us look at the connection between the “real” and the “apparent” nucleation rate.
Fig. 3 gives some illustration on the errors in Japp predicted by formula (13). It turned out that
V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622 619

under most atmospheric situations, the error in Japp is ¡ 10% for  up to about 1 nm, and ¡ 50%
for  up to about 10 nm, when using full equation (22) to calculate the factor . Slightly larger
errors result if the factor  is assumed to be constant and equal to 0:23 nm2 m2 h−1 . We conclude
that formulae (13) and (14) can be applied successfully to atmospheric nucleation events, provided
that the nuclei number concentration remains su6ciently low (¡ 105 –106 cm−3 ) to prevent e8ective
self-coagulation, and that there are no major 4uctuations in the pre-existing particle size distribution
or in condensable vapour concentrations during the nuclei growth (see Figs. 1 and 2).
The errors in nuclei number concentrations predicted using Eqs. (16) or (19) are more di6cult
to quantify. If the nucleation rate J is constant with time, the error in N[dp; 1 ; dp; 2 ] is comparable to
that in Japp (dp; 2 ) predicted by Eq. (13). If the nucleation rate varies with time, the errors becomes
comparable to the factor by which J varies over the period it takes for nuclei to grow up to the
size dp; 2 .
The initial size of fresh atmospheric nuclei is typically around 1 nm, whereas the minimum size
detected by current measurement techniques is around 3 nm. Fig. 4 shows the concentration of nuclei
in the size range 1–3 nm, as predicted using Eq. (16). We can see that in rural and remote regions,
nucleation rates as low as 1 cm−3 s−1 are su6cient to raise N[1 nm; 3 nm] above 103 nuclei cm−3 . In
polluted areas, N[1 nm; 3 nm] decreases rapidly with increasing  due to the e8ective scavenging of the
nuclei by pre-existing larger particles. Self-coagulation is more important in cleaner air, where it
starts to a8ect the accuracy of Eq. (16) at nucleation rates between about 10 and 100 cm−3 s−1 .

4. Summary and conclusions

While the production of new atmospheric aerosol particles by homogeneous or some other nucle-
ation mechanism has received increasing attention, there are still many practical problems in dealing
with the smallest particles (nuclei) associated with atmospheric nucleation events. In this paper, an
analytical formula between the “real” and the “apparent” nucleation rate has been derived. The latter
is equal to the rate at which the nuclei appear at some larger particle size as a result of their growth
by condensation, being always smaller than the “real” nucleation rate. For cases where the nucle-
ation rate is approximately constant with time, another formula was derived that connects the total
nuclei concentration in a desired size range and either the “real” or the “apparent” nucleation rate.
The results in this study are comparable with recent results obtained using aerosol dynamic model
(Kulmala, Pirjola, & MCakelCa, 2000) and the present results con;rm the idea that in the atmosphere
there can be plenty of very small aerosol particles (i.e. thermodynamically stable clusters) below the
detection limit of commonly-used aerosol instruments (see also Kulmala et al., 2000).
Since the derivation of the analytical formulae was based on several simplifying assumptions
and approximations, there are a couple of limitations regarding the atmospheric conditions under
which they can be applied. First, there must be no major 4uctuations (by more than a factor two
or three) in the pre-existing particle size distribution, or in the concentration of vapours responsible
for the nuclei growth. Second, the total nuclei number concentration must remain su6ciently low
(¡105 –106 nuclei cm−3 ) to prevent e8ective self-coagulation between the nuclei. The formulae are
therefore not applicable to very intensive nucleation bursts, to potential nucleation events associated
with cloud out4ows, or to nucleation occurring in plumes undergoing strong mixing with the ambient
air.
620 V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622

Fig. 4. Predicted concentration of nuclei in the size range 1–3 nm when the nuclei growth rate is equal to 2 nm h−1 (top) or
10 nm h−1 (bottom). The “real” nucleation rate is assumed to be equal to 1 (solid line), 10 (dotted line) or 100 cm−3 s−1
(dash–dotted line). The thick solid line in the top half of both pictures represent the limit at which self-coagulation causes
an error ¿ 10% in the predicted nuclei number concentration.

