RESEARCH FRONT
Rapid Communication
CSIRO PUBLISHING
K. Sellegri et al., Environ. Chem. 2005, 2, 260–270. doi:10.1071/EN05074
www.publish.csiro.au/journals/env
Quantification of Coastal New Ultra-Fine Particles Formation
from In situ and Chamber Measurements during
the BIOFLUX Campaign
K. Sellegri,A,B Y. J. Yoon,A,C S. G. Jennings,A C. D. O’Dowd,A,G L. Pirjola,D S. Cautenet,E
Hongwei Chen,F and Thorsten HoffmannF
A
Department of Physics, National University of Ireland, Galway, Ireland.
address: Laboratoire de Météorologie Physique, Université Blaise Pascal, Clermont-Ferrand, France.
C Present address: Korea Polar Research Institute/KORDI, Ansan, PO Box 29, Seoul 425-600 Korea.
D Helsinki Polytechnic, Department of Technology Helsinki, Finland.
E Laboratoire de Météorologie Physique, Université Blaise Pascal, Clermont-Ferrand, France.
F Johannes Gutenberg University of Mainz, Institute of Inorganic and Analytical Chemistry, Germany.
G Corresponding author. Email: [email protected]
B Present
Environmental Context. Secondary processes leading to the production of ultra-fine particles by nucleation are still poorly understood. A fraction of new particles formed can grow into radiatively active sizes,
where they can directly scatter incoming solar radiation and, if partly water soluble, contribute to the cloud
condensation nuclei population. New particle formation events have been frequently observed at the Mace
Head Atmospheric Research Station (western Ireland), under low tide and sunny conditions, leading to the
hypothesis that new particles are formed from iodo-species emitted from macroalgae.
Abstract. New particle formation processes were studied during the BIOFLUX campaign in September 2003 and
June 2004. The goals were to bring new information on the role of I2 in new particle formation from seaweeds and
to quantify the amount of I2 emitted and new particles formed by a given amount of seaweed. These two goals were
achieved by using a simulation chamber filled with selected species of seaweeds from the Mace Head area and flushed
with particle-free atmospheric air. It was found that total particle concentrations and particles in the 3–3.4 nm size
range produced in the chamber are positively correlated with gaseous I2 concentrations emitted by the seaweeds, with
a typical source rate of 2800 particles cm−3 ppt−1
(I2 ) in the 3–3.4 nm size range. In fact, I2 and particle concentrations
are also both directly positively correlated with the seaweed mass (64 300 particles cm−3 formed per kg of seaweed,
and 24 ppt of I2 per kg of seaweeds) until saturation was reached for a seaweed biomass of 7.5 kg m−2 . From the
chamber experiments, the flux of 3–3.4 nm particles was calculated to be 2.5 × 1010 m−2 s−1 for a seaweed loading
of 2.5 kg m−2 (representative of a typical seaweed field density), decreasing to 1 × 1010 m−2 s−1 for a seaweed
loading of 1 kg m−2 . At a seaweed loading of 2.5 kg m−2 , the growth rate of particles produced in the chamber was
calculated to be 1.2 nm min−1 . The source rates and growth rates determined from the chamber experiments were
used in conjunction with seaweed coverage in and around Mace Head to produce local emission inventories for a
meso-scale dispersion model. Comparison of the resulting aerosol size distributions from the model simulations
and those observed exhibited good agreement suggesting that the chamber fluxes and growth rates are consistent
with those associated with the tidal emission areas in and around Mace Head.
Manuscript received: 4 October 2005.
Final version: 27 October 2005.
Introduction
The global impact of atmospheric particles cannot be evaluated if their sources, both natural and anthropogenic, are not
understood. Natural atmospheric aerosols can be produced
through both primarily and secondarily processes. Sources
of primary particles, such as sea spray and desert dust have
been extensively studied and are usually correlated with the
© CSIRO 2005
260
wind speed and their production mechanisms are quite well
understood. However, secondary processes leading to the
production of ultra-fine particles by nucleation are still poorly
understood. A fraction of new particles formed can grow into
radiatively active sizes where they can directly scatter incoming solar radiation, and if partly water soluble, contribute
to the cloud condensation nuclei (CCN) population. New
1448-2517/05/040260
RESEARCH FRONT
Quantification of Coastal New Ultra-Fine Particles Formation from In situ and Chamber Measurements during the BIOFLUX Campaign
particle formation events have been frequently observed at
the Mace Head Atmospheric Research Station (western Ireland), under low tide and sunny conditions, leading to the
hypothesis that new particles are formed from iodo-species
emitted from macroalgae.[1,2] To confirm this hypothesis, it
has recently been shown in laboratory studies that ultra-fine
iodine-containing particles are produced by algae exposed to
scrubbed atmospheric air enriched with ozone.[3]
One of the main objectives of the scientific community
concerned with this topic is to be able to predict the number of
new particles formed by secondary processes, in order to evaluate their impact on a larger scale. Detailed models have been
developed in order to simulate new particles formation from a
certain amount of gaseous precursors that, under sunny conditions, will be oxidized into condensable species.[3] However,
significant uncertainties still remain, both on the nature of
the nucleating species and on the emission rate of their
precursors. Several chemical mechanisms implying iodocompounds have been suggested to explain the formation of
new particles in the coastal atmosphere of western Ireland.
Iodine was observed to be released from seaweeds mainly
as CH2 I2 , and in sometimes significant fractions of CHIBr
and CHICl.[4,5] In conjunction with these observations, it
was found that new particles were formed from CH2 I2 in
the presence of UV radiation and O3 .[1,6] Analysis, by mass
spectrometry, of the new particles formed during these experiments show that they were composed of iodine oxides and
oxy-acids. CH2 I2 readily photolyzes at ambient light levels, and I2 becomes the main gaseous specie after a few
minutes.[6] Depending on the uptake of I2 in the models, I2 is
not converted to the particulate phase and remains in the gas
phase until it is photolyzed,[6] or it becomes the major component of the final aerosol (uptake of 1).[6] Recent works show
that I2 is present at levels of the order of 10 ppt in the coastal
atmosphere of Mace Head, most likely emitted from macroalgae at low tide and that concentrations of this magnitude
could be significant in terms of contributing to the new particle formation process.[7] However, it is not yet elucidated if
I2 is directly emitted from seaweeds by liquid-to-gas transfer,
or if it is formed by oxidation of I− by H2 O2 upon stress on
the seaweed during low tide atmospheric exposure.[3]
The goal of this paper is two-fold: (1) to elucidate the
relationship between I2 and new particle formation from seaweeds; and (2) to quantify the source rates of new particles
formed in the vicinity of Mace Head and nearby biologicallyactive regions. We present the results of two measurement
campaigns (Quantifying coastal BIOgenic aerosol and gas
FLUX : BIOFLUX) on the coastal site of Mace Head and its
surroundings, conducted during September 2003 and June
2004. During BIOFLUX, a simulation chamber was built
to quantify I2 emissions along with new particle formation
and growth rate from selected species of seaweeds, representative of the area, under atmospheric conditions. A mobile
lab equipped with particle size distribution analyzers was
used to study in situ new particle formation at high-density
seaweed locations in order to relate chamber and in situ
measurements, by the mean of a mesoscale transport model
Regional Atmospheric Modelling System (RAMS). New
particle size distributions were simultaneously measured
at the Mace Head Atmospheric Research station (53.33 N,
9.90W, 5 m above sea level) and, with the mobile laboratory,
from several locations in the coastal areas of western Ireland
which are believed to have a higher particle formation activity
than Mace Head.
