Journal Archive

PM_2.5 samples were collected on the campus of the Center for Environmental Science in Saitama at Kazo (36°05ʹ10ʺN, 139°33ʹ37ʺE) in Saitama Prefecture, Japan (Fig. 1). Kazo is the inland Kanto plain and is approximately 40 km north-northwest of the Tokyo metropolitan area. The sampling site is surrounded by a paddy field; there are no specific emission sources such as large-scale factories or traffic-laden roads nearby (Yonemochi et al., 2011). The sampling instrument was installed on a building rooftop, about 6 m above the ground. During summer, the country’s highest temperature is often recorded at the Kumagaya Local Meteorological Observatory (Kumagaya, Saitama Prefecture, Japan), which is located about 15 km northwest of the sampling site. In summer, anthropogenic pollutants, mostly emitted from the surrounded Tokyo bay area, are transported to this area by sea breezes and undergo various photochemical reactions during their transportation. Open burning of biomass such as withered grass is carried out near the sampling site after the autumn harvest (Yonemochi et al., 2011, 2007).

Fig. 1. 
Location of the observation site (Kazo).

Sample collections were carried out in summer (July 26-August 6, 2014), and autumn (October 29-November 13, 2014). The average temperature during the sampling period was 28.2°C in summer and 13.8°C in autumn; the average amounts of solar radiation during the sampling periods in summer and autumn were 24.5 and 8.1 MJ/m²/day, respectively. We obtained meteorological data for the sampling site from the local monitoring network, which is managed by the Saitama Prefectural Government. In addition, for the concentration of PM_2.5, we used the monitoring data from the PM_2.5/PM_10-2.5/PM₁₀/OBC Dichotomous Monitor (PM-712, Kimoto Electric Co., Ltd., Osaka, Japan) located at this site.

PM_2.5 samples were collected on quartz fiber filters (20 cm×25 cm, Pallflex 2500QAT-UP, Pall Corp., Putnam, USA) with a high-volume air sampler (HV-1000R, Sibata Scientific Technology Ltd., Soka, Japan) attached to an impactor type PM_2.5 classifier (HVI-2.5, Tokyo Dylec Corp., Tokyo, Japan) (Kaneyasu, 2010). A glass fiber filter (φ 130 mm, HV-F, Tokyo Dylec Corp.) was used as trapping material on the collision collecting plate in the classifier. The sampling period was 9:00-17:00 (daytime) and 21:00-5:00 (nighttime), and the airflow rate was 740 L/min. All filters were baked at 600°C for 6 h before use to remove organic contaminants. In the US Environmental Protection Agency (EPA) WINS impactor and similar PM_2.5 sampling devices authenticated by the US EPA, such trapping material filters are impregnated with oil to prevent blow off of the separated coarse particles. However, in this research, we used an untreated trapping material filter in an effort to avoid subsequent analytical obstacles. All samples were stored at -30°C until analyzed.

Molecular marker compounds were assessed by using a derivatization-GC/MS method (Hu et al., 2008; Kleindienst et al., 2007; Hagino et al., 2006; Simoneit et al., 2004). The analytical target compounds were 2-methyltetrols (2-methylthreitol and 2-methylerythritol), cis-pinonic acid, and levoglucosan. 2-Methyltetrols and cis-pinonic acid are secondary molecular marker compounds derived from isoprene and α-pinene emitted from many plant species, respectively (Liang et al., 2012; Wagener et al., 2012; Fu and Kawamura, 2011; Fu et al., 2009; Bhat and Fraser, 2007; Kleindienst et al., 2007; Claeys et al., 2004). Levoglucosan is a primary molecular marker compound produced by biomass burning (Wagener et al., 2012; Offenberg et al., 2011; Kumagai et al., 2010; Lee et al., 2008; Bhat and Fraser, 2007; Hagino et al., 2006; Simoneit et al., 1999). Thirty punches of rounded pieces (φ 10 mm) were analyzed. Prior to extraction, the sample pieces were spiked with an internal recovery standard mixture (S-(+)-ketopinic acid and levoglucosan-d₇, for each μg of equivalency). They were ultrasonically extracted for 20 min each in 10 mL of a dichloromethane and methanol mixture (2 : 1, v/v). The extracts were then filtered and dried by a gentle stream of ultrapure nitrogen in a reaction vial. For the derivatization, 50 μL of the reagent N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA, including trimethylchlorosilane as a catalyst) was added to the dried samples. The samples were then reconstituted to 200 μL with a dichloromethane and hexane solution (1 : 1, v/v) and the reaction was carried out at 70°C for 2.5 h by using a dry bath. After the reaction, the resulting trimethylsilyl derivatives were evaluated by means of GC/MS (ISQ LT GC-MS, Thermo Fisher Scientific Inc., Waltham, USA). The analytical conditions of the GC/MS analysis are summarized in Table 1. Reference compounds were available for the target compounds, compounds were identified by comparing the GC retention times with the mass chromatograms of the pure reference compounds. Alternatively, they were identified by comparing their mass spectra with previously reported data (Hu et al., 2008; Claeys et al., 2007; Kleindienst et al., 2007; Jaoui et al., 2004) and by examining their retention times relative to those of other known compounds in GC chromatograms. A procedural blank was analyzed for each PM_2.5 filter sample, and the concentration estimate was adjusted accordingly.