When atmospheric conditions are such that the analytical formulae can be considered applica-
ble, the formulae are also relatively accurate. Comparisons to detailed numerical showed that errors
in analytical predictions were usually ¡ 10% for nucleation events occurring in clean or mod-
erately polluted air masses, and rarely exceeded 50% even in polluted air masses. These errors
are minor compared to the uncertainties in “real” nucleation rates predicted by current nucleation
theories.
The derived formulae are useful for both atmospheric modelers and people making ;eld experi-
ments. In measurements conducted in ;eld, for example, one is currently able to determine only the
“apparent” nucleation rate. The derived formulae provide a simple means to convert this rate to a
“real” nucleation rate (or to the total nuclei number concentration), allowing a more rigorous testing
of the viability of di8erent nucleation theories in the atmosphere.
V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622 621

In models, formation of new particles can be treated basically in two di8erent ways: (1) the
formation of critical clusters by nucleation and the subsequent growth of these clusters to larger
sizes is modelled explicitly, or (2) new particles, the number of which is estimated using some
speci;c algorithm, are inserted directly into the ;rst size section (usually ¿ 10 nm of the particle
diameter) of the model. The former approach is in principal very accurate, provided that we have a
proper nucleation theory. The drawback is that the modelled particle size range has to be extended
down to one nanometer with a ;ne resolution. This is not possible in current global or even regional
atmospheric models, so the latter approach has usually been applied. However, the latter approach
misses practically all aerosol dynamics occurring during the nuclei growth to the ;rst size section of
the model. The analytical formulae derived here allow atmospheric modelers to convert the “real”
nucleation rate, predicted by a desired nucleation theory, to the formation rate of larger particles
(5 –10 nm in diameter) in a way that is both robust and consistent with the real aerosol dynamics
taking place in the system.

Acknowledgements

This study has been funded by the Academy of Finland (contract number 67838) and by the Maj
and Tor Nessling foundation.

References

Arstila, H., Korhonen, P., & Kulmala, M. (1999). Ternary nucleation: kinetics and application to water-ammonia-
hydrochloric acid system. Journal of Aerosol Science, 30, 131–138.
Birmili, W., & Wiedensohler, A. (2000). New particle formation in the continental boundary layer: Meteorological and
gas phase parameter in4uence. Geophysical Research Letters, 27, 3325–3328.
Clarke, A. D., Eisele, F., Kapustin, V. N., Moore, K., Tanner, D., Mauldin, L., Litchy, M., Lienert, B., Carroll, M. A.,
& Albercook, G. (1999). Nucleation in the equatorial free troposphere: Favorable environments during PEM-Tropics.
Journal of Geophysical Research, 104, 5735–5744.
Fuchs, N. A., & Sutugin, A. G. (1971). Highly dispersed aerosols. In G.M. , Hidy, & J.R. Brock (Eds.), Topics in current
aerosol research. New York: Pergamon.
Harrison, R. M., Grenfell, J. L., Savage, N., Allen, A., Clemitshaw, K. C., Penkett, S., Hewitt, N., & Davison, B. (2000).
Observations of new particle production in the atmosphere of a moderately polluted site in eastern England. Journal
of Geophysical Research, 105, 17,819–17,832.
Heintzenberg, J., Covert, D. C., & Van Dingenen, R. (2000). Size distributions and chemical composition of marine
aerosols: a compilation and review. Tellus, 52B, 1104–1122.
Kavouras, I. G., Mihalopoulos, N., & Stephanou, E. G. (1998). Formation of atmospheric particles from organic acids
produced by forests. Nature, 395, 683–686.
Kerminen, V.-M., Virkkula, A., Hillamo, R., Wexler, A. S., & Kulmala, M. (2000). Secondary organics and atmospheric
cloud condensation nuclei production. Journal of Geophysical Research, 105, 9255–9264.
Korhonen, P., Kulmala, M., Laaksonen, A., Viisanen, Y., McGraw, R., & Seinfeld, J. H. (1999). Ternary nucleation of
H2 SO4 ; NH3 , and H2 O in the atmosphere. Journal of Geophysical Research, 104, 26349–26353.
Kulmala, M., Laaksonen, A., & Pirjola, L. (1998). Parameterizations for sulphuric acid=water nucleation rates. Journal of
Geophysical Research, 103, 8301–8308.
Kulmala, M., Pirjola, L., & MCakelCa, J. M. (2000). Stable sulphate clusters as a source of new atmospheric particles.
Nature, 404, 66–69.
622 V.-M. Kerminen, M. Kulmala / Aerosol Science 33 (2002) 609 – 622

