An air ion spectrometer (AIS) was deployed for the first time at the Concordia station at Dome C (75
Air ions, also known as atmospheric ions, are electric charge carriers in the atmosphere, ranging from primary ions (most have
a mobility diameter smaller than 0.8–1
Upon the ionization of air molecules by ionizing radiation, electric charges are created. After undergoing a series of chemical and
dynamic processes with trace gases and pre-existing aerosol particles, electric charges that survive the initial recombination and
other loss mechanisms are stabilized in the form of charged aerosol particles (Chen et al., 2016). Charged nanoparticles in the
mobility size range of 1.7–7
Carslaw et al. (2013) suggested that aerosol-related uncertainties in global models could be best reduced through the study of natural
aerosols in environments with negligible anthropogenic influence. Antarctica is such an environment. Long-term time series of particle
number concentrations have been published both from the coastal Antarctica, including the Neumayer station (Weller et al., 2011), and
from the upper plateau including the South Pole (e.g. Samson et al., 1990). Number size distributions of aerosol particles have been
measured during short-term campaigns, mainly at coastal stations (e.g. Ito, 1993; Koponen et al., 2003; Virkkula et al., 2007; Asmi
et al., 2010; Pant et al., 2011; Belosi et al., 2012; Kyrö et al., 2013; Weller et al., 2015), but also on the upper plateau at the
South Pole (e.g. Park et al., 2004). Hara et al. (2011) presented particle number size distributions measured on the coast of Queen
Maud Land at the Japanese station Syowa in the period 2003–2005. At the Norwegian Troll station in the inner region of Queen Maud Land, particle
number size distributions in the size range 30–800
In this work, we present a first set of results on air ion observation at Dome C. Our aims are to characterize the key features of air ions at this Antarctic site, including the seasonality of their concentrations, and to analyse the variability of air ions in relation to NPF. Previously, particle growth during NPF processes has been studied in terms of growth rates (GRs) using two methods: the appearance time method (Lehtipalo et al., 2014) and the mode-fitting method (Dal Maso et al., 2005). Here we compare GRs determined using these two methods.
The analyses in this work were based on ambient data collected from the Concordia station (75
The number size distribution of air ions was measured with an air ion spectrometer (AIS) during the campaign period. The AIS employs
two cylindrical multi-channel aspiration-type analysers and a high sample flowrate (60
The air sampled in the AIS is split into two equal streams. On the way to the analyser, each stream passes through a sample preconditioner, i.e. a corona charger coupled with an electrical filter. Sample preconditioners are turned on only during the measurement of background signals, where corona chargers produce charger ions of an opposite polarity to the subsequent analysers, so that clusters and particles in each sample stream are either neutralized or assigned an opposite polarity to the analyser and therefore generate no signal in the detection system. During the campaign, each AIS measurement cycle was composed of 1 min background probing and 4 min ambient sampling. Preconditioners are turned off for ambient sampling. Sample streams pass directly into the respective analysers, where air ions are segregated based on their electrical mobility into different channels. The analyser used in the AIS is a variant of the differential mobility analyser (DMA). Unlike common cylindrical DMA, in which ions are collected at the inner electrode and, via altering the voltage applied on the electrodes, ions of different mobility are measured (Hinds, 1999), the outer electrode in the AIS analyser serves the collecting role and ions of different mobility are collected simultaneously by different channels. The operation of the AIS analyser is based on electrical repulsion. The sample flow is introduced near the inner electrode and sheath flow near the outer one. The outer electrode of the AIS analyser is divided into 21 insulated sections, each of which is connected to an electrometer as the detector. Coupling the outer electrode design with a specially shaped inner electrode, which comprises several cylindrical sections biased at different potentials, the analyser is able to perform a concurrent classification of ions into the 21 measuring channels. More detailed technical descriptions of the instrument are presented by Mirme et al. (2007) and Mirme and Mirme (2013).
