Current Issue

Our study was conducted at two sites in the city of Saitama, which is located in the central Kanto region of Japan (Fig. 1); the city is the seat of Saitama Prefecture and has a population of more than 1 million. The two sites were chosen to represent a roadside and an urban background area. Route 57 (a main prefectural road with high traffic volumes of both diesel-and gasoline-powered vehicles, around 21,000 vehicles per weekday) was chosen as the roadside site. Samples were collected at curbside on a street corner near Saitama University at a height of 1.5 m above the ground. The urban background site was on the campus of Saitama University, about 250 m northeast of the roadside site. At this site, samples were collected at about 15 m above the ground, on the roof of a four-story building (Department of Applied Chemistry, Saitama University).

Fig. 1. 
Locations of sampling sites: roadside (dashed arrow) and urban background site (dotted arrow).

periods at each site: winter (December 16-28, 2004) and summer (July 20-August 2, 2005). During these periods, samples of PM₁₀ and PM_2.5 were collected using a pair of Federal Reference Method (FRM, Partisol sequential air sampler model 2025, Rupprecht & Patashnick Co., Inc., East Greenbush, NY, USA) air samplers and cyclone separators; the flow rate was maintained at 16.7 L/min for 12 h. Quartz fiber filters (47 mm diameter, Pallflex, 2500 QAST; Shibata Scientific Technology) were used as collecting media. Prior to sampling, the filters were heated at 800°C for 2 h to eliminate organic species, kept at 20°C and 50% relative humidity for 48 h, and then weighed. The filters were shielded from direct sunlight and rain. Background contamination was measured using operational blanks (unexposed filters), which were processed simultaneously with the field samples. In the laboratory, all filters were kept at 20°C and 50% relative humidity for 48 h, reweighed, and then stored in the refrigerator (-40±5°C) until they were chemically analyzed.

At the roadside, traffic volume was recorded on videotape for later manual counting of the numbers of diesel-and gasoline-powered vehicles. Vehicles were counted for 10 min at each hour, and the total volumes of vehicles in both directions were counted. During the sampling period, meteorological parameters (ambient temperature, relative humidity, wind speed, and wind direction) were monitored at the roadside with an EDO MET-9800 system (Shibata Scientific Technology).

During the winter, the data for nitrogen oxides(NO_x) and ozone (O₃) were obtained with an ambient air-quality monitoring station at a height of about 20 m above the ground, on the roof of the Saitama Institute of Public Health, which is 450 m away from Saitama University on northeast direction. In the summer, NO_x and O₃ were monitored at the roadside using a nitrogen oxides analyzer (DKK Corporation, Tokyo, Japan) and a UVAD-1000 O₃ analyzer (Shimadzu Corporation, Kyoto, Japan), respectively. Prior to monitoring, both analyzers were multipoint calibrated with a NO gas cylinder (Japan Fine Products Corporation, Kanagawa, Japan). The obtained calibration curves represent the best fits for all the data points (R²=1.00 and R²=0.998 for NO, and O₃ respectively).

All samples were analyzed for EC and organic carbon (OC) using a thermal optical carbon analyzer (DRI model 2001, Sibata Scientific Technology) with the IMPROVE thermal/optical reflectance protocol reported by Chow et al. (1993). Briefly, a 0.503 cm² (8 mm diameter) punch aliquot of a sample quartz filter was heated at 120 (OC1), 250 (OC2), 450 (OC3), and 550 (OC4)°C in a helium atmosphere, and at 550 (EC1), 700 (EC2), and 800 (EC3)°C in an oxidizing atmosphere of 2% oxygen and 98% helium. The analysis was repeated three times for each sample for better accuracy. For analysis of ions, an area of 1.90 cm² aliquot of each filter was extracted with 0.3 M NaHCO₃ and deionized water during the analysis of anions (NO₃^-, SO₄^2-, and Cl^-) and cations (K⁺, Na⁺, NH₄⁺, and Ca²⁺), respectively. All extracts were analyzed using an ion chromatograph (Dionex-100 IC) equipped with an electric conductivity detector. Mixed standards of five concentrations within the ranges of 0.1-2.5 ppm (w/v) for anions and 0.05-0.8 ppm (w/v) for cations were used to draw standard calibration curves.

