Total suspended particle (TSP) samples (n = 12) were collected during August and September 2014 at one of the tall buildings (20 meter above ground) on Minhang campus of Shanghai Jiao Tong University (SJTU), Shanghai, China, using a high volume sampler (Tisch environmental, USA). The sampling site is located in a residential area with a major emission of TSP from on-road vehicles, soil dust etc. It is 20 km far away from the city center, with no industries within 3000 meters radius. There is one expressway and a Shanghai Metro station, 300 meters and 1 km to the west of the sampling site, respectively. The sampling location is shown in Figure 1. PTFE filters (8′ × 10′) were used for collection of TSP, preconditioned (22°C, 33% RH) for at least 24 h in a dry box and weighed before field sampling. Sampling were conducted for 24 h at flowrate of 0.5–0.7 m³/min. Post-sample conditioning of PTFE filters was also conducted in the dry box under the same conditions before weighing the exposed filters laden with TSP samples. After the gravimetric determination of the TSP collected, the filters were kept in a −20°C freezer before chemical analysis.

Concentrations of total trace elements, along with their bioaccessible fractions, in the TSP were determined in this work. A portion (1/64 of the total filter area) of the PTFE filters was cut out using a pair of ceramic scissors and used for quantification of total trace elements. To analyze the total trace elements concentration, a closed vessel microwave digestion system (MLS-1200 mega, Milestone Inc., Italy.) was employed. The cut filter samples were added into the pre-cleaned Teflon vessels, together with 4 mL HNO₃, 2 mL H₂O₂, and 0.2 mL HF, and then processed using a microwave digestion system, operated in three stages (250 W for 5 min, 400 W for 5 min, and 600 W for 2 min). After digestion, the extracts were filtered with 0.45 μm Teflon membrane filters, diluted to 50 mL and stored in 4°C refrigerator before further analysis.

Gamble's solution and ALF were freshly prepared using the same chemical composition as reported in literature (Midander et al., 2007). For the bioaccessibility assessment, another portion of PTFE filter samples was added into falcon tubes containing 20 mL of SLFs to allow the dissolution to occur. A total dissolution time of 48 h was employed. To study the dissolution kinetics, 2 mL sub-samples were taken at each pre-defined time points (within 48 h) after centrifuge. The whole dissolution experiment was conducted in the 34°C incubator. All the sub-samples were filtered with 0.45 μm Teflon membrane filters and stored at 4°C. Lastly, concentrations of trace elements (Al, Co, Cd, Cr, Cu, Fe, K, Mg, Mn, Ni, Pb, and Zn) were quantified using an Inductively Coupled Plasma Mass Spectrometer (ICP-MS) (ELAN 6100 PerkinElmer, Inc., MA, USA).

For quality control and quality assurance purposes, a standard reference material (SRM) (NIST SRM 1649a, urban particulate matter) and three laboratory blank filters (three each for total elements, Gamble's soluble elements, and ALF soluble elements) were also analyzed. The reported concentrations of trace elements were calculated after subtraction of their concentrations in blank filters from the total concentrations in TSP samples. The recovery of different elements in SRM 1649a ranged from 85 to 111%. Other established quality control and quality assurance protocols as reported elsewhere were also strictly followed (Betha et al., 2014), such as the use of clean stainless steel forceps to handle filter samples and the use of spiked standards after analyzing 20 samples to check the validation of ICPMS calibration.

Health risk assessment was conducted for both adults (>20 years old) and children (10–20 years old) in this work. Four steps are involved in the risk assessment: hazard identification, exposure assessment, dose-response assessment and risk characterization. In this study, the trace elements chosen for risk assessment were: (1) carcinogenic Cd, Co, Ni and Cr; and (2) toxic Al, Co, Cr, Cu, Mn, Ni, Pb, and Zn. In second step, an average chronic daily intake (CDI, mg/kg/day) was calculated using Equation (1). C is the trace elements concentration (mg/m³) (bioaccessible amount of trace elements after 48-h dissolution). EF and ED are the exposure frequency and exposure duration, respectively (180 days exposure frequency was assumed). The inhalation rate (IR) and body weight (BW) used in the Equation were 12.3 m³/day and 57 kg for adults, and 13.6 m³/day and 46 kg for children (Wang et al., 2015), respectively. AT is the averaging life time (70 years, 365 days per year).

The cancer risk and toxic health risk were characterized by Excess Lifetime Cancer Risk (ELCR) and health quotient (HQ), respectively, using Equations (2) and (3). The slope factor (SF) and reference dosage (RfD) were derived from the Integrated Risk Information System (IRIS, USEPA; Lu et al., 2014).

