The concentration and chemical speciation of PM samples collected in four different locations are shown in Table 1. The mass concentration of the PM samples is shown in the third column of the table. The remaining columns describe the results of the chemical analysis for EC and OC, nitrate, sulfate and inorganic metals and trace elements. The last column in this table shows the DTT activity (in nmoles of DTT consumed per min and per μg of PM), which is discussed in the following section. Organic carbon is the most abundant material in PM_2.5 and PM_0.15 modes in most of the samples. Organic carbon species may originate either directly from vehicle exhaust, which is a more prevalent PM source next to the CA-110 and in the Caldecott tunnel, or from secondary particle formation, which would be more pronounced in the receptor site of Riverside. However, organic material can be also collected due to adsorption of gaseous organic species on the filter surface, a process that results to a positive mass artifact on the filter. This is particularly true for the quartz filters used for the EC/OC analysis and it is the reason that the mass reconstruction by chemical analysis is higher than the weighted mass for three of the samples collected (Caldecott Bore 2- PM_0.15, and CA-110 PM_0.15 and PM_2.5). Sampling artifacts (positive or negative) cannot thus be excluded for the other sampling locations. However, Figure 1 shows that with the exclusion of the two outliers from the CA-110 freeway, the mass concentration derived by filter weighing and the reconstructed PM mass from chemical analysis are in very good agreement, indicating that the effect of these artifacts is limited at all other sampling locations. The reconstructed mass varied between 79 and 95% of the measured mass, which confirms the consistency and overall reliability of the chemical measurements.

Elemental carbon is 10–25% of the PM_0.15 mode next to the freeway and in the tunnel, while it represents a much lower fraction (< 5%) at background and receptor sites. This is a strong indication that EC in the Los Angeles Basin mainly originates from road traffic emissions. Nitrate (most of which is in the form of ammonium nitrate), is particularly high in Downey and Riverside compared to the rest of the locations. The high nitrate levels at Riverside are of particular note and consistent with previous studies in that area, and reflect the result of atmospheric reactions of nitric acid with fugitive ammonia, largely emitted from the nearby upwind dairy farms in the area of Chino, CA 2728. Sulfate concentrations are generally low and do not show any clear trend with proximity to freeways or to receptor sites, an indication that sulfate in this area is mostly associated with the regionally dispersed, spatially homogeneous background aerosol.

The inorganic and trace elements detected accounted for up to ~50% of the total mass for coarse particles. Their fraction decreased in PM_2.5 and especially PM_0.15 samples. The most abundant single element detected was Si, with a mass fraction ranging between 1–17% of PM mass. The second most abundant element was iron (Fe) with a fraction reaching up to 15% in the coarse mode. The abundance of Fe was up to 10% also in the PM_0.15 mode, for samples collected in sites affected by traffic. Al, Ca and Cl were the third more abundant species, with relative abundances ranging from 0.5% (0.05% for Cl) to 10% of the PM mass in each size range. All other elements were found at lower concentrations. Na and S reached up to 3%, Mg, K, Ti, Cu, Zn and Ba reached up to 1% of the total PM mass per size mode. Finally, a number of other trace elements were detected, with a contribution at- or below 0.1% of the total mass. The most significant of those, with decreasing order of abundance (mean value for all samples shown in parentheses), were Sb (0.1%), Mn (0.1%), Pb (0.08%), Sn (0.05%), Zr (0.05%), P (0.04%), Cr (0.03%), Ni (0.04%) and V (0.03%).

PAH were also measured with the exception of samples collected in Riverside (Figure 2). The PAH species were divided into three groups for analysis, based on their molecular weight. The first group consisted of four species (Fluoranthene, Pyrene, Benz [a]anthracene and Chrysene) with a molecular weight between 202 and 228. The second group consisted of the PAHs (Benzo [k]fluoranthene, Benzo [b]fluoranthene, Benzo [a]pyrene) with a molecular weight of 252. The third group was comprised of PAHs with molecular weight in the range of 276–278 (Benzo [g, h , i]perylene, Indeno [1,2,3-cd]pyrene, Dibenz [a, h]anthracene). Generally, PAH concentrations increase with decreasing particle size, with the maximum concentrations observed for the two lighter species (Fluoranthene and Pyrene). The maximum total PAH concentrations were found for samples collected in the roadway tunnel. The concentration of heavier PAHs is higher in the gasoline-only tunnel, while the concentration of lighter components (PAH202-228) is higher in Bore 2, where diesel traffic is also permitted. This is consistent with previous studies 2930 which showed that the PAH profile of gasoline vehicles is shifted towards the heavier molecular weight species compared to diesel vehicles.

