Introduction
Tropospheric aerosol is an important environmental issue because it
dramatically reduces the visibility (Jacobson et al., 2000; Kanakidou et
al., 2005), affects the radiative forcing of climate (Seinfeld and Pandis,
1998), and causes a negative impact on human health (Pope and Dockery,
2006). All of these effects strongly depend on the abundances of aerosols
and their chemical and physical properties in different sizes. Particles with
diameters of 0.1–1.0 µm are very active in scattering and absorbing
incoming solar radiation and have a direct impact on climate (Ramanathan et
al., 2001; Seinfeld and Pankow, 2003). The knowledge of size distributions
of chemical components is thus essential to better understand their
potential contributions to climate change and pollution control. Their size
distribution also provides evidence for the sources and formation pathways
of atmospheric particles.
The emission sources and multiple secondary formation pathways of organic
aerosols are not well understood. Organic compounds account for up to 70 %
of fine aerosol mass and potentially control the physicochemical properties
of aerosol particles (Davidson et al., 2005; Kanakidou et al., 2005).
Low-molecular-weight diacids are one of the most abundant organic compound
classes in the atmosphere (Kawamura and Ikushima, 1993; Kawamura et al.,
1996; Kawamura and Bikkina, 2016). They are primarily derived from
incomplete combustion of fossil fuel and biomass burning (Kawamura and
Kaplan, 1987; Falkovich et al., 2005), and secondarily produced in the
atmosphere via photooxidation of unsaturated fatty acids and volatile
organic compounds (VOCs) from biogenic and anthropogenic sources (Kawamura
and Gagosian, 1987; Kawamura et al., 1996; Sempéré and Kawamura,
2003). The ability of organic aerosols to act as cloud condensation nuclei seems to be closely related to their mass-based size distributions
(Pradeep Kumar et al., 2003; Ervens et al., 2007).
The increasing atmospheric burden of organic aerosols is associated with
natural and anthropogenic emissions in the continental regions. Organic
aerosols are eventually transported to the oceanic regions. Rapid
industrialization in East Asia is expected to have an important impact on
global atmospheric chemistry over the next decades (Wang et al., 2013; Tao
et al., 2013; Bian et al., 2014). Large amounts of coal and biomass burning
in East Asia add more anthropogenic aerosols which alter the aerosol chemical
composition in the remote Pacific atmosphere (Mochida et al., 2007; Miyazaki
et al., 2010; Agarwal et al., 2010; Wang et al., 2011; Engling et al.,
2013). Water-soluble diacids and related compounds as well as major ions have
previously been studied for their size distributions in remote marine aerosols
(Kawamura et al., 2007: Mochida et al., 2007; Miyazaki et al., 2010),
whereas their size-segregated characteristics have not been studied in the
western North Pacific Rim.
We collected size-segregated aerosol samples with nine size ranges in spring
2008 in Cape Hedo, Okinawa, in the western North Pacific Rim. Cape Hedo is
located on the northern edge of Okinawa Island and can serve as a suitable
site for the observation of atmospheric transport of East Asian aerosols
with insignificant interference from local emission sources (Takami et al.,
2007). The samples were analyzed for dicarboxylic acids (C2–C12)
and related compounds such as ω-oxoacids (ωC2–ωC9), pyruvic acid (C3), and α-dicarbonyls
(C2–C3) to better understand the sources and processing of
water-soluble organic compounds at this marine receptor site.
Size-segregated samples were also analyzed for water-soluble organic carbon
(WSOC), organic carbon (OC), and major inorganic ions. The role of liquid
water content of aerosol in the size distribution of diacids and related
compounds is discussed. The potential factors responsible for their size
distributions are also discussed.
Materials and method
Site description and aerosol collection
The geographical location of Okinawa Island (26.87∘ N and
128.25∘ E) and its surroundings in East Asia are shown in Fig. 1. Okinawa is located in the outflow region of continental aerosols and on
the pathways to the Pacific. Cape Hedo has been used as a supersite of
the Atmospheric Brown Clouds project to study the atmospheric transport of
Chinese aerosols and their chemical transformation during long-range
transport from East Asia (Takiguchi et al., 2008; Kunwar and Kawamura,
2014). The sampling site at Cape Hedo is about 60 m a.s.l.
A map of East Asia with the location of Okinawa Island
(26.87∘ N and 128.25∘ E) and Asian countries.
Size-segregated aerosol samples were collected at Cape Hedo Atmosphere and
Aerosol Monitoring Station (CHAAMS) from 18 March to 13 April 2008. This
period was characterized by westerly wind in the lower troposphere, which
is the principal process responsible for the transport of both fossil fuel
combustion and biomass burning aerosols in East Asia to the western North
Pacific. 9-Stage Andersen middle volume impactor (Tokyo Dylec Company,
Japan; 100 L min-1) was used for the collection of size-segregated
samples. The sampler was equipped with quartz fiber filters (QFFs, 80 mm in
diameter) that were precombusted at 450 ∘C for 6 h in a furnace
to eliminate the adsorbed organic compounds. A total of five sets (OKI-1 to
OKI-5) of size-segregated aerosol samples were collected. Each sample set
consists of nine filters for the sizes of < 0.43, 0.43–0.65,
0.65–1.1, 1.1–2.1, 2.1–3.3, 3.3–4.7, 4.7–7.0, 7.0–11.3, and > 11.3 µm. The filter was placed in a preheated 50 mL glass vial with a
Teflon-lined screw cap and stored in a freezer at the station. The samples
were stored in darkness at -20 ∘C prior to analysis in Sapporo.
One set of field blanks was collected by placing a precombusted QFF into the sampler for 30 s
without sucking air before installing the real QFF.
Analytical procedures
Diacids and related compounds were determined by the method of Kawamura and
Ikushima (1993), and Kawamura (1993). Aliquot of the filters was extracted
with organic-free ultrapure water (specific resistivity > 18.2 MΩ cm) under ultrasonication. The extracts were passed through a
glass column packed with quartz wool to remove insoluble particles and
filter debris. The extracts were concentrated using a rotary evaporator
under vacuum and derivatized to dibutyl esters and dibutoxy acetals with
14 % BF3 in n-butanol at 100 ∘C. Acetonitrile and n-hexane
were added into the derivatized sample and washed with organic-free pure
water. The hexane layer was further concentrated using a rotary evaporator
under vacuum and dried to almost dryness by N2 blowdown and dissolved
in 100 µL of n-hexane. 2 µL of the sample was injected into a
capillary gas chromatograph (GC) (Hewlett-Packard HP6890) equipped with a flame
ionization detector. Authentic diacid dibutyl esters were used as external standards for the peak
identification and quantification. Identifications of diacids and related
compounds were confirmed by GC–mass spectrometry. Recoveries of authentic
standards spiked to a precombusted QFF were 85 % for oxalic acid
(C2) and more than 90 % for malonic to adipic (C3–C6)
acids. The detection limits of diacids and related compounds were ca. 0.002 ng m-3. The analytical errors in duplicate analyses are within 10 %
for major species.
To measure water-soluble organic carbon (WSOC), a punch of 20 mm diameter of
each QFF was extracted with organic-free ultrapure water in a 50 mL glass
vial with a Teflon-lined screw cap under ultrasonication for 15 min. The
water extracts were subsequently passed through a syringe filter (Millex-GV,
Millipore; diameter of 0.22 µm). The extract was first acidified with
1.2 M HCl and purged with pure air in order to remove dissolve inorganic
carbon and then WSOC was measured using a total organic carbon (TOC)
analyzer (Shimadzu TOC-VCSH) (Miyazaki et al., 2011). External
calibration was performed using potassium hydrogen phthalate before analysis
of WSOC. The sample was measured three times and the average value was used
for the calculation of WSOC concentrations. The analytical error in the
triplicate analysis was 5 % with a detection limit of 0.1 µgC m-3.
Organic and elemental carbon (OC and EC) was determined using a Sunset Lab
carbon analyzer following the Interagency Monitoring of Protected Visual
Environments (IMPROVE) thermal evolution protocol as described by H. Wang et al. (2005). A filter disc of 1.5 cm2 was placed in a quartz tube
inside the thermal desorption chamber of the analyzer and then stepwise
heating was applied. Helium (He) gas was applied in the first ramp and was
switched to mixture of He/O2 in the second ramp. The evolved CO2
during the oxidation at each temperature step was measured with a non-dispersive infrared detector system. The detection limits of OC
and EC were ca. 0.05 and 0.02 µgC m-3, respectively. The
analytical errors in the triplicate analysis of the filter sample were
estimated to be 5 % for OC and EC. EC was detected only in fine fractions.