The residual filter area remaining after the analysis of carbonaceous components and ions was used for extraction of PAHs. The residual area was cut into small pieces, which were ultrasonically extracted twice with 3.0 mL of acetonitrile (HPLC grade, Wako Pure Chemical Industries, Japan) for 1 h. The combined extracts were centrifuged at 2,500 rpm for 10 min and then filtered through a 0.2-μm syringe filter. The volume of the filtrate was reduced to 0.5 mL using a gentle stream of nitrogen gas (N₂), and the residue was transferred to 1-mL vials and analyzed by high-performance liquid chromatography (Shimadzu LC-10Avp series, Shimadzu Corporation, Kyoto, Japan). Table 1 summarizes the chromatographic conditions used during the analysis. The chromatographic peaks of the samples were identified by comparison of their retention times to those of standards. Selected PAHs were quantified by comparison of the measured peak areas with those of standards of known concentration. The quantified compounds were pyrene (PYR), benzo [a]anthracene (BaA), benzo[a]pyrene (BaP), benzo [b]fluoranthene (BbF), benzo[k]fluoranthene (BkF), indeno[1,2,3-cd]pyrene (IndP), and benzo[ghi]perylene (BghiP). PAH concentrations are reportedly lower during summer than during winter (Hayakawa et al., 2002; Smith and Harrison, 1996). For that reason and because of the time required for extraction and analysis, only samples collected during winter were analyzed for selected PAHs as described above. Finally, WRPLOT View software (Lakes Environmental Software Inc., Ontario, Canada) was used for drawing the rose diagrams of wind direction.

During all the sampling periods, the mass concentrations of PM₁₀ were below the air quality standards for suspended particulate matter in Japan (SPM, which corresponds to PM7, 100 μg/m³) and in the United States (PM₁₀, 150 μg/m³ daily average). In contrast, the average concentrations of PM_2.5 exceeded the United States Environmental Protection Agency (USEPA) standard for PM_2.5 (15 μg/m³, annual average, Table 2). Moreover, the USEPA daily average standard for PM_2.5 (35 μg/m³) was exceeded at both sites on a few days (25% and 18% of PM_2.5 data during winter and summer, respectively). The concentrations of PM_2.5 and PM₁₀ showed identical temporal variations (Fig. 2) and were strongly correlated (Pearson correlation coefficients: r=0.98 and 0.94 at the roadside and r=0.98 and 0.93 at the urban background site during the winter and summer periods, respectively). The average values of the PM_2.5/PM₁₀ ratio during winter were 0.83 and 0.81, whereas the values were 0.74 and 0.64 during summer at the roadside and the urban background site respectively. These values agree with those reported by Sasaki and Sakamoto (2006) for PM_2.5/PM₁₀ at Saitama sampling sites. These results indicate that PM_2.5 makes up the largest fraction of PM₁₀ at various places in Saitama.

Fig. 2. 
Temporal variation of PM_2.5 and PM₁₀ mass concentrations during winter and summer at the roadside and the urban background site.

Many other urban areas in Japan are affected by high concentrations of PM_2.5. Saitoh et al. (2002) reported an average concentration of 27.9 μg/m³ for PM_2.5 measurements carried out from December 26, 1998, to January 9, 1999, at Shinjuku, Tokyo; and PM_2.5 was the major contributing fraction of PM₁₀. Ma et al. (2004) reported three high-concentration episodes for PM_2.5 (>65 μg/m³) in the Kansai area.

Table 2 lists average mass concentrations of EC, OC, and total carbon (TC) observed during winter and summer. As expected, the concentrations of carbonaceous components were higher at the roadside than at the urban background site during each sampling period. However, the concentrations at the two sites were strongly correlated, especially for the PM_2.5 fraction (see Table 2).

The EC/TC ratios for PM_2.5 and PM₁₀ were very similar during the two study periods at both the roadside and the urban background. This observation seems to indicate that EC was associated with PM_2.5 rather than with larger particles. To confirm that this was the case, we investigated the correlation between the mass of particulate matters (PM_2.5 & PM₁₀) and the EC concentrations in both fractions. At both sites, PM_2.5 and EC were strongly correlated (Pearson correlation coefficients: r=0.88 and r=0.83, p<0.01, n=24 at the roadside and the urban background site, respectively) during winter (Fig. 3). However, they were only moderately correlated during summer. As for the relation between PM₁₀ and the EC, the correlation coefficients were: r=0.73 and r=0.72, p<0.01, n=24 at the roadside and the urban background site, respectively. These findings indicate that EC, which is mainly emitted from combustion sources (diesel-powered vehicles in the current study area), was associated with PM_2.5 at both sites.

Fig. 3. 
Correlation between PM_2.5 and ECPM_2.5 mass concentrations at the roadside and the urban background site during winter and summer.