Carbon component analysis was conducted using a thermal/optical carbon analyzer (DRI model 2001A, Atmoslytic Inc., Calabasas, CA, USA) with the IMPROVE Thermal-Optical-Reflectance Analysis method (Chow et al., 2001, 1993). A punch of a round piece (φ 8 mm) from each filter, sample was used to determine the carbon components. Blank filters were analyzed for all components, and blank values were subtracted from each detected component. We applied the definition of char-EC and soot-EC previously used by Han et al. (2007) to our results. The components of organic carbon (OC) and elemental carbon (EC), were classified as OC1, OC2, OC3, OC4, pyrolytic OC (PyOC), char-EC (=EC1-PyOC), soot-EC (=EC2) and EC3, and their behaviors were compared. Total carbon (TC) was defined as the sum of OC and EC.

Carbon isotopic analysis was carried out at the Paleo Labo Co. Ltd., Japan, using a compact 500 kV Accelerator Mass Spectrometer (1.5SDH, National Electrostatics Corp., Middleton, USA). To measure ¹⁴C by using an Accelerator Mass Spectrometer, 1 mg or more of carbon per sample is required. Therefore, the amount of carbon per filter was determined from the OC and EC concentrations of the PM_2.5 samples collected in summer. Consequently, eleven summer samples (seven daytime samples and four nighttime samples) that contained sufficient TC components were selected, and 1/4 to 1/2 sheets were analyzed. The percent modern carbon (pMC), an index of plant origin, was estimated based on the ¹⁴C concentration (pMC=100%, Δ¹⁴C=0‰) in 1950 as the reference value. Sample activity was corrected for isotopic fractionation to δ¹³C=-25.0‰.

The atmospheric concentrations of the molecular marker compounds (2-methyltetrols, cis-pinonic acid, and levoglucosan) are shown in Table 2. The concentration of the 2-methyltetols, derived from isoprene, was higher in the summer than in the autumn (November). Because the deciduous period overlaps with this decrease in the levels of the 2-methyltetrols, we considered the emission source to be broad-leaved tree. In contrast, the concentration of cis-pinonic acid, derived from α-pinene, remained almost constatnt throughout the summer and autumn. Therefore, we considered its emission source to be conifers, which undergo less defoliation. The concentration of levoglucosan was considerably increased in the autumn due to frequent biomass open burning. It is well known that the concentration of levoglucosan increases during harvest season because of the outdoor burning of rice husks and straw, and our results support this conventional wisdom.

The change over time in the atmospheric concentrations of the molecular marker compounds during the summer and autumn are shown in Fig. 2. In summer, the concentrations of the 2-methyltetrols and cis-pinonic acid increased during the daytime, but decreased during the nighttime. Similar, but weaker, trends were found in the autumn, suggesting weak secondary production.

Fig. 2. 
Changes over time in the concentrations of molecular marker compounds in PM_2.5 during summer and autumn. The concentration of the 2-methyltetrols is the sum of the concentrations of 2-methylthreitol and 2-methylerythritol. The white background indicates daytime, and the gray background indicates nighttime.

The change over time in the concentrations of total carbon (TC), elemental carbon (EC=char-EC+soot-EC+EC3), and organic carbon (OC=OC1+OC2+OC3+OC4+PyOC) during the summer and autumn are shown in Fig. 3. In summer, the weather was almost always fine, except for at the beginning of sample collection on the night of August 1. In contrast, it occasionally rained during the autumn sampling period (Oct. 31-Nov. 1, 6, 9, and 11-12), and high concentrations were observed in the intervals between the rain. Although the concentrations of the carbon components in the autumn may be slightly underestimated due to rainfall, they were higher than those in the summer and higher in the daytime than in the nighttime. In addition, on days when the concentrations of the carbonaceous components were high, the concentration of levoglucosan was also high. This increase in the concentrations of the carbon components may also be the result of biomass burning in autumn. Moreover, the high concentrations on days with no rainfall may also reflect a situation as outdoor burning.

Fig. 3. 
Changes over time in the concentrations of the carbonaceous components (total carbon: TC, elemental carbon: EC, and organic carbon; OC) of PM_2.5. The white background indicates daytime, and the gray background indicates nighttime.

