Introduction
Sulphur dioxide is an important tropospheric species, influencing air quality
as well as the acidity of precipitation (and therefore that of soil, lakes
and rivers). It also influences climate directly and indirectly through its
oxidation to sulphate and subsequent role in atmospheric new particle
formation e.g., and the growth of
aerosol particles e.g. and thus its effect on their
cloud condensation nuclei (CCN) properties e.g..
Global anthropogenic emissions of SO2 around the year 1990 were
estimated to be approximately 73 Tg S yr-1
, more than twice the total sulphur emissions from natural
sources. Similar values were given by in a review of
modelled SO2 emissions. In heavily industrialised regions, the
ratio of anthropogenic to natural emissions can be higher than 10. Although
air quality legislation in Europe and the USA has led to a significant
decline in industrial emissions of SO2 in the last couple of
decades, emissions from Asia and developing countries in other locations are
increasing .
The major sink of atmospheric SO2 is oxidation to sulphate,
estimated at approximately 51 Tg S yr-1
, and anthropogenic emissions of SO2 account for
approximately 72 % of sulphate aerosol in the atmosphere
.
SO2 is either oxidised in the gas phase (by reaction with the OH
radical), or it can be taken up by cloud droplets and undergo aqueous phase
oxidation. In the aqueous phase, oxidation is primarily by reaction with
H2O2 or O3, with the O3 reaction becoming
larger than typical gas phase reaction rates at pH higher than approximately
4 . Model studies suggest that aqueous oxidation
comprises a large majority of the global sulphate production, perhaps
80 % or more , however the range of model
estimates is rather large, reflecting the difficulty in reproducing cloud
processes in large-scale models.
The aqueous phase oxidation of SO2 by O3
proceeds by the absorption of SO2 into the cloud droplet,
and the establishment of equilibrium between SO2⋅H2O, HSO3-, and SO32-. Dissolved
O3 then oxidises the latter three species, forming
SO42-. The Henry's law coefficients for dissolution of
O3 and SO2, as well as the equilibrium constants for the
hydration of SO2 and the reaction rate constants for the subsequent reaction
with O3 have all been well studied in bulk solutions in the past
(predominantly at temperatures of approximately 20 ∘C or higher, see
Sect. ). The oxidation rate constants recommended by
are commonly adopted in models simulating cloud
chemistry for example, all of the seven models simulating SO2
oxidation in clouds, which were compared byused these
rate constants. However, to the best of our knowledge, no studies
have been performed at temperatures below 0 ∘C, thus values for
sub-zero temperatures are based on extrapolations of the temperature
dependence at higher temperatures. The temperature dependence recommended by
, and adopted in most modelling studies, is that of
, based on measurements at 25 and 16 ∘C. In that
study, experiments were only performed up to a pH of 4.02, where the vast
majority (i.e. 0.994) of dissolved S(IV) is present as HSO3-.
At higher pH, as encountered in cloud droplets, more of the S(IV) is present
as SO32-, for which the rate constant for reaction with O3 is approximately 4 orders of magnitude higher than that for HSO3-.
The temperature dependence of this reaction is therefore very important.
Measurements performed by , at 25 and 10 ∘C, using
buffered solutions at pH up to 6.2 suggest a slightly weaker temperature
dependence, and therefore greater oxidation rates at low temperatures than
suggested by .
A few previous cloud chamber studies have examined the aqueous phase
oxidation of SO2 in cloud droplets, hydrated aerosol, and fogs,
with the aim of reconciling reaction rate constants measured in bulk solutions with
those inferred from S(VI) production in a dispersed aqueous system. As
discussed below, the results are mixed, and it is not possible to exclude the
influence of un-measured contaminants on the rate of S(IV) oxidation.
The 6.7 m3, Desert Research Institute Dynamic Cloud Chamber was
described by , who performed experiments with the oxidation
of SO2 in droplets formed on a range of CCN, without the addition
of extra oxidants such as O3. The same chamber was then used
by to perform experiments with the addition of
O3. They used a Na2SO4 seed aerosol, and formed clouds
by performing adiabatic expansions, which produced temperature decreases from
approximately 21 to 10 ∘C, and cloud lifetimes of 440 to
490 s. Although the chamber could maintain a minimum wall temperature
of -40 ∘C, they only performed experiments beginning around
20 ∘C, representative of warm clouds. Similarly to the present
study, they used O3 mixing ratios of 120 to 150 ppbv, with
a wider range of SO2 mixing ratios (23 to 310 ppbv). They
found good agreement between the measured (dual channel ion chromatograph)
increase in S(VI) in the cloud water, and values calculated using the oxidation rate constants measured by
in bulk solutions, suggesting that such rate constants are indeed
applicable to the chemistry occurring in cloud droplets, at warm temperatures.
The aqueous phase oxidation of SO2 by H2O2 and by
O3 was also investigated in the Calspan 590 m3
environmental chamber by . The chamber was filled with
outside air that had been filtered through activated charcoal and aerosol
filters and experiments were performed at ambient temperature. After
humidifying the air and injecting O3 and SO2, they
achieved 4 min cloud periods by performing adiabatic expansions from
15 mb over atmospheric pressure to 5 mb below atmospheric
pressure. Aqueous phase oxidation rates determined from the resulting aerosol
growth were found to be much faster than those inferred using the bulk measurement based
rate constants
recommended by . This experiment was repeated by
, with improved instrumentation and an improved model, but
the measured oxidation rates were still found to be a factor of 5 higher than
those of . Small amounts of un-measured ammonia
contamination (increasing the droplet pH and raising the reaction rate) were
given as a possible explanation.
The CLOUD (Cosmics Leaving Outdoor Droplets) chamber at CERN provides an
essentially contaminant free, and precisely controlled environment in which
to perform experimental observations of aqueous phase reactions occurring in
cloud droplets. In this study, the aqueous phase oxidation rate of
SO2 by O3, in cloud droplets is examined. This study is
based on measurements performed during two experimental campaigns, in 2013
(CLOUD8) and 2014 (CLOUD9). Experiments were carried out at temperatures of
10 and -10 ∘C, with acidic (H2SO4) and partially to
fully neutral (ammonium sulphate) aerosol as CCN, and
a chemical-microphysical model of the chamber was written to compare commonly
accepted reaction rates with the formation of sulphate observed in the cloud
chamber droplets.
In the next section, the CLOUD chamber and associated instrumentation is
described. In Sect. , the details of the model are given, and
in Sect. we present a discussion of the results of the
experiments and a comparison with the modelled values. Conclusions are drawn
in Sect. .
Experimental setup
The experiments were conducted in the CLOUD chamber at CERN. The chamber
consists of a 3 m diameter electro-polished stainless steel cylinder,
with a volume of 26.1 m3. The temperature in the chamber can be
accurately controlled at any point between 183 and 300 K, by
regulating the temperature of air flowing between the chamber wall and its
thermal insulation housing . Experiments are performed in
a well-mixed flow chamber mode, with the sample air drawn off by the
instruments continually being replaced, and the mixing ratio of any added gas
phase species being held approximately constant. In CLOUD8, the sample flow,
and therefore the addition of gases to the chamber totalled
250 Lmin-1, leading to a dilution lifetime of 105 min.
In CLOUD9, the sample flow demands of the instruments were lower, leading to
a flow of 150 Lmin-1 and thus a dilution lifetime of
174 min. As species such as O3 and SO2 were
continually added to the chamber to maintain approximately constant mixing
ratios, the dilution lifetime only applies to the concentration of the
aerosol particles.
Gases in the chamber were mixed by two stainless steel fans, mounted at the
top and bottom of the chamber, and magnetically coupled to their gearboxes,
which are mounted outside the chamber. Between CLOUD8 and CLOUD9, the
gearboxes were upgraded to allow a fourfold increase in fan speed, but, for
the experiments presented here, the fans were set to the same speed in CLOUD8
and CLOUD9. At the beginning of a series of experiments, the chamber is
cleaned by heating to 373 K, flushing with ultra-pure water, and
drying with a pure air flow, resulting in extremely low levels of
contaminants. The pure air added to the chamber is created by the evaporation
of liquid N2 and liquid O2, at a ratio of 79 : 21. The
desired relative humidity (RH) in the chamber is achieved by passing the
necessary fraction of the inflow air through a Nafion humidifier, using water
which was purified by recirculation through a bank of Millipore Super-Q
filters and irradiated with ultraviolet radiation to suppress biological
activity (this treatment process is also used to create the water used to
clean the chamber). SO2 is added to the chamber from a gas cylinder
(998.2 ppmv ± 2 % in N2), as is NH3 for
the neutral seed experiments (1 % NH3 in N2).
O3 is created by irradiating a pure air flow at wavelengths below
200 nm, in an external O3 generator, from which it is piped
into the chamber. All gases have dedicated lines for injection into the
chamber to avoid contamination and reactions occurring in the gas lines.
Fittings and gas lines are made from stainless steel to avoid contamination.
More details of the CLOUD chamber are given in .
