We determined stable carbon isotopic composition (δ13C) of C2 to C9 dicarboxylic acids (DCAs) and some ketoacids in the marine aerosols from the western Pacific and Southern Ocean (35°N to 65°S). On average, oxalic (mean, −16.8‰), adipic (−17.2‰), and glyoxylic (−17.6‰) acids showed heavier δ13C, whereas maleic (−24.2‰), methylmalonic (−23.6‰), and phthalic (−23.1‰) acids were more depleted in 13C. Oxalic acid presented very diverse δ13C values (−27 to −7‰), which increased from midlatitudes toward the equator. A similar latitudinal trend of δ13C was also found for malonic, succinic, and adipic acids. However, such a trend was not observed for phthalic, maleic, and glyoxylic acids. We suggest that the δ13C increase toward the equator is associated with photochemically aged air masses, in which kinetic isotope effects for photochemical degradation of DCAs may be important. Differences in δ13C of some saturated DCAs were also found to increase from midlatitudes to the equator.
 Here we report the stable carbon isotopic compositions of LMW DCAs and one α-ketoacid and their latitudinal changes in the marine aerosols collected from the western Pacific and Southern Ocean. The data sets are discussed in terms of photochemical aging of the organic acids in the atmosphere. The results are also discussed using air mass back trajectories.
2. Experimental Methods
2.1. Aerosol Sampling and Analysis for Concentrations of DCAs and Ketoacids
 Marine aerosols were collected during a cruise between Tokyo and Antarctica conducted from 22 November 1994 to 11 February 1995 (KH94-4, R/V Hakuho Maru). The cruise track and surface wind conditions are shown in Figure 1. A high-volume air sampler loaded with a preheated (450°C, >3 hours) quartz fiber filter (20 × 25 cm2) was deployed for aerosol collection on an upper deck of the ship. The sampler was operated under the control of a wind sector (±45°) and wind speed (≥5 m s−1) system to avoid a potential contamination from ship exhausts. Meteorological data such as solar radiation and ambient temperature were collected continuously on board during the cruise using the facilities equipped for the ship. The samples were analyzed for water-soluble DCAs and ketoacids using the methods reported previously [Kawamura, 1993; Kawamura and Ikushima, 1993]. Briefly, an aliquot (typically one eighth) of a filter was cut in small pieces and extracted with ultrapure organic-free water in an ultrasonic bath. The water extracts were then concentrated to nearly dryness by a rotary evaporator under vacuum, and then derivatized to butyl esters and/or dibutyl acetals by reacting with 14% BF3 in n-butanol. Last the derivatives were determined using a HP 6890 gas chromatograph equipped with a fused silica capillary column (HP-5, 0.2 mm × 25 m × 0.52 μm) and an FID detector. Using this technique, aerosol concentrations were measured for 11 linear saturated α,ω-DCAs (C2 to C12) and 17 branched/unsaturated DCAs or ketoacids. Their aerosol concentrations and molecular distributions are presented elsewhere [Wang et al., 2006].
2.2. Determination of δ13C for DCAs and Ketoacids
 δ13C values of water-soluble DCAs and ketoacids relative to Pee Dee Belemnite (PDB) were measured using the method developed by Kawamura and Watanabe . Briefly, after an appropriate amount of internal standard (n-C13 alkane) was spiked to the derivatized fraction of each sample, δ13C of the derivatives were determined using GC/irMS (HP 6890 GC and Finnigan-MAT Delta plus irMS). δ13C of free organic acids in the sample were then calculated using a mass balance equation based on the measured δ13C of the derivatives and the derivatizing agent (1-butanol). Each sample was analyzed in replicate and the mean δ13C is reported. Difference in δ13C of free acids for replicate analyses is generally below 1‰. However, for minor species, the difference is sometimes up to 1.5‰ and occasionally over 2‰. Whether a reliable δ13C can be obtained for a compound in a sample depends on its concentration and properties of the sample. We thus only report δ13C for nine compounds (eight DCAs and one α-ketoacid) in this article although δ13C for some other organic acids were also determined.
2.3. Air Mass Back Trajectory
 Backward air mass trajectories were conducted using the 40-year reanalysis data of the European Centre for Medium-Range Weather Forecasts (ERA40). The trajectory calculation was based on backward tracking of selected air parcels, assuming that they were moving along the ambient airflow [Hatsushika and Yamazaki, 2003]. The flow pattern was updated every 6 hours.