The AIS assumes the normal temperature and pressure (NTP) condition and has a total sample flowrate of 60
In this work, to be comparable with particle data, air ion data are presented in Millikan mobility diameters. The conversion of
electrical mobility to sizes is based on the Stokes–Millikan equation (e.g. Hinds, 1999), using measured ambient temperature and
pressure. After the flow correction, the AIS has a measurable mobility size range of 0.9–48
The median size distribution of positive ions measured by the AIS on an event-free day (16 January 2011). The measured number size distribution of this day is shown in the contour plot.
The primary feature of an ambient AIS spectrum contains a persistent band of high ion concentrations at lowest sizes (Fig. 1), which is
typically known as the cluster ion band. The upper boundary of this band typically lies at around 1.7
A differential mobility particle sizer (DMPS), the same used by Järvinen et al. (2013), was responsible for recording the number
size information of total aerosol particles. The DMPS classifies particles of mobility sizes between 9 and 550
The ambient air temperature (
An automatic depolarization lidar (532
To assist our analyses, GRs, condensation sinks (CS), and formation rates of ions were determined from the measured air ion and
total aerosol particle spectra. The GR characterizes how rapidly particles enlarge in size due to condensational growth and
coagulation, which typically has a unit of
The concentration of ions/aerosol particles evolves in both time and particle size. We used two approaches in determining GRs: the mode-fitting method (Dal Maso et al., 2005) and the appearance time method (Lehtipalo et al., 2014). The mode-fitting method follows the concentration change in the time dimension and the appearance time method follows the change in the size dimension.
In the mode-fitting method, at each time stamp of the measurement, the representative size of the aerosol population is determined by
fitting a normal distribution to the measured concentration distribution along the logarithm of sizes with a base of 10. The mode of
the fitted curve is transcribed back to linear scale and taken as the representative size of the particle population measured at this
moment (a more detailed description of the method has been presented by Dal Maso et al., 2015). In contrast, in the appearance time
method, for each size (the geometric mean size of a measurement size bin), one determines the time (the appearance time) at which the
particle population is considered to reach this size, based on the measured concentration evolution in time (Lehtipalo et al.,
2014). This procedure is repeated for each measurement size bin. In this study, we defined the moment at which the concentration rises
to 75 % of the maximum as the appearance time for each size. The GR was then calculated as the slope of a linear fit to the size
data as a function of time. In addition, we also determined the instantaneous GR (
Sulfuric acid is a key chemical species in forming aerosol particles in the ambient air (Kulmala et al., 2014). We determined CS for
sulfuric acid vapours from the number size distribution measured by the DMPS, based on the method described by Pirjola et al. (1998)
and Dal Maso et al. (2002), using the following equation:
The diffusion coefficient of vapour molecules is determined using the Fuller's model (Poling et al., 2004; Tang et al., 2014), which
describes a binary gas system of species A diffusing in B:
The formation rates of 2 nm ions (
During the campaign period (330 days in total), there were 287 days with valid air ion measurements, i.e. valid air ion data were
collected on nearly 87 % of the measurement days. We were able to identify NPF, wind-induced ion formation,
and cloud activation events from the valid measurements. Altogether, NPF events were observed on 32 days, wind-induced ion formations
on 36 days, and cloud activations on 7 days with two certain cases. For the NPF events, 20 cases were classified as suppressed NPF events,
which were characterized by no clear particle growth beyond 10
Seasonality in the median
A clear seasonality was found on these event-free days in the cluster ion concentration (Fig. 2a). The cluster ion concentration was
highest during the warm months, with a maximum in February. The median cluster ion concentration was typically below
100
The daily-median cluster ion concentration at Dome C was observed to be higher on NPF event days compared with event-free days. A similar phenomenon was also seen in a boreal forest environment at SMEAR II station in southern Finland (Fig. S2). Such a connection between cluster ion concentrations and NPF occurrences may imply that compared with event-free days, NPF event days probably had higher concentrations of vapours that are able to contribute to both cluster ion formation and NPF. In the Antarctic autumn (February and March), the value of CS tended to be higher on NPF days compared with event-free days (Fig. 2b). Since a large CS is indicative of a higher atmospheric sink for low-volatility vapours, this observation suggests a particularly strong source rate of such vapours during this time of the year, especially on NPF days. Interestingly, the opposite pattern in CS was seen in the Antarctic spring (October and November, Fig. 2b).