Expansion system
By increasing the input flow of air beyond the sample flow drawn off by the
instruments, the chamber can be pressurised up to 220 hPa above
ambient pressure. This overpressure can be released through an exhaust valve,
resulting in an adiabatic cooling, and, at sufficiently high initial RH, the
activation of aerosol particles to form cloud droplets. At lower
temperatures, ice particles may form. The pressure-release valve is computer
controlled, and can be programmed to follow a linear decrease in pressure
over a given time period, or to follow any other prescribed pressure profile,
such as an initial rapid pressure decrease until a cloud is formed, followed
by a slower pressure reduction to maintain the cloud as long as possible.
A vacuum pump is mounted in the exhaust line, to ensure that the rate of
change of pressure does not decrease with the difference between the chamber
and ambient pressure. During and after the adiabatic cooling, the air in the
chamber is continually heated by the chamber walls, as the temperature
control system is maintained at the pre-expansion temperature, causing the
evaporation of the cloud after approximately 4–6 min.
Temperature and pressure measurement
The temperature inside the chamber was
measured with a string of six thermocouples (TC, type K) which were mounted
horizontally between the chamber wall and the centre of the chamber at
distances of 100, 170, 270, 400, 650, and 950 mm from the chamber
wall. The line of thermocouples was located midway between the top and bottom
of the chamber. The TC have a precision of below 0.1 ∘C, and were
calibrated by an in situ measurement with a string of well calibrated Pt100
sensors (one at each TC position). The temperature measured before, during
and after an expansion is shown in Fig. . During normal
operation, there was no systematic horizontal gradient across the chamber.
During expansions, a small oscillation in the temperature was observed. This
is caused by the mixing of air within the chamber, with the variation between
the temperature sensors being typically lower than approximately
0.5 ∘C. In the modelling work here, the temperature was taken as the
mean of the 3 innermost TCs (TC4–TC6).
(a) the evolution of temperature measured by the
6 TC before, during and after an expansion. Pressure above atmospheric
pressure is also shown (dashed line, right hand axis). (b) shows the
deviation of the temperature measured by each individual TC from the mean of
all TC.
![]()
The pressure in the chamber was measured by a Vegabar 51 pressure transmitter
with a precision of 0.1 bar.
Trace gas and aerosol measurement configuration
The gas phase concentrations of SO2 and O3 were measured
by trace gas monitors (Enhanced Trace Level SO2 Analyser, Model
43i-TLE, Thermo Scientific and O3 Analyser, Thermo Environmental
Instruments Inc., Model 49C, respectively). A proton transfer reaction time
of flight mass spectrometer (PTR-TOF-MS) with a mass resolving power of
4000–5000 (m/Δm, FWHM) and a mass accuracy within 10 ppm
was present, however the gas phase ammonia was typically
below the detection limit during the experiments. The aerosol particle number
size distributions were simultaneously measured by three different scanning
mobility particle sizer (SMPS) systems, consisting of a differential mobility
sizer (DMA) and a condensation particle counter (CPC, TSI model 3010). The
first of these was attached to a total sampling line that allowed the
measurement of all aerosols in the chamber. The second was attached to
a cyclone, with a cut off of 2 µm, which enabled the measurement
of the interstitial (i.e. non-activated) aerosol during the cloud periods.
During the subsaturated periods the SMPS attached to the total and the
interstitial lines measure essentially the same size distributions. The third
SMPS was attached to an additional sampling line, on which a pumped
counterflow virtual impactor (PCVI) was installed with the PCVI flow rate set
for a cutoff of 5 µm, thus sampling only the aerosol contained in
cloud droplets. The three SMPS systems scanned the size range between 17 and
415 nm approximately every minute. Unfortunately, the PCVI and
cyclone appear to have been influenced by pressure changes during the
expansions. It is, therefore, not possible to clearly identify the activated
fraction of aerosol from these measurements.
The total humidity in the chamber was measured by MBW dew point mirror
instruments (model MBW973 during CLOUD9 and MBW373LX during CLOUD8) attached
to a heated sampling line. During CLOUD9 a tuneable diode laser (TDL)
hygrometer, comparable to the APicT instrument as described by
, was used to measure the water vapour content in situ with
1 Hz time resolution using a single optical path once across the
middle plane of the CLOUD chamber. From the difference between total water
and water vapour, the liquid water content (LWC) or ice water content (IWC)
could be calculated. An optical particle sizer (WELAS Promo 2000, Palas GmbH)
measured the droplet size distributions during the cloud periods. An in situ
particle backscatter instrument the SIMONE, detected
the presence of droplets and phase transitions within the chamber. In
addition, the Particle Phase Discriminator mark 2, Karlsruhe edition (PPD-2K)
was used to monitor the phase of particles in the size range of 6 to
60 µm, and thus to detect ice formation. Ice particles and water
droplets were differentiated by the PPD-2K based on their forward scattering
signals .
The droplet size distributions were also measured during the cloud periods
with the cloud and aerosol spectrometer with polarisation detection (CASPOL),
which measures the forward and backward scattering from single particles in
the diameter range of 0.6 to 50 µm .
During both CLOUD8 and CLOUD9, the saturation relative humidity (100 %
with respect to liquid water) was determined from the chamber temperature at
the time that the SIMONE first detected an increased forward scattering
signal, indicating the beginning of droplet growth. As the SIMONE measured
in situ, and was not influenced by transmission effects or temperature
changes in the sampling line, it is expected that this provides the most
accurate way of determining the dew point and thus the LWC during the cloud
periods. From the dew point temperature, the water vapour mixing ratio was
calculated. A high and a low value of the dew point was also determined from
the SIMONE data, based on the uncertainty in the detection time of the cloud
formation with the scattering signal. The total water content of the air in
the chamber was assumed to be constant over the timescale of the experiments
modelled here. As aerosol growth was only observed during the cloud periods,
and the dew point was determined at the beginning of each 4–6 min
cloud period, the assumption of constant total water content is not expected
to influence the results. This approach was validated during CLOUD9 by
comparison with the TDL data. In Fig. , the SIMONE derived and
measured peak condensed water during CLOUD9 expansions is plotted. The SIMONE
derived values are calculated from the dew point determined as described
above, the measured condensed water is found by taking the difference of the
TDL (which only measures gas phase water) and the MBW (which measures total
water after the droplets have been evaporated). From this comparison, it can
be seen that the dew point and thus the total condensed water can be
accurately determined using the combination of SIMONE and temperature data.
A comparison of the SIMONE-derived and measured (MBW total water
minus TDL gas phase water) peak condensed water during CLOUD9 expansions.
![]()
During both CLOUD8 and CLOUD9, aerosol composition (ammonium to sulphate
ratio) was measured with an Aerodyne high-resolution time-of-flight aerosol
mass spectrometer HR-ToF-AMS,. To dry the sample flow,
a Nafion dryer (PermaPure) was attached to the sampling line prior to the
HR-ToF-AMS, and a pressure controlled inlet (PCI), as described in Bahreini
et al. (2008), was used to eliminate the effects of pressure variations in
the sampling line on the measured concentrations.
The ionic composition of the aerosol formed is derived from HR-ToF-AMS data
using the PIKA software package. A critical parameter required in the model
is the ammonium to sulphate ratio (ASR) in the aerosol phase, which
influences the pH of the cloud droplets and consequently SO2
reaction rates. The determination of this ratio heavily depends on the
relative ionisation efficiencies (RIE) of ammonium and sulphate. Ammonium
RIE, equal to 3.9±0.2, was determined during the IE calibration by
nebulising NH4NO3, whereas the sulphate RIE, equal to 1.38±0.07, was determined by nebulising (NH4)2SO4. Another
technical limitation that might affect the determination of ammonium
concentrations is related to the possible interference of water and oxygen
fragments: O+ and HO+. Figure displays the
intensities of the main ammonium ions normalised by that of
NH2+. The clustering of these ratios around a single value
(NH3+/NH2+=1.23±0.05 and
NH+/NH2+=0.072±0.008), irrespective of
NH4+ concentration indicates that the interference from
O+ and HO+ is negligible in our case.
AMS measurements of SO4 (red line) concentrations and
NH4/SO4 molar ratios (blue markers for raw data and black
line for smoothed data). Also shown are the ratios between the main ammonium
fragments, to inspect a possible interference from O+ and HO+ on the
determination of NH4 concentrations.
![]()
Measurements of glyoxal (CHOCHO) were performed with the Light Emitting Diode
Cavity Enhanced DOAS (LED-CE-DOAS). This instrument is a multispectral sensor
that selectively and simultaneously measures glyoxal, oxygen collision
complexes (O2-O2), methyl glyoxal (CH3COCHO), nitrogen
dioxide (NO2) and other gases utilising an Ocean Optics QE65000
spectrometer . The detection limit of this
instrument during the CLOUD9 experiment was determined to be
15–20 pptv at the native 1 min measurement resolution employed
for this study. The instrument has been extensively compared to
gravitational, UV-vis and IR absorption, phosphorescence and mass
spectrometric measurement techniques for alpha-dicarbonyls and NO2
and remote-sensing techniques .