3. Results and Discussion
3.1. Summary of δ13C for DCAs and Ketoacids
 δ13C values for nine organic acids are given in Table 1. On average, oxalic (C2), adipic (C6) and glyoxylic (ωC2) acids are more enriched in 13C while maleic (M), methylmalonic (iC4) and phthalic (Ph) acids are more depleted in 13C. δ13C values (mean: −17‰) of oxalic acid in this work are generally heavier than those (mean: −21‰) reported in the Bermuda aerosols collected during spring 1998 [Turekian et al., 2003], which might result from the differences in its source regions and/or atmospheric processes. It should be born in mind that unlike biomarkers that can faithfully reflect δ13C values in their sources because of their stable character, LMW organic acids are subjected to both photochemical production and degradation in the atmosphere [Chebbi and Carlier, 1996; Kawamura et al., 2005]. Consequently, significant carbon isotopic fractionation of these LMW molecules may have occurred during atmospheric transport (see discussion below). Cautions should thus be taken for the interpretation of their potential sources based on their δ13C values determined in samples. On the other hand, rich information on atmospheric chemical processes may exist behind the isotopic fractionation.
Table 1. The δ13C Values of Water-Soluble DCAs and One α-Ketoacid in Individual Samplesa
Collection Dates, LT
The δ13C values are relative to PDB and are given in permil. LT, local time (Japan standard time).
22–25 Nov. 1994
25–27 Nov. 1994
27–29 Nov. 1994
29 Nov. to 1 Dec. 1994
1–3 Dec. 1994
5–7 Dec. 1994
7–9 Dec. 1994
19–23 Dec. 1994
23–27 Dec. 1994
27–29 Dec. 1994
29 Dec. 1994 to 1 Jan. 1995
1–4 Jan. 1995
9–11 Jan. 1995
11–13 Jan. 1995
13–16 Jan. 1995
16–18 Jan. 1995
18–20 Jan. 1995
20–22 Jan. 1995
24–26 Jan. 1995
26–28 Jan. 1995
1–3 Feb. 1995
3–5 Feb. 1995
5–7 Feb. 1995
7–9 Feb. 1995
9–11 Feb. 1995
Mean ± standard error
−16.8 ± 0.8
−21.5 ± 0.6
−20.3 ± 0.5
−17.2 ± 0.7
−18.3 ± 1.2
−23.6 ± 1.6
−24.2 ± 0.9
−23.1 ± 2.6
−17.6 ± 2.3
3.2. Latitudinal Changes in δ13C of Individual Organic Acids
Figure 2 presents latitudinal variations of δ13C for the detected organic acids. Interestingly, δ13C of the saturated DCAs especially for C2, C3 (malonic) and C4 (succinic) increase from midlatitudes toward the equator with the exception of two samples (QFF659 and 687 collected near Papua New Guinea, further discussion is given later on them). Except for C2 in the Northern Hemisphere, all the correlations between δ13C of C2 to C4 DCAs and latitudes (from midlatitudes to the equator) are significant at a level of p < 0.05 or p < 0.08. Several possible atmospheric processes could contribute to the latitudinal changes in the isotopic composition. Those include evaporation, isotope exchange with inorganic carbon, and photochemical reactions. In addition, global distribution of C3 and C4 plants might also have some effects on the latitudinal changes in δ13C.
 Semivolatile properties of DCAs have been reported [Limbeck et al., 2001]. During evaporation isotopically lighter molecules are generally enriched in the vapor phase, the extent depending on the temperature [Hoefs, 1997]. Since the evaporation-related isotopic fractionation arises from the differences in the vapor pressures of isotopic compounds [Hoefs, 1997], it appears that this type of isotopic fractionation is only significant for very small molecules such as water molecules. For example, it has been found that carbon isotopic fractionation effects due to evaporation for monoaromatic hydrocarbons are very small (around +0.2‰) [Harrington et al., 1999]. Although it cannot be ruled out completely, evaporation-related isotopic fractionation for DCAs at ambient temperature is likely insignificant.
