Introduction
Frost flowers on sea ice are ice crystals that contain brine and sea salts.
They appear often during winter–spring on the surface of new and young sea
ice in polar regions. Frost flowers play an important role as an interface
among the atmosphere, sea ice, and ocean. Key conditions for formation and
growth of frost flowers are (1) a cold atmosphere (below -15 ∘C)
and (2) weak–calm winds (Perovich and Richter-Menge, 1994; Martin et al.,
1995, 1996; Style and Worster, 2009; Roscoe et al., 2011). Vertical gradients
of temperature and relative humidity near the sea-ice surface are crucially
important for the appearance of frost flowers. Frost flowers can be formed by
condensation of water vapor on nodules on new and young sea ice. Results of
earlier studies (e.g., Domine et al., 2005; Douglas et al., 2012) show that
water vapor is initially supplied to the atmosphere with sublimation or
evaporation occurring from the warm sea-ice surface. Strong vertical
gradients of air temperature above the sea-ice surface engender
supersaturation of water vapor near the sea-ice surface; the gradients then
induce condensation of water vapor (i.e., frost flower formation). The
concentrated seawater (i.e., brine) is present on new and young sea ice.
Consequently, brine with high salinity migrates upwardly and gradually on
frost flowers (Perovich and Richter-Menge, 1994; Martin et al., 1996; Roscoe
et al., 2011). Under cold conditions, solutes with lower solubility in brine
can be precipitated in and on sea ice and frost flowers depending on the
temperature. Earlier investigations (e.g., Marion, 1999; Koop et al., 2000;
Dieckmann et al., 2008, 2010; Geilfus et al., 2013) have revealed that
several salts can be precipitated at -2.2 ∘C (ikaite,
CaCO3⋅2H2O), -8.2 ∘C (mirabilite,
Na2SO4⋅10H2O), -15 ∘C (gypsum, CaSO4),
-22.9 ∘C (hydrohalite, NaCl⋅2H2O), -28 ∘C
(NaBr⋅5H2O), -33 ∘C (sylvite, KCl), -36 ∘C
(MgCl2⋅12H2O), and -53.8 ∘C (antarcticite,
CaCl2⋅6H2O). The salt precipitation occurring here at lower
temperatures causes changes to sea-salt ratios in brine and frost flowers,
known as sea-salt fractionation. In fact, early works have shown that
sea-salt ratios in frost flowers differed from those in brine and seawater
(Rankin et al., 2000, 2002; Alvarez-Aviles et al., 2008; Douglas et al.,
2012). It has been suggested that salt precipitation occurs in brine, because
sea-salt ratios cannot be changed during sea-salt fractionation on frost
flowers.
Frost flowers have a fine structure. Earlier field and laboratory experiments
indicated frost flowers as less fragile, even under strong winds, in spite of
their fine structure (Obbard et al., 2009; Roscoe et al., 2011). In addition,
results of model studies have implied that blowing snow contributes
importantly to atmospheric halogen chemistry (Yang et al., 2010; Abbatt et
al., 2012; Lieb-Lappen and Obbard, 2015). Because of the lower number density
of aerosol particles in polar regions, especially in Antarctic regions,
emission of sea-salt particles from sea-ice areas is an important aerosol
source during winter–spring (e.g., Wagenbach et al., 1998; Rankin et al.,
2000; Hara et al., 2004, 2011, 2012, 2013). In fact, sea-salt particles
released from sea-ice areas are dispersed from the boundary layer to the free
troposphere (up to ca. 4 km) over Syowa Station, Antarctica, through
vertical motion by cyclone activity (Hara et al., 2014), and into the
interior (Dome F Station and Concordia Station) of the Antarctic continent
(Hara et al., 2004; Udisti et al., 2012). Vertical transport of the sea-salt
particles originating from sea ice can act as a supply of cloud condensation
nuclei (CCN) and ice nuclei (IN) in the upper boundary layer–free
troposphere (Twohy and Poellet, 2005; Wise et al., 2012; DeMott et al.,
2016). Because of the horizontal transport of sea-salt particles into the
Antarctic Plateau, Na+ records in ice cores taken in the inland area
have been used recently as a proxy of the sea-ice extent (e.g., Wolff et al.,
2003, 2006).
As described above, sea-salt fractionation proceeds on new and young sea ice.
For that reason, sea-salt ratios in sea-salt particles (or aerosols) released
from sea-ice areas differ from those of the bulk seawater ratio (Hara et al.,
2012, 2013). For instance, a remarkable amount of Mg-rich sea-salt particles and aerosol particles
containing MgCl2 and MgSO4 were identified at Syowa Station during winter–spring because precipitation of mirabilite
and hydrohalite engenders Mg enrichment in sea-salt particles (Hara et al.,
2012, 2013). Because the relative humidity of Mg-salt deliquescence, such as
that of MgCl2, is lower than that of NaCl (e.g., Kelly and Wexler,
2005), sea-salt fractionation can engender modification of aerosol
hygroscopicity, which is closely related to phase transformation,
heterogeneous reactions, and abilities to act as cloud condensation nuclei
and ice nuclei. According to results of laboratory experiments conducted by
Koop et al. (2000), Br- can be enriched in frost flowers by sea-salt
fractionation. Reportedly, Br- enrichment occurs slightly in frost
flowers in the Weddell Sea, Antarctica (Rankin et al., 2002). Limited Br-
enrichment to Na+ has also been observed in a few samples collected at
Barrow, Alaska, although no Br- enrichment was detected in samples of
frost flowers and brine (Douglas et al., 2012). Additionally, results of some
earlier studies have described non-significant Br- enrichment in frost
flowers at Barrow and Hudson Bay (Alvarez-Aviles et al., 2008; Obbard et al.,
2009). Therefore, many issues remain with respect to the sea-salt and halogen
chemistry of aerosols and frost flowers. They demand further measurement
results and discussion. In addition to sea-salt fractionation,
sea-salt ratios in frost flowers and aerosols can be altered gradually by
heterogeneous reactions in a process known as sea-salt modification.
Furthermore, frost flowers have large specific surface areas:
63–299 cm2 g-1 (mean, 162 cm2 g-1) at Hudson Bay,
Canada (Obbard et al., 2009) and 185 (+80, -50) cm2 g-1 at
Barrow, Alaska (Domine et al., 2005). Because of their larger surface areas,
earlier studies have assessed the potential of frost flowers for use as
reaction sites (e.g., Kaleschke et al., 2004, references in Abbatt et al.,
2012). Sea-salt modification in sea-salt aerosols, and sea salts in and on
frost flowers and sea ice can act as potential sources of gaseous reactive
species. For instance, gaseous reactive halogen species (e.g., Br2,
HOBr, Br, and BrO) induce depletion of ozone and mercury near the surface in
both polar regions during the polar sunrise (Barrie et al., 1988; Schroeder
et al., 1998; Foster et al., 2001; Ebinghaus et al., 2002).
Locations of sampling and sea-ice conditions around Siorapaluk and
photographs of new sea-ice conditions off Siorapaluk (taken from helicopter
on 7 March 2014). Black, red, and blue dashed lines show locations of
sea-ice break in November, 2013, and on 10–14 February 2014 and 1 March 2014,
respectively. The letters A, B, and C illustrate the general locations in the corresponding maps and
photographs.
To elucidate the atmospheric impact of fractionated sea-salt particles and
their relationship with sea-salt particles in the atmosphere and frost flowers
on sea ice, one must ascertain (1) the chemical properties (e.g.,
concentrations, ratios, and pH) of frost flowers and brine, and (2) the
physical and chemical properties of aerosols (e.g., size distribution,
components, and mixing states) above seasonal sea ice with frost flowers.
Despite their importance, simultaneous observations and measurements of
aerosols and frost flowers over seasonal ice areas with known frost flower
occurrences have not been reported for polar regions, although sampling and
observations of frost flowers have been conducted in the Arctic (e.g.,
Alvaraz-Aviles et al., 2008; Douglas et al., 2012) and Antarctic (Rankin and
Wolff, 2000; Rankin et al., 2002). Using data from simultaneous measurements
and sampling of aerosols, frost flowers, and brine around northwestern
Greenland during winter–spring, this study was conducted to elucidate
sea-salt cycles in seasonal sea-ice areas and related phenomena, such
as sea-salt fractionation on the sea-ice surface including frost flowers,
brine, and snow, their aging processes, and the release of fractionated
sea-salt particles into the atmosphere.
Samples and analysis
Sampling sites and conditions
Figure 1 shows the locations where simultaneous observations of frost
flowers, brine, seawater, and aerosols were made on a new and young sea-ice
area in Robertson fjord near Siorapaluk in northwestern Greenland from
mid-December 2013 through mid-March 2014. In the fjord, the open sea surface
appears during summer. Sea ice formed gradually from October 2013 in the
fjord. Sea ice flowed out several times from the fjord by the action of sea
tides, heaving, and strong winds (Fig. 1) before and during the measurements
taken for this study. Moreover, sea ice with different ages was present in
the fjord because the locations of sea-ice breaks differed in each case, as
presented in Fig. 1. For the present study, we define sea-ice ages in the
fjord as new, young, old, and very old, depending on the sea-ice age. We
chose sites I–III (new–young sea ice; less than 1 cm to ca. 35 cm
thickness) as sampling sites of aerosols, frost flowers, brine, and seawater.