Experimental
The Mace Head Atmospheric Research Station,[8] located at the west
coast of Ireland, shows evidence of frequent nucleation events and
particle growth cases.[2] These particle formations are usually observed
when tides are low and sunshine is available. The simulation chamber was filled with seaweed collected from the Mace Head region. The
chamber was 2 m × 1 m × 1 m in dimension and was made of the Perspex with a 50% UV radiation transmittance. The chamber was operated
in two modes, one without flow through and another with flow of ambient coastal air scrubbed of background aerosols but not scrubbed of
gases. Chamber particle size distributions were measured using a nanoSMPS (Scanning Mobility Particle Sizer for nano-particles composed
of a 3085 TSI short column and a 3025 TSI CPC). Gaseous molecular iodine was collected using a denuder sampling technique based
on the specific reaction between starch and iodine.[9,10] An ethanolic
starch solution (2 mg mL−1 ) was used to coat to the inner surfaces of
brown glass tubes (inner diameter 6 mm, outer diameter 9 mm, length
50 cm). In order to produce a uniform starch coating of the inner denuder
walls, four 0.5 mL aliquots of the coating solution were slowly instilled
into the pre-cleaned rotating denuder. During the coating procedure the
denuder tubes were dried by flushing N2 at a flow rate of 0.5 L min−1 .
Finally, the denuders were sealed with polypropene end-caps. For the
chamber experiments, a volumetric air flow of 500 mL chamber air
per minute was sucked through the denuder tubes. The sampling time
varied between 15 and 30 min. Once the sampling was completed, the
open endings of the tubes were again sealed with PP-caps. In the laboratory, the iodine–amylose complex was eluted and extracted from the
denuder walls with 4 mL of a TMAH-solution (TetraMethylAmmonium
Hydroxide, 5% by weight) at 90◦ C for 3 hours. Afterwards, the solution
was diluted to 1% TMAH, with adding 200 ppb Tellurium as an internal standard and the iodine concentration was finally determined by
Inductively Coupled Plasma–Mass Spectrometry (ICP-MS, HP 4500,
Agilent, USA).
The Mace Head atmospheric research station is equipped with a twinCPC (3010 and 3025 TSI CPC) which detects particles with different
cut-off diameters (3 nm and 10 nm respectively) thus give information
on newly formed particles. Also, a Scanning Mobility Particle Sizer
(SMPS) is operated continuously at the station for the measurement of
the aerosol size distribution between 8 and 300 nm in diameter. The
Mobile lab (Helsinki Polytechnic, Finland),[11] was equipped with a
nano-SMPS and an Electrical Low Pressure Impactor (ELPI, Dekati).
The nano-SMPS was set to measure particle sizes from 3 to 50 nm with
a time resolution of 30 s and the ELPI measured particle sizes from 7 nm
to 10 µm with a 1 s resolution averaged to the same 30 s time resolution.
It was possible to compare the number of particles smaller than 10 nm
measured by the nano-SMPS in the mobile laboratory with the number
of particles smaller than 10 nm inferred from the twin CPC at the Mace
Head station, when the mobile laboratory was at the Mace Head research
station.
Results
New Particles from Seaweeds in the Simulation Chamber
Two genera of seaweeds widely found in the Mace Head tidal
area were tested, (1) Laminaria sp. and (2) Fucus sp. Both
Laminaria and Fucus seaweed species produced particles in
the simulation chamber with no distinguishable difference.
In the first experiment, no flow-through of coastal particlefree air was applied to the chamber (except the 1 LPM needed
261
RESEARCH FRONT
K. Sellegri et al.
for sampling, which was assumed to be negligible relative to
the 2 m3 chamber). Under these conditions, up to 5 × 106
3–3.4 nm particles cm−3 were produced. The newly formed
particles grew very rapidly to sizes of 50 nm where concentrations in this size range were of the order of 250 000 particles
cm−3 after 30 min. Particle formation during this experiment
was typical of a series of nucleation pulses where there is a
large burst of 3–3.4 nm particles followed by rapid growth
and where the formation of new particles stops shortly after
the initial pulse (Fig. 1a). After growth of the new particles to larger sizes, another nucleation pulse follows. This
pattern can be explained either by an increase in the condensation sink of the formed aerosol to a level which shuts
off nucleation or by a rapid consumption of O3 to shut off
the production of the nucleating species, or by a combination of both of these processes. The condensation sink due to
pre-existing particles during new particle formation events is
calculated using the full distribution of the Aitken, accumulation and coarse modes inferred from the SMPS distribution
Diameter (nm)
(a)
CS = 2π × D × β(i) × d(i) × N (i),
where D is the diffusion coefficient depending on density
and molecular volume of the condensable species OIO, β is a
coefficient depending on the Knudsen number and a sticking
coefficient of 1, the particle diameter is d and N the number
of aerosols of diameter d.