The concentration of total carbon (TC) was calculated by summing the
concentrations of OC and EC in each size fraction.
For the determination of major ions, a punch of 20 mm diameter of each
filter was extracted with organic-free ultrapure water under
ultrasonication. These extracts were filtered through a disc filter
(Millex-GV, Millipore; diameter of 0.22 µm) and injected into an ion
chromatograph (Compact IC 761; Metrohm, Switzerland) for measuring
MSA-, Cl-, SO42-, NO3-, Na+,
NH4+, K+, Ca2+, and Mg2+ (Boreddy and Kawamura,
2015). Anions were separated on a SI-90 4E Shodex column (Showa Denko;
Tokyo, Japan) using a mixture of 1.8 mM Na2CO3 and 1.7 mM
NaHCO3 solution at a flow rate of 1.2 mL min-1 as an eluent and 40 mM H2SO4 for a suppressor. A Metrosep C2-150 Metrohm column was
used for cation analysis using a mixture of 4 mM tartaric acid and 1 mM
dipicolinic acid solution as an eluent at a flow rate of 1.0 mL min-1.
The injected loop volume was 200 µL. The detection limits for anions
and cations were ca. 0.1 ng m-3. The analytical error in duplicate
analysis was about 10 %.
Field blanks were extracted and analyzed like the real samples. However,
blank levels were 0.1–5 % of real samples. The reported concentrations of
organic and inorganic species were corrected for the field blanks. All the
chemicals including authentic standards were purchased from Wako Pure
Chemical Co. (Japan), except for 14 % BF3/n-butanol (Sigma-Aldrich,
USA).
Seven-day backward air mass trajectories (NOAA HYSPLIT) at 500 m a.g.l.
(09:00 UTC) for the aerosol samples (OKI-1 to OKI-5) collected on
Okinawa Island. The dates given in each panel are the starting and ending
times of the collection of aerosol samples on Okinawa Island. The color scale shows
the altitude of the air parcel.
Backward air mass trajectories and meteorology
The backward trajectories of air masses were computed for the sampling
period using the Hybrid Single-Particle Lagrangian Integrated Trajectory
(HYSPLIT) model 4.0 developed by the National Oceanic and Atmospheric
Administration (NOAA) Air Resources Laboratory (ARL) (Draxler and Rolph,
2013). The 7-day trajectories at 500 m above the ground level for the
samples collected on Okinawa are shown in Fig. 2. Typical air mass
trajectories corresponding to 09:00 UTC for the samples collected on Okinawa
are shown in Fig. S1 in the Supplement.
Meteorological data including ambient temperature, relative humidity, and
wind speed for each sample period were obtained from Japan Meteorological
Agency (http://www.jma.go.jp). During our campaign, ambient
temperature, relative humidity, and wind speed ranged from 11.9 to 26.6 ∘C (ave. 20.0 ± 2.6 ∘C), 43.0 to 91.0 % (ave.
70.0 ± 12.0 %), and 0.10 to 10.2 m s-1 (ave.
3.73 ± 1.99 m s-1), respectively. The 7-day trajectories along with the
meteorological data, including precipitation and downward solar radiation
flux, are shown in Fig. S2.
Estimation of liquid water content (LWC) of aerosol
LWC of aerosol was calculated for the size-segregated samples collected on
Okinawa Island using the ISORROPIA II model (Fountoukis and Nenes, 2007).
ISORROPIA II is a computationally efficient and rigorous thermodynamic
equilibrium model that exhibits robust and rapid convergence under all
aerosol types with high computational speed (Nenes et al., 1998). ISORROPIA
II implies the Zdanovskii–Stokes–Robinson equation and treats only the
thermodynamics of K+–Ca2+–Mg2+–NH4+–Na+–SO42-–NO3-–Cl-–H2O
aerosol system to estimate the LWC. Therefore, the measured organic species
such as diacids and related compounds are not included in ISORROPIA II. The
model was run as a “reverse problem”, in which temperature, relative
humidity, and aerosol phase concentrations of water-soluble inorganic ions
were used as input for the estimation of aerosol LWC.
Concentrations (µg m-3) of major inorganic ions and
carbonaceous species in the fine- and coarse-mode aerosols on Okinawa Island
in the western North Pacific.
Inorganic ions
Fine modea
Coarse modeb
Mean
SDc
Min.d
Max.e
Mean
SD
Min.
Max.
Water-soluble inorganic ions
Cations
Na+
0.44
0.20
0.21
0.72
2.42
0.89
1.60
3.65
NH4+
2.40
1.18
0.74
3.69
0.03
0.01
0.03
0.05
K+
0.14
0.06
0.04
0.21
0.09
0.02
0.07
0.12
Mg2+
0.07
0.02
0.04
0.10
0.34
0.11
0.24
0.49
Ca2+
0.06
0.02
0.04
0.09
0.41
0.19
0.15
0.60
Total cations
3.12
1.22
1.28
4.37
3.29
1.02
2.55
4.82
Anions
MSA-
0.04
0.01
0.03
0.06
0.01
0.00
0.00
0.01
Cl-
0.12
0.13
0.02
0.29
4.27
2.25
1.77
7.25
NO3-
0.14
0.08
0.04
0.23
1.61
0.54
0.94
2.41
SO42-
10.1
4.85
2.88
14.9
1.46
0.44
0.69
1.81
Total anions
10.4
4.73
3.33
15.1
7.35
2.20
5.69
10.6
Total water-soluble ions
Total water-soluble ions
13.5
5.95
4.61
19.5
10.6
3.22
8.33
15.4
Carbonaceous components
WSOC
1.12
0.49
0.31
1.61
0.33
0.13
0.15
0.52
OC
1.62
0.59
0.62
2.12
0.60
0.17
0.36
0.82
OM
3.43
1.31
1.30
4.87
1.25
0.36
0.75
1.73
EC
0.05
0.03
0.00
0.09
–
–
–
–
TC
1.67
0.65
0.62
2.41
0.60
0.17
0.36
0.82
a Fine mode represents aerosol size
of Dp < 2.1 µm. b Coarse mode represents
aerosol size of Dp > 2.1 µm. c Standard
deviation. d Minimum. e Maximum.
Summarized concentrations (ng m-3) of
water-soluble dicarboxylic acids and related polar compounds in the fine- and coarse-mode
aerosols from Okinawa Island in the western North Pacific Rim.
Compounds
Abbreviation
Chemical formula
Fine modea
Coarse modeb
Mean
SDc
Min.d
Max.e
Mean
SD
Min.
Max.
