Huazhen Shen1,2, Chung-Shin Yuan This email address is being protected from spambots. You need JavaScript enabled to view it.2, Chun-Chung Lu2, Yubo Jiang2, Guohua Jing1, Gongren Hu1, Ruilian Yu1

College of Chemical Engineering, HuaQiao University, Xiamen 361021, China
Institute of Environmental Engineering, National Sun Yat-Sen University, Kaohsiung 80424, Taiwan


Received: August 24, 2019
Revised: September 24, 2019
Accepted: September 24, 2019

Download Citation: ||https://doi.org/10.4209/aaqr.2019.08.0410  


Cite this article:

Shen, H., Yuan, C.S., Lu, C.C., Jiang, Y., Jing, G., Hu, G. and Yu, R. (2019). Chemical Composition and Health Risk of PM2.5 from Near-ground Firecracker Burning in Micro Region of Eastern Taiwan. Aerosol Air Qual. Res. 19: 2252-2266. https://doi.org/10.4209/aaqr.2019.08.0410


Highlights

  • Downwind distance affects chemical compositions of PM2.5 from burned firecracker.
  • 80% of increments of PM2.5 conc. was ascribed to firecracker burning in Taitung.
  • Health risk caused by firecracker burning varied with the downwind distance.
 

ABSTRACT


The randomness of firecracker-burning site and the overlapping impact of multi-sources makes the source apportionment of PM2.5 during the firecracker burning events more difficult. To investigate the influences of the downwind distance to the firecracker-burning site on the temporospatial distribution of PM2.5 and public health risk, PM2.5 were sampled at three sites adjacent to a fixed firecracker-burning route accompanied with annual pilgrimage activity during the Lantern Festival in Taitung, Taiwan, which had a low background PM2.5 concentration. The metallic elements, water-soluble ions, carbonaceous contents were analyzed. The potential sources were identified using positive matrix factorization. Finally, the health risks were assessed by calculating the hazard quotient and incremental lifetime carcinogenic risk, respectively. The results showed that the average concentration of PM2.5 on the event days increased by approximately five-fold compared to the non-event days. The increase of chemical components varied significantly from the distance to the burning site. The concentrations of K, Fe, Al, Mg, K+, Cl and OC rose by 6–14 times at one site close to a site with intensive firecracker burning, while increased by 2–6 times at one site far away from the firecracker burning sites. The PM2.5 increment on the event days was mostly attributed to firecracker burning, kitchen fumes, and mobile sources. The health risk assessment results showed that the hazard index differed between the sampling sites. Furthermore, the cancer risk of one site close to the firecracker burning site was over the threshold, while that far away from the site was below the threshold.


Keywords: PM2.5; Firecracker burning; Chemical composition; PMF; Health risk.


INTRODUCTION


Near-ground firecracker burning could cause a rapid and dramatic increase in PM2.5 concentration, not only reducing ambient air quality and visual range but also posing hazards to residents' health. Moreover, other areas might suffer from PM2.5 pollution by such firecracker burning because of long-range transport (Alexandre et al., 2010; Tsai et al., 2012). Previous literature reported that Asian megacities such as Beijing, Tianjin, Nanjing, New Delhi often suffer from highly polluted PM2.5 episodes, accompanying more than several hundred µg m–3 during the traditional Chinese and Indian festivals (Zhang et al., 2017b; Jung et al., 2018; Mukherjee et al., 2018; Rastogi et al., 2019). During the festival, the concentration of various chemical indicators of firework displays increases dramatically. For instance, in Beijing, Ba, Cu, Pb as color agents, and propellants increased nearly100 times (Liu et al., 2019). Kong et al. (2015) found that organic carbon, elemental carbon, ammonium, potassium, and silicon accounted for nearly 70% of total PM2.5 in the Spring Festival in Nanjing. Different from the spatiotemporal distribution, chemical compositions, and sources of PM2.5 during traditional festivals in the megacities, a large number of near-gourd firecrackers are burned accompanying with blasting the God Handan or other folk performances as the pilgrimage activities occurred in the downtown area, a micro-region of Taitung City in Eastern Taiwan. The remarkable feature of PM2.5 episodes in a micro-region is that the PM2.5 influences in a short duration and distance. The distance from the sampling site to the firecracker burning site could cause large variation in the PM2.5 spatial distribution, making the estimation of the firecracker burning contributed to PM2.5 pollution more difficult. Additionally, regulated by local Environmental Protection Bureau, the pilgrimage routes are confined within the specified city blocks, and the firework burning sites are fixed as well. The areas near the pilgrimage are directly polluted by the firecracker burning, while the regions far away from the pilgrimage route are less affected by the firecracker burning. Therefore, it is of considerable significance to investigate the effects of the downwind distance to the burning site on the spatial distribution of PM2.5 by choosing appropriate PM2.5 sampling locations based on the pilgrimage routes.