After clear evidence of nucleation pulses from local
seaweed under atmospheric conditions, a steady-state experiment was setup to evaluate this phenomenon in more detail
and as a function of seaweed biomass. The chamber was
filled with a given amount of seaweeds, then the lid of
the chamber was closed and a flow rate of 800 L min−1
was applied to the chamber outlet, letting in ambient atmospheric air through the particle-filtered inlet. Hence, the mean
residence time being 2 min 30 s, temperature and relative
humidity inside the chamber are close to ambient conditions
(T = 16.7 ± 0.4◦ C and RH = 69.3 ± 1.6 over the measurement period). A steady state was reached within a few min
when new 3 nm particles were continuously produced (the
wave pattern was not observed when a constant flushing of
800 L min−1 was applied). After the steady state was reached,
sampling of the particles and gas phase molecular iodine
concentration were performed for about 15 min. The same
procedure was repeated for several different amounts of seaweeds. Size distributions of particles as a function of time
are shown Fig. 1b. For 16 kg of seaweeds, the total and
3–3.4 nm particle concentrations were 1 × 107 and 1.2 × 106
particles cm−3 respectively. Nanoparticles were continuously
produced whatever the amount of seaweed, irrespective of the
condensation sink present within the chamber. In fact, for the
highest chamber content of seaweed, the condensation sink
was as high as 0.7 s−1 and new particles were still produced
in significant quantities.
For each amount of seaweed biomass there corresponds
a steady state number of particles produced and steady
state gaseous I2 concentration resulting. We found that total
particle concentrations and 3–3.4 nm particles concentrations produced in the chamber are positively correlated
with the gaseous I2 concentrations emitted by the seaweeds
(Fig. 2a). The rate of new 3–3.4 particle production is of
2800 cm−3 ppt−1 . In fact, we also found that I2 and particle
concentrations are both directly positively correlated to the
seaweed mass until saturation is reached at approximately
15 kg (or 7.5 kg m−2 ) of seaweed (Fig. 2b). For densities of
seaweeds lower than 7.5 kg m−2 , the relation between seaweed and 3–3.4 nm particle is linear, with a concentration
of 64 300 particles cm−3 resulting from each kg of seaweed
present in the chamber. Similarly, concentrations of 24 ppt of
I2 per kg of seaweed biomass was observed. After 7.5 kg m−2
of seaweed, a limiting factor is preventing either I2 or the
nucleating species from being produced in the chamber. Possibly either the ozone-induced release of volatile precursor
compounds from the seaweed (e.g. I2 ) or the formation of
Nano-SMPS BIOFLUX chamber study: Fucus 03092011
50
10
3
263.54
10
Particle concentration/cm3
(10 nm < Dp < 266 nm) complemented by the particles concentration measured with the last 6 stages of the ELPI
(266 nm < Dp < 10 µm). Calculation of the condensational
sink was performed according to:[12]
263.545
263.55
263.555
100
263.56
1000
263.565
263.57
10 000
263.575
100 000
108
106
104
102
263.54
Diameter (nm)
(b)
263.545
263.55
263.555 263.56
Number of day
263.565
263.57
263.575
Nano-SMPS seaweeds chamber experiment 2004 06 25
50
10
26 kg
16 kg
9 kg
5 kg
3
177.6
Particle concentration (cm3)
10
177.61
100
177.62
1000
177.63
177.64
10 000
177.65
177.66
100 000
177.67
1e006
108
107
106
105
104
177.6
177.61
177.62
177.63 177.64
Number of day
177.65
177.66
177.67
Fig. 1. (a) Upper panel: size distribution of particles formed in the
simulation chamber when the lid is slightly opened but no flushing of
filtered air is applied to the chamber: a wave-like pattern is observed.
Lower panel: total particle concentration v. time for the same experiment. (b) Size distribution as a function of time (upper panel) and total
particle concentration as a function of time (lower panel) for different seaweed contents in the chamber when a flushing filter air flow
of 800 L min−1 is applied: a steady state is reached for each seaweeds
content.
262
RESEARCH FRONT
1.2 107
(a)
1.0
3–3.4 nm particles
Total number particles
0.8
0.6
R 2 0.975
a 70 900 80 000
b 2770 315
0.4
0.2
0.0
100
200
400
300
Particle conc. normalized to the 3–3.4 nm conc.
(a)
Particle concentration (number cm3)
Quantification of Coastal New Ultra-Fine Particles Formation from In situ and Chamber Measurements during the BIOFLUX Campaign
1.0
Measurements
Exp. fitting
0.8
0.6
0.4
0.2
0.0
20
10
500
(b)
12 106
Particle concentration (number cm3),
I2*104 (ppt)
(b)
3–3.4 nm particle concentration
Total particle concentration
I2
10
8
6
4
Particle conc. normalized to Ct0
I2 (ppt)
40
30
Dp (nm)
50
1.0
0.8
0.6
0.4
0.2
0
2
500
1000
1500
2000
Calculated residence time (s)
0
0
5
10
15
20
25
(c)
30
14
Seaweed mass (kg)
12
Dp (nm)
Fig. 2. (a) Correlation between total and 3–3.4 nm particle number and
I2 observed during chamber experiments when seaweeds mass is varied
and a constant flowrate of 800 L min−1 applied, R2 is the correlation
coefficient for the linear regression of the type y = ax + b, where a and
b are also given, calculated for the 3–3.4 nm particle number conc. v.
I2 . (b) Correlation between total and 3–3.4 nm particle number, I2 and
seaweed content under the same conditions.
10
8
6
4
2
0
0
condensable iodine vapours by subsequent gas phase reactions of iodine species with ozone is already limited by the
ozone concentration at the higher seaweed masses. Also, it is
possible that after a certain level of seaweed the flushing air
is not penetrating into the seaweed mass with the same efficiency (the flow is going over the surface of seaweed but not
through it). From these results, we suggest that ozone induce
the emission of I2 from seaweed (e.g. either directly by the
oxidation of iodide at the plant–air interface or indirectly
involving H2 O2 chemistry) as a first step, than I2 is photolysed, forming I followed by the reaction of I with ozone to
iodine oxides that condense.
On average, we picked up between 1 and 5 kg of seaweed per m2 in the field, hence, the most representative of
real atmospheric conditions in the chamber are for a seaweed content of 5 kg (the chamber has a 2 m2 floor area).
The growth rate of new particles can be calculated from
the chamber experiment for this most representative seaweed
density (2.5 kg m−2 ) typical of the area. The size distribution
of particles measured in the chamber show an exponential
decrease with increasing Dp (Fig. 3a), illustrating the fact
100
200
500
400
300
Calculated residence time (s)
600
700
Fig. 3. (a) Particle size distribution obtained in the chamber filled with
a field representative amount of seaweed and flushed with 800 L min−1 .
The particle concentrations have been normalized to the 3–3.4 nm
particle concentration in order to illustrate their decrease by number.
(b) Decrease in concentration in a Continuously Stirred Tank Reactor
(CSTR) that would be filled with particles that would be present at a
given concentration Ct=0 at t = 0 according to their residence time (c).