C Dal Maso, M.,


Kulmala, M., HCameri, K., Aalto, P. P., MCakelCa, J. M., Pirjola, L., Nilsson, E. D., Buzorius, G., Rannik, U.,
Seidl, W., Ho8mann, T., Janson, R., Hansson, H.-C., Viisanen, Y., Laaksonen, A., & O’Dowd, C. D. (2001a). Overview
of the international project on Biogenic aerosol formation in the boreal forest (BIOFOR). Tellus, 53B, 324–343.
Kulmala, M., Dal Maso, M., MCakelCa, J. M., Pirjola, L., VCakevCa, M., Aalto, P., Miikkulainen, P., HCameri, K., & O’Dowd,
C. D. (2001b). On the formation, growth and composition of nucleation mode particles. Tellus, 53B, 479–490.
MCakelCa, J. M., Dal Maso, M, Pirjola, L., Keronen, P., Laakso, L., Kulmala, M., & Laaksonen, A. (2000a). Characteristics
of the atmospheric particle formation events observed at boreal forest site in southern Finland. Boreal Environment
Research, 5, 299–313.
MCakelCa, J. M., Koponen, I., Aalto, P., & Kulmala, M. (2000b). One year of data of submicron size modes of tropospheric
background aerosol in southern Finland. Journal of Aerosol Science, 31, 595–611.
Nyeki, S., Li, F., Weingartner, E., Streit, N., Colbeck, I., GCaggeler, H. W., & Baltensperger, U. (1998). The background
aerosol size distribution in the free troposphere: An analysis of the annual cycle at a high-alpine site. Journal of
Geophysical Research, 103, 31,749–31,761.
O’Dowd, C. D. (2001). Biogenic coastal aerosol production and its in4uence on aerosol radiative properties. Journal of
Geophysical Research, 106, 1545–1549.
Raes, F., Van Dingenen, R., Cuevas, E., Van Velthoven, P. F. J., & Prospero, J. M. (1997). Observations of aerosols in the
free troposphere and marine boundary layer of the subtropical Northeast Atlantic: Discussion of processes determining
their size distribution. Journal of Geophysical Research, 102, 21,315–21,328.
Raes, F., Van Dingenen, R., Vignati, E., Wilson, J., Putaud, J.-P., Seinfeld, J. H., & Adams, P. (2000). Formation and
cycling of aerosols in the global troposphere. Atmospheric Environment, 34, 4215–4240.
Seinfeld, J. H., & Pandis, S. N. (1998). Atmospheric chemistry and physics: From air pollution to climate change. New
York: Wiley.
Weber, R. J., & McMurry, P. H. (1996). Fine particle size distributions at the Mauna Loa Observatory, Hawaii. Journal
of Geophysical Research, 101, 14,767–14,775.
Weber, R. J., Marti, J. J., McMurry, P. H., Eisele, F. L., Tanner, D. J., & Je8erson, A. (1997). Measurements of new
particle formation and ultra;ne particle growth rates at a clean continental site. Journal of Geophysical Research, 102,
4375– 4385.
Weber, R. J., McMurry, P. H., Mauldin, L., Tanner, D. J., Eisele, F. L., Brechtel, F. J., Kreidenweis, S. M., Kok, G. L.,
Schillawski, R. D., & Baumgardner, D. (1998). A study of new particle formation and growth involving biogenic trace
gas species measurement during ACE 1. Journal of Geophysical Research, 103, 16,385–16,396.
Woo, K. S., Chen, D. R., Pui, D. Y. H., & McMurry, P. H. (2001). Measurement of Atlanta aerosol size distributions:
Observations of ultra;ne particle events. Aerosol Science and Technology, 34, 75–87.
Zhang, L., Gong, S., Padro, J., & Barrie, L. (2001). A size-segregated dry deposition scheme for and atmospheric aerosol
module. Atmospheric Environment, 35, 549–560.
Zufall, M. J., Bergin, M. H., & Davidson, C. I. (1998). E8ects of non-equilibrium hygroscopic growth of (NH4 )2 SO4 on
dry deposition to water surfaces. Environmental Science and Technology, 32, 584–590.

You might also like