During winter, the contribution of TC to the mass of PM_2.5 was 47% (EC: 17%, OC: 30%) at the roadside, whereas it was 43% (EC: 13%, OC: 30%) at the urban background site. Similar percentages were observed during summer. Moreover, at both sites the temporal variations of TC were similar to those of PM_2.5. The high levels of carbonaceous components could be attributable to traffic emissions. Particularly, the higher EC concentrations observed at the roadside compared with those observed at the urban background site are attributable to the exhaust of diesel-powered vehicles. This is supported by the observation that EC was significantly correlated (r=0.50, p<0.02, n=24 and r=0.45, p<0.05, n=22 during winter and summer, respectively) with the percentage of diesel-powered vehicles traveling on the road. However, the narrow range of the diesel-powered vehicles percentage (19-32%) as well as the meteorological conditions could explain the moderate correlation coefficients that were obtained. Furthermore, EC was strongly correlated (r=0.91, n=22, p<0.01) with the NO/NO_x ratio measured at the roadside during summer (Fig. 4). The NO/NO_x ratio represents the primary emitted NO_x which are mainly emitted by diesel-powered vehicles. These correlations indicate that diesel-powered vehicles traveling on the road may have been the major source of EC measured during our study.

Fig. 4. 
Correlation between NO/NO_x and ECPM_2.5 at the roadside during summer.

Using the analytical data for the chemical components measured, we constructed a mass closure model (Harrison et al., 2003) for the chemical composition of the particulate matter measured at the two sites. We focused on the chemical composition of PM_2.5. The model is described briefly as follows:

• Sulphate : Sulphate concentrations were increased to the equivalent mass of ammonium sulphate using a factor of 1.38. A further 29% increase was applied to allow for water content.
• Nitrate : Nitrate concentrations were increased to the equivalent mass of ammonium nitrate using a factor of 1.29. A further 29% increase was applied to allow for water content.
• Chloride : Chloride concentrations were increased to the equivalent mass of ammonium chloride using a factor of 1.51. A further 32% increase was applied to allow for water content.
• EC : EC mass was unchanged because EC exists in the form of graphitic carbon.
• OC : Factors of 1.3 and 1.7 were used to convert carbon masses to organic compound masses at the roadside and the urban background site, respectively.

Harrison et al. (2003) used a factor of 1.4 to convert OC to organic compounds for the background site in their study. However, in our study we used a factor of 1.7, which was applied during a similar study in Japan (Takahashi et al., 2008). This factor was derived from the observation of a study conducted in Tokyo by Takegawa et al. (2005). For soil particles, Harrison et al. (2003) assumed the contribution from soil particles and road dust on the basis of Fe analysis. However, in our study we did not analyze Fe. Therefore, we multiplied PM_2.5 mass concentrations by the factors mentioned by Takahashi et al. (2008) to estimate the contribution of soil particles to PM_2.5 (0.051 and 0.010 during winter and summer, respectively). In Fig. 5, the mass concentrations calculated with the model are plotted against the measured mass concentrations during both winter and summer. The good agreement between the calculated and observed concentrations of PM_2.5 at the investigated sites indicates the validity of our model. The chemical composition of PM_2.5 was quite similar at both sites. Fig. 6 shows the chemical composition of PM_2.5 at both the roadside and the urban backgrounds site. The most abundant components were OC, EC, sulphate, and nitrate. Carbonaceous compounds showed high contributions to the calculated chemical composition of PM_2.5 (on average 56% at the roadside and 55% at the urban background site). Sulphate and nitrate were associated with carbonaceous components. However, during certain periods (July 29-30, 2005), sulphate was more abundant than the other components. In contrast, nitrate contributions were lower on the same dates. The wind direction on these dates was not stable, but it tended to parallel the road. Usually, high sulphate concentrations and low nitrate concentrates indicate that an aerosol has traveled long distances (Hagino et al., 2007; Baek and Aneja, 2004). Therefore, we suspect that an air mass that included pollutants from other areas passed through the atmosphere at our investigation area on those dates.

Fig. 5. 
Observed mass concentration of PM_2.5 plotted against calculated mass concentration at the roadside and the urban background site during winter and summer.

Fig. 6. 
Calculated chemical composition of PM_2.5 at the roadside (R) and the urban background site (U.B.), along with the difference between the two (R-U.B.) on July 25, 2005.

The highest contribution of carbonaceous components was observed during days when the wind was coming mainly from the southwest, that is, from the roadside toward the urban background site. Therefore, we strongly believe that at the latter site, the contribution of carbonaceous components can be attributed to motor vehicle emissions. Moreover, at the roadside, the concentrations of several chemical components such as EC and OC were markedly lower on the weekend than during the week. This reduction was associated with the reduction in the number of diesel-powered vehicles on the weekend. The difference between data for the weekday and weekend was marked in the case of the EC concentration, whereas for the OC, the weekend concentrations were closer to the weekday concentrations.