From (a) and (c) calculated diameter of particles for a given residence
time.
that the bigger the particle, the more time it spent in the
chamber, and the more chances it had to be flushed hence
the less concentrated it is. The exponential distribution is in
fact in agreement with the theoretical residence time distribution of a continuously stirred tank reactor (CSTR). For
such a reactor the residence time distribution of particles
of a given concentration Ct=0 , given the 2 m3 volume and
800 L min−1 flow rate would be as shown Fig. 3b. By fitting
Fig. 3a, and Fig. 3b, we obtain Fig. 3c, i.e. the residence time
corresponding to each particle size bin. Larger particles have
spent more time in the chamber, the relationship between
263
RESEARCH FRONT
K. Sellegri et al.
Fig. 4. Map of the sampling area showing the Mace Head atmospheric research station and
its surroundings, including the higher seaweed density field measurement site of Mweenish.
Wind sectors defined as the 25ile and 75ile of the wind direction over the two hours around
low tide are shown for three study cases (27, 28 and 29 September) are also shown, indicating
westerly winds and the seaweeds field crossed.
their size and residence time is their growth rate. The growth
rate is observed to be constant with size and is calculated to
be of 1.2 nm min−1 .
This implies that 3 nm particles have only spent 20 s in
the chamber before they reach the next size bin (they are
sampled during the size bin width of 0.4 nm). Concentrations
of 3–3.4 nm particles are 500 000 cm−3 for 5 kg (2.5 kg m−2 )
of seaweeds, which corresponds to a flux of new particle
formation of 2.5 × 1010 m−2 s−1 .
Consequently, for the range of 1–5 kg m−2 seaweed density, we expect the production areas to release between
1 × 1010 and 5 × 1010 particles m−2 s−1 . It is important to
note that these experiments have been performed under
closely realistic atmospheric conditions of typical light levels
and ambient O3 (34 ppb) concentrations. The main deviation
from true atmospheric conditions was the 50% reduction in
light intensity (which does not seem to significantly influence the system) and the removal of background aerosol.
These close-to-real atmospheric conditions were required in
order to quantify realistic I2 and new particle production from
seaweeds.
If the particle production events observed regularly over
these coastal areas are driven by seaweeds activity during low
tide, we should be able to apply the chamber-derived source
rates and growth rates to all regions around the coast. In the
following sections, we evaluate whether this is feasible to
an acceptable degree of accuracy through comparison with
growth rates of newly formed particles measured at the nearby
strong source region Mweenish.
New Particle Growth Rate from Mweenish
The Mace Head atmospheric station is located about 100 m
from the tidal zone and surrounded by rocky areas where
seaweeds are relatively less abundant compared with spots
nearby (Fig. 4). As a result, it is believed that areas of higher
seaweed density nearby, such as the Mweenish area, have
greater potential in terms of particle formation. Mweenish
is located to the south east of Mace Head and is characterised by multiple source regions in all wind sectors except
for the northerly sector. In particular, prevailing westerly
winds advect air over multiple tidal emission regions. This
is illustrated on Fig. 4 by the wind sectors calculated over the
minimum tide height 2-hour period for three case study days
(27, 28 and 29 September).The Helsinki Polytechnic mobile
lab was mobilised to sample particle size distributions from
Mweenish.
264
RESEARCH FRONT
Quantification of Coastal New Ultra-Fine Particles Formation from In situ and Chamber Measurements during the BIOFLUX Campaign
Nano-SMPS BIOFLUX 03092700
200 103
27/09/2003
150
10
3
270
270.1
270.3
270.2
270.4
270.5
270.6
270.7
270.8
270.9
dN/dlogDp
Diameter (nm)
50
271
100
10:00–10:30
10:30–11:00
11:00–11:30
11:30–12:00
50
100
Diameter (nm)
10 000
100 000
dN/dlogDp (cm3)
106
0
4
105
6
7 8 9
2
10
3
4
5
Dp (nm)
103
102
270
28/09/2003
150 103
270.1
270.4 270.5 270.6
Number of day
270.3
270.2
270.7
270.8
270.9
271
Nano-SMPS BIOFLUX 03092800
100
11:00–11:30
11:30–12:00
12:00–12:30
12:30–13:00
50
10
0
3
271
4
271.1
271.2
10
271.3
271.4
100
271.5
271.6
271.7
1000
271.8
10 000
271.9
5
6
7 8 9
2
10
272
3
4
5
Dp (nm)
100 000
400 103
dN/dlogDp (cm3)
106
29/09/2003
300
dN/dlogDp
104
102
200
12:00–12:30
12:30–13:00
13:00–13:30
13:30–14:00
100
100
271
271.1
271.2
271.3
271.4
271.5
271.6
271.7
271.8
271.9
272
Number of day
0
Nano-SMPS BIOFLUX 03092900
50
Diameter (nm)
5
104
50
Particle concentration (cm3)
1000
dN/dlogDp
Particle concentration (cm3)
10
4
5
6
7 8 9
2
10
3
4
5
Dp (nm)
Fig. 6. Particle size distributions during 27, 28 and 29 September
calculated over 30 min, for the two hours around low tide.
10
3
272
272.2
Particle concentration (cm3)
10
272.4
272.6
272.8
273.2
100 000
10 000
1000
100
273
produced at various distances covering the source region.
During a particle formation event, particle number concentrations of all sizes increase with the surface of seaweed exposed
to the atmosphere, and reach a maximum at low tide (Fig. 6).
With the exception of 27 September, we observe a ‘characteristic’ size distribution representative of the tidal area and
wind direction, with only the intensity of this size distribution
evolving with tidal height. On 27 September, an increase in
the mean size is observed from 1130 hours to 1200 hours,
which would correspond to a slight change of wind direction
from NW to N, hence bringing to Mweenish larger particles
from further away. No growth of the newly formed particles
is observed when the wind speed and direction is stable and
is characteristic of being a fixed distance from a point, or
multiple point sources.
At maximum particle concentrations, we can suppose that
the largest seaweed area is exposed, which is confirmed by a
synchronization with the minimum tidal height. During the
low tide period, the largest particles measured at the sampling
dN/dlogDp (cm3)
106
104
102
100
272
272.2
272.4
272.6
Number of day
272.8
273
273.2
Fig. 5. New particle formation events during the three successive
study case days (27, 28 and 29 September) in the high-density seaweed
area of Mweenish. Particle concentration v. size distribution v. time.