Seed aerosol formation
Two kinds of seed aerosol were used in these experiments, pure
H2SO4, and partially to fully neutralised ammonium sulphate
aerosol. The pure H2SO4 aerosol was formed in an external CCN
generator, which comprised a temperature controlled stainless steel
vessel holding a ceramic crucible filled with concentrated
H2SO4. After heating the vessel to between 150 and
180 ∘C, depending on the desired characteristics of the
aerosol population, a flow of N2 was passed through the
vessel, above the crucible to transport the hot H2SO4 vapour
into the chamber. In addition, during CLOUD9, a humidified flow of
N2 was added to the aerosol injection line immediately
downstream of the H2SO4 vessel, to create more
reproducible size distributions. As the vapour cooled in the injection
line, H2SO4 droplets formed. The partially or fully
neutralised aerosol was formed by using the same aerosol generator,
and injecting NH3 directly into the chamber, where it was
taken up by the acidic aerosol. The mode diameter of the aerosol
distribution produced by this method was approximately
65–75 nm, with a full width at half maximum (FWHM) of
approximately 50–70 nm.
Experimental procedure
A typical experiment began by pressurising the chamber to 220 hPa
above ambient pressure, and injecting SO2 and O3 so as to
reach the desired mixing ratios. Seed particles were then added to the
chamber, and observed for approximately 40 min to detect any growth
under subsaturated conditions, before the pressure in the chamber was rapidly
reduced to form the cloud. After a 15–30 min waiting time, during
which the chamber temperature re-stabilised, the pressure was increased
again, and further expansions were performed. This was repeated until the
seed aerosol numbers decreased below a few hundred per cubic centimetre. With
initial particle number densities of approximately
6000–8000 cm-3, two to four cloud cycles could usually be
performed. The properties of the aerosol, cloud droplets, and gas phase
species were continually measured during all stages of the experiment.
The aerosol size distribution measured by the SMPS attached to the
total sampling line, for experiments performed on the 17 November 2013. The
white line of points shows the mode diameter. Aerosol growth is clearly
observed during the cloud periods (marked by the purple vertical bars).
![]()
In order to better illustrate the experimental procedure, it is useful to
examine some basic measurements. The SMPS measurements of the aerosol size
distribution during one experiment are shown in Fig. . At
approximately 13:00 UTC, sulphuric acid aerosol was injected into the
chamber (not shown). The conditions in the chamber were held constant for
approximately 2 hours, during which time the number of aerosol particles
reduced due to dilution. At approximately 15:10 UTC, the first expansion was
performed, and a cloud formed (marked in purple in the figure). The resulting
aerosol growth can clearly be seen, both in the distribution, and from the
mode diameter (line of white dots), which increases from 70 to 90 nm.
The chamber was re-pressurised and a second expansion was performed shortly
after 17:00 UTC, also resulting in substantial aerosol growth. No growth was
observed during sub-saturated periods.
Measured parameters during the second expansion performed on
17 November 2013. (a) the time series of pressure and
temperature, with the expansion occurring at approximately
600 s. The resulting temperature decrease below the dew
point and increase in liquid water content are shown in
(b). Vertical lines in (a) and (b)
indicate the time at which the SIMONE first detected enhanced
forward scattering, indicating the presence of droplets. The
temperature measured at this time was taken as the dew point and
thus the total water content could be calculated. (c)
and (d) show gas mixing ratios and the total density of particles,
while (e) shows the diameter of the 25th and 75th percentile of
particles, as well as the median diameter (dashed, dot-dashed and solid lines
respectively).
![]()
The chamber conditions for the second expansion in Fig. , which
were used as input for the model, are shown in Fig. . Before
the expansion, the temperature and pressure were approximately constant at
283 K and 1160 hPa respectively, and the experiment was
performed with approximately 18–20 ppbv SO2 and
120 ppbv O3. At approximately 600 s the expansion
begins, and the pressure dropped, leading to an adiabatic cooling. The
vertical line denotes where the SIMONE first detected the presence of water
droplets, indicating that the dew point was approximately 1 K below
the chamber temperature before the expansion started. During this expansion,
the temperature reached approximately 1.5 K below the dew point, and
the liquid water content of the air (LWC) reached a total of
0.7 gm-3. In panel d, it can be seen that the total aerosol
number concentration (the integral of the total SMPS size distribution)
decreased due to the approximately 20 % pressure reduction during the
expansion, and due to the sedimentation and deposition of cloud droplets
containing aerosol. However, before and after the expansion, the measured
values remained relatively stable, with a slight decrease due to dilution. In
the bottom panel of Fig. the diameter of the 25th and 75th
percentile of the particles, as well as the median diameter are plotted as
a function of time. The growth in the dry aerosol diameter resulting from
aqueous phase chemistry during the cloud is clearly visible. The aerosol
growth occurred rapidly as the LWC increases, with growth ending soon after
the LWC reached its peak.
Model description
A microphysical and chemical box model was constructed to simulate the
experiments in the CLOUD chamber. The model was initialised approximately
10 min before each expansion with the dry aerosol size distribution
measured by the SMPS attached to the total sampling line. The aerosol sizes
measured by the SMPS were not adjusted to account for any remaining water
present at the measurement RH (approximately 3 % during CLOUD8 and
approximately 20 % during CLOUD9). This will lead to a slight
overestimate of the non-water volume of aerosol during CLOUD9, however this
effect is similar to the magnitude of the measurement uncertainty of the SMPS
instruments (approximately 14 % in volume for particles larger than
50 nm diameter and approximately 22 % in volume for particles
smaller than 50 nm in diameter). Subsequently, the model was
integrated along a time series of data measured during the chamber
experiments. Temperature and pressure, as well as the mixing ratios of
SO2 and O3 were input at 1 s resolution, and while the
actual resolution of the aerosol NH4+ fraction derived from the
AMS data was approximately 10 s, this was also interpolated to
a 1 s resolution for model input. The water vapour mixing ratio for
each experiment was set at the beginning of the simulation (calculated from
the dew point temperature, as described in Sect. ), and the sum
of the gas and liquid phase water was held constant. The gas phase ammonia
mixing ratio was determined for the starting time of the simulation and held
constant throughout. As the gas phase ammonia mixing ratio was below the
detection limit of the PTR-TOF-MS, this mixing ratio was determined using the
Extended Aerosol Inorganic Model (E-AIM) , by
summing the total amount of sulphate and ammonium present in the aerosol
population, and calculating the corresponding equilibrium gas phase ammonia
mixing ratio under the RH and temperature conditions of the chamber. The
validity and limitations of this approach will be discussed further in
Sect. . As described below, E-AIM is also used to calculate
the water activity and vapour pressure over the droplets. E-AIM has been
written with the aim of reproducing the thermodynamics of the aerosol system
as precisely as possible, without making compromises for the sake of
computational efficiency. It is therefore considered a benchmark model
e.g..
Aerosol water content and droplet growth
At sub-saturated conditions, the RH in the chamber during the simulated time
periods was typically above 90 %. At RH below 93 %, the hydrated
aerosol was assumed to be in equilibrium with the gas phase, and the water
activity was calculated, at the maximum model time step of 1 s, using E-AIM.
This calculation took into account the amount of SO42- and
NH4+ in the aerosol, as well as the temperature, while the
pressure was held constant at 1 atm.
Once the RH reached or exceeded 93 %, a full kinetic calculation of the
water uptake into the aerosol or droplets was performed. The Kelvin effect
was accounted for in both the equilibrium and the kinetic calculations. At
RH ≥ 93 %, the equilibrium vapour pressure of water over the
aerosol or droplets, as well as the surface tension the latter
calculated in E-AIM following and the solution density were
determined based on interpolation of a pre-calculated lookup table. The
lookup table was created using E-AIM, and provides the equilibrium vapour
pressure as a function of temperature,
NH4+ : SO42- ratio and
H2O : SO42- ratio. The lookup table covers the
range of H2O : SO42- ratios from 12 to pure water,
temperature from 250 to 300 K and
NH4+ : SO42- ratio from 0 to 2. The vapour
pressure determined from the lookup table was then adjusted to account for
the Kelvin effect. The equilibrium vapour pressure over very dilute
NH4+/SO42-/H2O solution droplets (assumed
to be pure water) was calculated using the approach of , who
parameterised the equation of .
The physics of water uptake by aerosol and cloud droplets is well discussed
in standard atmospheric physics textbooks, however, as this is the first time
the model is described, we believe it beneficial to provide as detailed
a description as possible.