Dicarboxylic acids
Saturated normal-chain diacids
Oxalic
C2
HOOC–COOH
135
37.4
76.0
176
40.2
14.7
22.1
60.0
Malonic
C3
HOOC–CH2–COOH
19.5
6.84
7.56
23.6
12.4
3.52
6.87
15.5
Succinic
C4
HOOC–(CH2)2–COOH
13.4
4.98
5.08
17.5
8.02
2.21
4.66
10.1
Glutaric
C5
HOOC–(CH2)3–COOH
3.30
1.54
1.00
4.75
1.89
0.57
1.07
2.66
Adipic
C6
HOOC–(CH2)4–COOH
3.49
1.09
2.47
4.98
2.50
1.24
1.45
4.23
Pimelic
C7
HOOC–(CH2)5–COOH
0.46
0.24
0.04
0.63
0.32
0.11
0.20
0.44
Suberic
C8
HOOC–(CH2)6–COOH
0.07
0.07
0.00
0.16
0.04
0.02
0.02
0.07
Azelaic
C9
HOOC–(CH2)7–COOH
1.20
0.72
0.51
2.41
1.15
0.60
0.49
2.10
Decanedioic
C10
HOOC–(CH2)8–COOH
0.17
0.11
0.01
0.30
0.08
0.07
0.03
0.19
Undecanedioic
C11
HOOC–(CH2)9–COOH
0.47
0.33
0.13
0.76
0.25
0.10
0.14
0.38
Dodecanedioic
C12
HOOC–(CH2)10–COOH
0.07
0.03
0.03
0.09
0.05
0.02
0.02
0.07
Branched-chain diacids
Methylmalonic
iC4
HOOC–CH(CH3)–COOH
0.43
0.23
0.09
0.71
0.47
0.37
0.09
0.99
Methylsuccinic
iC5
HOOC–CH(CH3)–COOH
0.81
0.27
0.37
1.00
0.59
0.13
0.45
0.80
2-Methylglutaric
iC6
HOOC–CH(CH3)–(CH2)2–COOH
0.35
0.24
0.05
0.70
0.19
0.20
0.04
0.53
Unsaturated aliphatic diacids
Maleic
M
HOOC–CH = CH–COOH - cis
0.81
0.25
0.41
1.05
0.73
0.23
0.37
0.95
Fumaric
F
HOOC–CH = CH–COOH - trans
0.31
0.09
0.20
0.42
0.21
0.08
0.12
0.30
Methylmaleic
mM
HOOC–C(CH3)= CH–COOH - cis
0.34
0.27
0.11
0.76
0.57
0.48
0.11
1.37
Unsaturated aromatic diacids
Phthalic
Ph
HOOC–(C6H4)–COOH - o-isomer
6.29
2.85
1.99
9.3
2.79
0.81
1.85
3.9
Isophthalic
iPh
HOOC–(C6H4)–COOH - m-isomer
0.46
0.07
0.35
0.55
0.17
0.06
0.09
0.22
Terephthalic
tPh
HOOC–(C6H4)–COOH - p-isomer
2.21
1.15
0.32
3.30
0.64
0.38
0.09
1.17
Multifunctional diacids
Malic
hC4
HOOC–CH(OH)–CH2–COOH
0.14
0.05
0.11
0.21
0.14
0.06
0.07
0.20
Ketomalonic
kC3
HOOC–C(O)–COOH
4.92
3.79
0.46
9.28
0.49
0.17
0.32
0.77
4-Ketopimelic
kC7
HOOC–CH2–CH2–HC(O)(CH2)2–COOH
2.57
0.83
1.26
3.20
0.43
0.16
0.26
0.69
Total diacids
196
58.1
98.3
253
74.1
24.3
41.4
105
ω-Oxocarboxylic acids
Glyoxylic
ωC2
OHC–COOH
14.1
5.92
4.77
20.2
4.81
2.00
2.23
7.20
3-Oxopropanoic
ωC3
OHC–CH2–COOH
0.08
0.05
0.00
0.12
0.05
0.04
0.02
0.12
4-Oxobutanoic
ωC4
OHC–(CH2)2–COOH
2.23
1.12
0.86
3.56
0.68
0.35
0.41
1.22
9-Oxononanoic
ωC9
OHC–(CH2)7–COOH
0.74
0.20
0.54
1.07
1.06
0.34
0.57
1.41
Total oxoacids
17.1
7.04
6.27
25.0
6.60
2.33
3.26
9.52
Ketoacid
Pyruvic
Pyr
CH3–C(O)–COOH
2.61
0.76
1.67
3.48
2.32
1.20
0.76
4.09
α-Dicarbonyls
Glyoxal
Gly
OHC–CHO
2.74
1.12
1.45
4.40
0.84
0.26
0.50
1.17
Methylglyoxal
MeGly
CH3–C(O)–CHO
1.09
0.98
0.25
2.53
0.65
0.16
0.45
0.87
Total α-dicarbonyls
2.83
1.59
1.03
4.68
1.49
0.37
0.96
1.86
Aromatic monoacid
Benzoic acid
C6H5–COOH
16.5
11.0
4.57
28.3
1.98
1.01
0.70
3.38
a Fine mode represents aerosol size of Dp < 2.1 µm. b
Coarse mode represents aerosol size of Dp > 2.1 µm. c Standard deviation. d Minimum. e
Maximum.
Results and discussion
Size-segregated aerosol chemical characteristics
We use 2.1 µm as a split diameter between the fine- and coarse-mode
particles. Table 1 presents the concentrations of inorganic and carbonaceous
species in the fine- and coarse-mode aerosols. Abundances of organic matter
(OM) in the atmosphere are generally estimated by multiplying the measured
OC mass concentrations with the conversion factor of 1.6 for urban aerosols
and 2.1 for aged aerosols (Turpin and Lim, 2001). CHAAMS is located in the
outflow region of East Asian aerosols and local anthropogenic activities are
insignificant. Because the aerosols reaching Okinawa undergo the atmospheric oxidation during the long-range transport, the
fraction of oxygenated organic species is often high (Takami et al., 2007;
Irei et al., 2014; Kunwar and Kawamura, 2014). Therefore, we used the
conversion factor of 2.1, instead of 1.6 for the calculation of OM.
OM was more enriched in fine-size fractions than in the coarse-size fractions (Table 1). The elevated level of OM in fine fractions on Okinawa suggests a
substantial contribution of organic aerosols primarily from combustion
sources, and secondarily from photochemical processes during long-range
atmospheric transport. The OM in fine-mode aerosol on Okinawa may consist of
oxygenated organic compounds such as diacids, ω-oxoacids, and α-dicarbonyls.
Okinawa was strongly affected by long-range transport of
continental air masses from Siberia and Mongolia as well as North China and
Korea (Fig. 2). It is difficult to specify the source regions of air
masses for each sample set because the sampling duration was 3–5 days. Each
sample contains mixed continental and oceanic air masses. Precipitation may
have an insignificant effect on the transport of pollutants from the source
region to Okinawa because air masses did not experience serious
precipitation events during transport (Fig. S2a).
Sulfate is the most abundant anion in fine mode, whereas chloride is the
dominant anion in coarse mode. The cation budget is largely controlled by
ammonium in fine mode, whereas sodium is the most abundant cation in coarse
mode. The high abundance of SO42- in fine particles suggests a
significant contribution of anthropogenic sources including industrial
emissions in East Asia via long-range transport of aerosols over the western
North Pacific Rim. SO42- is an anthropogenic tracer of industrial
activities whereas NH4+ is the secondary product of NH3 that
is largely derived from the agricultural usage of nitrogen-based fertilizers
(Pakkanen et al., 2001) and volatilization from soils and livestock waste in
East Asia (Huang et al., 2006). The dominant presences of Na+ and
Cl- in coarse mode suggest a substantial contribution from sea salt.
Na+ and Cl- are emitted from the ocean surface as relatively
larger particles. A substantial amount of NO3- was detected in
coarse mode, suggesting a formation of Ca(NO3)2 or NaNO3 in
coarse fractions through the reactive adsorption of gaseous HNO3 onto
preexisting alkaline particles.
Average molecular distributions of water-soluble dicarboxylic
acids and related compounds in size-segregated aerosols collected on Okinawa
Island.
Pearson correlation coefficientsa (r) matrix among
the selected chemical species/components measured in the fine- and coarse-mode aerosols
from Okinawa Island in the western North Pacific Rim.