Nucleation bursts were detected at Mweenish and are presented for three consecutive days (Fig. 5). 30 min averages
of the particle size distribution over 2 hours around low tide
(Fig. 6) show that, with constant wind direction and speed, the
particle mode is stable at diameters between 5 and 6 nm. Particles from 3 to 10 nm detected at the sampling site have been
265
RESEARCH FRONT
K. Sellegri et al.
53.57N
site are originating from the farthest spot of air-exposed seaweed. From the time needed to advect over this distance,
and noting the largest particle size observed, we can infer
a growth rate. Because on 27 and 29 September, more distant sources can contribute to the size distribution observed,
we chose to focus 28 September as a case study: the wind
direction is purely west and no other source than the immediate coastline could contribute to the new ultra-fine particles
observed. On 28 September, the largest diameter for which
particle concentrations were significantly increased during
nucleation compared to background level is 14.2 nm (Fig. 6).
The 25th percentile and 75th percentile of the wind speed are
calculated to be 6.2–7.2 m s−1 , respectively, for the period
corresponding to the maximum seaweed exposure around low
tide. From Fig. 4, 14.2 nm particles have traveled between
5 and 6 km before reaching the sampling site, which took
between 12 and 16 min. Hence, the growth rate is estimated
to be between 0.8 and 1.2 nm min−1 , which corresponds to
the lower side of what has been measured in the chamber for
a 2.5 kg m−2 of seaweeds density (1.2 nm min−1 ). However,
the path length of the particles up wind of the sampling site
is not evenly and fully represented by seaweed fields of 1
to 5 kg m−2 density (Fig. 4). If we estimate that in this area,
about half of the path way is covered by seaweed fields, this
growth rate would correspond to a mean 1.25 kg m−2 , to compare with the 2.5 kg m−2 in the chamber. Hence, the growth
rate of particles produced in the chamber is consistent (within
a factor two underestimated) with the growth rate of particles
that would have been produced by the seaweed fields around
Mweenish and transported there.
0.25E10
y (km)
0.24E10
0.22E10
8
0.21E10
0.20E10
6
0.18E10
0.16E10
4
0.15E10
2
0.12E10
0.10E10
0
0.90E9
0.13E10
0.75E9
0.60E9
2
0.45E9
0.30E9
4
0.15E9
53.28N
4
10.3W
2
0
2
4
0.00E00
6
8
x (km)
9.48W
Fig. 7. Local emission inventory, in number of 3–3.4 nm particles
m−2 s−1 , calculated from the new particle flux measured in the chamber for a 5 kg m−2 seaweeds density, and the spatial repartition of the
intertidal zones on the western side of the measurement area (from
Fig. 4), used as source input in Regional Atmospheric Modelling System
for the 28 September simulation. Mace Head is at coordinates 0,0.
which the emissions are initially confined.[18] The meteorological fields have been initialized and nudged every 6 hours
by ECMWF (European Center for Medium-Range Weather
Forecasts) database. The simulation starts on 27 September
2003 at 00:00 UTC and ends 28 September 2003 at 13:00
UTC. The observed and simulated values of the speed and
the direction of wind, the temperature and the humidity are in
good agreement. An aerosol inventory (source map) is introduced as an input in RAMS (Fig. 7) and the emission at each
time step is activated. This source map was built by calculating the source rate of each km−2 from (1) the flux of particles
m−2 of seaweed measured in the simulation chamber for
2.5 kg of seaweed per m2 and (2) the percentage of seaweed
field in the km−2 inferred from a detailed map of the area
(Fig. 4). Only sources situated west of the sampling site were
considered, in agreement with the wind direction. From 12:00
UTC and 12:15 UTC, emissions of new particles are constant:
tide level oscillates around the maximum low tide. Intramodal coagulation was introduced with a coagulation sink
coefficient 10−4 s−1 .[19] The modeled particle concentrations
are illustrated in Fig. 8 for 12:00 UTC from which it was
calculated that the predicted total particulate concentration at
Mweenish is 0.9 × 104 cm−3 . This is about five times lower
than the observed total concentration of 5.5 × 104 cm−3 at
12:00 UTC. Considering the uncertainty in the seaweeds field
spatial density, the agreement is quite reasonable, although
the chamber fluxes might be underestimated (as the growth
rate was). Lower emission rates in the chamber could be
explained by a surface of exposure to total mass of seaweeds,
which is not representative of the ambient conditions when
seaweed contents are higher than 2.5 kg m−2 (the fresh air
flow is flushed over the seaweed and not through). Also,
the geometry of the chamber (wall effects) possibly slightly
inhibits new particle formation and growth.
New Particle Emissions from the Coastal Area
around Mweenish
In this section we test if the new particle formation rate measured in the chamber is also realistic in comparison with the
ambient observations at Mweenish. A mesoscale transport
model (RAMS) was used to simulate particle concentration
that would be found in Mweenish, based on the new particle
fluxes measured in the chamber and the seaweed field around
the area.
The RAMS (http://www.atmet.com/, verified November
2005)[13] model is a parallel mesoscale model allowing the
simulation of meteorological fields with horizontal scale
spanning from one km to a thousand km. It includes nested
grids. Many investigations on regional pollution with an
online coupling of emission codes which can be associated to chemical mechanism, were previously made using
RAMS model.[14–17] For this study, simulations have been
performed using three nested grids simultaneously to take
the synoptic and local circulation into account. Grid1 covers the region with a horizontal resolution of 25 km. Grid2,
the intermediate domain, has a resolution of 5 km, and Grid3
consists of the experimental domain with a resolution of 1 km
(17 meshes). We used a time step of 10 s and 35 levels in the
vertical dimension (the same in the three grids) with 15 levels
from surface to 1500 m, which ensures a fine description
of the boundary layer. However, the boundary layer contains an initial surface mixing layer of 100 to 300 m, into
266
RESEARCH FRONT
Quantification of Coastal New Ultra-Fine Particles Formation from In situ and Chamber Measurements during the BIOFLUX Campaign
0.5 nm min−1 (if about half the distance is on average covered by seaweed fields). The zero-distance particles (within
the grid sector where Mweenish is situated) are considered to
be 3–4.4 nm sized particles. Modeled and measured size distributions are shown Fig. 10 where quite close agreement is
seen using the chamber fluxes and growth rates applied to the
observed seaweed biomass concentration in the area, if higher
seaweed densities are used for the simulations (5 kg m−2 and
7.5 kg m−2 instead of the usual 2.5 kg m−2 ). A better agreement is achieved when the intertidal area density is higher
(7.5 kg m−2 ). The seaweed fields density being typically
lower than 7.5 kg m−2 , this confirms that the fluxes of 3–
3.4 nm particles derived from our chamber experiment might
be slightly underestimated. However, the shape of the new
particle size distribution could be relatively well simulated
based on the hypothesis that all particles were produced by
the seaweed field.