 Inorganic carbon species (CO2, HCO3− and CO32−) have much heavier δ13C values than most organic compounds [Hoefs, 1997]. Under high-temperature conditions (above several hundred degrees Celsius), kinetic isotope effects (KIEs, ratios of reaction rate constants for 12C and 13C and commonly expressed as (k12/k13 − 1) × 1000, ‰) will be very small and isotope exchange between organic and inorganic carbon species may be significant [Dias et al., 2002a, 2002b]. However, we have not found any reports in which such isotope exchange takes place significantly at ambient temperature and pressure.
 In general, C4 plants have much heavier δ13C than C3 plants (with a mean of −13 and −27‰, respectively) [Hoefs, 1997]. Although there is a very dense coverage of C4 plants in northern Australia, the coverage of C4 plants around the western Pacific equatorial is quite low [Still et al., 2003]. Ten days' backward air mass trajectories found that during the sampling period no significant air masses flowed out of northern Australia to the sampling areas. Thus the distribution of C3 and C4 plants cannot be used to successfully explain the latitudinal changes in δ13C of the saturated DCAs particularly in the Northern Hemisphere.
 During the sampling, ambient temperature maximized around the equator and solar radiation generally increased from midlatitudes toward the equator except for the areas where the two “outliers” (i.e., QFF659 and 687) were collected. The δ13C increase of the saturated DCAs is possibly associated with photochemically aged air masses since photochemical reactions should be more active toward the equator. It is generally accepted that KIEs seem more plausible than thermodynamic equilibrium effects being responsible for carbon isotopic fractionation at least in most biochemical reactions [Hoefs, 1997]. Unidirectional chemical reactions generally show a preferential enrichment of the lighter isotope in the reaction products with the remaining reactants being heavier [Hoefs, 1997]. Laboratory experiments and ambient measurements have found that remaining small aliphatic and aromatic hydrocarbons as well as isoprene become more enriched in 13C after the photochemical reactions with OH radicals and the δ13C increase has been ascribed to the KIEs for their photochemical destruction [Rudolph et al., 2000, 2002, 2003]. Hence the increasing trends of δ13C toward the equator for the saturated DCAs (Figure 2) may result mainly from KIEs for their photochemical degradation. This is seemingly supported by a positive correlation between δ13C values of C2 to C4 DCAs and solar radiation obtained for the samples collected in the areas from midlatitudes to the equator (see Figure 3).
Figure 4 plots δ13C of C2 to C4 and C6 DCAs in the western Pacific aerosols as a function of relative abundance (in terms of molar ratio to total DCAs) of oxalic acid. The relative abundance of C2 has been proposed as a measure of photochemical processing in the remote marine atmosphere [Kawamura and Sakaguchi, 1999]. As seen in Figure 4, isotopic values of C2 to C4 and C6 DCAs positively correlate with relative abundance of C2. Although relative abundance of oxalic acid increases with the increment of its δ13C, its concentrations decrease toward the equator (see Figure 5). The latitudinal change patterns for aerosol concentrations of all other organic acids in this work are very similar to oxalic acid. On average, in the Northern Hemisphere, concentrations of C2, C3 and C4 decreased from midlatitudes toward the equator by 94%, 94% and 98%; in contrast, their δ13C increased by 13‰, 6‰ and 4‰, respectively. Similarly, the concentrations in the Southern Hemisphere decreased by 39%, 63% and 90% for C2, C3 and C4, respectively; whereas their δ13C increased by 9‰, 7‰ and 5‰. This further suggests that photochemical decomposition of the saturated DCAs is one of key factors responsible for the increase in their δ13C toward the equator.
 No latitudinal trend was observed in δ13C of the saturated DCAs for the samples from the Southern Ocean (south of 50°S). In this region average ambient temperatures were below 6°C during the sampling and the latitudinal change is small (within 15 degrees). Backward air mass trajectories found that the air masses for the Southern Ocean samples originated from the Southern Ocean and Antarctica. This suggests that aerosol organic acids over the Southern Ocean may have been produced mainly via in situ photochemical reactions with insignificant contribution from polluted terrestrial air masses. Their precursors may be predominantly derived by sea-to-air emissions of marine organics including unsaturated fatty acids, phenolic compounds [Kawamura et al., 1996] and olefins [Warneck, 2003].