Fine modeb
Na+
NH4+
K+
Mg2+
Ca2+
MSA-
Cl-
NO3-
SO42-
WSOC
OC
C2
C3
C4
C5
C9
Ph
ωC2
ωC9
Gly
Benzoic
LWC
Na+
1.00
NH4+
-0.25
1.00
K+
-0.32
0.99
1.00
Mg2+
0.98
-0.16
-0.23
1.00
Ca2+
-0.21
0.62
0.33
-0.15
1.00
MSA-
-0.32
0.92
0.92
-0.17
0.53
1.00
Cl-
0.65
-0.85
-0.85
0.58
-0.33
-0.78
1.00
NO3-
0.65
-0.56
-0.55
0.68
0.22
-0.36
0.76
1.00
SO42-
-0.10
0.99
0.98
-0.02
0.59
0.89
-0.78
-0.49
1.00
WSOC
0.10
0.91
0.93
0.16
0.30
0.79
-0.57
-0.27
0.96
1.00
OC
0.12
0.91
0.95
0.16
0.25
0.80
-0.57
-0.32
0.93
0.99
1.00
C2
0.12
0.89
0.85
-0.13
0.22
0.80
-0.53
-0.30
0.92
0.99
0.98
1.00
C3
-0.05
0.90
0.89
-0.05
0.20
0.66
-0.68
-0.53
0.90
0.93
0.96
0.89
1.00
C4
-0.12
0.96
0.95
-0.09
0.15
0.76
-0.75
-0.55
0.96
0.95
0.96
0.92
0.99
1.00
C5
-0.12
0.99
0.96
-0.05
0.33
0.87
-0.80
-0.53
0.99
0.93
0.93
0.91
0.95
0.97
1.00
C9
0.64
0.01
0.02
0.61
0.42
-0.16
0.46
0.47
0.10
0.20
0.39
0.38
0.33
0.23
0.09
1.00
Ph
0.41
0.78
0.73
0.46
0.42
0.63
-0.40
-0.16
0.87
0.92
0.93
0.90
0.83
0.83
0.86
0.23
1.00
ωC2
0.11
0.92
0.90
0.19
0.19
0.82
-0.57
-0.25
0.96
0.99
0.99
0.99
0.90
0.93
0.95
0.36
0.93
1.00
ωC9
0.23
0.22
0.12
0.18
-0.56
-0.01
-0.32
-0.53
0.29
0.13
0.22
0.05
0.31
0.26
0.32
0.80
0.02
0.16
1.00
Gly
0.01
0.86
0.86
0.15
0.09
0.92
-0.52
-0.07
0.86
0.89
0.82
0.93
0.70
0.78
0.85
0.21
0.85
0.92
-0.11
1.00
Benzoic
-0.13
0.99
0.99
-0.05
-0.23
0.90
-0.27
0.46
0.99
0.96
0.99
0.93
0.91
0.96
0.99
0.12
0.85
0.96
0.21
0.90
1.00
LWC
0.16
0.87
0.83
0.30
0.53
0.88
-0.53
-0.13
0.92
0.90
0.87
0.92
0.82
0.83
0.89
0.18
0.90
0.95
0.19
0.95
0.91
1.00
Coarse modec
Na+
NH4+
K+
Mg2+
Ca2+
MSA-
Cl-
NO3-
SO42-
WSOC
OC
C2
C3
C4
C5
C9
Ph
ωC2
ωC9
Gly
Benzoic
LWC
Na+
1.00
NH4+
0.60
1.00
K+
0.96
0.77
1.00
Mg2+
0.98
0.63
0.33
1.00
Ca2+
-0.12
0.03
-0.06
-0.29
1.00
MSA-
-0.15
-0.66
-0.03
-0.25
-0.02
1.00
Cl-
0.98
0.59
0.90
0.98
-0.27
-0.22
1.00
NO3-
-0.30
-0.23
-0.15
-0.39
0.98
0.28
-0.55
1.00
SO42-
0.33
0.32
0.56
0.28
0.63
0.25
0.16
0.67
1.00
WSOC
-0.18
-0.26
0.06
-0.20
0.23
0.55
-0.36
0.92
0.72
1.00
OC
-0.11
-0.10
0.13
-0.10
0.21
0.36
-0.28
0.92
0.72
0.97
1.00
C2
-0.05
0.26
0.30
0.15
0.63
0.09
-0.08
0.88
0.76
0.93
0.82
1.00
C3
0.32
0.33
0.53
0.31
0.68
0.18
0.15
0.75
0.92
0.88
0.82
0.93
1.00
C4
0.33
0.39
0.60
0.35
0.53
0.16
0.33
0.32
0.88
0.31
0.55
0.36
0.63
1.00
C5
0.05
0.05
0.22
-0.06
0.62
0.32
-0.05
0.43
0.75
0.28
0.38
0.22
0.45
0.91
1.00
C9
0.85
0.20
0.25
0.91
-0.16
-0.59
0.85
-0.31
0.18
-0.08
-0.25
0.25
0.30
0.19
-0.23
1.00
Ph
-0.52
-0.54
-0.29
-0.54
0.73
0.59
-0.66
0.93
0.54
0.56
0.33
0.63
0.58
0.21
0.40
-0.58
1.00
ωC2
0.23
0.37
0.85
0.68
0.12
0.42
0.59
0.23
0.73
0.53
0.52
0.53
0.76
0.60
0.32
0.23
0.21
1.00
ωC9
0.83
0.53
0.82
0.87
-0.33
0.03
0.80
-0.22
0.21
0.07
0.16
0.28
0.38
0.08
-0.31
0.93
-0.28
0.33
1.00
Gly
0.26
0.26
0.78
0.57
0.05
0.52
0.58
0.06
0.69
0.28
0.33
0.22
0.55
0.76
0.57
0.24
0.12
0.89
0.13
1.00
Benzoic
-0.40
-0.60
-0.57
-0.36
-0.70
0.17
-0.29
-0.43
-0.88
-0.40
-0.35
-0.57
-0.73
-0.91
-0.77
-0.37
0.19
-0.48
-0.07
-0.51
1.00
LWC
0.61
0.03
0.53
0.56
-0.70
0.48
0.63
-0.51
-0.10
-0.19
-0.13
-0.29
-0.08
-0.03
-0.22
0.23
-0.31
0.57
0.25
0.63
0.31
1.00
See Tables 1 and 2 for abbreviations. a Correlation
is significant at 0.05 level for the values where r is > 0.80. b Fine
mode represents aerosol size of Dp < 2.1 µm. c Coarse mode
represents aerosol size of Dp > 2.1 µm.
The molecular distributions of detected diacids and related compounds in
size-segregated aerosols are shown in Fig. 3. Table 2 presents the
summarized concentrations of those compounds in fine and coarse modes.
Oxalic acid (C2) was found to be the most abundant diacid followed by
malonic (C3) and succinic (C4) acids in all size-segregated
aerosols. The predominance of C2 in size-segregated aerosols is due to
the fact that it can be secondarily produced by the photooxidation of
anthropogenic and biogenic organic precursors in gas and aqueous phase
(Kawamura and Sakaguchi, 1999; Warneck, 2003; Carlton et al., 2006). C2
can also be produced primarily from fossil fuel combustion (Kawamura and
Kaplan, 1987) and biomass burning (Kundu et al., 2010) in East Asia
and is long-range-transported to Okinawa.
Phthalic (Ph) and adipic (C6) acids are the next abundant diacids,
whereas ketomalonic acid (kC3) is more abundant than C6 diacid in
the size ranges of 0.43–0.65 to 0.65–1.1 µm (Fig. 3). Ph
and C6 diacids originate from various anthropogenic sources and thus
they can be used as anthropogenic tracers. Ph primarily originates from coal
burning and vehicular emissions, whereas photooxidation of aromatic
hydrocarbons such as naphthalene and o-xylene derived from incomplete
combustion of fossil fuel form Ph via secondary processes (Kawamura and
Kaplan, 1987). Moreover, abundant presence of Ph may also be caused by
enhanced emission of phthalates from plastics used in heavily populated and
industrialized regions in China and the subsequent long-range atmospheric
transport to Okinawa. Phthalic acid esters are used as plasticizers in
resins and polymers (Simoneit et al., 2005). They can be released into the
air by evaporation because they are not chemically bonded to the polymer.
Kawamura and Usukura (1993) reported that C6 diacid is an oxidation
product through the reaction of cyclohexene with ozone (O3). The high
abundances of Ph and C6 diacids on Okinawa suggest a significant
influence of anthropogenic sources in East Asia via long-range transport of
aerosols over the western North Pacific Rim.
Azelaic acid (C9) is more abundant than adjacent suberic (C8) and
decanedioic (C10) acids in all the size-segregated aerosols (Fig. 3
and Table 2). Kawamura and Gagosian (1987) proposed that C9 is a
photooxidation product of biogenic unsaturated fatty acids such as oleic
acid (C18:1) containing a double bond at C-9 position. Unsaturated
fatty acids can be emitted from sea surface microlayers and from local
vegetation on Okinawa (Kunwar and Kawamura, 2014). Moreover, Okinawa was
affected by long-range transport of air masses from Siberia and Mongolia as
well as North China and Korea (Fig. 2). Such continental air masses can
also deliver C9 via atmospheric processing of unsaturated fatty acids
during long-range transport. The abundant presence of C9 indicates that
atmospheric oxidation of unsaturated fatty acids also occurs in Okinawa
aerosols during long-range transport. ω-Oxocarboxylic acids and
α-dicarbonyls were detected in the Okinawa aerosols. Glyoxylic acid
(ωC2) was identified as the most abundant ω-oxoacid,
whereas glyoxal (Gly) was more abundant than methylglyoxal (MeGly) in all
the sizes. ωC2 and Gly are the oxidation products of several
anthropogenic and biogenic VOCs and are primarily generated by fossil fuel
combustion and biomass burning (Zimmermann and Poppe, 1996; Volkamer et al.,
2001), and are further oxidized to C2 diacid (Myriokefalitakis et al.,
2011). The predominance of ωC2 and Gly indicates their
importance as key precursors of C2 in Okinawa aerosols.