To summarize, growth rates, new particle formation
rates and new particle size distributions measured in situ
at Mweenish, although slightly higher, are consistent with
(1) the growth rate and new particle formation rate simulated
in the chamber with seaweeds, and (2) the spatial repartition of the seaweed fields relative to the sampling site. In an
attempt to evaluate the role of these high-density seaweed
fields in the coastal particle productions, we used a mobile
van that allowed us to detect new particle production on a
lager spatial scale.
The vertical extension of the new particle plume can also
be simulated with RAMS. Figs 9a and 9b illustrate the vertical dispersion of the total particle concentrations for the two
transects shown on Fig. 8. The vertical extension of the new
particle plume is enhanced when the coastline rises above
the sea level (section B), but, in general, the plume rapidly
reaches 300 m (in less than 4 km) and 600 m after 10 km.
These simulations are in agreement with the LIDAR observations performed during PARFORCE (PARticle FORmation
and fate in the Coastal Environment, 1998–1999, Mace
Head).[18]
From the growth rate and traveling speed, each size bin
of particle can be spatially associated to a source distance.
Hence the size of the particles at Mweenish can be simulated by keeping in memory the distance at which they have
been produced. This can be done by entering in the source
map a different tracer for each grid sector, and calculating
their growth during transport using the growth rates of
53.57N
y (km)
8
6
4
2
0
2
4
53.28N
4
2
0
2
8
6
4
10.3W
0.37E5
0.36E5
0.33E5
0.32E5
0.30E5
0.29E5
0.27E5
0.25E5
0.22E5
0.20E5
0.18E5
0.16E5
0.15E5
0.13E5
0.11E5
0.90E4
0.75E4
0.60E4
0.37E4
0.18E4
0.00E00
x (km) part/m3
Section B
Spatial Extension of New Ultra-Fine Particle Production
in the Coastal Area
Section A
The hypothesis of new particle formation from high productivity spots around Mace Head was mentioned during
QUEST (Quantification of Aerosol Nucleation in the European Boundary Layer) airborne measurements.[20] In April
2002, airborne particle concentration measurements using
condensation particle counters show that nano-particles were
9.48W
Fig. 8. Spatial repartition of particle total concentrations (cm−3 )
simulated using Regional Atmospheric Modelling System at ground
level.
0.37E5
0.36E5
0.33E5
0.32E5
0.30E5
0.29E5
0.27E5
0.25E5
0.22E5
0.20E5
0.18E5
0.16E5
0.15E5
0.13E5
0.11E5
0.90E4
0.75E4
0.60E4
0.37E4
0.18E4
0.00E00
z (km)
1400
1200
1000
800
600
400
200
0
4
2
0
2
4
6
8
0.37E5
0.36E5
0.33E5
0.32E5
0.30E5
0.29E5
0.27E5
0.25E5
0.22E5
0.20E5
0.18E5
0.16E5
0.15E5
0.13E5
0.11E5
0.90E4
0.75E4
0.60E4
0.37E4
0.18E4
0.00E00
z (km)
1600
x (km)
1400
1200
1000
800
600
400
200
0
4
2
0
2
4
6
8
x (km)
Vertical cross-section (B) at y 2 km
28/09/12HUT
Vertical cross-section (A) at y 2 km
28/09/12HUT
Fig. 9. Vertical spatial distribution of the total new particle concentration (cm−3 ) for the two sections shown on Fig. 8, calculated with Regional
Atmospheric Modelling System.
267
RESEARCH FRONT
K. Sellegri et al.
formed over Mweenish and Roundstone area (North of Mace
Head), Co Galway, whereas the Mace Head station simultaneously showed no particle formation. Fig. 11 is showing the
respective I2 concentrations at two different locations around
Mace Head, which have different seaweed surroundings (with
north-easterly winds, the bridge location is downwind of a
larger seaweed field than MRI-Mweenish). These results, also
compared with the chamber I2 concentrations, indicate that
I2 concentrations are a function of the seaweed field density.
Considering the I2 -to-new particle concentrations pointed out
in this work, they would confirm the QUEST observations
that Mace Head is experiencing less intense new particle
formation events than its surroundings during the prevailing westerly winds. Simultaneous measurements of ultra-fine
particles in Mace Head and Mweenish have been performed
during BIOFLUX, which can help us address this matter. Previously, a classification of the new particle formation events
at Mace Head according to 4 wind sectors was performed.[21]
Type I events are locally produced particles observed during clean marine wind sector; type II events show particles
produced locally and from more distant sources transported
by northerly winds; during type III events, distant sources
produce new particles in more polluted airs masses coming
from the south and south-eastern sector, and finally no event is
observed for type IV, when the wind sector is north-east. Comparison of the particle formation intensity observed in Mace
Head with its intensity in Mweenish was performed over a
number of days during the BIOFLUX campaign (on 14, 25,
26, 27 and 30 September and 1 October 2003). Table 1 shows
for the different wind sectors the comparison of particle
<10 nm number concentration for both sites under different
environmental conditions (low tide occurrence, wind speed
and direction and condensational sink).
Particle formation is more intense in Mweenish compared
to Mace Head when the wind sector is of type I (western
marine air masses), which is consistent with the lack of
high-density seaweed areas west of Mace Head. Particle
formation is more intense in Mace Head compared to
Mweenish when the wind sector is of types III and IV
(south–south-east): there is an immediate production zone
Mweenish t11h45
Mweenish t12h00
Mweenish t12h15
150 103
Mweenish t12h30
dN/dlogDp
Predicted from chapter (7.5 kg m2)
Predicted from chapter (5 kg m2)
100
50
0
4
5
6
7
8
9 10
2
3
4
5
Dp (nm)
Fig. 10. Particle size distribution calculated from the Regional Atmospheric Modelling System simulation of each size bin of particles corresponding to a distance from Mweenish to the source map shown
in Fig. 7, and a growth rate of 0.5 nm min−1 , and comparison with the measured size distributions at
Mweenish around 12 UTC.