Site I was approximately 2 km distant from Siorapaluk. The sea ice on site I
flowed during 10–12 February 2014 (shown as a red dashed line); then it
refroze. Several days to a week prior (approximately mid-February), an
open lead appeared around site II. Thereafter, new sea ice was formed. The
refrozen lead width was 2–6 m near sampling sites. The sea surface appeared
off Siorapaluk on approximately 1 March from sea-ice breakage and strong winds
(shown as a blue dashed line). Then, new sea ice formed again (photographs,
Fig. 1). On 3 and 5 March, we traversed the new sea-ice area (blue
dashed line in Fig. 1) to observe the sea-ice conditions and the appearance
of frost flowers on the sea ice. Then, we chose sampling sites (IIIa and
IIIb) that were safely accessible. The new sea ice at sites IIIa and IIIb
(less than 1 cm thickness) was a few days old. We accessed sampling sites
with frost flowers on foot and by dog sledge from Siorapaluk.
Sampling of frost flowers, brine, snow, and seawater
Sampling of frost flowers, brine (slush), and seawater was conducted from
20 February 2014 through 3 March 2014 on sea ice near Siorapaluk. Frost
flowers were taken from the sea-ice surface using a clean stainless steel
shovel. During the campaign, snowfall and blowing snow occurred occasionally.
On the frost flower and slush layer at sites I and II, snow was present,
slightly, to the extent to that the fine structure of frost flowers was
identified clearly. All bodies of the frost flowers above the sea-ice surface
were collected if the forms of frost flowers were mature and dry. Young frost
flowers were collected if the frost
flowers were also coated with brine water. When snow was present on the
frost flower and slush layer, some frost flower samples can contain slightly
snowy pieces because of the difficulty of separation. Brine samples were
collected by shaving off a thin layer of the sea-ice surface, and coating the
brine water in proximity to the frost flowers. With the exception of new
sea ice at site III with thicknesses of a few centimeters, slush was sampled on
sea ice where frost flowers formed. Because it was difficult to collect
only brine from the slush layer, the slush layer was sampled as “brine
samples” in this study. Snow on sea ice was also taken using a clean
stainless steel shovel from the location with snow accumulation (< 3 cm
depth) without frost flowers at sites I and II. Pieces of frost flowers,
brine (slush) samples, and snow were moved into individual polyethylene bag
(Whirl-pak; Nasco). Using a dropper, seawater samples were collected in
polypropylene bottles from a crack in the sea ice or through a hole we made in the
sea ice. All samples were melted at ambient temperature. The H+
concentration (i.e., pH) of the liquid sample was measured using a portable
pH meter (B-212; Horiba Instruments Ltd.). Then residue of the sample was
transferred to polypropylene bottles. The samples in the bottles were kept in
conditions below -20 ∘C in Greenland. Then, samples were
unfrozen during transport (ca. 3 days) to our laboratory in Japan because of
the difficulty of carrying frozen samples in an airplane. After the bottled
liquid samples were transported to Japan, all were kept frozen in a cold room
until chemical analyses were conducted.
Aerosol sampling and measurements
Aerosol measurements and direct sampling were conducted over seasonal sea ice
around Siorapaluk, Greenland from 17 December 2013 through 7 March 2014.
Aerosol number concentrations were measured at flow rate of
2.83 L min-1 using a portable optical particle counter (OPC, KR12A;
Rion Co. Ltd.). The measurable size range was Dp>0.3, > 0.5,
> 0.7, > 1.0, > 2.0, and > 5.0 µm. The OPC packed in an
insulator box was set at ca. 1 m above the sea-ice surface using a tripod.
Aerosol number concentrations were recorded every 23–25 s, corresponding to
1 L of air volume, during direct aerosol sampling. Details of OPC
measurements were presented by Hara et al. (2014).
Direct aerosol sampling was done using a two-stage aerosol impactor.
Carbon-coated collodion thin films supported by a Ni micro-grid (square 300
mesh; Veco Co.) were used as sample substrates in this study. The cut-off
diameters (aerodynamic diameter) of the impactor were 2.0 and
0.2 µm at a flow rate of ca. 1.2 L min-1. The impactor was
set at ca. 1 m above the seasonal sea-ice surface, similarly to OPC
measurement. Direct aerosol sampling was conducted for 10–15 min depending
on the aerosol number concentration. Aerosol samples were kept in
polyethylene capsules immediately after aerosol measurements and sampling.
The polyethylene capsules with aerosol samples were packed into polyethylene
zipper bags. All bags with aerosol samples were put into an airtight box
together with a desiccant (Nisso-Dry M; Nisso Fine Co., Ltd.) until analysis
at room temperature to prevent humidification that can engender morphological
change and efficient chemical reactions, as described by Hara et al. (2002,
2005, 2013, 2014). All aerosol samples described as a result of this study
were analyzed and observed within 1 year of sampling.
Meteorological data (air temperature, relative humidity, air pressure, wind
direction, and wind speed) were measured using an automatic weather station (AWS; HOBO
U30-NRC Weather Station, Onset Computer Corp.), which was set on the coast
near Siorapaluk and ca. 1 km away from the village. Meteorological data were
recorded on the data logger of the automatic weather station with time resolution
of 5 min. A Thermo Recorder (TR-7Wf; T and D Corp.) and thermosensor (TR1106;
T and D Corp.) were used for measurements of the temperatures of seawater,
slush layer (brine) on sea ice, base of frost flowers on the slush layer, and
in the atmosphere above the top of the frost flowers.
Variation of air temperature, relative humidity, and wind speed at
Siorapaluk.
Sample analysis
Analysis of frost flower, brine, snow, and seawater
Re-frozen samples of frost flower, brine, snow, and seawater were melted at
ambient temperature before chemical analysis. Concentrations of ion species
(Na+, K+, Mg2+, Ca2+, Cl-, NO3- and
SO42-) in frost flowers, brine, and seawater were measured using ion
chromatography (ICS 2100; Thermo Fisher Scientific Inc.) after 103-fold
dilution by ultrapure water, whereas those in snow were determined without
dilution. A guard column (IonPac CG12; Thermo Fisher Scientific Inc.), column
(IonPac CS12; Thermo Fisher Scientific Inc.), and a 20 mM CH3SO3H
eluent were used for the cation measurement, and a guard column (IonPac AG14;
Thermo Fisher Scientific Inc.), column (IonPac AS14; Thermo Fisher Scientific
Inc.), and 3.5 mM NaOH eluent were used for anion measurements.
Concentrations of Br- in frost flowers and brine were measured using an
ion chromatograph–mass spectrometer (IC-MS) after 106-fold dilution
using ultrapure water. Ionic contents in samples were separated using ion
chromatography with a guard column (IonPac AG11-HC; Thermo Fisher Scientific
Inc.), a column (IonPac AS11-HC; Thermo Fisher Scientific Inc.), and gradient
KOH eluent (4–36 mM), and were injected into a mass spectrometer (6100
series single quadrupole LC/MS; Agilent Technologies Inc.). The detection
limit of Br- in IC-MS was 0.9 ng L-1. Iodine concentrations in
frost flowers and brine were measured after 103-fold dilution using
ultrapure water with an inductively coupled plasma mass spectrometer (ICP-MS,
7700 series single quadrupole ICP-MS; Agilent Technologies Inc.). The
radio frequency (RF) power and flow of carrier Ar gas were, respectively, 1550 W and
0.80 L min-1 in ICP-MS analysis. The detection limit of iodine used
for this study was 17 ng L-1. Samples of frost flowers and brine were
injected into ICP-MS after melting. Because ICP-MS can provide only the
elemental concentrations in samples, iodide (I-), and iodate
(IO3-) were not separated in this study. Details of IC-MS analytical
procedures are described elsewhere by Motohiro Hirabayashi (personal communication, 2017). We
calculated concentration factors (CF-X) with molar concentrations of chemical
species (X) in samples divided by the concentration of X in seawater taken at
Siorapaluk to evaluate the concentration processes of the chemical species
during the formation and aging of frost flowers. Analytical errors of
respective analytical methods were estimated from the reproducibility of
determination of standard solutions with concentrations similar to those of
the field samples.
Analysis of individual aerosol particles
Individual aerosol particles on the sample substrate were observed and
analyzed for this study using a scanning electron microscope equipped with an
energy-dispersive X-ray spectrometer (SEM-EDX, Quanta FEG-200F, FEI, XL30;
EDAX Inc.). The analytical conditions were 20 kV accelerating voltage and
30 s counting time. Details of analytical procedures were described by Hara
et al. (2013, 2014). We analyzed 1261 particles in coarse mode (mean, 41
particles per sample) and 6337 particles in fine mode (mean, 192 particles
per sample). In this study, most aerosol-sampled areas on the substrates were
analyzed in coarse mode. Although we attempted to analyze as many coarse
particles as possible, the lower aerosol number concentrations in coarse mode
limit the number of the aerosol particles analyzed in this study.