The changes over time in the concentrations of the carbon fractions during summer and autumn are shown in Fig. 4 and Fig. 5, respectively. In summer, the components of the carbon fraction were, in order of daytime concentration (highest to lowest), OC2 [3.2±0.4 μgC/m³ (daytime), 1.4±0.3 μgC/m³ (nighttime)], OC3 [2.7±0.4 μgC/m³ (daytime), 1.5±0.5 μgC/m³ (nighttime)], char-EC [2.3±0.6 μgC/m³ (daytime), 1.1±0.4 μgC/m³ (nighttime)], and PyOC [1.9±0.4 μgC/m³ (daytime), 0.8±0.3 μgC/m³ (nighttime)]. In contrast, in autumn, the order became OC3 [3.5±1.4 μgC/m³ (daytime), 2.0±0.7 μgC/m³ (nighttime)], char-EC [3.4±1.6 μgC/m³ (daytime), 1.9±0.8 μgC/m³ (nighttime)], OC2 [2.5±1.0 μgC/m³ (daytime), 1.7±0.5 μgC/m³ (nighttime)], and PyOC [2.1±1.0 μgC/m³ (daytime), 1.0±0.4 μgC/m³ (nighttime)]. With the exception of OC1 and OC3, the fraction concentrations showed a significant correlation with the average concentration of PM_2.5 during the daytime in summer [r=0.60-0.90 (n=12, p< 0.05)]. This trend was almost the same for the maximum concentration of photochemical oxidant [r=0.63-0.78 (n=12, p< 0.05)], suggesting the influence of photochemical secondary production. In the nighttime, the correlation was weaker as a whole, but the concentrations of OC2, PyOC, and soot-EC each showed a significant correlation with the average concentration of PM_2.5 [r=0.55-0.65 (n=12, p<0.05)]. In contrast, during autumn, there was a significant strong correlation between all fractions and the average concentration of PM_2.5 during both the daytime [r=0.52-0.97 (n=15, p<0.05)] and nighttime [r=0.70-0.97 (n=16, p<0.01)]. However, the highest concentration of photochemical oxidant showed a moderate correlation with only the PyOC concentration [r=0.61 (n=16, p<0.05)] in the daytime, and the influence of photochemical secondary production was considered to be small.

Fig. 4. 
Changes in the concentrations of the carbonaceous fractions during the summer. The white background indicates daytime, and the gray background indicates nighttime.

Fig. 5. 
Changes in the concentrations of the carbonaceous fractions during the autumn. The white background indicates daytime, and the gray background indicates nighttime.

The changes over time in the ratio of the char-EC concentration to the soot-EC concentration are shown in Fig. 6. The ratio was 2.0±0.4 (daytime) and 1.6±0.4 (nighttime) in summer, and 2.8±0.8 (daytime) and 2.6±1.0 (nighttime) in autumn. During rainfall, when the influence of outdoor burning would be considered small, the soot-EC, which was likely derived from automobiles, would be expected to be dominant, and this ratio was around 1 in some cases. On the other hand, when high concentrations of carbon components are observed in autumn, the ratio is significantly higher (5 and above). The one exception to this finding, on Oct. 30 when the ratio was low despite the high concentration of levoglucosan, does not negate the impact of biomass burning. This low ratio is due to a considerably high concentration of soot-EC. Consequently, we consider that one of the emission source was car exhaust gas. Although other emission sources are not clear, soot-EC may be formed by outdoor burning of plastic products (e.g., agricultural materials, domestic waste). Han et al. (2007) define char-EC as derived from low-temperature incomplete combustion, and soot-EC as particleized after gasification by high-temperature incomplete combustion. In other words, the higher ratio of char-EC concentration to soot-EC concentration means that the contribution of low-temperature incomplete combustion (e.g., biomass burning), is greater than that of high-temperature incomplete combustion (e.g., discharge from an internal combustion engine), so the value of this ratio can be considered an indicator of outdoor open burning.

Fig. 6. 
Changes over time in the ratios of the char-EC concentration to the soot-EC concentration during the summer and autumn. The white background indicates daytime, and the gray background indicates nighttime.

The ¹⁴C analysis revealed that the pMC for TC in the PM_2.5 collected in summer ranged from 50.7% to 74.4 % in the daytime (average, 62%), and from 60.3% to 92.9% in the nighttime (average, 77%). In other words, the ratio of plant-derived carbon to TC in the PM_2.5 was higher than that of fossil fuel-derived carbon during the daytime and nighttime, indicating that the plant-derived carbon contributes greatly to PM_2.5. The difference between the daytime and the nighttime values was likely due to anthropogenic sources from social activities during the daytime. A similar trend was seen in a summer survey in Maebashi, Japan (Fushimi et al., 2011).

Diurnal changes in the carbonaceous concentrations derived from fossil fuel (C_FF) and non-fossil fuel (C_NFF) in summer PM_2.5 samples are shown in Fig. 7. The C_FF and C_NFF increased during the daytime, but decreased during the nighttime. This variation in C_FF demonstrates the influence of anthropogenic sources from social activities during the daytime. The variation in C_NFF was in good agreement with the changes in the atmospheric concentrations of the 2-methyltetrols and cis-pinonic acid in summer (Fig. 2). There was one exception to this finding: C_NFF during the nighttime on July 29 was higher than that during the daytime. This was consistent with the changes in the ratios of the char-EC concentration to the soot-EC concentration (Fig. 6) but was not very consistent with the change in the concentration of levoglucosan. Since the cause is unclear, it should be investigated further by extending the analysis to new samples.

Fig. 7. 
Diurnal changes of the carbonaceous concentrations derived from fossil fuel (C_FF) and non-fossil fuel (C_NFF) in summer PM_2.5 samples. The white background indicates daytime, and the gray background indicates nighttime.