When a droplet is large compared to the mean free path of water in air
(continuum regime), the total flux of water to a droplet, and thus the uptake
(in moles per second), is given by
Jc=4πRpDg(pH2O-pvap)×10-4RT,
where Rp is the droplet radius in cm, Dg the diffusivity of water in
air (in cm2 s-1), pH2O the partial
pressure of water vapour (in hPa), Pvap the vapour pressure of water
at the droplet surface (in hPa, calculated from the product of the
equilibrium vapour pressure of water at a given temperature
and the activity of water in the droplet), R is the gas constant
(8.314 Jmol-1K-1), and T is the temperature (K). Dg is
calculated as
Dg=0.211×1013.0PT273.151.94,
where P is the pressure in hPa. On the other hand, small particles find themselves in the kinetic regime, with the flux (in moles per second) described by
Jk=πRp2να(pH2O-pvap)×10-4RT,
where α is the mass accommodation coefficient for water (assumed to be 1.0). The mean speed of the water molecules (in cms-1) is given by:
ν=8×10-7RTπMH2O12,
where MH2O is the molar mass of water (in gmol-1).
The transition from the kinetic to continuum regime is accounted for in the model by the flux matching approach of :
JJc=1+Kn1+1.71Kn+1.33Kn2,Kn=λRp,λ=3Dgν,
where Kn is the Knudsen number (dimensionless) and λ is the mean free path (cm).
The Kelvin effect is accounted for by multiplying pvap by a correction factor, C:
C=exp2MH2Oσ1×107RTρRp,
where σ is the surface tension and ρ is the density of the H2O/H2SO4 solution at the temperature of interest.
The time step for the calculation of the water flux to or from the droplets
was calculated according to the ratio between the flux and the total droplet
water content, so that the droplet water content could change by no more than
2 % in a single time step. The maximum time step was restricted to
1 s.
The aerosol and the water droplets were assumed to be in thermal equilibrium
with the gas phase at all times. The aerosol concentration varies linearly
with the pressure in the chamber, with concentrations decreasing by around
20 % during an expansion due only to the pressure change. During some
expansions, particularly when the number of seed aerosol was low, the
observed change in particle number concentration was larger than the modelled
change, indicating that processes other than the pressure change influenced
the particle loss. Corrections for both these effects are applied in the
analysis.
Chemistry
The partitioning of SO2 and O3 to the cloud droplets, as
well as the subsequent aqueous phase chemistry was calculated on a time step
100 times smaller than that for the water partitioning. Performing the
analysis described by showed that, under the conditions
of the experiments presented here (even at -10∘C), the partitioning
of SO2 to the cloud droplets can be represented with an effective
Henry's law approach, neglecting mass transport limitations. Thus the total
amount of S(IV) in the droplet is given by
[S(IV)]=HSO3-+SO32-+SO2⋅H2O
with
HSO3-=HSO2K1pSO2[H+],SO32-=HSO2K1K2pSO2[H+]2
and
[SO2⋅H2O]=HSO2pSO2,
where pSO2 is the gas phase partial pressure of SO2,
K1 and K2 are equilibrium constants (Table ), and
HSO2 is the Henry's law coefficient for the dissolution of
SO2 in water (Table ). H2SO4 completely
dissociates to HSO4- and H+. The equilibrium constants
given in Table are taken from , who, in
turn, reported the values recommended by , based on an
evaluation of experimental studies conducted between 1910 and 1974. The heat
of dissolution for HSO2 is based on measurements between 25 and
50 ∘C.
Equilibrium constants used in the model. The temperature dependence
of the equilibrium constants is given by K=K298exp-ΔHR1T-1298, where K is the
equilibrium constant at temperature T in Kelvin. K1 refers to the
equilibrium SO2⋅H2O⇌HSO3-+H+, K2 refers to the equilibrium
HSO3-⇌SO32-+H+
and K3 to the equilibrium HSO4-⇌SO42-+H+.
K298 [M]
-ΔH/R [K]
Reference
K1
1.3×10-2
1960
,
K2
6.6×10-8
1500
,
K3
1.02×10-2
2720
,
KNH3
1.7×10-5
-4353.09
KH2O
1.0×10-14
-6710
,
The pseudo first order rate coefficient for the aqueous phase oxidation of S(IV) by O3 is
kS(IV)=(k0×[SO2⋅H2O])+k1×HSO3-+k2×SO32-,
with k0, k1 and k2 given in Table . Ozone, on the other
hand, is subject to mass transport limitations under the experimental
conditions described here, therefore, similarly to , we
follow the approach of , with the change in aqueous phase
O3 concentration being given by
dCaqdt=kmtRTpO3-kmtHO3RTCaq-QRp,
with
kmt=[Rp2RT3Dg+Rp(2πMO3RT)(1/2)3α]-1,Q=3cothqq-1q2,
and
q=RpkS(IV)Daq.
Henry's law coefficients and heats of dissolution. The temperature dependence of the coefficients is given by
HA(T2)=HA(T1)expΔHAR1T1-1T2.
H [M atm-1 at 298 K]
ΔHA [kcal mol-1 at 298 K]
Reference
HO3
1.1×10-2
-5.04
HSO2
1.23
-6.25
Here, pO3 is the partial pressure of O3,
HO3 is the Henry's law coefficient for O3, measured by
at temperatures between 5 and 30 ∘C
(Table ) and Caq is the aqueous phase concentration at
the surface of the droplet. The coefficient kmt accounts for the
gas and interfacial mass transport limitations. Rp is the
radius of the droplet or aerosol particle, MO3 is the molar mass
of O3 and α is the mass accommodation coefficient
4.0×10-2,. In Eq. (), Q, given
by Eq. (), is a correction factor to account for the lower aqueous
phase concentrations of O3 caused by diffusion limited transport
within the droplet. Daq is the aqueous phase diffusion
coefficient of O3, taken as a generally representative value of
1×10-5 cm2 s-1 , and assumed to be temperature
independent. Dg is the gas phase diffusion coefficient of
O3, where the typical value of 0.1 cm2s-1, for
a temperature of 298 K, given by is used as
a starting point, and scaled for a particular temperature and pressure as
follows. According to , the value of the gas phase
diffusion coefficient of a gas A in a second gas B can be calculated as
DAB=0.00266T3/2PbarMAB1/2σAB2ΩD,
where Pbar is the pressure in bar, σAB2 is
a characteristic length in Å, and ΩD is a dimensionless
diffusion collision integral. Knowing the value of Dg at a particular
temperature and pressure (298 K, 1 bar) permits the
calculation of a constant to represent all terms in Eq. (),
except T and Pbar, enabling scaling to other temperatures and
pressures:
Dg(Tf,Pf)=1.94×10-5Tf3/2Pf,
where Tf and Pf are the temperature and pressure of interest.
At the beginning of the chemistry time step, the partial pressures of
O3 and SO2, as well as the temperature and pressure and
the AMS derived NH4 mass fraction of the aerosol from the input
file were interpolated to the model time. The total NH4 mass in
each aerosol particle or droplet was calculated relative to the S(VI) mass,
and the NH4 and S(VI) concentrations were calculated using the
water volume calculated for a particle or droplet in that size bin.
Subsequently, the concentration of H+ in each droplet, and thus
the concentration of the other ions was calculated by iteratively solving the
electroneutrality equation,
H++NH4+AMS+NH4+part=OH-+HSO3-+2SO32-+2SO42-+HSO4-,
where NH4+AMS and
NH4+part are the NH4+ determined
from the AMS mass fraction and from the partitioning of gas phase
NH3 into the cloud droplet, respectively. The latter is given by
NH4+part=HNH3×pNH3×KNH3×H+/KH2O,
with pNH3 being the gas phase partial pressure of NH3,
and the remaining constants defined in Tables and .
In Eq. (), HSO3- is given by
Eq. (), SO32- by
Eq. (), SO42- by
SO42-=K3S(VI)H++K3
and HSO4- by
HSO4-=H+S(VI)H++K3.
Once the concentrations of the ions are known, the production of S(VI) is
calculated from
ΔS(VI)=QkS(IV)[O3]aqdt.
The new aqueous phase O3 concentration is calculated using
a backward Eulerian approach to solve Eq. (), so that a relatively
long time step can be used without large fluctuations in
[O3]aq which could lead to values that are negative, or
exceed the maximum concentration determined by Henry's law.
Results
Data from a total of 31 expansions were analysed, 15 at 10 ∘C, and
16 at -10 ∘C. Table lists the most important data
describing the different experiments and model runs. SO2 mixing
ratios ranged from 0.9 to 26.4 ppbv, and O3 from 63.5 to
137.2 ppbv.
Aerosol activation and cloud droplets
Peak LWCs of up to 1.5 gm-3 were seen during the 10 ∘C
experiments, and the largest size reached by the median droplets in the
modelled size distribution was 17 µm, with the smallest being
4.9 µm. Most of the aerosol particles were activated during the expansions,
with modelled activated fractions generally being around 0.9. In one case,
however, only 34 % of the aerosol particles were activated.
The number of cloud droplets formed in the model (orange stars) and
the number of aerosol immediately before the expansion (black diamonds),
plotted against the maximum number of droplets detected by WELAS during the
presence of the cloud. The error bars on the modelled droplet numbers
correspond to simulations using the high and low values of the dew point
calculated from the SIMONE data. Data are shown for all CLOUD9 experiments
where data were available.