Photographs of (a) sea-ice and frost flower conditions
on 20 February at site I, (b) frost flowers at site I on
22 February, (c) frost flowers at site II, (d) sea-ice
conditions at site IIIa, (e) frost flowers on 4 March at site IIIa,
and (f) old sea-ice conditions on 2 March immediately after the
storm. Red dotted lines indicate approximate location of margin between old
sea ice and young sea ice.
Relationship among each component with Na+ concentrations of frost
flowers, brine, and seawater taken in this study. Black and red lines
respectively present regression lines of frost flowers and brine. Open black
and red circles respectively present concentrations in frost flowers and
brine. Filled blue stars represent the concentrations of seawater taken
around Siorapaluk. Open blue stars represent the concentrations of seawater
in the literature presented in Table 1. Error bars indicate standard
deviation (1σ) of analytical errors.
Results
Meteorological conditions during the campaign
Figure 2 depicts time variations of air temperature, relative humidity, and
wind speed during our measurement period. The air temperature was -34.2 to
+1.8 ∘C. Conditions below -20 ∘C occurred on days of
year (DOY) 47.5–58.6 (17–28 February 2014) during our intensive sampling
and observations of aerosols and frost flowers. Because of an approaching
cyclone, several strong wind events occurred during the campaign.
Particularly strong winds on DOY 39.7–41.2 (9–11 February) caused breaks
and outflows of sea ice from the front of Siorapaluk (ca. 1 km distant).
Then the seawater started refreezing immediately after the weather calmed.
Sea ice off Siorapaluk (ca. 5–6 km distant) broke and flowed out again off
Siorapaluk due to storm conditions on DOY 58.9–60.3 (28 February–1 March).
Sea ice formed after normal weather conditions returned. Frost flowers
appeared on the new sea ice in both cases.
Conditions of frost flowers and sea ice
Sea ice conditions where frost flowers formed can be categorized as three
types for this study: young ice (site I), younger ice in refrozen leads
between young ice (site II), and new sea ice (sites IIIa and IIIb). Figure 3
depicts photographs of frost flowers observed at site I, categorized as young
ice (Fig. 3a) on 20 February. The sea-ice surface was partly covered with
thin snow. Frost flowers were formed patchily on the ice surface with little
or no snow cover. The sea-ice surface underneath frost flowers was wetted by
brine (sherbet-like, i.e., slush layer). Figure 3b presents a close-up
photograph of frost flowers observed on 22 February at the same site as that
observed on 20 February. Salt crystals deposited on the branches of frost
flower crystals were identified. Frost flowers at site I were covered
completely with snow after the storm on 28 February–1 March. Figure 3c shows
frost flowers observed at site II (Fig. 1). The sea-ice surface underneath
frost flowers was wetted by brine. Figure 3d shows frost flowers observed at
site IIIa (Fig. 1) on 4 March. The site was at the edge of sea ice, close to
open water. The frost flower diameter was approximately 5 mm. The sea-ice
surface was covered with brine water (Fig. 3e). The frost flower crystals
were partially submerged in the brine water. During our campaign
(December 2013 to March 2014), frost flowers were absent on old and very old
sea ice. Although snow covered dry surface of old and very old sea ice
patchily before the storm, bare sea ice without a slush layer was apparent
after the storm (Fig. 3f).
Concentrations of sea salts in frost flowers, brine, seawater, and snow
on sea ice
Figure 4 depicts relationships among the respective components of frost
flowers, brine, snow, and seawater found in this study. Logarithmic plots of
Fig. 4 are shown in Fig. S1. Concentrations of Na+ in frost flowers and
brine were 48–154 mmol L-1, which greatly exceeded the concentration
of seawater collected at Siorapaluk (Na+, 42.5 mmol L-1). The
ratios of Cl- / Na+ in seawater collected at Siorapaluk were
similar to ratios reported in the literature (e.g., Lide, 2005; Millero et
al., 2008), although the SO42- / Na+ ratio at Siorapaluk was
similar to the ratio in Millero et al. (2008) and slightly higher than the
ratios in Lide (2005), as shown in Fig. S2. Moreover, the ratios of
K+ / Na+ and Ca2+ / Na+ in seawater at Siorapaluk
were slightly lower (by ca. 20 %) than those of the literature values.
Differences among the ratios found at Siorapaluk and those of the literature
(e.g., Lide, 2005; Millero et al., 2008) were larger than our analytical
errors (less than 5–6 %), as estimated from the reproducibility of the
determination of standard samples with concentrations similar to those of the
analyzed samples of frost flowers, brine, and snow. In this study, seawater
ratios at Siorapaluk were used in the following analysis and discussion
except for Br- and I, which were referred from earlier reports (e.g.,
Lide, 2005; Millero et al., 2008; Millero, 2016).
The concentration factors of Na+ (CFNa) in frost flowers and
brine were 1.14–3.67, which roughly approximated the results reported in
previous studies (e.g., Rankin et al., 2002; Alvarez-Aviles et al., 2008). By
contrast, the Na+ concentrations in snow samples collected on sea ice
were 1–2 orders lower than that of seawater. The Na+ concentrations of
fresh snow on sea ice were lower than 0.1 mmol L-1. By contrast,
Na+ concentrations were 0.4–3.2 mmol L-1 in the aged snow on
sea ice. High correlation among components was identified in frost flowers,
brine, and snow, as shown below.Frost flowers:
[Cl-]=1.302[Na+]+2.158(R2=0.969)[Br-]=0.0022[Na+]+0.025(R2=0.910)[I]=1.17×10-6[Na+]+5.867×10-4(R2=0.791)[Mg2+]=0.122[Na+]-0.761(R2=0.982)[K+]=0.023[Na+]-0.063(R2=0.988)[Ca2+]=0.023[Na+]-0.019(R2=0.996)
Brine:
[Cl-]=1.421[Na+]-21.35(R2=0.964)[Br-]=0.0020[Na+]+0.06(R2=0.962)[I]=1.10×10-6[Na+]+3.25×10-5(R2=0.842)[Mg2+]=0.114[Na+]-0.8267(R2=0.982)[K+]=0.022[Na+]-0.140(R2=0.988)[Ca2+]=0.021[Na+]-0.023(R2=0.972)
Snow:
[Cl-]=1.315[Na+]+0.02(R2=0.972)[Mg2+]=0.035[Na+]+0.02(R2=0.751)[K+]=0.024[Na+]-0.002(R2=0.997)[Ca2+]=0.026[Na+]+2×10-5(R2=0.994)
Statistics of molar ratios of different sea salts relative to Na+ and Cl-
concentrations in frost flowers, brine, and seawater collected in this study.
K+ / Na+
Mg2+ / Na+
Ca2+ / Na+
Cl- / Na+
SO42- / Na+
Br- / Na+
I / Na+
K+ / Cl
Mg2+ / Cl
Ca2+ / Cl
SO42- / Cl-
Br- / Cl-
I / Cl-
Frost
Mean
0.022
0.113
0.023
1.326
0.029
0.0025
1.83×10-6
0.017
0.086
0.017
0.022
0.0019
1.38×10-6
flower
SD
0.001
0.008
0.001
0.052
0.020
0.0003
3.38×10-7
0.000
0.004
0.000
0.016
0.0002
2.42×10-7
n= 23
median
0.023
0.116
0.023
1.351
0.019
0.0025
1.83 ×10-6
0.017
0.086
0.017
0.014
0.0019
1.35 ×10-6
min
0.020
0.091
0.022
1.217
0.008
0.0015
1.26 ×10-6
0.016
0.074
0.017
0.006
0.0012
9.65 ×10-7
max
0.023
0.121
0.024
1.387
0.070
0.0031
2.95 ×10-6
0.017
0.093
0.018
0.058
0.0023
2.20 ×10-6
Brine
Mean
0.021
0.105
0.022
1.170
0.056
0.0020
1.46 ×10-6
0.019
0.096
0.020
0.049
0.0019
1.35 ×10-6
n= 11
SD
0.001
0.008
0.001
0.246
0.026
0.0002
2.47 ×10-7
0.008
0.035
0.008
0.022
0.0008
5.77 ×10-7
median
0.021
0.109
0.022
1.230
0.056
0.0020
1.37 ×10-6
0.017
0.085
0.017
0.049
0.0017
1.13 ×10-6
min
0.019
0.091
0.020
0.451
0.021
0.0018
1.21 ×10-6
0.016
0.081
0.016
0.016
0.0014
9.56 ×10-7
max
0.022
0.115
0.024
1.346
0.101
0.0024
1.85 ×10-6
0.044
0.202
0.045
0.089
0.0043
2.97 ×10-6
Snow
Mean
0.023
0.076
0.026
1.321
0.037
0.018
0.058
0.020
0.030
n= 15
SD
0.001
0.030
0.002
0.188
0.029
0.002
0.024
0.002
0.026
median
0.024
0.070
0.026
1.303
0.025
0.018
0.054
0.020
0.019
min
0.020
0.032
0.021
1.065
0.010
0.012
0.024
0.014
0.007
max
0.027
0.122
0.029
1.953
0.108
0.020
0.095
0.022
0.101
Seawater
0.020
0.091
0.020
1.227
0.0613
0.0018a
2.76 ×10-7–
0.016
0.074
0.016
0.0500
0.0015a
2.32 ×10-7–1.15 ×10-6 b
n= 2
1.37 ×10-6b
n indicates sample number. a Molar
ratios of Br- were listed using bulk seawater ratio (Lide, 2005).
b Iodine (I-+ IO3-) concentration in seawater was
estimated from the concentrations of I- and IO3- measured by
Ito (1997, 1999), Hirokawa et al. (2003), Ito et al. (2003), Chen et
al. (2007), and Horikawa et al. (2016). The estimated iodine concentrations
in seawater are 0.130–0.647 µmol L-1. Upper and lower molar
ratios of I / Na and I / Cl were calculated using the estimated
iodine concentrations and the concentrations of N+ and Cl- from
Lide (2005).