Chamber
Fucus
900
Samples from seaweed
chamber
800
I2 Mixing ratios (ppt)
700
Chamber
Fucus
846
740
Chamber
Laminaria
600
519
500
Samples at Mweenish
Bridge
400
300
200
100
Samples at Mweenish
(low tide)
30/09
14:30⬃15:00
01/10
15:00⬃15:30
35
34
30/09
16:30⬃17:30
125
121
0
Fig. 11. Comparison of I2 mixing ratios for two different locations around Mace Head and in the seaweed
chamber for two types of seaweed (Laminaria and Fucus).
268
RESEARCH FRONT
Quantification of Coastal New Ultra-Fine Particles Formation from In situ and Chamber Measurements during the BIOFLUX Campaign
Table 1. Comparison of the new ultra-fine particle formation intensity between the Mace Head atmospheric research station and the
high-density seaweed area Mweenish for various environmental conditions
Particle concentrations are integrated (average 1) over the whole period indicated in the ‘Time’ column. Particle mean concentrations (average 2)
are also calculated over the same time period as the I2 sampling period (around low tide during 15 to 30 min)
Date
I
14 September
11:40–15:00
310
II
III
25 September
27 September
27 September
14 September
25 September
30 September
10:45–12:00
10:20–13:30
09:20–10:20
10:30–11:30
08:30–10:45
13:00–15:00
300
71 700
43 350
4250
1575
155 700
600A
69 650
2300
60
500
32 700
1 October
12:10–14:30
450 000
13 800
IV
A At
Time (UTC)
ConcPart<10 nm (number per cm3 )
Event
type
Mace Head
Mweenish
Average 1
Average 2
1950
1100
487
382
50 000
–
–
–
35 000
85 000
213 000
Wind direct (◦ )/
speed (m s−1 )
I2 (ppt)
Mweenish
CS
Low tide
(UTC)
30
24
96
28
–
–
–
12
36
34
0.003–0.005
13:00
205/6.5
0.0003
0.0003
0.003–0.005
0.006
0.002
10:00
11:30
12:00
11:30
10:00
13:30
197/10
270/6
320/3
192/8.2
170/10
50/5.5
0.006
14:00
40/6
Mweenish but also 20 km down wind (Glinsk).
south-east of Mace Head. Intense new particle formation
(up to 1 million particles cm−3 ) occurs in Mace Head, even
though the air masses originating from sectors III and IV
are usually polluted (condensation sink ranging from 0.002
to 0.006 s−1 ). When the wind sector is north (type II), it
happened that both Mace Head and Mweenish experienced
strong particle formation events (on 26 and 27 September).
These observations show that particle formation events are
consistent with the location of seaweed fields influenced
by wind direction during sampling. They corroborate airborne measurements showing that particle production was
more intense in Mweenish than in Mace Head during clean
marine western winds,[2] which are the most frequent wind
directions observed in the area. New particle formation can
be detected irrespective of wind direction and environmental condition (polluted on 25 September and 1 October, and
cloudy/rainy on 14 September), provided that sampling is
performed downwind of a seaweed field during low tide.
Also shown on Table 1 are the ambient concentrations of
I2 measured over 15 to 30 min at Mweenish, sampled in the
same place as the corresponding 3–3.4 nm concentrations.
Concentrations of I2 varied from 20 to 100 ppt during low
tide, in agreement with the concentrations measured in the
simulation chamber from the seaweeds representative density
of 2.5 kg m−2 (75 ppt). Denuders were placed less than 2 m
from the seaweed field where the dispersion and dilution of
I2 would be low and comparable with that seen in the chamber
measurements.The relationship between I2 and ultra-fine particle concentrations is not so clear. For the cases presented, the
highest concentration of I2 (100 ppt) corresponds the lowest
ultra-fine particle concentration (400 cm−3 ), and the highest particle concentrations (213 000 cm−3 ) were sampled,
whereas I2 was at an average level (34 ppt) for a nucleation
event. In fact, it is not a surprise not to be able to relate
the local I2 concentration with the total concentration of
particles smaller than 10 nm, since we have shown earlier
that the whole history of seaweed field repartition on the
wind trajectory should be taken into account. This suggests
that the dynamics of air masses, wind direction and seaweed
repartition are major parameters necessary to predict a new
particles concentration at a given location.
Summary and Conclusions
The goals of this paper were to investigate the relationship
between I2 and new particle formation and to provide the
first quantitative estimate of the flux of coastal new particles
as a function of exposed seaweed biomass.
In a 2 m3 simulation chamber filled with various amounts
of seaweeds, we showed that under near-real atmospheric
conditions (natural UV light, ambient O3 level and particlefree flushing un-scrubbed air), new particle formation was
occurring at high rates (up to 1 million particles cm−3 ), in proportion to high I2 concentrations (up to 400 ppt) with a ratio of
2800 particles cm−3 ppt−1
(I2 ) . I2 is produced proportionally to
the amount of seaweeds, up to a maximum seaweeds level
of 7.5 kg m−2 . Above this level of seaweeds, I2 and subsequent new particles formed start to be produced with a
lower efficiency, presumably due to a lack of O3 reaching
the seaweed surface or available for oxidation. At an ‘in situ
representative’ seaweed content, new particles were formed
at a rate of 2.5 × 1010 m−2 s−1 , whereas the CS was about
0.015 s−1 . In fact, for seaweeds densities in the range of what
is usually found in the area, new particle concentrations were
found to be linearly proportional to the amount of seaweeds,
which provides a relatively simple parameterization of the
new particle formation in the coastal area of western Ireland.
This parameterization could be tested against in situ measurements using a mobile laboratory around Mace Head. The
seaweeds mass-to-new particle flux relationship measured in
the chamber was applied to the intertidal zone map of the
area, fed as a source map in a mesoscale model (RAMS), and
compared with real atmospheric particle size distributions.
Simulations were in agreement with the measurements, in
terms of total particle concentrations, within a factor 5. In
the non-spatially-homogeneous environment of costal areas,
growth rates could not be calculated on the real atmospheric
measurements with the usual method of tracking the temporal
269
RESEARCH FRONT
K. Sellegri et al.
evolution of the nucleation mode from 3 to 25 nm. Instead,
ambient growth rates were calculated from the largest new
particle detected and the farthest new particle intertidal zone,
under stable wind speed and direction conditions. This average growth rate of 0.8–1.2 nm min−1 about twice higher than
the growth rate measured in the simulation chamber if the seaweed field was composed of 2.5 kg seaweed m−2 over about
half of the total area (0.5 nm min−1 ). When a 0.5 nm min−1
growth rate is introduced in the meso-scale simulation of
particle growth over the coastal area, we obtain simulated
size distributions in fairly good agreement with the observed
size distributions.