The overlapping of multi-sources including local coal-fired power plants, metal processing factories, incinerators, as well as PM2.5 from long-range transport in addition to firecracker burning during the festivals has made it even more challenging to clarify the contribution of firecracker burning to PM2.5 pollution (Feng et al., 2012; Yang et al., 2014; Feng et al., 2016). Kong et al. (2015) indicated that PM2.5 pollution in Nanjing mainly comes from fireworks (50.0%), coal combustion (15.6%), soil (14.8%), vehicular exhaust (14.4%). Mukherjee et al. (2018) observed that in New Delhi firecracker burning in urban areas increased PM2.5 concentration by more than 25% during the Diwali festival in 2016. Moreover, fine particles emitted from remote biomass burning through long-range transport also dramatically promote ambient PM2.5. Compared to several industrial cities located in the western Taiwan, Taitung City has no heavy industries such as steelmaking, power generation, or waste incineration, and thus local PM2.5 emission sources including coal-burning boilers, road dusts, traffic exhausts, kitchen fumes, and etc. dominate the PM2.5 pollution. Furthermore, due to the barrier of the Central Range in Taiwan Island, Taitung City is rarely affected by PM2.5 transported from outside the region (Fang et al., 2010; Lee et al., 2018). Therefore, Taitung City has low PM2.5 concentration with an annual average PM2.5 level of 9.7 µg m–3 for the past five years, which is even lower than the most stringent criteria of the annual average concentration of 10 µg m–3 (1T-3) allowed by United Nations (Fang et al., 2013). Due to its low PM2.5 background, Taitung City is selected for investigating the contribution of near-ground firecracker burning on PM2.5 concentration.

A variety of metallic elements could be released in the firework displays (Tian et al., 2014; Jiang et al., 2015), which are used to generate colorful flame. Although the health risk assessment of heavy metals emitted from firework displays has been extensively investigated (Lin et al., 2016; Song et al., 2017; Zhang et al., 2017a; Pong et al., 2018; Zhang et al., 2018b; Greven et al., 2019), the risk assessment of heavy metals from near-ground firecracker burning has rarely been conducted. During the pilgrimage time in Taitung, “Blasting the God Handan”, blessings and other exorcisms activities often attract an abundance of attendees at a pretty short distance. Therefore, firecracker burning may have a tremendous impact on the health of spectators but has mostly been overlooked (Wehner et al., 2000; Ravindra et al., 2003; Song et al., 2006; Moreno et al., 2007). Consequently, it is very imperative to assess the health risk of heavy metals emitted from firecracker burning on the spectators.

To investigate the impacts of the downwind distance to the burning site on the spatiotemporal distribution and the chemical composition of PM2.5 during the Lantern Festival in Taitung City, we collected PM2.5 samples in three periods (before, during, and after the firecracker burning) at three different sites, i.e., a site close to the burning site with intensive firecracker burning (site A), a site close to burning site but with non-intensive firecracker burning (site B), and a site far away from burning sites without firecracker burning (site C). The PM2.5 concentrations and their metal components, water-soluble ionic (WSI), and carbonaceous species at the three sampling sites were analyzed. The potential sources and their contributions to PM2.5 were analyzed by backward trajectory simulation and positive matrix factorization (PMF). The health risk of heavy metals in PM2.5 emitted from firecracker burning was further assessed as well.