In this study, the Br- and I concentrations in the snow samples were not
determined. High coefficients of determination imply strongly that these
components are sea salts (derived from seawater). When the intercepts are
close to zero, the slopes of the relationships might be close to the ambient
molar ratios. Furthermore, the slopes of the relationships can be biased
positively in the cases of contamination/mixing of NSS species such as
minerals and anthropogenic species, which can be deposited onto surfaces of
frost flowers, brine, and snow. In contrast, the ratios can be biased
negatively in the cases of sea-salt fractionation on sea ice and
depletion/release of the contents in frost flowers, brine, and snow into
the atmosphere. The molar ratios in frost flowers, brine, and snow are
presented in Table 1. With the exception of Mg2+ / Na+ in snow
and I / Na+ in frost flowers and brine, the molar ratios conform to
the slopes. The intercept values and the coefficients of determination in
these ratios are, respectively, larger and smaller than the other ratios.
Higher slopes and molar ratios relative to Na+ and considerable
SO42- depletion were observed clearly in frost flowers (Fig. 4). It
is expected that this SO42- depletion was caused by mirabilite
precipitation. To evaluate the contribution of mirabilite precipitation to
changes in the molar ratios in frost flowers, snow, and brine, sea-salt
ratios relative to Na+ after mirabilite precipitation were compared.
Clear negative slopes (ca. -0.041) for molar concentrations were identified
for the relationship between Na+ and NSS SO42- (not shown), as
discussed by Wagenbach et al. (1998) and Hara et al. (2004). The negative
slope found in this study resembles the slope for sea-salt aerosols in
Antarctica (e.g., Wagenbach et al., 1998; Hara et al., 2004). Sea-salt ratios
relative to Na+ after mirabilite precipitation were estimated using the
negative slope values (R=-0.041). Assuming that mirabilite precipitation
occurred only in sea-salt fractionation, the amount of the depleted
SO42- ([SO42-]SO4-depleted) and the concentration
of Na+ ([Na+]SO4-depleted) after mirabilite
precipitation were calculated using the Na+ concentration in seawater or
brine ([Na+]seawater) and the following
equations.
SO42-SO4-depleted=Na+seawater×RNa+SO4-depleted=Na+seawater-2×SO42-SO4-depleted
Then, the sea-salt ratios after mirabilite precipitation were estimated as
follows.
[X]Na+SO4-depleted=XseawaterNa+SO4-depleted
In Eq. (3), [X]seawater denotes sea-salt concentrations in
seawater or brine other than Na+ and SO42-. The estimated
ratios (i.e., Cl- / Na+, 1.33; Mg2+ / Na+, 0.099;
K+ / Na+, 0.022; Ca2+ / Na+, 0.021) were
consistent with the ambient molar ratios in frost flowers and snow samples.
This coincidence strongly suggests that mirabilite precipitation causes a change
in sea-salt ratios in frost flowers and brine.
Results of t test of the molar ratios of sea-salt components
among frost flowers, brine, and snow.
Frost flowers–brinea
Ratios
Cl- / Na+
K+ / Na+
Mg2+ / Na+
Ca2+ / Na+
SO42- / Na+
Br- / Na+
I / Na+
t value
2.939
3.585
2.920
4.113
-3.449
4.668
3.186
p value
0.006
0.001
0.006
2.54 ×10-4
1.60 ×10-3
5.20 ×10-5
0.003
Ratios
Na+ / Cl-
K+ / Cl-
Mg2+ / Cl-
Ca2+ / Cl-
SO42- / Cl-
Br- / Cl-
I / Cl-
t value
-3.925
-1.501d
-1.470d
-1.515d
-4.035
0.079d
0.176d
p value
0.00045
0.143d
0.151d
0.139d
3.18 ×10-4
0.937d
0.861d
Frost flowers–snowb
Ratios
Cl- / Na+
K+ / Na+
Mg2+ / Na+
Ca2+ / Na+
SO42- / Na+
t value
0.114d
-2.981
5.835
-6.698
-1.083d
p value
0.910d
0.005
1.16 ×10-6
8.20 ×10-8
0.286d
Ratios
Na+ / Cl-
K+ / Cl-
Mg2+ / Cl-
Ca2+ / Cl-
SO42- / Cl-
t value
-0.648d
-3.248
5.492
-5.844
-1.120d
p value
0.521d
0.003
3.32 ×10-6
1.122 ×10-6
0.270d
Snow–brinec
Ratios
Cl- / Na+
K+ / Na+
Mg2+ / Na+
Ca2+ / Na+
SO42- / Na+
t value
1.271d
4.936
-3.175
6.228
-1.730e
p value
0.217d
4.88 ×10-5
0.004
1.95 ×10-6
0.096e
Ratios
Na+ / Cl-
K+ / Cl-
Mg2+ / Cl-
Ca2+ / Cl-
SO42- / Cl-
t value
-1.332d
-0.621d
-3.328
-0.079d
-1.962e
p value
0.196d
0.541d
0.003
0.938d
0.062e
a Thirty-two degrees of freedom. b Thirty-six degrees of
freedom. c Twenty-four degrees of freedom. d, e indicates
“insignificant” and “slightly significant”, respectively.
In addition to mirabilite precipitation, hydrohalite precipitation can change
the molar ratios in aerosols and frost flowers (e.g., Marion et al., 1999;
Koop et al., 2000; Hara et al., 2012). Next, we attempted to compare the
relationships of respective sea salts to Cl- and Mg2+
(Figs. S3–S5, in the Supplement) to understand sea-salt fractionation by
precipitation of mirabilite and other salts, and distribution of the
fractionated sea salts on sea ice. Relationships among Mg2+, K+,
Ca2+, and Cl- in frost flowers well matched those in brine
(Figs. S3–S4). The molar ratios to Cl- were not changed by mirabilite
precipitation without Cl- loss by heterogeneous reactions. The
relationships of sea salts implies that mirabilite precipitation made
important contributions to the change in sea-salt ratios in frost flowers.
Furthermore, Student t tests were applied to assess differences in the
molar ratios among frost flowers, brine, and snow (Table 2). The molar ratios
relative to Na+ in frost flowers were significantly higher than those in
brine (p<0.01), although K+ / Cl-,
Mg2+ / Cl-, Br- / Cl-, and I / Cl- were
not significantly different (p>0.1). When comparing frost flowers and
snow, we found the ratios of K+ and Ca2+ to Na+ and Cl-
in frost flowers to be significantly lower than those in snow (p<0.01).
However, the ratios of Mg2+ / Na+ and Mg2+ / Cl-
in frost flowers were significantly higher than those in snow (p<0.01).
Differences between frost flowers and snow for Cl- / Na+ (or
Na+ / Cl-), SO42- / Na+, and
SO42- / Cl- were not significant (p>0.1). Although
K+ / Na+ and Ca2+ / Na+ in snow were higher than
those in brine, Mg2+ / Na+ and Mg2+ / Cl- in snow
were significantly lower than those in brine (p<0.01). The
SO42- / Na+ and SO42- / Cl- ratios in snow
were slightly lower (though statistically significantly) than those in brine
(0.05<p<0.1). Between snow and brine, the ratios of
Cl- / Na+ (or Na+ / Cl-) were not significantly
different (p>0.1). The statistical analysis indicated that distribution of
the fractionated sea salts was highly heterogeneous in frost flowers, snow,
and brine.
Figure 4 and Table 1 show that the molar ratios in frost flowers, brine, and
snow differed among sampling sites, conditions such as temperature, and the
age of frost flowers (details are discussed herein). The molar ratios in
frost flowers resembled the ratios found by previous studies of frost flowers
and aerosols (Rankin et al., 2002; Douglas et al., 2012; Hara et al., 2012).
The Br- / Na+ ratios found from previous investigations,
however, differed considerably among frost flower sampling sites (Rankin et
al., 2002; Alvarez-Aviles et al., 2008; Douglass et al., 2012). Br-
enrichment in frost flowers was observed in this study, and in previous
studies conducted in the Weddell Sea, Antarctica, by Rankin et al. (2002),
although Alvarez-Aviles et al. (2008) and Obbard et al. (2009) reported that
Br- was not enriched in frost flowers (similar to the seawater ratio) at
Barrow, Alaska, or at Hudson Bay, Canada. Furthermore, the slope of
Mg2+–Na+ in surface snow on sea ice was lower than the seawater
ratio, although Mg2+–Na+ ratios from fresh snow samples with
Na+ concentrations lower than 0.1 mmol L-1 were similar to the
seawater ratio.