Size distributions of water-soluble inorganic ions in the aerosol
samples collected on Okinawa Island.
Inorganic species
The particle size distributions of major ions are shown in Fig. 4. Pearson
correlation coefficients (r) among the measured ions in different size modes
are given in Table 3. Na+ and Cl- are mainly derived from the
ocean surface as sea salt particles in the marine atmosphere (Kumar et al.,
2008; Geng et al., 2009). The size distributions of Na+ and Cl-
were found to be bimodal with two peaks in coarse mode (Fig. 4a and b).
Their peaks at 2.1–3.3 or 3.3–4.7 µm and at > 11.3 µm suggest that they are of marine origin due to bubble bursting of surface
seawater. Andreas (1998) suggested that the sea spray falls into two types
that are defined as film and jet bubbles; film bubbles correspond to the
size of 0.5–5 µm, whereas jet bubbles are produced the size of 5–20 µm.
Their coarse-mode peaks at 2.1–3.3 µm or 3.3–4.7
and > 11.3 µm in Okinawa aerosols were associated with film and jet
bubbles. We found that size distribution of Mg2+ is similar to those of
Na+ and Cl- with a significant positive correlation to coarse-mode
Na+ and Cl- (r= 0.98), suggesting their similar origin and
sources.
A high concentration of Ca2+ in coarse-mode particles demonstrates its
contribution from soil dust (Kerminen et al., 1997a; Tsai and Chen, 2006).
A lifting of soil dust in continental sites followed by subsequent
long-range atmospheric transport to a remote marine site is also proposed as
an important source of Ca2+ (Y. Wang et al., 2005). Ca2+ showed
unimodal distribution with a peak at either 2.1–3.3 or 3.3–4.7 µm
(Fig. 4c). The coarse-mode Ca2+ is mostly derived from crustal
CaCO3, which heterogeneously reacts with acidic gases (HNO3 and
SO2) (Kerminen et al., 1997a). This formation mechanism is further
supported by a strong correlation of coarse-mode Ca2+ with
NO3- (r= 0.98). There is no correlation between Ca2+ and
Na+ or Cl- (r= -0.12 or -0.27), revealing that sea salt
contribution of Ca2+ is negligible in Okinawa aerosols. This result
suggests that long-range transport of soil dust is an important contributor
of Ca2+ in the marine aerosols from the western North Pacific Rim.
There are natural limestone caves formed by elevated coral reefs on Okinawa
Island. Although local limestone dust may also be resuspended in the
atmosphere by wind (Shimada et al., 2015), the local dust contribution to
the ambient level of Ca2+ on Okinawa may be small. This interpretation
can be supported by the fact that Ca2+ peaked in lower coarse size
range of 2.1–3.3 or 3.3–4.7 µm. It has been suggested that Ca2+
is likely associated with upper coarse size range when the contribution of
locally produced soil particles is significant (Bian et al., 2014). Smaller
coarse-mode Ca2+ is likely associated with long-range-transported Asian
dust to Okinawa. Moreover, concentrations of Ca2+ in coarse mode were
found to be much higher in OKI-1 (0.51 µg m-3) and OKI-2 (0.60 µg m-3) than that in the OKI-5 sample (0.15 µg m-3).
Backward trajectories also indicated that the air masses which originated from
Mongolia and Siberia were transported to Okinawa during the collection of
OKI-1 and OKI-2 samples, whereas the OKI-5 sample has an influence of marine air
masses. Such air mass origin again indicates the long-range transport of Asian
dust from East Asia to the western North Pacific.
Potassium is enriched in biomass burning aerosols and therefore its
abundances in fine particles can serve as a diagnostic tracer of biomass
burning (Yamasoe et al., 2000). Moreover, contributions of K+ from sea
salt and dust sources are highly variable in regional case studies, with its
dominance in coarse-mode particles. Fresh biomass burning particles mostly
reside in the condensation mode at 0.1–0.5 µm in diameter (Kaufman
and Fraser, 1997; Kleeman and Cass, 1999). A unimodal size distribution of
K+ was observed in most sample sets (OKI-1 to OKI-4), with a peak at
0.65–1.1 µm in diameter (Fig. 4e). The peak of K+ at 0.65–1.1 µm suggests that biomass burning particles emitted in East Asia might
have undergone growth to a relatively large size by absorbing water vapor
from the atmosphere during long-range transport to Okinawa. This
interpretation is supported by the fact that K+ showed a positive
correlation with LWC (r= 0.83) in fine mode. The fine-mode nss-K+
accounted for 95 % of total K+ in the OKI-2 sample set and 88 % of that
in the OKI-3 sample set when air masses come from Siberia and Mongolia as
well as North China. The abundant presence of fine-mode nss-K+ in the
OKI-2 and OKI-3 samples further indicates a long-range atmospheric transport
of biomass burning aerosols from the Asian continent to the western North
Pacific Rim.
NOx is a precursor of NO3-, which can be converted to
HNO3 and then react with NH3 to form NH4NO3. A unimodal
size distribution of NO3- was observed with a peak at 2.1–3.3 or
3.3–4.7 µm in diameter (Fig. 4f). It should also be noted that the
NO3- concentration in coarse mode is much higher than that in fine
mode (Table 1). This result suggests that either dust or sea salt particles
are the source of coarse-mode NO3- on Okinawa. Coarse-mode
NO3- is produced by heterogeneous reaction of gaseous NO2 or
HNO3 with alkaline metals such as Na+ and Ca2+ as shown in
Reactions (R1) and (R2) (Kouyoumdjian and Saliba, 2006; Seinfeld and Pandis,
2006).
HNO3(g)+NaCl(aqands)→NaNO3(aqands)+HCl(g)2HNO3(g)+CaCO3(s)→Ca(NO3)2(s)+H2O+CO2(g)
As discussed earlier, the air masses which originated from Siberia are transported
over Mongolia and North China. Asian dust can be transported from the Asian
continent to Okinawa. Therefore, it is possible that the gaseous HNO3
might already have reacted with CaCO3 (mineral dust particle) to form
NO3- before arriving to Okinawa through R-2. We found that coarse-mode Na+, which is derived from sea salts, is negatively correlated
(r= -0.30) with coarse-mode NO3-. Although this correlation is
not significant (p= 0.51), the negative correlation may indicate some
reactive loss of NO3- from sea salt particles in coarse mode on
Okinawa. NO3- peaked at the same particle size of Ca2+.
Therefore, NO3- in Okinawa coarse-mode aerosols probably
resulted from the uptake of HNO3 gas by soil dust particles enriched
with Ca2+ via heterogeneous reactions near the source regions. This
process is further supported by a good correlation between NO3-
and Ca2+ (r= 0.98) in coarse mode.
The particle size distributions of SO42-, which is a major source
of acid deposition (Pakkanen et al., 2001), have been the subject of
numerous studies in the past few decades (Huang et al., 2006; Kouyoumdjian
and Saliba, 2006). Condensation-mode SO42- arises from gas-phase
oxidation of SO2 followed by gas-to-particle conversion, whereas fine-mode SO42- is formed through aqueous-phase oxidation of SO2
in aerosols and cloud droplets (Seinfeld and Pandis, 1998). SO42-
on coarse mode can be attributed to a combination of sulfate and
heterogeneous reactions of SO2 on soil dust or sea salt particles
(Seinfeld and Pandis, 1998; Pakkanen et al., 2001). A unimodal size
distribution of SO42- was observed with a peak at 0.65–1.1 µm (Fig. 4g). Gao et al. (2012) suggested that an in-cloud process produces
SO42- as larger particles by aqueous-phase oxidation of SO2
in cloud droplets. Therefore, the peak of SO42- at 0.65–1.1 µm on Okinawa may be involved with aqueous-phase oxidation of
SO2 in aerosols.
Size distribution of methanesulfonate (MSA-) is similar to that of
SO42- (Fig. 4i) on Okinawa. MSA- showed a strong
correlation with SO42- (r= 0.89) in fine mode, suggesting that
MSA- should have similar origin with SO42- in fine mode.
Although MSA- is produced by gas-to-particle conversion via the
oxidation of dimethyl sulfide (DMS) emitted from the ocean (Quinn et al.,
1993; Kerminen et al., 1997b), there is some indirect evidence that
liquid-phase production might also be possible (Jefferson et al., 1998).