 As mentioned above, there are two “outliers” near the equator (QFF659 and 687; see Figure 2) showing lighter δ13C than what might be expected for the saturated DCAs in terms of latitudinal change. It was found that the two “outliers” experienced weaker solar radiation (22.2 and 23.9 MJ/m2/day) than the nearby Southern Hemisphere samples (25.2 and 26.3 MJ/m2/day). This may be a major reason for their lighter δ13C. This also suggests that the two samples were photochemically relatively fresh (less aged). Interestingly, the two samples were found to contain significantly higher concentrations of fatty acids than the nearby samples, especially C18 unsaturated fatty acids being over ten times more abundant [Niwai, 1996]. Unsaturated fatty acids are major lipid components in marine algae and enriched in the microlayers of sea surfaces [Marty et al., 1979] but are subjected to quick photooxidation after emitted to the atmosphere [Kawamura and Gagosian, 1987]. Air mass back trajectories did not show any significant air outflow from Papua New Guinea during the sampling (see Figure 6). Although some air masses for the two samples came over nearby small islands, a significant amount of organics might have been emitted into the air from local ocean sources.
3.3. Possible Effect of Decarboxylation on δ13C
Figure 7 plots the differences in δ13C (Δδ13C) of C2 to C4 DCAs as a function of latitude. Interestingly, Δδ13C (C2 – C3), i.e., δ13C of oxalic acid minus δ13C of malonic acid, and Δδ13C (C2 – C4), i.e., δ13C of oxalic acid minus δ13C of succinic acid, over the western Pacific increase from midlatitudes toward the equator. Actually, we found that Δδ13C (C2 – C6) also have a similar trend (R2 = 0.90, the data are not shown since only four data points are available).
 Oxalic acid (C2) contains only carboxyl carbon but larger saturated DCAs have additional alkyl carbon. KIEs for alkyl carbon have been studied for the photochemical oxidation of methane (+5.4‰) [Cantrell et al., 1990] and some small nonmethane saturated hydrocarbons (<+4‰) [Rudolph et al., 2000]. In contrast, KIEs of carboxyl carbon in photochemical decarboxylation of organic acids are not available at present. However, thermal or spontaneous decarboxylation under various conditions has been extensively studied for the KIEs, which are in the range of +30 to +60‰ at room temperature [Lewis et al., 1993]. When an organic acid undergoes decarboxylation, it loses one carboxyl group and forms a carbon dioxide molecule. It has been concluded that rupture of a carbon-carbon bond to the carboxyl carbon is the rate-determining step for the decarboxylation [Fry, 1970].
 It is unlikely to use the small isotopic fractionation of alkyl carbon to successfully explain the significant increase in δ13C of the saturated DCAs. We thus speculate that KIEs for the photochemical decarboxylation of the saturated DCAs might be one of the key factors controlling their stable carbon isotopic composition. This is because larger DCA molecules have less opportunity than oxalic acid to undergo reactions of the carboxyl carbon and would consequently have smaller isotopic fractionation during photochemical degradation in the atmosphere. However, it should be noted that since the mechanisms for photochemical decarboxylation of the saturated DCAs are not clearly understood and the KIEs are not available at present, the above hypothesis is subject to further research.
4. Summary and Conclusions
 Marine aerosols collected from the western Pacific and Southern Ocean have been analyzed using GC/irMS for the stable carbon isotopic composition of saturated DCAs and ketoacids. δ13C values of the saturated DCAs especially oxalic, malonic and succinic acids were found to increase from midlatitudes toward the equator although such a trend was not observed for phthalic, maleic and glyoxylic acids. Interestingly, difference in δ13C between oxalic and malonic acids and that between oxalic and succinic acids were also found to increase from midlatitudes to the equator. On the basis of these observations, we propose that the δ13C increase may be associated with photochemically aged air masses and might have resulted mainly from kinetic isotope effects (KIEs) for their photochemical degradation.
 We thank Takeji Niwai for his help in the sample collection during the R/V Hakuho Maru cruise (KH94-4) and the cruise members for their support. We also thank Koji Yamazaki for air mass back trajectory analysis and Tomomi Watanabe for her technical support. This study is in part supported by the Japanese Ministry of Education, Science, Sports and Culture through grant-in-aid 14204055. Financial support from the Japan Society for the Promotion of Science to H.W. is also acknowledged. We appreciate the very useful suggestions from two anonymous reviewers.