Variations of (a) concentration factor of Na+
(CFNa) and (b) molar ratios of
SO42- / Na+ in frost flowers at each sampling site. The red line
indicates the bulk seawater ratio. Error bars indicate standard deviation
(1σ) of analytical errors.
Short-term features of (a) air temperature measured by AWS
(TAWS), air temperature above frost flowers (Tair, ca.
10 cm above the sea-ice surface), temperature of base of frost flowers
(TFF), and (b–h) molar ratios of sea salts in frost
flowers and brine at site I. Tair and TFF were not
measured on 20–22 February. Error bars indicate standard deviation (1σ) of analytical errors. Blue and green dashed lines show the molar ratios after
precipitation of mirabilite and hydrohalite, respectively.
Aging of frost flowers and brine on sea ice
Figure 5 presents variations of CFNa and molar ratio of
SO42- / Cl- in frost flowers at site I. As described above,
aged frost flowers, young frost flowers, and fresh frost flowers were
collected respectively at sites I, II, IIIa, and IIIb. The
values of CFNa of all samples exceeded 1.0, even on new ice at
sites IIIa and IIIb, indicating that concentrated brine was excluded from sea
ice to the sea-ice surface during sea-ice formation before the frost flower
formation (Fig. 5a). Some samples with low CFNa at sites I and II
could have been contaminated with minimal snowfall. The CFNa of frost flowers
at sites IIIa and IIIb was lower than that at sites I and II, which suggests
that sea-salt concentrations and sea-salt ratios of frost flowers varied
depending on the age of frost flowers and sea ice.
To ascertain the changes in sea-salt components that occur along with the
aging of frost flowers, we attempted to monitor the sea-salt components of
frost flowers and brine at site I on 20–28 February 2014. Figure 6 depicts
short-term variations of air temperatures measured by AWS (TAWS),
air temperature above frost flowers (Tair, ca. 10 cm above the
brine/sea-ice surface), temperatures at the base of frost flowers
(TFF), and molar ratios of sea salts
(SO42- / Na+, SO42- / Cl-,
Br- / Cl-, I / Cl-, Mg2+ / Cl-,
K+ / Cl-, and Ca2+ / Cl-) in frost flowers and
brine. Unfortunately, we did not measure Tair and TFF
on 20–22 February. TAWS during 20–28 February was lower than
the temperature for mirabilite formation (ca. -9 ∘C). In addition,
TFF was -18.9 to -21.3 ∘C on 24–28 February.
Figure 6a shows that TAWS and Tair were lower than
-25 ∘C beginning on 23 February. Sea-salt ratios in brine at site I were
distributed similarly to seawater ratios. In contrast to the ratios in brine,
molar ratios of SO42- / Na+ in frost flowers decreased
considerably relative to seawater ratios. Although the molar ratios of
SO42- / Na+ in frost flowers did not change greatly on 26
February, other sea-salt ratios (Mg2+ / Cl-,
K+ / Cl-, and Ca2+ / Cl-) increased
simultaneously. In addition, Na+ / Cl- ratios in frost flowers
dropped slightly on 26 February. Molar ratios of Br- / Cl- and
I / Cl- in frost flowers were higher than the seawater ratio (values
in literature) and brine (Fig. 6d, e). As described above, Br- and I
were enriched in frost flowers at site I. Although ratios of
Mg2+ / Cl-, K+ / Cl-, Ca2+ / Cl-,
and Br- / Cl- increased on 26–28 February, increases in the
ratio of I / Cl- were not clear.
Although the molar ratios of Mg2+ / Cl-,
K+ / Cl-, and Ca2+ / Cl- cannot be changed by
mirabilite precipitation, these ratios increased on 26 February when
TFF dropped approximately to the temperature for hydrohalite
precipitation. Because of analytical errors 2–4 times smaller than the
differences of their molar ratios, it is expected that hydrohalite precipitation results in changes
in the molar ratios. To ascertain the contribution of hydrohalite
precipitation, we attempted to estimate the sea-salt ratios using the
following assumptions.
Only mirabilite precipitated over
20–24 February.
There was no hydrohalite precipitation over 20–24 February.
Hydrohalite precipitation began on 26 February.
Sea salts other than mirabilite and hydrohalite were not fractionated on
20–28 February.
Assuming mirabilite precipitation on 20–24 February,
sea-salt ratios in the residual brine are given as shown in Eq. (3). Under
this assumption, Mg2+ was not precipitated. Therefore,
Mg2+ / Cl- ratios did not change in mirabilite precipitation,
as follows.
Mg2+Cl-Seawater=Mg2+Cl-SO4-depleted
Here, ([Mg2+] / [Cl-])seawater and
([Mg2+] / [Cl-])SO4-depleted respectively denote the
molar ratios of seawater collected at Siorapaluk and the residual brine after
mirabilite precipitation. When hydrohalite precipitation occurred, the
Cl- concentrations in the residual brine
([Cl-]hydrohalite-depleted) decreased gradually with
hydrohalite precipitation as follows.
Cl-hydrohalite-depleted=Cl-seawater-Cl-hydrohalite
Here, [Cl-]hydrohalite and [Cl-]seawater
respectively stand for the amount of Cl- in hydrohalite and the Cl-
concentration in seawater or brine. Therefore, Mg2+ / Cl-
ratios in the residual brine after hydrohalite precipitation
(([Mg2+] / [Cl-])hydrohalite-depleted) are given as
shown below.
Mg2+Cl-hydrohalite-depleted=Mg2+seawaterCl-hydrohalite-depleted
[Mg2+]seawater represents the Mg2+ concentration in
seawater or brine. In fact, [Cl-]hydrohalite in Eq. (6) can
be estimated by substitution of the ambient Mg2+ / Cl- ratio in
frost flowers on 28 February (Mg2+ / Cl-≈0.103) with
[Mg2+] / [Cl-])hydrohalite-depleted. Then, the
other sea-salt ratios after hydrohalite precipitation
(([X] / [Cl-])hydrohalite-depleted) can be given as the
equation below.
XCl-hydrohalite-depleted=XseawaterCl-hydrohalite-depleted
The amount of Na+ in hydrohalite ([Na+]hydrohalite) was
the same as that in [Cl-]hydrohalite. Therefore, the other
sea-salt ratios relative to Na+ after hydrohalite precipitation
(([X] / [Na+])hydrohalite-depleted) can be estimated
using the same procedure. As presented in Fig. 6, the ratios of
Na+ / Cl- in frost flowers were close to the estimated ratios
in hydrohalite precipitation.
SEM image of aerosol particles collected on 1 March 2014 above the
sea-ice area with frost flowers.
Morphology of sea-salt particles
Figure 7 depicts SEM images of aerosol particles collected above the sea-ice
area with frost flowers. Most coarse aerosol particles collected on
1 March 2014 had structures with cuboid-crystal-like materials (bright color
in SEM image) and non-crystal materials around the cuboid-crystal-like materials (gray–dark gray color in SEM image).
Furthermore, most aerosol particles had stains around the particles. The
presence of stains is direct evidence that the aerosol particles had a liquid
surface in the atmosphere. We detected Na, Cl, and Mg in aerosol particles of
this type. Therefore, the particles might be identified as sea-salt
particles. Strong peaks of Na and Cl were identified from cuboid-crystal-like
materials, whereas strong peaks of minor sea salts such as Mg, K, and S were
also obtained from non-crystal materials. Depending on the amount (mass) of
sea salts and water in a sea-salt particle, salts with lower solubility can
exist in a state with a solid core. In SEM observations, however, aerosol
particles were exposed to high-vacuum conditions. They dried up. The
localization of each sea salt in a particle might proceed in a high-vacuum
chamber. Therefore, it is worth noting that the salt distribution in an SEM
image differed from the state in the ambient atmosphere.
EDX spectra of sea-salt and sea-salt-related particles collected
over the sea-ice area. Aerosol particles were collected (a) in
coarse mode on 23 February 2014, (b) in fine mode on 20 February
2014, (c) in coarse mode on 21 February 2014, (d) in fine
mode on 15 January February 2014, (e) in fine mode on 26 February
2014, (f) in coarse mode on 23 February 2014, (g) in coarse
mode on 3 March 2014, (h) in fine mode on 19 December 2013,
(i) in fine mode on 15 January 2014, and (j) in coarse mode
on 19 January 2014. Asterisks denote background peaks derived from the sample
substrate.
Variations of (a) wind
speed, (b) aerosol number concentrations, relative abundance of each
aerosol type in (c) coarse mode and (d) fine mode. Red plus
(+) marks indicate the date when the low clouds (fogs) were identified off
Siorapaluk.