Consecutive NPF events observed during the period 9–15 March 2011.
Consecutive multi-mode formation and growth events observed during 12–16 February 2011.
Comparison of GR for nine growth modes (GR
One of the major features observed in the AIS spectra at Dome C was the process of NPF and subsequent growth of newly formed charged
clusters/particles. The smooth growth that lasts for several hours can imply a homogeneous condition in the sampled air (Hirsikko
et al., 2007; Manninen et al., 2010). Seven clear NPF events were seen on consecutive days during the period 9–15 March 2011, with the initial
step traceable down to the cluster ion band (Fig. 3b). All these NPF events occurred during westerly winds, apart from the first one on 9
March (Fig. 3e). This NPF event was associated with winds from the contaminated sector (10–90
Figure 4 shows examples of multi-mode NPF events that were observed during 12–16 February 2011. One of them (14 to 15 February) had
three concurrent NPF and growth events. The first of them was initially captured by the DMPS at around 03:00 UTC, and at a size of
around 15
The GRs determined by the appearance time method tended to be higher than those determined by the mode-fitting method (Fig. 5). This difference probably originates from the foundations that these two methods rest on. Both methods were developed to treat the measured number size distribution data of ions/particles that vary in time, size, and ion/particle number concentrations. The appearance time method follows the concentration change as a function of time along the particle size dimension, whereas the mode-fitting method tracks the concentration change as a function of particle size along the time dimension. Accordingly, the appearance time method is able to preserve the growth features related to particle size, whereas the mode-fitting method characterizes better the evolution of particle growth with time.
The NPF event that occurred on 12 March between 00:00 and 06:00 UTC had clearly two simultaneously growing modes, with corresponding
particle GRs marked by
The GR dependence on size. Sizes are in mobility diameters and GRs of ions and particles are presented as
discrete time derivatives of the change in mobility diameters (
A cloud activation event observed on 20 January 2011.
A wind-induced ion formation event observed on 3–4 July 2011.
Even though the GR is often calculated as the slope of a linear fit to the size data as a function of time (Yli-Juuti et al., 2011;
Lehtipalo et al., 2014), like in Fig. 5, it is not always appropriate to express the change in sizes along time by a linear
proportionality, e.g.
GRs estimated from linear fittings with root-mean-square errors expressed as uncertainties for the nine growth modes
shown in Fig. 5.
The formation rate of 2 nm positive ions (
A cloud activation event initiated at around 14:00 UTC was observed on 20 January 2011 (Fig. 7). In general, such events are
characterized by a disappearance of aerosol particles from the measured particle size range (Komppula et al., 2005; Kyrö et al.,
2013). Additionally, a sudden drop in the cluster ion concentration has been reported as a feature for cloud activation events
(Lihavainen et al., 2007). We observed similar connections between cluster ions and cloud activation at Dome C (Fig. 7a–c). Moreover,
we found that the cloud activation event was accompanied by a burst of ions in the 8–42
Ion concentrations as a function of wind speed:
By following the approach introduced by Komppula et al. (2005) based on DMPS measurements, it can be estimated that particles larger
than about 110
Ion formation events during strong wind episodes have been observed at Aboa in Antarctica (Virkkula et al., 2007), as well as at the
high-altitude site on Jungfraujoch in Switzerland (Manninen et al., 2010). At Dome C, we observed wind-induced ion formation especially
during the dark months (15 cases during May–August). An example of such an event, observed during 3–4 July 2011, illustrates the
close connection between the ion formation and wind speed (Fig. 8): ions generated by a strong wind were mainly in the cluster ion size
range, even though a large number of ions were also apparent in the 1.9–10
Under strong wind conditions, small snowflakes and ice crystals in the surface layer of the accumulated snow on the ground can be
resuspended by turbulence and be shattered further by their collisions (Pomeroy and Jones, 1996). Vapours adsorbed on and trapped in
these snowflakes and ice crystals can be released into the air to replenish vapours in the air that are capable of participating in
cluster ion formation and possibly also NPF. This resuspension process also assists the escape of vapours trapped beneath
the surface snow layer on the ground. Moreover, owing to the sudden drop of the surrounding vapour pressure, gaseous molecules of water