Biomass burning also produces DMS in the atmosphere (Meinardi et al., 2003;
Geng and Mu, 2006). MSA- showed high correlation with K+ or
NH4+ (r= 0.92) in fine mode, indicating that an enhanced
emission of DMS from biomass burning followed by the subsequent oxidation
during long-range transport may have contributed significantly to fine-mode
MSA- on Okinawa. Moreover, MSA- can also be produced in fine mode
by the oxidation of DMS that is emitted from marine phytoplankton in the
surrounding ocean. It is noteworthy that East Asian aerosols traveled over
the marine regions including the East China Sea, Sea of Japan, and Pacific
Ocean during long-range atmospheric transport. The size distribution of
MSA- observed over Okinawa is consistent with previous studies from the
China Sea by Gao et al. (1996), who suggested that MSA is produced through
the oxidation of S-containing species in the marine atmosphere.
Size distributions of selected water-soluble dicarboxylic acids
and related compounds in the aerosol samples collected on Okinawa Island.
NH4+ in the Okinawa aerosols shows a unimodal size distribution
with a peak at 0.65–1.1 µm (Fig. 4h), indicating that
NH4+ is mainly formed by gas-to-particle conversion via the
reaction with H2SO4 and HNO3. Interestingly, the size
distribution of NH4+ is similar to that of SO42- and
diacids such as oxalic acid (Figs. 4g and 5a). We also found a strong
correlation between SO42- and NH4+ on fine mode (r= 0.99). Ion balance calculations are commonly used to evaluate acid-base
balance of aerosol particles. Average equivalent ratios of total cations
(Na+, NH4+, K+, Mg2+, and Ca2+) to anions
(Cl-, NO3-, and SO42-) in fine fractions varied from
0.75 for the size bin of 0.65–1.1 µm to 0.86 for the size bin of
1.1–2.1 µm, indicating that fine-mode aerosols on Okinawa were
apparently acidic.
NH3 is an alkaline gas that neutralizes the acidic particles in the
atmosphere. Kerminen et al. (1997a) proposed that particulate NH4+
is secondarily formed via heterogeneous reactions of gaseous NH3 with
acidic species (H2SO4 and HNO3). The reaction of NH3
with H2SO4 is favored over its reaction with HNO3. The
average NH4+/SO42- equivalent ratios in fine-mode
particles on Okinawa varied from 0.36 for the size bin of 1.1–2.1 µm
to 0.81 for the size bin of 0.43–0.65 µm, indicating that NH3
was not abundant enough to neutralize all SO2. The aerosol chemical
composition data obtained from the ISORROPIA II model revealed that
significant amounts of SO42-, HSO4-, and NH4+
in fine mode were present in liquid phase, whereas SO42- and
NO3- were mainly present as solid phase in the coarse-mode
aerosols in the forms of CaSO4 and Ca(NO3)2, respectively.
Interestingly, the average NH4+/SO42- equivalent ratios
in coarse-mode particles ranged from 0.01 for the size bin > 11.3 µm to 0.09 for the size bins of 2.1–3.3 and 3.3–4.7 µm,
suggesting that coarse-mode aerosols on Okinawa were also
NH4+-poor. This result further indicates that there was not enough
NH3 to neutralize HNO3, and thus the shortfall of NH3 may be the
restrictive factor for the formation of NH4NO3 in Okinawa
aerosols. Therefore, NO3- reacts with coarse particles that
contain alkaline species (Ca2+) in Okinawa aerosols.
The size distribution of SO42- depends on the concentration of
NH4+, richness of NH3 in the air, and the presence of coarse-mode particles. SO42- and NH4+ often coexist in fine
mode because H2SO4 condenses on this mode as fine particles that
have more surface area (Jacobson, 2002). Although NH3 was not abundant
enough to neutralize all SO42-, most of SO42- might be
neutralized by NH3 in fine mode. Hence, SO42- is enriched in
fine mode rather than being associated with dust particles. An enrichment of
NO3- in the dust fraction in our study is supported by the
laboratory studies of Hanisch and Crowley (2001a, b), who found a large
and irreversible uptake between HNO3 and various authentic dust samples
including samples from the Chinese dust region.
Water-soluble organic carbon (WSOC) and organic carbon (OC)
The mass-based size distribution of WSOC is characterized by a major peak at
0.65–1.1 µm in fine mode and by a small peak at 3.3–4.7 µm in
coarse mode (Fig. 6a and Table 1). Huang et al. (2006) observed that fine-mode WSOC was primarily derived from combustion sources and secondarily
produced in the atmosphere by the photochemical oxidation of VOCs. The WSOC
concentrations showed a strong correlation with fine-mode SO42-
(r= 0.96). Because production of SO42- is closely linked to
photochemical activity, this result suggests an important secondary
production of WSOC in fine-mode particles during long-range atmospheric
transport from East Asia. WSOC concentrations also showed high correlation
with K+ (r= 0.93) and NH4+ (r= 0.91) in fine mode. This
result suggests that direct emission from biomass burning or fast oxidation
of biomass-burning-derived precursors significantly contributes to the
formation of fine-mode WSOC in Okinawa aerosols during long-range transport.
Size distributions of water-soluble organic carbon (WSOC) and
organic carbon (OC) in the aerosol samples collected on Okinawa Island.
The mass size distribution pattern of OC is similar to that of WSOC with a
major peak in the size range of 0.65–1.1 µm, whereas a small peak
appeared in the size range of 3.3–4.7 µm in diameter (Fig. 6b).
Primary emission from biomass burning and/or photooxidation of biomass-burning-derived
precursors might be a dominant source of fine-mode OC in
Okinawa aerosols. This interpretation is supported by the fact that OC
showed a strong correlation (r= 0.95) with K+ in fine mode. The
fine-mode OC showed significant positive correlations with SO42-
(r= 0.93) and NH4+ (0.91), suggesting a secondary photochemical
formation of OC in the fine mode of Okinawa aerosols.
A significant portion of OC may be oxidized to WSOC during the atmospheric
transport from East Asia to the western North Pacific. The mass ratio of
WSOC / OC has been proposed as a measure of photochemical processing or aging
of organic aerosols especially in long-range-transported aerosols (Aggarwal
and Kawamura, 2009). The WSOC / OC ratios varied from 0.51 to 0.76 with an
average of 0.67 ± 0.09 in the fine mode and from 0.43 to 0.63 with an average
of 0.55 ± 0.09 in the coarse mode. The higher WSOC / OC ratio in fine
mode suggests that organics are significantly subjected to photochemical
processing in fine aerosols during long-range transport from the Asian
continent to Okinawa compared to coarse-mode aerosols.
Source contributions and secondary processes that may convert VOCs to more
soluble forms on the surface area of fine particles could cause higher
WSOC / OC ratios in fine mode. Biomass-burning-derived OC is highly
water-soluble and usually resides in fine mode, whereas coarse-mode OC
contains high molecular weight organic compounds emitted by soil
resuspension and emissions of pollen and fungal spores, which are less
water-soluble (Wang et al., 2011; Mkoma et al., 2013). Biomass burning
significantly contributed to fine-mode WSOC on Okinawa as discussed above.
Moreover, accumulation of gas-phase precursors of WSOC may occur
preferentially in the particle size, with the greatest surface area
(Kanakidou et al., 2005). It has been proposed that fine particles offer
more surface area, and thus the reaction rate is more on the surface of fine
particles than coarse particles (Kanakidou et al., 2005). The higher WSOC / OC
ratio in fine particles than coarse particles has also been observed in
long-range-transported East Asian aerosols over northern Japan (Agarwal et
al., 2010).
The WSOC / OC ratio in fine mode showed a weak positive correlation with
downward solar radiation flux (r= 0.39). This weak correlation is probably
due to the fact that fine-mode WSOC can be produced in the aqueous phase of
aerosols during long-range transport. Based on the year-round measurements
of total suspended aerosols from Okinawa Island, Kunwar and Kawamura (2014) documented
higher WSOC / OC ratio in winter (ave. 0.60) and spring (ave. 0.45) than
summer (ave. 0.28). These observations demonstrate that WSOC can be produced
from OC under a weak solar radiation condition on the transport pathway from
the source region to Okinawa possibly via aqueous-phase processing.