Elemental compositions of sea salts and relating salts in each aerosol
particle
Figure 8 depicts EDX spectra of sea-salt and sea-salt-related
particles in aerosol particles collected over the sea ice. In accordance with
procedures of aerosol classification presented by Hara et al. (2013, 2014)
and atomic ratios of the respective particles, the mixing states of sea-salt
particles and related salt particles were classified into the following
types: (1) sea-salt particles having atomic ratios similar to those of
seawater (Fig. 8a); (2) Mg-rich sea-salt particles (Fig. 8b); (3) K-rich
sea-salt particles (Fig. 8c); (4) modified sea-salt particles with a slight
Cl loss (Fig. 8d); (5) wholly modified sea-salt particles (Fig. 8e);
(6) sea-salt particles internally mixed with mineral elements such as Al and
Si (Fig. 8f); (7) Na2SO4 particles without Mg (Figs. 8g, and S6);
(8) MgCl2 particles (Fig. 8h); (9) MgSO4 particles (Fig. 8i); and
(10) KCl particles (Fig. 8j). Figure S7 shows that Mg in sea-salt particles
might be present as MgCl2. In fact, Mg and S were detected from aerosol
particles in Fig. 8i. Atomic ratios of Mg and S of the aerosol particles
containing only Mg and S were approximately compatible with MgSO4.
In fact, Mg-rich sea-salt particles, K-rich sea-salt particles, Mg-salt
particles, and K-salt particles were identified also in the boundary layer
over Syowa Station, Antarctica, during winter–spring (Hara et al., 2013) and
near the surface on the Antarctic Plateau during summer (Hara et al., 2014).
Additionally, aerosol particles with atomic ratios similar to CaCO3 were
observed in the same aerosol samples (shown in Supplement,
Fig. S8). In this study, these particles were observed only in aerosol
samples collected near new sea ice. In addition, aerosol particles containing
sulfates, minerals, soot, and anthropogenic metals were observed in this
study.
Ternary plots of Na–Mg–S of sea-salt particles and the modified
sea-salt particles collected over the sea-ice area: (a) 3
March 2014, (b) 14 January 2014, (c) 21 February 2014, and
(d) 3 January 2014. Letters A, B, C, and D within each panel denote,
respectively, the ratios of seawater and those fully modified with
SO42-, MgCl2, and MgSO4. Red open circles and blue
triangles respectively present ratios of particles in coarse and fine modes.
The notation “At %” stands for atomic ratio (in percent).
Abundance of sea-salt and related salt particles
For quantitative discussion, the relative abundance was estimated from the
results of EDX analyses (Fig. 9). In this study, sea-salt and modified
sea-salt particles were major components of the coarse mode. High abundance
of sea-salt particles corresponded to strong winds, high aerosol number
concentrations, and appearance of low clouds (fog) above open sea off
Siorapaluk. Although a few samples showed low relative abundance of sea-salt
particles and modified sea-salt particles, this result corresponded to high
relative abundance of minerals, including those containing sulfates. However,
it is worth noting that the low number concentrations in coarse mode in the
atmosphere engender low number density on the sample substrates and engender
high uncertainty. For number concentrations of less than 10 L-1 in
Dp>2.0 µm, the number of the analyzed particles in coarse mode was 3 to 16 particles.
Aerosol particles containing sulfates were dominant in fine mode, although
sulfate particles were observed rarely in coarse mode. The relative abundance
of NSS sulfate particles in fine mode was roughly equivalent to that of the
Arctic boundary layer around Svalbard (Hara et al., 2003). The relative
abundance of sulfate particles in fine mode was higher under conditions with
low wind speeds and low aerosol number concentrations. Sea-salt particles and modified sea-salt particles in fine mode showed
relative abundance greater than 40 % under conditions with strong winds or high
aerosol number concentrations in coarse mode.
Sea-salt fractionation of aerosol particles in coarse and fine
modes
Figure 10 presents ternary plots of sea salts (Na, Mg, and S) and Mg-rich
sulfates in coarse and fine modes. Internal mixtures of sea salts and
minerals were excluded from the ternary plots. The sum of the atomic ratios
of Na, S, and Mg in each sea-salt particle was not 100 % in the most
cases. Therefore, we converted the sum of the atomic ratios to 100 % for
ternary plots. Labels A, B, C, and D respectively denote the bulk seawater
ratio, sea-salt particles (close to MgCl2) in which chloride was
completely displaced by sulfate, sea-salt particles in which Na was
completely replaced by Mg, and sea-salt particles (close to MgSO4) in
which chloride and Na were completely displaced by sulfate and Mg,
respectively. When the sea-salt particles are modified by sulfates and are
not fractionated, they are distributed around the stoichiometric line of
A–B. Magnesium in sea-salt particles can be enriched gradually with sea-salt
fractionation by precipitation of mirabilite
(Na2SO4⋅10H2O) and
hydrohalite (NaCl⋅2H2O; Hara et al., 2012). When sea-salt
fractionation (replacement between Na and Mg) occurs without sea-salt
modification by sulfate, sea-salt particles are distributed around the
stoichiometric line of A–C. When sea-salt fractionation and sea-salt
modification by sulfate occur stoichiometrically and simultaneously, sea-salt
particles are distributed around the stoichiometric line of A–D. The ratios
of Na–Mg–S of frost flowers and brine were distributed around the bulk
seawater ratio (point A), although limited Mg enrichment was recognized in
this study.
The Mg ratios in coarse aerosol particles collected near new sea ice (site
IIIa) on 3 March were distributed mainly around the bulk seawater ratio and
NaSO4 ratio (Fig. 10a). Magnesium was enriched slightly in some sea-salt
particles, even in coarse modes. In contrast to the sea-salt particles,
Mg-free particles were distributed around the Na2SO4 ratio, as
depicted in Fig. 8g. As described above, these particles distributed around
the Na2SO4 ratio (point B) might be mirabilite particles. In fine
mode, most of the sea-salt particles were distributed between the
stoichiometric lines of the seawater ratio–MgSO4 and seawater
ratio–MgCl2. Unlike with the Mg ratio in coarse mode, Mg enrichment was remarkably observed in fine sea-salt particles.
Although most of the sea-salt particles in coarse mode were distributed
around the bulk seawater ratio on 14 January and 21 February 2014
(Fig. 10b, c), some coarse sea-salt particles had strong Mg enrichment and
were distributed around the stoichiometric line of seawater ratio–MgCl2. A few particles showed atomic ratios that were roughly equal to
that of MgCl2. Similarly to coarse sea-salt particles, Mg enrichment was
also identified in fine mode on 14 January and 21 February 2014. However,
Mg-rich sea-salt particles lay approximately midway between stoichiometric
lines of seawater ratio–MgCl2 and seawater ratio–MgSO4. Moreover,
MgSO4 particles were identified occasionally in this study (e.g.,
21 February 2014).
In contrast to sea-salt particles on 14 January, 21 February, and
3 March 2014, sea-salt particles in both modes were distributed mostly around
the seawater ratio under storm conditions with blowing and drifting snow on
1 March 2014 (Fig. 10d), although some sea-salt particles in coarse and fine
modes had limited Mg enrichment. Winds came not from young sea-ice area with
frost flowers (sites I and II) but from old and very old sea-ice areas.
Variations of the fractionated sea-salt particles during winter
Figure 11 presents variations of Mg / Na ratios in sea-salt particles and
wind speed. Sea-salt particles internally mixed with mineral particles were
excluded from Fig. 11 to avoid misunderstanding of the sea-salt chemistry.
The ratios of Mg / Na were higher than the bulk seawater ratio
(Mg / Na of approximately 0.09 in seawater at Siorapaluk, and 0.11 in
reports of seawater of the literature, Lide, 2005; Millero et al., 2008) in
coarse and fine sea-salt particles during measurements. Higher Mg / Na
ratios and their large variation in sea-salt particles were observed in both
coarse and fine modes under calm wind conditions. In conditions with blowing
snow or strong winds (> 5 m s-1), the Mg / Na ratios and their
standard deviation tended to decrease in both modes (particularly in fine
mode). For instance, median Mg / Na ratios in strong winds were ca. 0.18
in both modes on DOY 40 (10 February), and ca. 0.16 in coarse mode and 0.22
in fine mode on DOY 59 (1 March). Furthermore, Mg / Na ratios in coarse
sea-salt particles increased gradually on DOY 52–57 (22–27 February), when
the air temperature was below -25 ∘C and the wind speed was less
than 4 m s-1. After the storm on DOY 59 (1 March), Mg / Na ratios
of sea-salt particles were distributed around the seawater ratio, although
the ratios varied. In contrast to coarse sea-salt particles, Mg / Na
ratios were higher in fine modes in this study. Similar tendencies were
observed for aerosol particles over Syowa Station, coastal Antarctica (Hara
et al., 2013), and on the Antarctic continent (Hara et al., 2014). This size
dependence of Mg enrichment of sea-salt particles is important to elucidate
the processes of sea-salt particle release from sea ice and frost flowers.
Variations of wind speed and atomic ratios of Mg / Na in
sea-salt particles and the modified sea-salt particles collected over the
sea-ice area. The box plots show values of 90, 75, 50 (median), 25, and
10% denoted, respectively, with the top bar, top box line, black middle
box line, bottom box line, and bottom bar. The red lines show mean values of
the seawater Mg / Na ratio.
Discussion
Sea-salt fractionation on sea ice
The molar ratios of Cl- / Na+ in seawater collected at
Siorapaluk were roughly equal to the ratios reported in the literature (Lide,
2005; Merrlet et al., 2008), although slight differences were identified in
other ratios such as K+ / Na+ and Ca2+ / Na+.