EXPERIMENTAL METHODS



Sampling Protocol

Three sites were selected for PM2.5 manual sampling. As illustrated in Fig. 1, site A was a balcony on the second floor of a building neighboring an intensive firecracker burning site on the route of pilgrimage activities. Site B was a balcony on the third floor of a building close to a non-intensive firecracker burning site on the boundary of the path of pilgrimage activities, while site C was a balcony on the third floor of a building far away from sites A and B without firecracker burning at all. The distance between sites A and B was approximately 300 m, while that between sites A and C was about 450 m. The sampling periods set on February 21st and 24th, before and after the firecracker burning event, respectively, represented the non-event days (NDs) and the PM2.5 samples were collected continuously for 24 h (from 8:00 to 8:00 of the next day). PM2.5 samples collected on February 21st were denoted as A1, B1, and C1, and those received on February 24 were denoted as A4, B4, and C4. On February 22nd and 23th, which were the event days (EDs) during the Lantern Festival when pilgrimage activities were held, the samples were collected from 14:00 to 24:00. From the midnight (0:00) to the next morning (8:00), no firecracker burning was conducted. To avoid low PM2.5 concentration from the midnight to the early morning dragging down the average PM2.5 level of the event day, PM2.5 samples were not collected at that time. PM2.5 samples collected on February 22nd were denoted as A2, B2, and C2, while those samples collected on February 23rd were denoted as A3, B3, and C3. PM2.5 was sampled using the PM2.5 sampler (BGI, PQ200) with WINS impactors under an air flow rate of 16.7 L min1. Each quartz fiber filter of 47 mm diameter was conditioned in the desiccators at constant room temperature (25 ± 3°C) and relative humidity (45 ± 5%) for at least 24 h and further weighted by an analytical microbalance (Sartorius MC 5) with a mass precision of 10–6 gram before and after PM2.5 sampling.


Fig. 1. Location of three sampling sites selected for collecting PM2.5 at Taitung, Taiwan.Fig. 1. Location of three sampling sites selected for collecting PM2.5 at Taitung, Taiwan.

In addition to the three aforementioned sampling sites, we also conducted on-line monitoring of PM2.5 and PM10 concentrations by a real-time particulate monitor (Met-one, Aerocet 531S) along the pilgrimage route during the firecracker burning activities on the EDs. 


Chemical Composition Analysis

Prior to analyzing the metallic elements in PM2.5, a sheet of 1/2-sized quartz filter paper was placed in a 120 mL vessel for conducting microwave digestion with 30 mL of an acid mixture [HNO3 (v/v% = 65%): HClO4 (v/v % =70%) = 3:7, v/v] at 150–200°C for 2 h. After that, 25.0 mL of ultrapure water was added to rinse off the residual acids, then the digestion and rinse processes were repeated for three times, and finally, the solution was diluted to 30 mL with 0.5 mol L–1 HNO3 for further metallic analysis. Sixteen metallic elements (Mg, Al, Ca, Ti, Mn, Fe, Ni, Zn, Pb, Cr, Cd, K, As, Ni, Cu, and Na) were analyzed using an inductively coupled plasma-atomic emission spectrometer (ICP-AES) (PLASMA 400, PERKIN ELMER) (Wu et al., 2015).

Before analyzing the WSIs, a quarter of the quartz filter was placed in a 50 mL polyethylene (PE) bottle, 30 mL deionized water (DI H2O) was added, and the mixture was sonicated for 120 min. The solution was further filtered with a 0.45 µm acetate fiber filter, and the filtrate was analyzed for ionic species using an ion chromatography (Dionex, Series 120). The measurement method with operating parameters were described in detail by previous research (Li et al., 2013a).