The average mass concentration of TSP during the field campaign period was 61.73 ± 14.86 μg/m³. Chan and Yao (2008) reported that there was a declining trend of the annual average TSP concentration in Shanghai (256 μg/m³ in 1996 ~156 μg/m³ in 2000). However, Wang Y. et al. (2006) reported a very high TSP concentration (231 μg/m³) during 2003–2005. The possible reasons of the low TSP concentration reported in the current study include a short sampling duration (12 selected days over a period of 2 months) and the frequent rain events during the sampling period, leading to the washout of the airborne PM through wet deposition. The measured concentrations of trace elements are presented in Table 1. Mg, Al, K and Fe were the most abundant elements (>1000 ng/m³), while Ni, Co, Cd, and Cr were the less abundant elements (<20 ng/m³). Mg, Al, K and Fe in TSP mainly originated from crustal dust and also possibly from resuspension of soil particles (Wang et al., 2013). Local traffic emissions represent another major emission source of particulate-bound trace elements near the sampling site, since an express way was located within 500 meters from the sampling site. This pollution source may explain the presence of Cu, Pb, and Zn in TSP with relatively high concentrations. Enrichment factors were calculated to find out the enrichment of trace elements, using Equation (4).

Where EF_x is the enrichment factor of an element x, C_xp and C_xc are the concentrations of the element x in TSP and crustal matters, respectively, and C_Alp and C_Alc are the concentrations of Al in the TSP samples and crustal materials, respectively. The elemental concentrations in crustal matters used in the above calculation represent the average concentrations of crustral elements reported by Mason (1966) and Balasubramanian and Qian (2004). According to the EF results (Table 2), K, Mn, and Fe had EFs close to 1, suggesting that they mainly originated from crustal materials. Mg, Cr, Co, and Ni were contributed by air pollution sources of mixed origin. Other than natural emission sources, fuel oil combustion and metal production could also emit Co, Cr, and Ni (Hagler et al., 2007; Ye et al., 2014). The large EFs of Cu, Zn, Cd, and Pb (larger than 10) suggested a major anthropogenic contribution to these elements. As mentioned, Cu, Pb, and Zn were mainly released from the local traffic as the sampling site was situated near an express way. Cd can be emitted from high temperature processes, such as metal smelting and fuel combustion.

Different particulate-bound chemical species behave differently in the human body after their inhalation and deposition, depending on their solubility in lung fluids and toxicity. In general, highly soluble chemical species are easily taken up by lung fluids and then get introduced to the body circulation system through absorption. Thus, bioaccessibility of trace elements is a very important factor to be considered for the assessment of the overall health risk associated with inhalation of PM. Bioaccessibility (%) of the 12 trace elements was estimated as the ratio of the soluble fraction of trace elements in SLFs to the total concentration. The results are shown in Figure 2. Bioaccessibility was then categorized as very high (>50%), high (31–50%), intermediate (16–30%) and low mobility (<15%). The resultant bioaccessibility ranged from 7% (Al) to 47% (K) in the Gamble's solution and 32% (Mn) to 67% (K) in ALF. All the trace elements showed higher solubility in ALF than in the Gamble's solution (p < 0.05), which can be attributed to the more acidic condition of ALF (Colombo et al., 2008). Mg, K and Cu were the most soluble elements in both the Gamble's solution and ALF. No trace elements showed very high solubility in the Gamble's solution. On the contrary, all the trace elements presented a high or very high solubility in ALF. Zn was highly soluble in both the Gamble's solution and ALF. Pb showed a very high bioaccessbility in ALF, but low solubility in the Gamble's solution. Cr, Mn, Co, Ni, and Cd presented high bioaccessibility in ALF, and intermediate solubility in the Gamble's solution. The relatively high accessibility of four carcinogenic trace elements (Cr, Co, Ni, and Cd) may enhance the potential of their health effects. The difference in the bioaccessibility in ALF and the Gamble's solution ranged from 13% (Zn) to 50% (Pb). There was no correlation between bioaccessibility and the total concentrations of trace elements present in PM, suggesting that the bioaccessible fraction of trace elements is not directly related to their abundance in PM. The high bioaccessibility of Cu observed in this study is of concern. According to Fuchs and Packer (2001), chronic copper poisoning can cause damage to eye, liver, and kidney. Though the copper concentration measured in Shanghai is much lower than the guideline value (0.54 μg/m³), the concentration of Cu should still be closely monitored to make sure it is within the safe range. Cu could be released to the atmosphere from sources such as brake pads, rubber tires, concrete, antifouling paints, metal industry, and roofing materials (Midander et al., 2007).