![]()
The number of modelled cloud droplets (defined as the number of droplets
larger than 1 µm in diameter; orange symbols) is compared with
the number measured by the WELAS, in Fig. , for several
of the CLOUD9 experiments. WELAS data from two of the CLOUD9 experiments, and
from the CLOUD8 experiments were not available. At aerosol numbers below
1000 cm-3, all aerosol are activated, and the modelled number of
droplets matches the number of droplets detected by the WELAS. At higher
aerosol concentrations, the modelled and measured numbers of droplets
diverge, with the modelled number being higher than that measured by the
WELAS. The LWC calculated from the WELAS data is sometimes lower than the LWC
derived from the SIMONE, TDL and MBW data, which may explain part of the
discrepancy between modelled and WELAS-measured droplet numbers. Further,
cooling in the chamber is unlikely to be totally homogeneous, possibly
resulting in slightly differing activated fractions of aerosol in different
regions of the chamber in the initial stages of the expansion. Much of the
clear relationship between the modelled droplet number and that measured by
the WELAS can likely be explained by the fact that both depend strongly on
the number of aerosol in the chamber (black diamonds in
Fig. ), nevertheless it is encouraging that the modelled
and measured droplet numbers are well correlated.
Ammonia in the CLOUD chamber
At this point, it is useful to examine the assumptions made with regard to
the gas phase ammonia, and the amount of ammonium in the aerosol and the
cloud droplets in greater detail. In Sect. , it was stated
that the ASR of the aerosol were measured using an AMS, that these ratios
were then used to determine the gas phase ammonia in the chamber, using
E-AIM, and that the uptake of this ammonia to cloud droplets was calculated
assuming that the droplets were in an effective Henry's law equilibrium with
a constant gas phase ammonia concentration. For this approach to be accurate,
a number of conditions must be met, each of which are discussed in the
subsections below. We find that the pre-expansion determination of the gas
phase ammonia mixing ratio is likely to be correct, as is the assumption that
the ammonium concentration in the droplets is in equilibrium with the gas
phase ammonia. However, the gas phase ammonia mixing ratio during the cloud
periods cannot generally be assumed to be constant.
Accuracy of AMS measurement of ammonia in hydrated aerosol
The equilibrium gas phase ammonia mixing ratio, calculated with
E-AIM, as a function of the ammonium to sulphate ratio of the aerosol. Data
shown are for -10 and 10 ∘C,
atRH=95 %.
![]()
Firstly, the ammonium to sulphate ratio in the particles must be accurately
measured by the AMS. Some technical aspects of this measurement were already
discussed in Sect. . The NH4+ : SO42-
ratio used to determine the gas phase ammonia mixing ratio was measured
during the subsaturated period immediately preceding each expansion, when
the RH in the chamber was approximately 95 %. The calculation of the
gas phase ammonia was carried out with E-AIM, using the AMS determined
NH4+ : SO42- ratio, at a pressure of one atmosphere
and the temperature and RH conditions measured in the chamber. Any reduction
in NH4+ : SO42- ratio due to processes such as the
drying of the air on the way to the AMS would lead to a lower calculated gas
phase ammonia mixing ratio (Fig. ). Calculations with E-AIM
show that under the assumption of a metastable aqueous phase, without the
formation of solids, the NH4+ : SO42- ratio can be
reduced from 2 to 1.5 during a change in RH from 95 to 35 %. If solid
phases are allowed to form however, the
NH4+ : SO42- ratio is much less sensitive to
changes in RH. It has been observed that ammonium sulphate particles bounce
from the AMS vapouriser, therefore we expect that ammonium sulphate is in
a solid state and that there are minimal ammonium losses during the drying
process.
Rate constants for the aqueous phase oxidation of S(IV) with
dissolved ozone, used in Eq. (). The temperature dependence of
the rate constants is given by
k(T)=k298exp-ER1298-1T.
The activation energies given for the k1 and k2 values are those of
, based on measurements at 16 and 25 ∘C.
K298 [M-1s-1]
-E/R [K]
Reference
k0
2.4×104
0
,
k1
3.7×105
-5530
,
k2
1.5×109
-5280
,
In a recent study, nebulised mixtures of ammonium sulphate and ammonium
nitrate were measured with an AMS (nitrate activity coefficients of 0, 0.3,
0.5, 0.7 and 1). It was found that the AMS reliably measures the predicted
ammonium content of the mixtures of these internally mixed ammonium nitrate
and ammonium sulphate particles (Xu et al., 2016). We are not aware of any
analogous study for aerosol containing only varying ratios of ammonium and
sulphate, however if ammonia is not lost from the mixtures described above
during the measurement process, we do not expect that it will be lost from
aerosol formed from pure ammonium and sulphate solution. Therefore it is
likely that the AMS determined ammonium to sulphate ratios are correct.
Uptake of ammonia by a droplet
The aqueous phase concentration of NH4+ in
a 7 µm diameter water droplet, exposed to a 22 pptv
NH3 gas phase, similar to the conditions of experiment
CLD8_20_11_1b.
![]()
Since it is possible to estimate the gas phase ammonia before the expansion
and cloud formation from the AMS measurements with a reasonable degree of
accuracy, the next question is whether or not our assumption of an effective
Henry's law equilibrium between droplet and gas phase ammonia is accurate.
Using an experiment with a relatively high gas phase ammonia concentration as
an example (CLD8_20_11_1b), the uptake of ammonia by a 7 µm
diameter water droplet was calculated, under the assumption that the droplet
initially contained no ammonium, and was exposed to a 22 pptv ammonia
gas phase. The development of the droplet ammonium concentration with time is
plotted in Fig. , for three different solution pH. At the
lowest, pH=4, equilibrium is reached after approximately
600 s, which is comparable with the time scale of the clouds formed
in the chamber (400–600 s). At higher pH, the timescale becomes
shorter very rapidly. As shown in Table , for most of the
experiments, the pH lay between 4 and 5. We therefore expect that the
ammonium concentration in the droplets will be essentially equal to the
values predicted by assuming an effective Henry's law equilibrium with the
gas phase.
A list of the measured and modelled conditions during each
experiment. Ti is the temperature just before the expansion, and
Tmin is the minimum temperature measured during the expansion.
Peak LWC is the maximum condensable water during the expansion, derived from
the dew point temperature (Tdew) and Tmin. The median
drop diameter (modelled) is the largest size reached by the median drop
during the expansion and the median dry diameter is taken from the model
output just prior to the expansion. The maximum activated fraction is the
maximum modelled number of droplets larger than 1 µm in diameter
during the expansion, divided by the total number of aerosol. The change in
measured volume divided by the change in modelled volume is the numerical
value for the points plotted in Fig. . The maximum pH is
the maximum pH reached by the median sized droplet during the expansion
(modelled), the NH4+/SO42- molar ratio is calculated from
the AMS data just prior to the expansion. The gas phase ammonia mixing ratio
was calculated from the AMS data immediately prior to the expansion, as
described in Sect. and the number of aerosol is the modelled
number of aerosol just prior to the expansion.
Run
Stage
SO2
O3
Ti
Tmin
Tdew
Peak LWC
Median drop
Median dry
max.