The DTT activities of the 18 samples are shown in the last column of Table 1. The DTT activity is highest in the PM_0.15 mode, followed by the PM_2.5 and the coarse modes at all sampling sites, with averages of 0.088 (± 0.040), 0.038 (± 0.022) and 0.023 (± 0.009) nmoles DTT per min per μg PM for ultrafine, PM_2.5 and coarse PM, respectively. The PM_2.5 fraction contains all PM less than 2.5, including the PM_0.15 mode. Similar observations regarding the effect of particle size on per mass DTT activity were also made by Cho et al. 22 in their tests in various locations in the Los Angeles Basin. More importantly, the redox activity becomes maximum for PM_0.15 sampled in the road tunnel, which is directly influenced by the on-road fresh emissions. Interestingly, the highest DTT activity per mass of PM was associated with the sample collected in the gasoline only tunnel bore (B2) of the Caldecott tunnel. This result is also consistent with the dynamometer study by Geller et al. 26, that showed higher PM redox activity per mass of PM emitted by a gasoline over a diesel vehicle.

As a first step in our exploratory data analysis, we attempted to identify correlations between the DTT redox activity measured in PM samples and their composition in EC, OC ions, elements and PAHs. Table 2 shows the Pearson correlation coefficients (R) and the associated coefficient of significance (p). All particle size ranges (PM_0.15, PM_2.5 and coarse PM) have been combined in this correlation. This analysis shows limited and not statistical significant correlation of DTT activity with EC, NO₃, SO₄ and the sum of metals and elements determined. None of the individual metals and elements measured was significantly correlated with DTT activity, with the exception of Cr, which led to a Pearson coefficient of 0.65 albeit not statistically significant at a significance level of 0.05. The correlation between DTT and OC levels was not significant when all 18 samples are taken into account. However, this includes the two unrealistically high OC values determined in the PM_0.15 and PM_2.5 samples next to the CA-110 (Table 1), a likely result of positive sampling artifacts, as discussed earlier. Exclusion of these two values leads to a much improved, and statistically significant correlation (R = 0.87). All measured PAHs were significantly correlated with DTT activity at the p = 0.05 level. Most importantly, DTT activity is highly correlated with the heavier classes of PAH species, with Pearson coefficients as high as 0.95 for the 13 samples for which PAH analysis was available. The difference between the correlation coefficient for the lighter PAHs and the heavier species may reflect differences in the volatility of the redox active species. Also, lighter PAHs may also be prone to sampling artifacts, similar to total OC. However, in our case, removal of the two CA-110 samples did not significantly change the correlation. Despite these differences, the results of Table 2 show that the organic component of the PM samples is an important factor in determining their redox activity. These correlations were established independently of particle size, i.e. they are applicable for the whole size range of inhalable particles.

We further explored correlations of DTT activity with individual species within each particle size mode. Table 3 presents the results of these correlations for the PM_0.15 and PM_2.5 modes, for which an adequate sample size was available. The results in this table demonstrate that there is no correlation with EC and a significantly negative correlation at the p = 0.05 level with NO₃ and SO₄. None of these species can be considered responsible for the redox activity of the PM samples, even within each particle size mode. In contrast, the positive DTT correlation with OC is evident even within each particle size mode. Table 4 also shows high Pearson coefficients between the DTT activity and several transition metals, such as Mn, Fe, Cu, Zn, and metals Pb and Ba. The correlations with Mn and Zn are only significant in the PM_0.15 mode. The DTT correlation with transition metals is also shown in Figure 3.