Calculated LWC for each sample from Okinawa and average LWC in
size-segregated aerosols are shown in Fig. 7. The highest LWC was found at
the size of 0.65–1.1 µm in the fine mode in Okinawa samples. WSOC can
also contribute to aerosol LWC, although its ability to absorb water is
significantly less than that of inorganics (Ansari and Pandis, 2000; Speer
et al., 2003; Engelhart et al., 2011). Moreover, organic species are not
taken into account in ISORROPIA II for the calculation of LWC. It is
noteworthy that WSOC / OC ratio and LWC in fine mode significantly correlate
with r= 0.87, whereas negative correlation was found in coarse mode (r= -0.19),
suggesting a possible production of WSOC from OC in aerosol
aqueous phase in the fine mode of Okinawa aerosols. There may also be another
important source of fine-mode WSOC in Okinawa aerosols such as primary
emission from biomass burning and secondary formation via gas-phase
photochemical reactions during long-range atmospheric transport (Hagler et
al., 2007; Lim et al., 2010). This result may indicate that shorter-chain
diacids and related polar compounds can contribute more to fine-mode WSOC
via oxidation of various organic precursors during long-range transport
(Carlton et al., 2007; Kawamura et al., 2005, 2007; Miyazaki et al., 2010).
Aerosol liquid water contents for each sample in size-segregated
aerosols and average liquid water contents in size-segregated aerosols on
Okinawa Island.
Dicarboxylic acids and related compounds
The size distributions of selected diacids and related compounds are shown
in Fig. 5. Based on the sources and formation processes, their size
distributions fall into two groups: a group with a dominant fine mode and a
group with a dominant coarse mode, as discussed in the following sections.
C2, ωC2, Gly, Ph, and benzoic acid
The first group, including C2, ωC2, Gly, Ph, and benzoic
acid, showed similar size distributions to maxima in fine mode.
C2 showed a peak at 0.65–1.1 µm in fine mode (Fig. 5a). The
size distribution of C2 on Okinawa is different from that observed off
the coast of East Asia by Mochida et al. (2003a, 2007), who found a strong
bimodal pattern of C2 with a peak in the coarse mode. They suggested
that the coarse-mode peak of C2 emerged by the uptake of gaseous
diacids or heterogeneous oxidations of organic precursors on the dust and
sea salt particles during long-range transport. The unimodal distribution of
C2 on Okinawa with maxima in fine mode suggests that the heterogeneous
uptake of C2 on dust and sea salt particles did not occur.
The condensation mode of C2 is likely produced photochemically in the
gas phase followed by condensation onto preexisting particles at 0.1–0.5 µm (Huang et al., 2006). The fine-mode peak of C2 at the size of
0.65-1.1 µm in Okinawa aerosols suggests a preferential production of
C2 via the oxidation of precursors in aerosol aqueous phase during
long-range atmospheric transport. We found that size distribution of C2
diacid is similar to that of SO42- (Figs. 4g and 5a), suggesting
a secondary formation of C2, possibly in aerosol aqueous phase. The good
correlations of C2 with SO42- (r= 0.92) and NH4+
(0.89) in fine mode further supports that C2 is a secondary
photochemical product. Fine-mode C2 can also be produced primarily from
fossil fuel combustion and biomass burning in East Asia and long-range-transported to Okinawa. C2 diacid showed a significant positive
correlation with fine-mode K+ (r= 0.85), indicating that biomass
burning contributed significantly to fine-mode C2 in Okinawa aerosols.
Lim et al. (2005) and Legrand et al. (2007) reported the formation of
diacids in aqueous phase. Here we investigate the impact of LWC on the
formation of diacids in Okinawa aerosols. LWC of a particle can influence
the production of C2 via the changes in gas/particle partitioning of
organic precursors and subsequent heterogeneous reactions in aerosol
aqueous phase. A strong positive correlation (r= 0.92) of C2 with LWC
was found in fine mode, whereas the correlation was negative in coarse mode
(r= -0.29), indicating a possible aqueous-phase production of C2 via
the oxidation of C2 precursors in fine mode. Several secondary
formation pathways are known to C2 in atmospheric aerosols. C2 is
produced by the decay of its higher homologues (C3–C5 diacids) or
oxidation of unsaturated fatty acids such as oleic acid (C18:1)
followed by the degradation to shorter-chain diacids in aqueous phase
(Kawamura and Ikushima, 1993; Kawamura and Sakaguchi, 1999; Pavuluri et al.,
2015). C2 can also be produced by the aqueous-phase oxidation of
ωC2, which can be formed by aqueous oxidation of Gly and MeGly,
produced by the oxidation of various VOCs including toluene, ethene, and
isoprene (Zimmermann and Poppe, 1996; Volkamer et al., 2001; Lim et al.,
2005; Carlton et al., 2006; Ervens et al., 2008).
The scatter plots of C2 with C3–C5 diacids in fine and coarse
modes are shown in Fig. S3. The robust correlations of C2 with
C3–C5 diacids (r= 0.89–0.92) were found in fine mode, indicating
that they might have similar sources and origin or C2 may be produced
via the decay of its higher homologues (C3–C5 diacids) during
long-range transport. The differences in the slopes of linear regression of
C2 with C3 and C4 diacids between fine and coarse modes are
not significant but slopes are slightly higher in fine mode than the coarse
mode (Fig. S3a–d and Table S1). Interestingly, a significantly higher slope
was observed for the regression line between C2 and glutaric (C5) acid
in fine mode than coarse mode (Fig. S3e–f and Table S1). It is also
noteworthy that the slope of the regression line of C2 with C5 diacid
is significantly higher than that for C3 and C4 diacids in fine
mode (Fig. S3a, c, e and Table S2). These results indicate that fine-mode
oxalic acid may be produced from oxidation of glutaric acid during
long-range transport via succinic and malonic acids as intermediates. The
laboratory studies of Hatakeyama et al. (1985) and Kalberer et al. (2010)
have documented that glutaric acid is produced by the oxidation of
cyclohexene by O3, which can be further oxidized in aqueous phase to
result in oxalic acid (Kawamura and Sakaguchi, 1999; Legrand et al., 2007).
This interpretation is further supported by the fact that C3–C5
diacids were enriched in the fine mode of most samples (Fig. 5b–d) and
showed good correlations with LWC (r= 0.82–0.89), possibly due to the
enhanced secondary production by the oxidation of its precursor compounds in
aerosol aqueous phase.
The size distribution of ωC2 and Gly is similar to that of
C2 diacid in the Okinawa samples (Fig. 5e and f). The enrichment of
ωC2 and Gly in fine mode may be associated with enhanced
secondary formation via aqueous-phase processing of their precursors during
long-range transport. This interpretation is evidenced by the fact that
strong correlations of ωC2 and Gly were found with
SO42- (r= 0.96 and 0.86, respectively) and LWC (0.95) in fine
mode. The fine-mode ωC2 and Gly can also be produced primarily
from biomass burning in East Asia and be long-range-transported to Okinawa.
Significant positive correlations between ωC2 and K+ (r= 0.90), and Gly and K+ (0.86) suggest that biomass burning contributed
significantly to the fine-mode ωC2 and Gly in Okinawa aerosols.
Gly is a well-known precursor of ωC2 and C2 in atmospheric
aerosols (Lim et al., 2005; Ervens and Volkamer, 2010; Myriokefalitakis et al.,
2011). The preferential enrichment of Gly and ωC2 in fine
mode can form C2 in Okinawa aerosols by aqueous-phase processing.
High correlations among C2, ωC2, and Gly in fine mode (r= 0.92–0.99) also indicate their similar sources and formation processes and
that C2 diacid may be produced by the oxidation of ωC2 and
Gly in fine mode. There is no significant difference in the slope of
regression line of C2 with ωC2 between the fine and coarse
modes (Fig. S3g–h and Table S1), whereas the slope of the regression line of
C2 with Gly is significantly higher in fine mode than coarse mode
(Fig. S3i–j and Table S1). It is also remarkable that the slope of linear
regression of C2 with Gly is significantly higher than that with
ωC2 in fine mode (Fig. S3g–i and Table S2). This result may
indicate a possible formation of fine-mode oxalic acid from glyoxal via
glyoxylic acid as an intermediate during long-range atmospheric transport in
the western North Pacific.