Δ Meas. vol/
max pH
NH4+/ SO42-
NH3(g)
Naerosol
number
[ppbv]
[ppbv]
[K]
[K]
[K]
[g m-3]
Diameter
Diameter
activated
Δ Mod. vol
molar ratio
pptv
[ µm]
[nm]
fraction
CLD8_17_11_1a
1243.04
17.4
82.3
282.3
277.6
281.0
0.9
9.8
66.1
0.92
0.84
4.37
0.03
0.0142
2100.35
CLD8_17_11_1b
1243.06
18.3
125.8
282.4
278.4
281.2
0.8
13.4
79.6
0.92
0.53
4.28
0.03
0.0123
732.06
CLD8_18_11_1b
1245.03
19.9
126.5
282.5
278.5
280.8
0.5
8.6
66.5
0.73
0.69
4.24
0.02
0.0377
2288.45
CLD8_20_11_1a
1249.03
1.3
136.3
282.3
278.0
281.3
0.9
8.7
76.4
0.87
2.33
5.55
1.58
13.0602
3041.92
CLD8_20_11_1b
1249.09
2.3
137.2
282.3
278.5
281.7
0.9
6.9
88.2
0.95
1.25
5.55
1.67
21.5740
6323.98
CLD8_20_11_2a
1250.11
23.7
136.8
262.7
258.4
261.8
0.3
5.1
71.0
0.95
1.75
4.49
1.87
1.4404
5628.41
CLD8_20_11_2b
1251.04
26.6
128.9
262.3
258.5
261.5
0.3
4.9
91.4
0.90
0.40
4.57
1.88
2.5927
6309.59
CLD8_20_11_2c
1251.09
21.0
126.9
262.4
258.5
261.4
0.3
7.0
113.3
0.90
1.01
4.47
1.85
1.1498
1874.42
CLD8_21_11_1a
1252.04
18.3
115.3
262.3
258.3
261.6
0.3
6.0
81.9
0.96
0.66
4.48
1.87
1.0562
3213.07
CLD8_21_11_1b
1252.12
18.5
115.4
262.3
258.5
261.7
0.3
5.0
82.1
0.98
0.56
4.52
1.80
1.3962
4322.27
CLD8_21_11_1c
1252.17
19.8
118.8
262.5
258.5
261.5
0.3
7.2
101.8
0.96
0.83
4.42
1.69
0.7903
1632.21
CLD8_05_12_1a
1306.20
1.1
92.4
282.7
278.8
281.6
0.6
9.8
53.2
0.34
0.94
5.79
1.75
21.1771
3920.10
CLD8_05_12_1b
1306.27
0.9
131.0
282.8
278.9
281.6
0.6
11.6
71.1
0.59
0.65
5.70
1.74
24.2970
1341.82
CLD8_07_12_1a
1310.14
19.6
107.1
262.6
259.0
261.5
0.2
5.5
71.2
0.92
6.15
4.25
1.24
0.2614
2672.03
CLD8_07_12_1b
1310.17
10.3
178.1
262.6
259.2
261.6
0.2
7.3
109.6
0.96
3.62
4.23
1.21
0.1665
1086.43
CLD8_07_12_1c
1310.23
10.2
203.9
262.5
259.0
261.5
0.2
6.8
88.5
0.90
2.66
4.21
0.85
0.1085
1440.50
CLD9_24_09_2b
1418.11
18.5
112.8
283.4
279.2
282.4
0.8
8.1
66.0
0.86
0.41
4.40
0.61
0.5036
3492.84
CLD9_24_09_2c
1418.15
19.4
115.6
283.4
279.1
282.9
1.0
10.6
76.2
0.90
0.25
4.44
0.84
0.6745
1794.43
CLD9_25_09_1a
1422.05
29.2
90.8
283.7
278.3
282.2
0.9
10.1
98.8
0.98
0.39
4.41
0.92
1.7512
1856.36
CLD9_25_09_1b
1422.08
26.4
85.8
284.0
278.7
282.3
1.0
12.1
112.0
0.97
0.66
4.36
0.89
0.5860
998.07
CLD9_25_09_1c
1422.12
25.4
73.9
284.8
278.7
282.6
1.1
17.0
136.4
1.00
0.63
4.32
0.55
0.2159
410.01
CLD9_25_09_2a
1422.15
25.0
63.5
283.8
278.6
283.1
1.4
8.6
75.7
0.86
0.43
4.35
0.77
0.6497
5028.45
CLD9_25_09_2b
1422.17
25.1
89.2
284.2
277.9
282.9
1.5
10.7
91.6
0.89
0.39
4.35
0.58
0.7761
2756.07
CLD9_25_09_2c
1422.20
25.0
84.9
284.1
277.8
282.9
1.4
12.7
113.0
0.92
0.42
4.42
1.01
1.3842
1476.15
CLD9_28_09_1a
1434.11
17.5
85.3
262.3
257.7
260.8
0.3
5.7
82.2
0.94
2.06
3.91
0.28
0.0098
3385.48
CLD9_28_09_1b
1434.13
20.4
92.8
263.3
257.9
261.7
0.4
6.8
88.1
0.97
1.64
3.97
0.32
0.0093
1970.34
CLD9_28_09_1c
1434.16
22.9
109.4
263.3
258.0
261.9
0.4
8.7
99.1
0.98
1.31
4.02
0.43
0.0124
1149.79
CLD9_28_09_1d
1434.19
21.8
112.2
263.4
258.1
261.9
0.4
10.5
113.6
0.99
1.27
4.04
0.63
0.0174
595.93
CLD9_28_09_2a
1435.04
15.5
107.2
263.5
258.1
261.8
0.3
8.3
49.7
0.92
0.56
4.17
0.10
0.0081
1445.49
CLD9_28_09_2b
1435.06
7.6
109.1
263.1
257.9
261.8
0.4
9.4
73.9
0.96
0.31
4.29
0.92
0.0176
782.26
CLD9_28_09_2c
1435.09
4.0
111.9
263.2
257.7
261.7
0.4
11.9
88.1
0.95
0.40
4.37
0.71
0.0077
373.06
Data from the model simulations, showing the gas phase concentration
of ammonia, the total particle phase ammonium contained in all particles in
one cm3 of air before cloud formation, and the maximum additional
amount of ammonium taken up into all droplets in one cm3 of air
during the cloud formation. Values in brackets indicate the ratio of the
number of moles of ammonium contained in the particles or droplets to the gas
phase ammonia concentration. The first three columns are reproduced from
Table , for convenience. Note that the aqueous phase ammonium is
in addition to the particle phase ammonium, which is also contained in any
droplets that form.
Run
Stage
NH4+ / SO42-
NH3 (g)
NH3 (g)
NH4+ (particle)
NH4+ (aq, peak)
number
[molar ratio]
[pptv]
[mol cm-3]
[mol cm-3], (ratio)
[mol cm-3], (ratio)
CLD8_17_11_1a
1243.04
0.03
0.0142
7.05×10-19
5.86×10-16 (830.32)
3.86×10-16 (547.13)
CLD8_17_11_1b
1243.06
0.03
0.0123
6.11×10-19
2.62×10-16 (428.41)
2.63×10-16 (430.42)
CLD8_18_11_1b
1245.03
0.02
0.0377
1.88×10-18
9.32×10-16 (496.51)
9.83×10-16 (523.54)
CLD8_20_11_1a
1249.03
1.58
13.0602
6.50×10-16
4.14×10-14 (63.70)
1.65×10-14 (25.37)
CLD8_20_11_1b
1249.09
1.67
21.5740
1.07×10-15
1.41×10-13 (131.59)
3.26×10-14 (30.38)
CLD8_20_11_2a
1250.11
1.87
1.4404
7.71×10-17
7.23×10-14 (937.81)
3.89×10-14 (504.72)
CLD8_20_11_2b
1251.04
1.88
2.5927
1.39×10-16
1.87×10-13 (1342.01)
6.33×10-14 (455.78)
CLD8_20_11_2c
1251.09
1.85
1.1498
6.16×10-17
7.20×10-14 (1167.93)
3.18×10-14 (515.48)
CLD8_21_11_1a
1252.04
1.87
1.0562
5.66×10-17
7.22×10-14 (1274.75)
3.20×10-14 (564.51)
CLD8_21_11_1b
1252.12
1.80
1.3962
7.48×10-17
7.55×10-14 (1008.44)
2.93×10-14 (391.31)
CLD8_21_11_1c
1252.17
1.69
0.7903
4.23×10-17
4.04×10-14 (953.98)
2.43×10-14 (574.69)
CLD8_05_12_1a
1306.20
1.75
21.1771
1.07×10-15
4.85×10-14 (45.31)
1.33×10-14 (12.38)
CLD8_05_12_1b
1306.27
1.74
24.2970
1.23×10-15
5.58×10-14 (45.46)
1.43×10-14 (11.67)
CLD8_07_12_1a
1310.14
1.24
0.2614
1.42×10-17
4.03×10-14 (2829.00)
1.19×10-14 (839.10)
CLD8_07_12_1b
1310.17
1.21
0.1665
9.07×10-18
2.88×10-14 (3172.37)
6.87×10-15 (757.97)
CLD8_07_12_1c
1310.23
0.85
0.1085
5.91×10-18
2.64×10-14 (4464.63)
6.56×10-15 (1109.05)
CLD9_24_09_2b
1418.11
0.61
0.5036
2.55×10-17
1.15×10-14 (450.25)
1.76×10-15 (68.79)
CLD9_24_09_2c
1418.15
0.84
0.6745
3.42×10-17
8.48×10-15 (247.82)
9.17×10-15 (268.19)
CLD9_25_09_1a
1422.05
0.92
1.7512
8.87×10-17
2.44×10-14 (275.10)
2.40×10-14 (270.27)
CLD9_25_09_1b
1422.08
0.89
0.5860
2.97×10-17
1.20×10-14 (403.80)
1.31×10-15 (44.29)
CLD9_25_09_1c
1422.12
0.55
0.2159
1.09×10-17
5.05×10-15 (463.43)
6.33×10-16 (58.07)
CLD9_25_09_2a
1422.15
0.77
0.6497
3.29×10-17
2.66×10-14 (808.86)
3.81×10-15 (115.77)
CLD9_25_09_2b
1422.17
0.58
0.7761
3.92×10-17
2.00×10-14 (509.26)
1.73×10-14 (442.08)
CLD9_25_09_2c
1422.20
1.01
1.3842
7.00×10-17
2.32×10-14 (331.90)
1.47×10-14 (210.51)
CLD9_28_09_1a
1434.11
0.28
0.0098
5.35×10-19
9.05×10-15 (16 911.79)
1.37×10-15 (2555.99)
CLD9_28_09_1b
1434.13
0.32
0.0093
5.08×10-19
5.19×10-15 (10 223.71)
1.05×10-15 (2064.01)
CLD9_28_09_1c
1434.16
0.43
0.0124
6.76×10-19
6.35×10-15 (9391.70)
2.66×10-16 (393.23)
CLD9_28_09_1d
1434.19
0.63
0.0174
9.49×10-19
4.92×10-15 (5184.55)
1.44×10-15 (1519.16)
CLD9_28_09_2a
1435.04
0.10
0.0081
4.40×10-19
1.17×10-15 (2664.48)
7.30×10-16 (1657.98)
CLD9_28_09_2b
1435.06
0.92
0.0176
9.60×10-19
4.21×10-15 (4390.07)
1.00×10-15 (1045.57)
CLD9_28_09_2c
1435.09
0.71
0.0077
4.19×10-19
2.02×10-15 (4814.28)
1.40×10-16 (333.88)
Assumption of constant gas phase ammonia
The assumption that the gas phase ammonia is constant during the cloud
formation is more difficult to constrain. As shown in Table ,
under the assumption of a constant gas phase, and droplet equilibrium with
the gas phase, the amount of ammonia that was taken up into the cloud
droplets suspended in a cubic centimetre of air was between approximately 12
and 2500 times the gas phase amount in that same air volume. The experiments
with the highest values of this ratio were the ones performed under
essentially acidic conditions, such that ammonia was almost absent from the
gas phase. The results of these experiments are therefore not sensitive to
uncertainties in the gas phase ammonia concentration.