and other trace species may also be freed from the resuspended particles by sublimation (Pomeroy and Jones, 1996). Ionizing radiation
produces primary ions, which are either lost through ion–ion recombination or transformed into more stable air ions by nucleation or
condensation (Chen et al., 2016). A small concentration of ions slightly larger than the cluster sizes could be observed in connection
to the high wind speeds between 06:00 and 12:00 UTC (Fig. 8b and c). As the wind speed increased further after 12:00 UTC, the vapour
replenishment was probably amplified, leading to an ion burst in the size range of 0.9–10
Turbulent conditions might enhance the collection of electric charges by the shattered snowflakes and ice particles via a charge
transfer from initial charge carriers, contributing to the formation of an ion burst. In addition, the shattered particles might gain
electric charges through friction charging. However, we think that these two pathways of ion formation are not likely to contribute to
the ion burst captured by the AIS. In principle, the shattering of resuspended snowflakes and ice particles mechanically by turbulence
results in the formation of particles of smaller but random sizes. If this mechanism had produced nano-sized particles that
subsequently became electrically charged either by charge transfer or friction charging, our AIS should have detected some of them and
have shown an unsystematic spectrum, i.e. ions of random sizes and concentrations. Yet, conversely, the AIS showed high
concentrations of ions of only small sizes, and hardly anything of sizes larger than 2–3
Putting together all the 36 wind-induced ion formation events, the logarithm of the ion concentration exhibited linear relations to
the wind speed (Fig. 9), as also observed at Aboa (Virkkula et al., 2007). For both cluster ions and ions in the size range of
1.9–10
Based on 1 year of air ion observations with an AIS at Dome C, Antarctica, we found that this site has a rich
set of ion processes, especially when considering its inland location on the largest ice desert on the Earth – the Antarctic
Plateau. NPF, wind-induced ion formation, and ion production and loss associated with cloud/fog formation were
the main processes that were found to modify the number size distribution of air ions at this high-altitude site. On event-free days,
i.e. on days without the above-mentioned processes or other anomalies, concentrations of cluster ions (0.9–1.9
GRs determined using the mode-fitting method and appearance time method were used to characterize the NPF processes. Comparison between these two methods suggests that the GRs derived from the appearance time method work better in depicting the cases
with a fast particle growth, whereas GRs determined from the mode-fitting method appeared to be more suitable for describing cases with
a slow particle growth. We found that the change in particle diameters did not usually increase linearly with the time. Therefore, we
derived the instantaneous GR (
Ion production in relation to cloud/fog formation in the size range of 8–42
The air ion data used in this work were limited to the positive polarity due to a technical malfunctioning of the negative analyser. Further ambient measurements on air ions would be valuable to be carried out at Dome C and other sites on the Antarctic Plateau, not only to reveal possible differences between positive and negative ion properties and their connections to the ion and aerosol processes, but also to understand the mechanisms behind the ion formation related to the cloud/fog formation or wind episodes and to acquire a better characterization of atmospheric NPF in Antarctica. In the future in addition to air ions, the properties of neutral clusters and particles also need to be probed in order to understand the relative importance of ions and neutrals in atmospheric NPF at Dome C, and to characterize the comparability of the roles of ions and neutrals in atmospheric NPF observed at Dome C and at other sites around the globe.
Data used in this work can be found
via
The authors declare that they have no conflict of interest.
This work received financial support from the Academy of Finland (project nos. 264375 and 264390), the NordForsk funded Nordic Centre
of Excellence CRAICC (Cryosphere–atmosphere interactions in a changing Arctic climate, project no. 26060), and the Academy of
Finland's Centre of Excellence Programme (Centre of Excellence in Atmospheric Science – From Molecular and Biological Processes to the
Global Climate, project no. 272041). Funding for this research was also provided by Consiglio Nazionale delle Ricerche and PNRA
(projects 2009/B.04 and 2010/A3.05). We appreciate the support of the IPEV/PNRA Project “Routine Meteorological Observation at
Station Concordia”,