Considering the reproducibility of determination of sea-salt components in
our analytical procedures, the differences (more than 10 %) between the
seawater ratios collected at Siorapaluk and the literature values cannot be
explained by analytical errors. According to Millero (2016), seawater
concentrations and seawater ratios were varied in each sampling site. Indeed,
differences in the seawater ratios (lower K+ / Na+ ratio) were
observed at the Antarctic coast (Hara et al., 2012 and references therein).
Therefore, the difference might result from differences of seawater ratios at
the sampling location.
Coincidence of the ambient molar ratios in frost flowers and snow with the
estimated molar ratios in mirabilite precipitation implies that the molar
ratios relative to Na+ in frost flowers and snow were affected strongly
by mirabilite precipitation. Salt precipitation in sea-salt fractionation
depends on temperature (Marion et al., 1999; Koop et al., 2000). Therefore,
we attempted to compare the relationships of the respective sea salts to
Cl- and Mg2+ (Figs. S1 and S2) for identification of sea-salt
fractionation other than mirabilite precipitation. The molar ratios to
Cl- were not changed by mirabilite precipitation without Cl- loss
by heterogeneous reactions. Relationships among Mg2+, K+,
Ca2+, and Cl- in frost flowers well matched those in brine. Indeed,
t test results (Table 2) suggest that molar ratios of Mg2+, K+,
and Ca2+ relative to Cl- in frost flowers were not significantly
different from those in brine. Although frost flowers were more likely to be
richer in Br- and I- than in Cl- and Mg2+ in most samples
(Table 1, Figs. S3–S5), Student t test results indicated no significant
difference because of high ratios in a few brine samples. Considering that
the first step of precipitation of Na salt is mirabilite precipitation
approximately at -9 ∘C, the coincidence of the relationships of
among Mg2+, K+, Ca2+, and Cl- suggests that enrichment of
Mg2+, K+, and Ca2+ in frost flowers was driven mainly by
mirabilite precipitation. It is worth noting that t tests were applied to all measurements of frost flowers and snow of different ages. Therefore,
we must consider changes in molar ratios carefully if sea-salt fractionation
other than mirabilite precipitation had proceeded with aging of frost flowers
(details are discussed in the next section).
Considering the presence of MgCl2 and MgSO4 in aerosol particles,
Mg salts might be present in frost flowers and the slush layer. According to
earlier laboratory and model studies (e.g., Mairon et al., 1999), MgCl2⋅6H2O and KCl (sylvite) can precipitate respectively at approximately
-36 and -34 ∘C. During the measurements, the minimum air
temperature (-34.1 ∘C) and temperature at the surface of the slush
layer (TFF) were higher than the temperature at which MgCl2⋅6H2O precipitates. Therefore, MgCl2⋅6H2O precipitation
might not have occurred during the measurements. In other words, Mg2+
might be distributed in the residual brine on frost flowers and in the slush
layer and snow.
Aging of frost flowers and sea-salt fractionation
Molar ratios of SO42- / Cl- in frost flowers at sites I and
II (Fig. 5b) were considerably lower than the seawater ratio. This change in
SO42- / Cl- ratios might be attributed to sea-salt
fractionation by sulfate depletion (i.e., mirabilite precipitation). By
contrast, SO42- / Cl- ratios at site III were roughly
equivalent to the seawater ratio. Considering direct evidence indicating that
mirabilite-like particles were identified in aerosols only at sites IIIa and
IIIb, we believe that mirabilite might be precipitated in the early sea-ice stage. Details of
mirabilite-like particles are discussed in Sect. 4.3.
At site I, sea-salt ratios changed gradually with growth and aging of frost
flowers, as shown in Fig. 6. As discussed in Sect. 4.1, lower
SO42- / Na+ ratios might result from the mirabilite
precipitation. Moreover, the ratios of Mg2+ / Cl-,
K+ / Cl-, and Ca2+ / Cl- increased simultaneously
and Na+ / Cl- decreased slightly on 26–28 February when
TFF dropped approximately to the temperature for hydrohalite
precipitation (ca. -22 ∘C). This simultaneous change might not
result from analytical errors because the differences were 2–3 times larger
than analytical errors (1σ). The ambient molar ratios of
Na+ / Cl- in frost flowers were similar to the estimated ratios
in hydrohalite precipitation, although the ratios were slightly lower than
the estimated ratios. The cation Ca2+ can be fractionated by precipitation of
ikaite (-2.2 ∘C, Dieckmann et al., 2008, 2010) and gypsum
(-15 ∘C, Marion et al., 1999). Therefore, lower ratios of
Ca2+ / Cl- might result from precipitation of these salts.
Presence of K-rich sea-salt particles and K-salt particles in sea-salt
particles in the atmosphere implies that sea-salt fractionation with K+
occurred in our measurements, although sylvite can be precipitated at
-33 ∘C (Marion et al., 1999). The coincidence implies that the
molar ratios in frost flowers on 26–28 February were attributed to
hydrohalite precipitation. In contrast to TFF, Tair
increased slightly on 26–27 February. Then it increased greatly on
28 February–1 March. In spite of the slight increase of TAWS and
Tair, TFF tended to decrease slightly during
24–27 February. This decrease of TFF might be attributed to
reduction of heat conduction by sea-ice growth (larger thickness).
Consequently, it is expected that the sea-ice thickness was a fundamentally
important factor for sea-salt fractionation on sea ice, in addition to
Tair.
It is worth noting that molar ratios in frost flowers cannot change if
sea-salt fractionations such as mirabilite and hydrohalite precipitation
occur on frost flowers after brine migration onto frost flowers. Regardless
of whether mirabilite and hydrohalite were precipitated on frost flowers or
not, the total amount (mass) of precipitated salts and sea salts in residual
brine did not change without liberation by heterogeneous reactions. The
cations Mg2+, K+, and Ca2+
cannot be released into the atmosphere by heterogeneous reactions. Results
show that TFF was lower than the temperature at which mirabilite
and hydrohalite precipitates and higher than those for sylvite,
MgCl2⋅6H2O, and NaBr⋅5H2O (e.g., Marion, 1999; Koop
et al., 2000). Therefore, Mg might be enriched in the residual brine.
Precipitation of mirabilite and hydrohalite might be driven near the surface
of brine on the sea ice, as suggested by (1) a change in the molar ratios in
frost flowers, (2) non-significant change in those in brine, and
(3) TFF close to ca. -21 ∘C. Then, the residual brine
might be migrated vertically onto frost flowers. The ratios of
Br- / Cl- and I / Cl- in frost flowers were mostly
higher than those in brine, except for a few brine samples. It is expected
that Br- and I were richer in frost flowers because of sea-salt
fractionation. However, the concentrations of Br- and I in seawater
sampled at Siorapaluk were not determined in this study.
Fractionated sea-salt particles in the atmosphere
From single-particle analysis of aerosols, several types of salt particles,
likely related to sea-salt fractionation, were identified in aerosols:
(1) Mg-rich sea-salt particles, (2) Na2SO4 particles, (3) Mg-salt
particles (MgCl2 and MgSO4), (4) K-rich sea-salt particles, (5) KCl
particles, and (6) CaCO3 particles. With sea-salt fractionation in frost
flowers, brine, and surface snow on sea ice, Mg2+, K+ and Ca2+
were enriched in frost flowers and surface snow (excluding Mg2+ in
surface snow). Therefore, Mg-rich sea-salt
particles (Fig. 8b), K-rich sea-salt particles (Fig. 8c), Mg-salt particles
(Figs. 8h–i), and K-salt particles (Fig. 8j) might originate from the
sea-ice area and might be associated with sea-salt fractionation. As
presented in Fig. 8g, Mg was not detected in aerosol particles containing Na
and S. Magnesium ratios in coarse sea-salt particles usually exceed the
detection limit of single particle analysis by EDX. Therefore, the aerosol
particles might have an extremely low Mg ratio. The atomic ratios of Na and S
of the Mg-poor particles imply strongly that the particles were in the form
of Na2SO4. If the sea-salt particles were modified with
SO42- by heterogeneous reactions (Supplement, Figs. S9–S10), then
the modified sea-salt particles contained sea-salt Mg. Consequently, the
presence of Na2SO4 particles cannot be explained by their release
from the sea surface and then sea-salt modification. Particles of
Na2SO4 were observed only at new sea ice, sites IIIa and IIIb,
where CaCO3-like particles were also identified. In the early stage of
sea-ice formation, ikaite (CaCO3⋅6H2O) and mirabilite
(Na2SO4⋅10H2O) can be precipitated respectively at -2
and -8.8 ∘C on/in sea ice (Dieckmann et al., 2008, 2010; Marion et
al., 1999). Therefore, mirabilite-like and ikaite-like particles might be
released into the atmosphere in the new sea-ice area.