For the experiments with near neutral seed aerosol (for example
CLD8_20_11_2a), the droplets would have needed to take up around 500 times
the gas phase ammonia in order for our assumption of droplet equilibrium with
a constant gas phase to hold. The only possible sources of ammonia in the
chamber during the cloud formation were the chamber walls, any water in the
bottom of the chamber or the particles themselves. The total wall area of the
chamber is approximately 42 m2, which is comparable to the total
surface area of the droplets (8.5 m2 for 10 µm
diameter droplets at a concentration of 1000 cm-3,
34 m2 for 20 µm diameter droplets at a concentration
of 1000 cm-3). Before the expansion, the walls were in equilibrium
with the gas phase and the particles (no changes in the
NH4+ : SO42- ratio of the particles were observed
before the expansion, unless further ammonia was injected into the chamber),
being coated in a mixture of ammonium, sulphuric acid and water. During the
expansion, the walls maintained a constant temperature, while the temperature
of the gas in the chamber decreased.
The uptake of gases on chamber walls was investigated by
, who pointed out that when the accommodation coefficient
of the gas on the chamber walls is high (in their case higher than
approximately 6×10-6), transport is diffusion-limited, with the
rate of diffusion also depending on the turbulence in the chamber
(parameterised by an eddy diffusion coefficient). For their Teflon chamber,
they find an accommodation coefficient of 1.2-4.8×10-8 for
ammonia, suggesting that exchange with the walls occurs even more slowly than
the rate of diffusion. For the CLOUD chamber, previous measurements have
shown wall loss rates for sulphuric acid of 1.7×10-3s-1, corresponding to a lifetime of approximately
10 min. Assuming ammonia to behave in a similar way to sulphuric acid in the
CLOUD chamber, this suggests that the transport between the walls and the gas
phase is orders of magnitude too slow to maintain a constant gas phase
ammonia mixing ratio during the cloud formation.
A further possible source of ammonia is the un-activated particles. Once the
cloud forms, the ammonia vapour pressure over these deliquesced aerosol will
be higher than over the droplets, and ammonia may be transferred via the gas
phase from the aerosol to the droplets. However, from the data shown in
Table , it can be seen that the ammonia contained in the
particles was only 2–4 times greater than the additional ammonia required
for the droplets to reach their equilibrium value. Using CLD8_20_11_1b as
an example again, the maximum ammonium loss from a non-activated aerosol can
be calculated, such that the ammonia vapour pressure over that aerosol
reaches the vapour pressure over a droplet. In this case, only approximately
16 % of the ammonia can be removed from a non-activated aerosol, making
it necessary that as little as 30 % of the aerosol activate if the
remaining aerosol were to supply the missing ammonia to the droplets. The
aerosol in CLD8_20_11_1b had an NH4+ : SO42-
ratio of 1.67. As shown in Fig. , the vapour pressure over
the aerosol is highly sensitive to the NH4+ : SO42-
ratio, particularly as this approaches 2. This means that in other
experiments where the NH4+ : SO42- ratio was
higher, an even smaller fraction of the ammonia in the unactivated aerosol
can be given up. As shown in Fig. , the model likely
overestimates the activation, particularly at larger aerosol numbers, however
in order to provide enough ammonia, only a minority of the aerosol could have
been activated. A visual inspection of the aerosol diameters measured with
the SMPS showed that in all experiments, the majority of the aerosol grew
during the cloud periods, and must therefore have activated to form cloud
droplets. To summarise, the unactivated aerosol likely released some ammonia
during the cloud periods, however this would not have been sufficient to
supply the missing ammonia to the cloud droplets.
Glyoxal as an indicator of exchange between chamber walls and the gas phase
The change in glyoxal mixing ratio during an expansion performed in
the CLOUD chamber as part of a series of isoprene oxidation experiments.
Pressure and temperature are also shown, for reference.
![]()
During the analysis of other experiments being carried out in the CLOUD
chamber (ozone initiated oxidation of isoprene), high precision measurements
of gas phase gyoxal were performed. During these measurements, it was found
that as soon as the pressure decrease associated with an expansion began, the
gas phase glyoxal increased rapidly from around 50 pptv to almost
500 pptv (Fig. ). Glyoxal is relatively soluble,
with an effective Henrys law coefficient of approximately 4×105 Matm-1 at 298 K in pure water , a
value that increases by orders of magnitude for solutions containing sulphate
. Therefore this gas phase increase occurs in spite
of the simultaneous uptake by the droplets. Ammonia is comparably soluble,
with the effective Henry's law coefficient ranging between approximately
1×104 Matm-1 for a solution pH of 7, and 1×107 Matm-1 when the pH is 4 . It is
possible that the pressure change and increased turbulence during the
decompression of the chamber lead to a better ventilation of the chamber
walls. This would be characterised by a higher eddy diffusion coefficient,
increasing the rate of exchange between the gas phase and the walls above the
value that is observed during non-decompression periods.
Although high-precision ammonia measurements are not available for the CLOUD8
and CLOUD9 experiments, we expect any soluble gas adsorbed onto the chamber
walls to respond to the pressure change in a similar way as glyoxal has been
shown to behave. In the experiments with neutral or semi-neutral seed
aerosol, a large amount of ammonia was injected into the chamber, which was
partly taken up by the initially acidic seed, and partly deposited on the
chamber walls. A re-mobilisation of ammonia from the chamber walls similar to
that observed for glyoxal would lead to a large amount of ammonia in the gas
phase, which could be taken up by the droplets.
In summary, it is not possible to support the assumption of a constant gas
phase ammonia mixing ratio. It is likely that the droplets take up a large
amount of ammonia from the gas phase, due to a rapid mobilisation of ammonia
that was previously deposited on the chamber walls. However, it is not
possible to quantify the size of this source. It is also likely that
a certain amount of ammonia is transferred from the unactivated aerosol to
the cloud droplets during the cloud formation.
As mentioned above, the amount of gas phase ammonia is very small when the
NH4+ : SO42- ratio is low (below approximately
1.0). Below, we show that the more acidic experiments are uninfluenced by the
uncertainty in gas phase ammonia, as for these experiments it was only
present at trace levels. We believe it is still interesting to include the
more neutralised experiments in the analysis to investigate the sensitivity
of the results to ammonium in the droplets and aerosol. In order to perform
the model simulations, we maintain the assumption of a constant gas phase, as
a base case, and discuss the implications further below.
Aerosol growth and the effect of ammonia
In this section, the modelled and measured growth of the dry aerosol volume
before and after the expansion is discussed.
The pH of droplets formed during the expansions was seldom above 5, due to
the uptake of SO2 and rapid production of SO42- at
higher pH, and the resulting reduction in pH. The only exceptions were
experiments such as CLD8_05_12_1a and CLD8_05_12_1b where there were
comparatively high gas phase NH3 mixing ratios (approximately 21
and 24 pptv respectively, determined with the E-AIM model, as
described in Sect. ). A total of 12 of the experiments were
performed at NH4+ : SO42- ratios of greater than 1.
The modelled and measured change in total volume of the aerosol population
was calculated by integrating the dry aerosol size distribution (as measured
by the SMPS attached to the total sampling line), and subtracting the total
volume before the cloud period from that afterwards. Both modelled and
measured data were corrected for particle losses during the expansion, by
multiplying the volume after the expansion by the ratio of the number density
of aerosol before the expansion to that after the expansion.