Although the redox activity of transition metals in biological reactions is well-established 3132, the DTT assay in general does not reflect metal-based redox activity 22 especially for transition metals such as Cu and Fe. An important finding of the current study, however, is that while transition metals are not correlated with DTT activity in the pooled samples of particles from different size ranges, a strong correlation exists for PM_2.5 and PM_0.15 samples. As evidenced in the results shown in Table 3, several of the transition metals in the PM_0.15 and PM_2.5 ranges are highly correlated with PAH, possibly due to their common sources (e.g. vehicle exhaust emissions). For example, the Pearson correlation coefficients between PAH with molecular mass above 252 and Mn, Fe, Cu, and Zn are 0.86, 0.95, 0.96 and 0.99, respectively. Hence, the very high correlations of transition metals and DTT activity in Figure 3 might more probably be attributed to the high correlation of metals and PAH in the PM samples, especially in the PM_0.15 mode.

Sampling took place at four diverse sites, during the period of June 2003 to July 2005. These sites were at Downey, Riverside, the vicinity of the CA-110 freeway, and the two bores of the Caldecott tunnel. The first three sites are located in the Los Angeles basin, whereas the Caldecott tunnel is located in the metropolitan area of San Francisco, at Orinda, CA. Detailed information about the Los Angeles Basin sites is given by Sardar et al. 28. Briefly, Downey, located in central Los Angeles, is downwind of the "Alameda corridor", a narrow industrial zone and transportation route between the Ports of Los Angeles/Long Beach and Downtown Los Angeles. The area is characterized by a high density of diesel trucks, which serves to transfer overseas cargo from the port to industrial sites, warehouses, and the rail yards near downtown Los Angeles. The Downey site is approximately 10 km downwind of some oil refineries, 1–2 km downwind of two major freeways, and is heavily impacted by vehicular sources. Riverside is about 90 km east of downtown Los Angeles. The site is also about 25 km downwind of the Chino area dairy farms, a strong ammonia source leading to high concentrations of ammonium nitrate 39. The area is upwind of surrounding freeways and major roads. The predominantly westerly wind transports particles generated near central Los Angeles toward Riverside, resulting in an aged and photochemically processed aerosol. Riverside is also characterized by some of the highest PM levels in the Basin. Measurements at the CA-110 freeway were conducted at 2.5 m from the edge of the freeway (as described in Kuhn et al. 40). This freeway connects downtown Los Angeles and Pasadena, CA. On this stretch of the freeway, only light-duty vehicles are permitted, thus affording a unique opportunity of studying emissions from pure light-duty traffic under ambient conditions. Finally, the 1.1-km long Caldecott Tunnel includes three two-lane bores with a 4.2% incline from west to east. Bores 1 and 3 allow both light-duty vehicles and heavy-duty vehicles, while Bore 2 is restricted to light-duty vehicle traffic only. Traffic flows from west-to-east in Bore 1, east-to-west in Bore 3, and the direction of traffic switches from westward in the morning to eastward in the afternoon and evening in Bore 2. Field sampling was conducted in the afternoon in Bores 1 and 2 (B1 and B2) for 4 days each from approximately 12 p.m. to 6 p.m., when all traffic in the two bores traveled eastward.

In each site, size fractionated PM samples were collected over a period of 2–3 weeks, for about 5 days per week, and 6–7 hrs/day. Thus each sample is a composite of some 70–100 hrs of collection.

Coarse (i.e. particles with aerodynamic diameter 2.5–10 μm), fine (PM_2.5, < 2.5 μm), and ultrafine (PM_0.15 < 0.15 μm) particles were collected at the these sites by the Versatile Aerosol Concentration Enrichment System (VACES), in a process described in greater detail by Li et al. 5 and Cho et al. 22. Briefly, the VACES uses three parallel sampling lines (concentrators) to simultaneously collect coarse and fine particles at a flow rate of 120 liters per minute (lpm) into a liquid impinger (BioSampler™, SKC West Inc., Fullerton, CA) at 5 lpm. Particles are injected into the BioSampler in a swirling flow pattern so that they can be collected by a combination of inertial and centrifugal forces. This inertia-based collection mechanism, coupled with the short residence time on the order of 0.2 s for particles and gases in the Biosampler precludes any inadvertent trapping of gaseous co-pollutants in the particulate layer.