Next, we specifically examined Mg / Na ratios in sea-salt particles to
elucidate the sea-salt cycles in seasonal sea-ice areas. Magnesium-rich
sea-salts and Mg
salts cannot be evaporated or vaporized under ambient conditions: these
particles must be released through physical processes. Sea-salt fractionation
can occur if sea-salt particles are fractured in the atmosphere. However,
direct evidence of the fracture of sea-salt particles in the atmosphere has
not been obtained (Lewis and Schwartz, 2004). The following evidence is
important to elucidate the origins of Mg-rich sea-salt particles and Mg-rich
salt particles in the atmosphere: (1) the presence of highly Mg-rich
particles (Mg-rich sea-salts, MgCl2, and MgSO4), (2) TFF
lower than the temperature at precipitation of mirabilite and hydrohalite,
(3) higher Mg / Na ratio in fine mode, (4) small variation of the
Mg / Na ratio in strong winds and blowing snow, (5) high
Mg2+ / Na+ ratios in frost flowers, and (6) Mg depletion in the
aged surface snow on sea ice.
Schematics of sea-salt cycles in the sea-ice area. Dotted arrows
indicate the speculated processes.
The Mg / Na ratios in sea-salt particles differed greatly depending on
the sampling site and meteorological conditions (e.g., winds and temperature)
as presented in Fig. 11. It is worth noting that sea-salt particles at sites
IIIa and IIIb were distributed around seawater ratios from DOY 60 (2 March).
Therefore, most of the sea-salt particles, except mirabilite-like and ikaite-like particles at sites IIIa and IIIb, might have been released from the sea surface. Moreover, the presence of
ikaite-like and mirabilite-like particles in the atmosphere implies that
these particles were released into the atmosphere from the early stage of sea
ice after precipitation of ikaite and mirabilite on sea ice and frost
flowers. In contrast to Mg / Na ratios in sea-salt particles at sites
IIIa and IIIb, higher Mg / Na ratios than seawater ratios were identified
at sites I and II. Higher Mg / Na ratios in sea-salt particles suggest
strongly that Mg-rich sea-salt particles in the atmosphere were supplied from
the sea-ice area with sea-salt fractionation. As shown in Fig. 4,
Mg2+ / Na+ ratios in frost flowers at site I increased
gradually during 20 February to 1 March under colder conditions. The
correlation
between high Mg / Na ratios in coarse mode and the coldest conditions
implies strongly that Mg / Na ratios in coarse sea-salt particles
responded rapidly to sea-salt fractionation on sea ice and frost flowers.
Because of the aerosol number concentrations, relative abundance of sea-salt
particles, and high Mg / Na ratios relative to seawater ratios in strong
winds, the fractionated sea-salt particles might have been dispersed from sea ice into
the atmosphere by strong winds. Although high aerosol number concentrations
were observed occasionally at Siorapaluk under calm winds, the features might
result from transport of (1) sea-salt particles released elsewhere by strong
winds and (2) anthropogenic aerosols (i.e., sulfates and Arctic haze).
Because of the high abundance of sea-salt particles, most cases of higher
aerosol number concentrations in calm winds were likely associated with the
release from sea-ice area and transport of sea-salt particles. Similar
phenomena (aerosol enhancement) were identified also at the Antarctic coast
(Hara et al., 2010). As discussed above, Mg was likely richer in frost
flowers and the residual brine on sea ice. Earlier investigations (Obbard et
al., 2009; Roscoe et al., 2011) revealed, however, that no aerosol particles
were released by the breakage of frost flowers under strong winds.
Considering the direct evidence of Mg depletion in aged surface snow on
sea ice, Mg-rich sea-salt particles and Mg-salt particles were likely
released from surface snow mixed with the residual brine on sea ice. The
variations of Mg / Na ratios in sea-salt particles were smaller in both
coarse and fine modes under storm conditions (DOY 40 and 59), although
the Mg / Na ratios were higher than the seawater ratio. Winds passed from
the old and very old sea-ice area to the sampling sites in the storm
conditions. Consequently, Mg-rich sea-salt particles in the storms might be
released also from the snow layer on old and very old sea ice through erosion
of snow by strong winds because the slush layer was absent on old and
very old sea ice. By contrast, Mg / Na ratios varied largely under
calm wind conditions. To explain the presence of highly Mg-rich sea-salt
particles and Mg-rich salt particles, we inferred that these particles were
released from the aged surface snow and the residual brine on slush layer and
frost flowers through erosion of snow with the residual brine and splashing
and shattering of the residual brine film. Higher Mg / Na ratios in fine
sea-salt particles are eminently explainable if the processes proceeded on
seasonal sea-ice areas. To elucidate these points, we must accumulate more
information related to the salt distribution on and in frost flowers and sea
ice at the nanometer–micrometer scale.
Sea-salt cycles in seasonal sea-ice area
From the evidence and results from this work and earlier works, we propose
the following as hypotheses for the sea-salt fractionation processes on
sea ice and the release of sea-salt particles into the atmosphere (Fig. 12).
Initial stage – open sea surface:
Before sea-ice formation, sea-salt particles are released from the sea
surface through bubble bursting and breaking waves (e.g., Lewis and Schwartz,
2004). Sea-salt ratios in the particles released by bubble bursting are
similar to the seawater ratio (Keene et al., 2007).
First stage – seawater freezing:
Seawater starts freezing at lower air temperatures. In this stage, sea ice
was likely present in conditions of grease ice, frazil ice, and sludge
(Comiso and Steffen, 2001; Brandt et al., 2005; Comiso, 2010). Considering
that sea-salt particles with ratios similar to seawater were found to be
present only at sites IIIa and IIIb, these particles must be released from
the sea surface in the initial stage and first stage. Depending on the
temperature at the sea-ice surface, ikaite can start precipitation at
temperatures lower than -2 ∘C (Dieckmann et al., 2008, 2010).
Second stage – sea-ice formation and sea-salt fractionation:
Then, the sea-surface was covered with thin sea ice (i.e., nilas) at sites
IIIa and IIIb. The presence of sea ice prevents the release of sea-salt
particles from the sea surface into the atmosphere. A strong vertical gradient
of air temperature near the sea-ice surface might cause frost flower
formation on sea ice (e.g., Perovich and Richter-Menge, 1994; Martin et al.,
1995, 1996; Style and Worster, 2009; Roscoe et al., 2011). Some brine can be
migrated vertically on frost flowers. Cooling of surface of the frost flowers
and brine on sea ice can engender precipitation of ikaite and mirabilite.
The presence of ikaite-like particles and mirabilite-like particles in the
atmosphere suggests that these particles are released into the atmosphere
through physical processes. Mirabilite-like and ikaite-like particles were
identified in aerosols collected only at sites IIIa and IIIb. Therefore,
these particles might be released from fresh sea-ice areas. However,
specific release processes remain unclear.
Third stage – frost flower growth and sea-salt fractionation:
With sea-ice growth, the temperature on sea ice (TFF) might
decrease gradually by reduction of heat conduction from seawater to the
sea-ice surface. Lower temperatures on and in the slush layer can induce
sea-salt fractionation by precipitation of mirabilite and hydrohalite.
Precipitation of mirabilite and hydrohalite can engender sea-salt enrichment
(e.g., Mg2+, K+, and Ca2+) in frost flowers and the residual
brine. The residual brine having Mg enrichment is migrated vertically on
frost flowers.
Fourth stage – strong winds and snowfall on frost flowers:
Under conditions with strong winds, snowfall, and blowing snow, snow
particles were attached on frost flowers and slush layers. As suggested by
laboratory experiments (Roscoe et al., 2010), no aerosol particles are
released from frost flowers. However, Mg-rich sea-salt particles and
Mg salts might be released from the slush layer and surface snow on sea ice.
Fifth stage – frost flower and slush layer covered with snow:
When snowfall and blowing snow are sufficient to cover frost flowers and the
slush layer on new–young sea ice, frost flowers and slush layer are buried
completely in snow after the storm. After snow deposition onto new–young
sea ice, the residual brine with Mg2+ enrichment might be migrated
vertically and gradually into the snow layer. As a result, the snow layer on
new–young sea ice was wet, as observed in this study. Sea salts in the
migrated brine, frost flowers and snow can be redistributed through snow
metamorphosis, although distributions of sea salts might be heterogeneous in
the snow layer. Magnesium-rich sea-salt particles and Mg-salt particles might be
released from the surface snow on sea ice because of limited loss of Mg2+
in the aged surface snow on sea ice as shown in Figs. S1 and S4. Therefore,
we speculate that splash and erosion of the residual brine on snow and frost
flowers by winds are plausible release processes of Mg-rich sea-salt
particles and Mg-salt particles.
Sixth stage – snow erosion by strong winds:
With sea-ice growth, snow and slush layers can be frozen gradually. Then,
strong winds (i.e., storm condition) engender erosion of aged surface snow on
sea ice and release of Mg-rich sea-salt particles into the atmosphere. A dry
and hard surface of sea ice appears after snow layers are removed from old
and very old sea ice. Because of wet conditions in the snow and slush layer,
a large amount of surface snow remained on the young sea ice.
We observed frost flowers and aerosols on sea ice in a fjord near Siorapaluk,
Greenland. Therefore, we were able to compare respective sea-ice stages
easily. If sea-ice areas having frost flowers are present in the locations
affected strongly by winds, waves, tides, and ocean currents (e.g., Arctic
Ocean and Antarctic coast), then sea ice can flow out and have many cracks,
which can occur often in polar regions. Under such conditions, some stages in
Fig. 12 might duplicate and proceed simultaneously.