Figure 3 shows the representative kinetic results of dissolution of trace elements in the Gamble's solution and ALF. It was observed that the dissolution processes of all the trace elements were fast during the first 2 h and gradually reached an equilibrium afterwards. It appears that a duration of total 48 h of dissolution is sufficient for a complete transfer of trace elements from the particulate phase to the liquid phase. Julien et al. (2011) reported similar dissolution profiles when analyzing the dissolution of SRM s in the Gamble's solution. In general, the trend in the concentration (mg/L) of dissolved trace elements (Mg > K > Fe > Zn > Cu > Ni) is similar to that of total concentrations of particulate-bound trace elements in ambient air (Mg > K > Fe > Zn > Cu > Ni). This observation implied that higher concentrations of trace elements in TSP would also result in their higher dissolved concentrations in lung fluids after inhalation and deposition. However, this trend was not observed for all the trace elements (such as Mn), suggesting that the concentration of total trace elements was not the only factor that governs the leaching process. For instance, Al had a higher total concentration than those of K and Fe, but lower dissolved concentrations in Gamble's solution. All the trace elements had higher dissolved concentrations in ALF than in the Gamble's solution, as well as higher bioaccessibility in ALF. Many studies reported that a lower pH of the liquid media would enhance the dissolution of trace elements (Colombo et al., 2008; Joshi et al., 2009; Wiseman and Zereini, 2014). The higher dissolution of trace elements in ALF suggests that the inhaled PM is likely to undergo phagocytosis before significant amounts of trace elements are dissolved. Phagocytosis is one of the major clearance pathways of particles in the lung (Gehr and Heyder, 2000). It is also suspected that the dissolved metals may hamper the immune function of macrophages (Jalava et al., 2008), and thus lower down the clearance rate of particles through phagocytosis. From Figure 3, it can be observed that different trace elements dissolved in the lung fluids at different rates. For example, it was observed that the dissolution of Cr was slower than that of Cd and Ni. Moreover, the slope of the dissolution profiles of trace elements in ALF was not always larger than that of trace elements in the Gamble's solution. The dissolution rate is determined by the specific chemical speciation forms of trace elements, for example, As (III) vs. As(V) and salts vs. oxidates and origins of the trace elements (crustal or combustion sources; Midander et al., 2007).

The dissolution of nearly all elements proceeds in a curvilinear trend, when exposed to ALF and the Gamble's solution, with their rapid release occurring at the beginning of a protracted period of incubation (48 h). However, there are variations in the solubility patterns of individual trace elements which are not unexpected because of differences in their origins, chemical states, physical and chemical aging as a function of time during the sampling period, consequently leading to high heterogeneity nature of PM samples collected.

The bioaccessible concentrations used in this study provide the upper limit of risk assessment estimation since they represent the availability of toxic elements for absorption when dissolved in SLFs (Mukhtar and Limbeck, 2013). The purpose of using bioaccessible concentrations instead of the total elemental contents in PM samples for health risk calculations is to make a realistic estimate of the actual health risks due to inhalation of PM and to prevent overestimation as the particulate-bound elements are not fully bioaccessible (Luo et al., 2012).

Health risks were conducted for both adults and children. The resultant ELCR and HQ values are presented in Tables 3, 4, respectively. ELCR higher than 10⁻⁶ (1 in a million) indicates a significant level of cancer risk, whereas HQ higher than 1 implies significant toxic health effects on exposed human population. As can be seen from Table 4, the HQ values calculated from the bioaccessible fraction of trace elements in both the Gamble's solution and ALF were far <1 (0.084–0.253), suggesting that no significant toxic effects on both adults and children could be expected with the measured concentrations of toxic trace elements. However, only trace elements were considered in the risk assessment. Other toxic compounds presenting in TSP such as PAHs were not considered in this study. Consequently, the HQ may be underestimated. The HQ values estimated for children using both bioaccessible fractions in the Gamble's solution and in ALF were higher than those for adults (p < 0.05). A similar trend was found in the estimated cumulative ELCR (Gamble's solution: 13.4 × 10⁻⁶ for adults and 20.4 × 10⁻⁶ for children; ALF: 31.1 × 10⁻⁶ for adults and 47.4 × 10⁶ for children) (p < 0.05). The results suggested that children that are exposed to outdoor particulate air pollution in Shanghai may experience higher cancer risk and non-cancer risk than adults. All the resultant cumulative ELCRs were higher than 10⁻⁶, indicating that adults and children are likely to experience significant cancer risk due to inhalation of particulate-bound trace elements. Mn was the key contributor to HQ while Cr made a significant contribution to the cumulative ELCR. The concentration of Mn was not the highest among all the trace elements considered in the risk assessment. However, it should be of concern because of its high toxicity. It was assumed all the Cr dissolved was Cr(VI), which may overestimate the ELCR for Cr. Both HQ and ELCR calculated using the bioaccessible fraction in ALF were about two times of that calculated using the Gamble's solution (p < 0.05). This was due to the high solubility of trace elements in ALF (pH = 4.5) under more acidic conditions compared to the Gamble's solution (pH = 7.4).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.