In each sampling line of the concentrator, coarse PM, PM_2.5 and PM_0.15 were concentrated from a flow of 120 lpm to a flow of 6 lpm, thereby, being enriched by a factor of 20. From the 6 lpm of concentrated flows samples, 4 lpm was drawn through the BioSampler connected to the respective minor flow, while 2 lpm passed through diffusion dryer for PM_2.5 and PM_0.15 only to remove excess water and dry the aerosol. Diffusion drying of coarse PM was not considered necessary since it is concentrated without hydration of the aerosol. The dry concentrated aerosol flow was then split into two equal halves of 1 lpm, each diverted into a filter sampler consisting of either a 47-mm Teflon filter (2-mm pore PTFE; Gelman Science, Ann Arbor, MI) or a 47 mm prebaked quartz filter (Pallflex Corp., Putnam, CT). The PTFE filters were used to determine particle mass and the metal content, whereas quartz filters were used to determine the PM content of elemental and organic carbon (EC-OC), inorganic ions, and PAHs.

For measurement of mass concentrations, the PTFE filters were weighed before and after each field test using a Mettler 5 Microbalance (MT 5, Mettler-Toledo Inc., Highstown, NJ), under controlled relative humidity (40–45%) and temperature (22–24°C) conditions. At the end of each experiment, filters were stored in the control humidity and temperature room for 24 h prior to weighing to ensure removal of particle-bound water. The concentration of trace elements and metals was determined by means of X-ray fluorescence subsequent to filter weighing. The quartz filters were cut into two unequal parts, 1/4 and 3/4 of the total filter. The smaller piece was analyzed by means of ion chromatography to determine particle-bound sulfate and nitrate concentrations. A small area (1 cm²) of the remaining filter (3/4) was removed to determine the EC and OC content of PM. EC and OC were determined using a thermal optical transmittance method as specified in NIOSH method 5040 (Birch 41). The remaining portion from the filter above was used to determine the concentrations of PAH using procedures described elsewhere 42. In brief, the filters corresponding to each size range were ultrasonically extracted with dichloromethane and the PAH content of the dichloromethane extract was analyzed by high-pressure liquid chromatography fluorescence using NIST SRM 1649a as the positive control.

The DTT assay is described in greater detail by Cho et al. 22. This assay provides a measure of the overall redox activity of the sample based on its ability to catalyze electron transfer between DTT and oxygen in a simple chemical system. We have used this assay because it provides a quantitative measure of redox activity that can be normalized to mass or air volume such that samples from different sources can be compared. Other assays used for this purpose include the consumption of ascorbate 43, oxidation of dichlorofluorescin 44 and an ESR procedure 45 to monitor the levels of free radical species. We have not compared the DTT assay directy to other assays, but have observed that ascorbate consumption by PM is sensitive to metal chelating agents, whereas DTT consumption is not, suggesting that former procedure monitors redox activity of particle-bound metals, whereas the latter the redox activity of mostly organic species, as it is discussed in this paper.

The electron transfer is monitored by the rate at which DTT is consumed under a standardized set of conditions and the rate is proportional to the concentration of the catalytically active redox-active species in the sample. In brief, the Biosampler PM samples of known mass are incubated at 37°C with DTT (100 mM) in 0.1 M potassium phosphate buffer at pH 7.4 (1 mL total volume) for times varying from 0 to 30 minutes, and the reaction quenched at preset times by addition of 10% trichloroacetic acid. An aliquot of the quenched mixture is then transferred to a tube containing Tris HCl (0.4 M, pH 8.9), EDTA (20 mM) and 5,5'-dithiobis-2-nitrobenzoic acid (DTNB, 0.25 mM). The concentration of the remaining DTT is determined from the concentration of the 5-mercapto-2-nitrobenzoic acid formed by its absorption at 412 nm. The DTT consumed is determined from the difference between the mercaptobenzoate formed by the blank and that formed by the sample. The data collected at the multiple time points is used to determine the rate of DTT consumption, which is normalized to the quantity of PM used in the incubation mixture.