New particle formation depends on the balance between
the source rates and sinks of condensable precursor gases.
However, in these chamber experiments, we found a direct
link between the amount of seaweeds and new particle production that would integrate both parameters into a simple
relationship. This relationship has been found to be realistic regarding the amount and size of new particles formed at
Mace Head and its surroundings. In fact, in this area, because
new particle formation occurs with high intensities also in
polluted air masses (types III and IV), we suggest that the
condensable iodo-compound sources are so important that
they are the one parameter determining the intensity of new
particle formation.
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
Acknowledgements
[14]
This work was supported by Irish Research Council for Science and Engineering Technology (IRCSET) and European
Commission under the QUEST contract. For the RAMS modeling, computer resources were provided by CINES (Centre
Informatique National de l’Enseignement Supérieur) project
amp2107. The authors also wish to thank the computer team
of the laboratoire de Météorologie Physique de l’Université
Blaise Pascal (France); A.M. Lanquette, S. Banson and Ph.
Cacault.
[15]
[16]
[17]
[18]
References
[19]
[1] C. D. O’Dowd, J. L. Jimenez, R. Bahreini, R. C. Flagan,
J. H. Seinfeld, K. Hameri, L. Pirjola, M. Kulmala,
S. G. Jennings, T. Hoffmann, Nature 2002, 417, 632.
doi:10.1038/NATURE00775
[2] C. D. O’Dowd, K. Hämeri, K. M. Mäkelä, L. Pirjola, M. Kulmala,
S. G. Jennings, H. Berresheim, H.-C. Hansson, G. de Leeuw,
G. J. Kunz, A. G. Allen, C. N. Hewitt, A. Jackson, Y. Viisanen,
T. Hoffmann, J. Geophys. Res. 2002, 107, 8108.
[3] G. McFiggans, H. Coe, R. Burgess, J. Allan, M. Cubison,
M. R. Alfarra, R. Saunders, A. Saiz-Lopez, J. M. C. Plane,
[20]
[21]
270
D. J. Wevill, L. J. Carpenter, A. R. Rickard, P. S. Monks, Atmos.
Chem. Phys. 2004, 4, 701.
L. J. Carpenter, G. Malin, P. S. Liss, F. C. Küpper, Glob. Biogeochem. Cycles 2000, 14, 1191. doi:10.1029/2000GB001257
J. M. Mäkelä, T. Hoffmann, C. Holzke, M. Väkevä, T. Suni,
T. Mattila, P. Aalto, U. Tapper, E. I. Kauppinen, C. D. O’Dowd,
J. Geophys. Res. 2002, 107, 8110. doi:10.1029/2001JD000580
J. L. Jimenez, R. Bahreini, D. R. Cocker III, H. Zhuang,
V. Varutbangkul, R. C. Flagan, J. H. Seinfeld, C. D. O’Dowd,
T. Hoffmann, J. Geophys. Res. 2003, 108, 4318. doi:10.1029/
2002JD002452
A. Saiz-Lopez, J. M. C. Plane, Geophys. Res. Lett. 2004, 31,
L04112. doi:10.1029/2003GL019215
S. G. Jennings, C. Kleefeld, C. D. O’Dowd, C. Junker, T. G. Spain,
P. O’Brien, A. F. Roddy, T. C. O’Connor, Boreal Environ. Res.
2003, 8, 303.
A. F., Holleman, E. Wiberg, in Inorganic Chemistry (Ed.
N. Wiberg) 2001, pp. 402–469 (Academic Press: Berlin).
V. T. Calabrese, A. Khan, J. Polym. Sci. A 1999, 37,
2711–2717.
doi:10.1002/(SICI)1099-0518(19990801)37:15
<2711::AID-POLA4>3.0.CO;2-6
L. Pirjola, H. Parviainen, T. Hussein, A. Valli, K. Hämeri,
P. Aalto, A. Virtanen, J. Keskinen, T. Pakkanen, T. Mäkelä,
R. Hillamo, Atmos. Environ. 2004, 38, 3625. doi:10.1016/
J.ATMOSENV.2004.03.047
M. Kulmala, M. Dal Maso, J. M. Mäkelä, L. Pirjola, M. Väkevä,
P. Aalto, P. Miikku-lainen, K. Hämeri, C. D. O’Dowd, Tellus
2001, 53B, 479.
W. R. Cotton, R. A. Pielke, R. L. Walko, G. E. Liston,
C. J. Tremback, H. Jiang, R. L. Mcanelly, J. Y. Harrington,
M. E. Nichols, G. G. Carrio, Meteorol. Atmos. Phys. 2003, 82,
5. doi:10.1007/S00703-001-0584-9
S. Cautenet, D. Poulet, C. Delon, R. A. Delmas, J. M. Gregoire,
J. M. Pereira, S. Cherchali, O. Amram, G. Flouzat, J. Geophys.
Res. 1999, 104, 30 641. doi:10.1029/1999JD900393
M. Taghavi, S. Cautenet, G. Foret, Atmos. Chem. Phys. 2004,
4, 825.
M. Taghavi, S. Cautenet, J. Arteta, Atmos. Res. 2005, 74, 65
(Special issue ESCOMPTE). doi:10.1016/J.ATMOSRES.2004.
06.007
F. Minvielle, G. Cautenet, M. O. Andreae, F. Lasserre, G. Foret,
S. Cautenet, J.-F. Léon, O. L. Mayol-Bracero, R. Gabriel,
P. Chazette, R. Roca, Atmos. Environ. 2004, 38, 1811.
doi:10.1016/J.ATMOSENV.2003.07.017
G. Kunz, G. de Leeuw, E. Becker, C. D. O’Dowd, J. Geophys.
Res. 2002, 107, 8106. doi:10.1029/2001JD001240
M. Dal Maso, M. Kulmala, K. E. J. Lehtinen, J. M. Mäkelä,
P. Aalto, C. D. O’Dowd, J. Geophys. Res. 2002, 107, 8097.
doi:10.1029/2001JD001053
C. D. O’Dowd, ICNAA 2004 (International Conference on
Nucleation and Atmospheric Aerosols, Kyoto).
C. D. O’Dowd, K. Hämeri, K. M. Mäkelä, M. Väkeva, P. Aalto,
G. de Leeuw, G. J. Kunz, E. Becker, H.-C. Hansson, A. G. Allen,
R. M. Harrison, H. Berresheim, C. Kleefeld, M. Geever,
S. G. Jennings, M. Kulmala, J. Geophys. Res. 2002, 107,
8107. doi:10.1029/2000JD000206