The enrichment of C2, ωC2, and Gly in fine mode on Okinawa
was probably due to the enhanced oxidation of anthropogenic precursors
emitted in East Asia during long-range transport because their size
distributions are consistent with that of Ph and benzoic acid (Fig. 5g and h), which are tracers of anthropogenic sources. The strong correlations of
fine-mode C2, ωC2, and Gly with Ph (r= 0.85–0.93) and
benzoic acid (r= 0.90–0.96) further suggest that anthropogenic precursors
are their important sources in fine mode. Ph and benzoic acid are directly
emitted from combustion sources and secondarily produced in the atmosphere
by the photooxidation of aromatic hydrocarbons emitted from the incomplete
combustion of fossil fuel (Kawamura et al., 1985; Kawamura and Kaplan, 1987;
Ho et al., 2006).
Aromatic hydrocarbons such as naphthalene and toluene have been suggested as
major precursors of Ph and benzoic acid, respectively (Schauer et al., 1996;
Kawamura and Yasui, 2005). Based on the high levels of naphthalene and
toluene in China (Liu et al., 2007; Tao et al., 2007; Duan et al., 2008), Ho
et al. (2015) recently suggested that oxidation of naphthalene and toluene
in the atmosphere is one of the major sources of Ph and benzoic acid,
respectively. High levels of precursors in the source regions might favor
the significant secondary production of Ph and benzoic acid during
long-range transport in the western North Pacific. It may be possible that
their precursors emitted in East Asia were taken up by aqueous-phase aerosol
and oxidized to result in Ph and benzoic acid in fine mode during long-range
transport. Moreover, enrichment of Ph and benzoic acid in fine mode further
suggests that these species are associated with combustion sources either by
primary emission and/or secondary production from the precursor compounds,
being consistent with other anthropogenic SO42-, NH4+,
and K+. Fine-mode Ph can also be produced from evaporation of
phthalates from plastics used in populated and industrialized regions in
East Asia and long-range-transported to Okinawa as discussed earlier. This
explanation is consistent with the enrichment of terephthalic acid (tPh) in
fine mode (Fig. 5i), which is a tracer of plastic burning (Kawamura and
Pavuluri, 2011).
C9 and ωC9
The second group of organic compounds, including C9 and ωC9, showed bimodal size distribution with a major peak on coarse mode
at 3.3–4.7 µm and minor peak on fine mode at 0.65–1.1 µm
(Fig. 5j and k). The strong correlations were found between C9 and
Na+ (r= 0.85), and ωC9 and Na+ (0.83) in coarse
mode, indicating that C9 and ωC9 may be emitted into the
atmosphere from the sea surface microlayers together with sea salt particles
on Okinawa. Kawamura and Gagosian (1987) suggested that C9 and ωC9 are also derived from the photooxidation of unsaturated fatty
acids such as oleic acid (C18:1) that are produced by phytoplankton and
emitted from sea surface microlayers as sea salt particles. The laboratory
experiments also documented the formation of C9 and ωC9
due to photooxidation of C18:1 (Matsunaga et al., 1999; Huang et al.,
2005; Ziemann, 2005; Tedetti et al., 2007). Sea surface microlayers in the
surroundings of Okinawa can also emit unsaturated fatty acids together with
sea salts. Therefore, the major peaks of C9 and ωC9 on the
coarse mode may be derived from heterogeneous oxidation of unsaturated fatty
acids of marine phytoplankton origin on the sea salt particles.
Wang et al. (2011) suggested that unsaturated fatty acids can be directly
emitted as fine particles from food cooking emissions in urban areas in China
and be oxidized to C9 diacid in fine mode. The minor peak of C9
and ωC9 in fine mode can be explained by the oxidation of
fine-mode unsaturated fatty acids derived from food cooking or gaseous
unsaturated fatty acids during long-range transport to the western North
Pacific.
Ratios of selected diacids
Kawamura and Ikushima (1993) proposed that the malonic to succinic acid ratio
(C3 / C4) is a tracer to evaluate the extent of photochemical
processing of organic aerosols. Because C4 is oxidized to C3, an
increase in the C3 / C4 ratio indicates an increased photochemical
processing. The average C3 / C4 ratio in sum of all the size
fractions was found to be 1.5 ± 0.1 in Okinawa aerosols. This result
suggests that the extent of photochemical processing is much greater on
Okinawa than Los Angeles (0.35) (Kawamura and Kaplan, 1987) but similar to
that of urban Tokyo (1.5) (Kawamura and Ikushima, 1993), whereas it is lower
than those of marine aerosols at Chichijima Island in the western North
Pacific (2.0) (Mochida et al., 2003b) and the remote Pacific including
the tropics (3.9) (Kawamura and Sakaguchi, 1999). Figure 8a shows changes in the
C3 / C4 ratios as a function of particle size. The C3 / C4
ratios exhibit higher values at 1.1–2.1 µm in fine mode and at
2.1–3.3 and 3.3–4.7 µm in coarse mode. This result suggests that
C3 production via C4 decomposition occurs more efficiently at
these size ranges by aqueous-phase processing.
Mass concentration ratios of malonic to succinic acid and phthalic
to azelaic acid in size-segregated aerosols collected on Okinawa Island.
Ph diacid originates from various anthropogenic sources, whereas C9
diacid is specifically produced by the oxidation of biogenic unsaturated
fatty acids (Kawamura and Gagosian, 1987; Kawamura and Ikushima, 1993).
Therefore, Ph / C9 ratio is most likely used as a tracer to understand
the source strength of anthropogenic vs. biogenic sources of diacids.
A higher Ph / C9 ratio shows more influence of anthropogenic sources,
whereas a lower ratio shows more influence of biogenic sources. Figure 8b
presents changes in the ratios of Ph / C9 as a function of particle
sizes. The higher Ph / C9 ratios were obtained on fine-mode particles
rather than coarse-mode particles. These results suggest that fine aerosols on
Okinawa are significantly influenced by anthropogenic sources whereas the
coarse aerosols are more influenced by biogenic sources. A significant
contribution of Ph on fine mode further supports that anthropogenic sources
are an important source of diacids and related compounds in the fine mode of
Okinawa aerosols.
Summary and conclusions
Nine-stage atmospheric particles from < 0.43 to > 11.3 µm in
diameter, collected in spring 2008 at Cape Hedo, Okinawa, in the
western North Pacific Rim, were analyzed for water-soluble diacids and
related compounds as well as water-soluble organic carbon (WSOC), organic
carbon (OC), and inorganic ions. The molecular distributions of diacids were
characterized by the predominance of oxalic acid (C2) followed by
malonic (C3) and succinic (C4) acids in all stages, suggesting
that they are most likely produced by the photooxidation of VOCs and
particulate organic precursors in the source region and/or during long-range
atmospheric transport. The abundant presence of SO42- as well as
phthalic and adipic acids in Cape Hedo suggested a significant contribution
of anthropogenic sources including industrial emissions in East Asia to
Okinawa aerosols via long-range atmospheric transport.
SO42-, NH4+, and diacids up to 5-carbon atoms as well as
glyoxylic acid (ωC2) and glyoxal (Gly) showed good correlations
with peaks in fine mode (0.65–1.1 µm). WSOC and OC also peaked on
fine mode with an additional minor peak on coarse mode. Similar size
distributions and strong correlations of diacids (C2–C5), ωC2 and Gly with SO42- in fine mode suggest their secondary
formation possibly in aerosol aqueous phase. Their strong correlations with
LWC in fine mode further suggest an importance of the aqueous-phase
production in Okinawa aerosols. They may have also been directly emitted
from biomass burning as supported by strong correlations with K+ in
fine mode. The robust correlations of C2 with C3–C5 diacids
as well as ωC2 and Gly indicate that they are the key
precursors of C2 diacid in Okinawa aerosols.
Longer-chain diacid (C9) and ω-oxoacid (ωC9)
showed bimodal size distribution with a major peak on coarse mode,
suggesting that they were directly emitted and/or produced by photooxidation
of unsaturated fatty acids mainly derived from sea surface microlayers via
heterogeneous reactions on sea spray particles. We observed that WSOC and OC
in fine particles are photochemically more processed in the atmosphere than
in coarse particles during long-range transport. This study demonstrates
that anthropogenic and biomass burning aerosols emitted from East Asia have
significant influence on the molecular compositions of water-soluble organic
aerosols in the western North Pacific Rim.
Data availability
The data from this paper can be obtained by contacting the authors of this article.
Meteorological data including ambient temperature, relative humidity, and
wind speed for each sample period were obtained from the Japan Meteorological
Agency (http://www.jma.go.jp).