(a) The measured growth in the total aerosol volume as
a function of NH4+/SO42- ratio in the seed
aerosol, at 10 and -10 ∘C. The growth is determined as the
difference in aerosol volume before and after the cloud. The error bars on
the measured growth stem from the uncertainty in diameter in the SMPS
measurements (which translates to approximately 14 % in volume for
particles with a diameter greater than 50 nm and 22 % for those
with a diameter less than 50 nm). (b) As in (a),
except here the modelled data are presented. The error bars stem from
simulations carried out using the high and low limits of the dew point, as
determined from the SIMONE data.
![]()
The dependence of the aerosol growth on the seed aerosol composition is shown
in Fig. a and b, for the measured and modelled growth
respectively. The aerosol growth during the supersaturated periods clearly
depends on the aerosol seed composition, with the strength of the effect
increasing rapidly at NH4+ : SO42- ratios above
approximately 1. This is due to the fact that the more neutral seeds are in
equilibrium with a higher gas phase NH3 amount, which will
partition into the cloud droplets when they form. This leads to a higher pH
during the cloud phase and accelerated SO2 oxidation. For the
aerosols with an NH4+ : SO42- ratio above 1.2, the
additional amount of ammonium which would partition to the droplets under the
constant gas phase assumption would have more than balanced the sulphate in
the droplets, leading to a very large increase in pH, and correspondingly,
a very large increase in aerosol growth. In contrast, at
NH4+ : SO42- ratios below approximately 1, there is
little ammonia in the gas phase, and it does not influence the aerosol
growth. The modelled dependence of aerosol growth on seed composition is very
similar to the measured dependence, however Fig. b shows
that the model systematically predicts greater aerosol growth at
10 ∘C than at -10 ∘C, an effect that is not seen in the
measured data (in the experimental data it appears that the lower reaction
rates at lower temperatures are balanced by the greater solubility of the
gases). This may point to a slightly too large temperature dependence of the
reaction rate constants, and indeed, using the temperature dependence of
leads to a slight closing of the gap between the
-10 ∘C and the 10 ∘C modelled points (not shown).
However, as the rate constants given in the studies of and
were based on measurements at 25 ∘C, weakening the
temperature dependence also increases the modelled growth of aerosol at
10 ∘C. During the CLOUD8 experiments at -10 ∘C, the
presence of an ice phase was detected from the PPD-2K measurements. The onset
of ice formation was also seen in the change of SIMONE forward scattering and
depolarisation signals. This ice formation was always detected after the
pressure decrease had ended, by which time the aerosol growth had also
ceased. No further growth of the aerosol was observed during or after the
formation of ice. An upgrade in the chamber expansion system between CLOUD8
and CLOUD9 prevented formation of ice during the CLOUD9 experiments.
(a) The modelled vs. observed total increase in aerosol
volume due to the aqueous phase oxidation of SO2 to
SO42- at 10 and -10 ∘C. Note that the
masses are corrected to account for any particle loss during the
expansion (see text). The dashed lines indicate the range of
a factor of two higher or lower than the observed values. The error
bars are as for Fig. . (b) As for
(a), but with doubled gas phase NH3. (c) As for (a), but with no gas phase NH3.
![]()
A comparison of the modelled and measured aerosol volume growth is shown in
Fig. a. Here, the gas phase ammonia mixing ratio was
calculated with E-AIM, as described above, and was held constant during the
cloud period. In general, the modelled volume growth matches the measured
values to within a factor of two (dot dashed lines in
Fig. ). This indicates that observed aerosol growth in the
CLOUD chamber matches the growth predicted by the reaction rate constants specified by
, at both 10 and -10 ∘C. The model appears to
slightly overestimate the aerosol growth at 10 ∘C, in the case of
the more acidic seed aerosols, while the -10 ∘C points are more
evenly distributed about the 1 : 1 line. Applying the temperature
dependence of has the effect of increasing the modelled
growth of both the -10 and the 10 ∘C points (the latter to
a slightly lesser degree), and rather decreases the agreement between the
modelled and measured values (not shown).
It should once again be noted that the assumption of a constant gas phase
ammonia cannot be supported. While the walls almost certainly provide a large
source of ammonia, this source cannot be quantified and the relatively good
agreement of modelled and measured growth during the 12 partially to fully
neutralised experiments must be ascribed to coincidence in each case. As the
modelled results of the more neutralised experiments depend on the gas phase
NH3 mixing ratio, two further model runs were performed, the first
with doubled gas phase NH3 (Fig. b), the second
with no gas phase NH3 (Fig. c), to assess the
sensitivity. From panel c, it is clear that the gas phase NH3 is
necessary for the model to reproduce the observed growth of the more
neutralised seed aerosol. Particularly in the 10 ∘C experiments at
NH4+ : SO42- ratios of close to 2, the modelled
aerosol growth is far too low if gas phase NH3 is neglected. On the
other hand, doubling the gas phase NH3 with respect to the original
values determined with E-AIM leads to an overestimation of the aerosol
growth. In general however, it can be seen that although the modelled growth
is clearly influenced by the gas phase NH3, small variations do not
greatly impact aerosol growth. It is also clear that for the more acidic
aerosol, the results are essentially identical if one assumes no gas phase
ammonia, or a constant gas phase ammonia mixing ratio in equilibrium with the
aerosol, as the latter is almost negligible.
Influence of ions on aqueous phase oxidation of <inline-formula><mml:math display="inline"><mml:mrow class="chem"><mml:msub><mml:mi mathvariant="normal">SO</mml:mi><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:mrow></mml:math></inline-formula>
One of the features of the CLOUD chamber is the ability to perform
experiments under conditions with varying concentrations of gas phase ions.
By applying a 30 kV electric field across the chamber, ions can
effectively be swept from the chamber, providing a neutral environment. In
the absence of the clearing field, natural radiation sources such as galactic
cosmic rays (GCR) lead to the ionisation of gases, creating ion pair
concentrations of approximately 650 cm-3, representative of the
number concentrations found in the atmospheric boundary layer. Higher
ion-pair concentrations can be achieved by using a pion beam from the CERN
proton synchrotron, however for the time period in which the experiments
described here were conducted, this was not available. Instead, the CIGAR (Corona Ion Generator for Aerosol Research) was
used to create higher ion concentrations. The CIGAR is a stainless steel
wire, 100 mm in length and 100 µm in diameter, situated
in an argon surrounding atmosphere in direct contact with the air in the
CLOUD chamber. A voltage of 3.5 kV is applied across the wire, which
is just below the voltage necessary to produce a corona discharge. The CIGAR
produced ion pair concentrations of approximately 1.5×105 to
2.0×105cm-3.
As for Fig. a, except that the colours indicate
the ion conditions under which the experiments were performed: ion free
(neutral), natural ion concentrations (GCR) and enhanced ion concentrations
(CIGAR).
![]()
SO2 mixing ratios of 20 ppbv correspond to approximately
5×1011 molecules cm-3, at 1013.25 hPa
and as the aqueous phase oxidation proceeds rapidly, one would not
necessarily expect an influence of, at most, a few hundred thousand ions
cm-3 on the aerosol growth. Nevertheless, a few experiments were
performed to confirm that this was the case, and that no unexpected surface
charge effects could influence the results. The modelled and measured aerosol
growth are again plotted in Fig. , with the colours
indicating under which conditions the experiments were performed. All except
five of the experiments were performed under natural GCR conditions, however
the CIGAR and neutral points show no sign of any systematic bias, confirming
that ions do not have a measurable influence on SO2 uptake and
oxidation.
Conclusions
Experiments have been performed to investigate the aqueous phase oxidation of
SO2 by O3 in cloud droplets. The observed aerosol growth
was compared with the growth predicted by a model using the reaction rate
constants recommended by , which are widely used in the modelling
of these reactions. It was shown that the modelled growth under acidic seed
aerosol conditions generally agrees with the observed growth, suggesting that
the rate constants of , which were measured in bulk solutions, do
indeed accurately represent the chemistry occurring in dispersed aqueous
systems. This contrasts with the findings of at least two previous studies
, and seems to confirm the suggestion made by
those authors that the observed disagreement resulted from the presence of
contaminants such as ammonia. Furthermore, we have performed what we believe
to be the first laboratory-based measurements of the aqueous phase oxidation
of SO2 in a population of super-cooled cloud droplets, and
confirmed that the generally accepted temperature dependence of the oxidation
reactions, measured by , is consistent with our
experimental results at temperatures of -10 ∘C.
Experiments were also performed for partially to essentially fully
neutralised aerosol (ammonium sulphate), however as there is likely to be
a large, unquantifiable source of ammonia from the chamber walls during the
cloud formation, these experiments can only be used to illustrate the
sensitivity of modelled aerosol growth to ammonia.
The agreement between the model and the experimental data presented here
illustrates that chamber experiments, performed under well defined
conditions, may be used to determine aqueous phase reaction rate constants.
Additionally, such experiments may be used to determine reaction rate constants in
super-cooled droplets, which are important for atmospheric applications but
impossible to measure in bulk solutions.