The carbonaceous species in PM2.5 were analyzed using an elemental analyzer (CHNS/O 1108, Carlo Erba) with an autosampler (Model AS 200) and a DP 700 integrator. The elemental carbon (EC) and total carbon (TC) in PM2.5 were determined, and organic carbon (OC) was estimated by subtracting EC from TC. The preparation method was described detailly in previous research (Yang et al., 2017).


Quality Assurance (QA) and Quality Control (QC)

The filters and other consumables were supplied with backups. All instruments and filters were carefully installed prior to sampling. All items involved sampling operation such as the starting and ending time of PQ200, sampling flow rate, weather conditions, and etc. were recorded so that the operating parameters were available for checking to accurately determine the PM2.5 concentration. To ensure the accuracy of chemical analysis experiments, the spiking method was performed in which a specific amount of standard reagent was spiked into the samples for the detection of chemical compositions under the same analytical conditions. Each measurement was repeatedly analyzed to obtain the recovery rate. The closer the recovery rate approached to unity, the higher the accuracy of the analytical instrument is (Tsai et al., 2010; Lin et al., 2013; Yuan et al., 2006).


Relative Abundance Analysis

Relative abundance analysis (Ra) is the ratio of the percentage of the target metallic elements in PM2.5 of the EDs over that in the NDs, as shown in Eq. (1) to identify the enriched metal influenced by the firecracker burning but not by other pollution sources (Kong et al., 2015; Liu et al., 2019).

 

where CiFC and CPM2.5FC (µg m–3) are the average concentrations of a target metallic element i and PM2.5, respectively, in the EDs; Cin-FC and CPM2.5n-FC (µg m–3) are the average concentrations of the corresponding metallic element i and PM2.5 in the NDs, respectively.


Backward Trajectory Simulation

In order to trace the air masses transported toward Taitung City, backward trajectories from the reception sites are commonly used to identify their potential source regions. The Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) was employed to trace the transport routs of the air parcel arriving at Taitung City on February 21st–24th in this study. Three-dimensional (height, longitude, and latitude) and 72-h backward trajectories were simulated every 6 h from 18:00 UTC, February 20th (Li et al., 2017a). Air masses with low altitude of 100, 300, and 500 m were simulated. Finally, a total of 24 simulated back trajectories were plotted for this study.


Positive Matrix Factorization

PMF5.0 was employed to quantify the mass contribution of the chemical constitutes during the EDs and NDs as shown in Eq. (2).

 

To evaluate the fitting degree of PMF, Q is the goodness-of-fit parameter which can be obtained by Eq. (3),

 

where Unc is the uncertainty matrix of each chemical species and can be calculated by Eq. (4) (U.S. EPA, 2014),

 

where EF is the determined error fraction, which was set as 10% in this study (Yang et al., 2015; Li et al., 2017b). A total of twelve samples were divided into two groups representing the PM2.5 concentrations of EDs and NDs, respectively. Twenty-six chemical species were introduced into the PMF model for analyzing their contribution. A total of 20 times was run with the number of factors ranging from 3 to 6 and determined by the Q value, residual analysis, and correlation analysis between the observed and predicted values. To increase the Q value, the rotations were performed by selecting an appropriate number of the factors with the strength ranged from –1.0 to 1.0. The simulated results of source apportion with the highest values of Q (Robust) and Q (True) was chosen as the PM2.5 source profiles during the NDs or EDs.


Health Risk Assessment of Heavy Metals in PM2.5

To evaluate the non-carcinogenic health risk, the inhalation (Dinh) for adults and children was determined based on an individual's body weight and exposure span as shown in Eq. (5) (Wang et al., 2007; Zhang et al., 2018a):

 

where C (mg m–3) is the average concentration of metallic species in PM2.5 sampled on February 22nd and 23rd; Dinh (mg kg–1·day–1) is the exposure dosage by respiratory inhalation; InhR is the inhalation rate of 7.6 and 20 m3 day–1 for children and adults, respectively; EF is the exposure frequency and once every year (day year–1) was adopted for representing the EDs in this study; ED is the exposure duration of 6 and 24 years for children and adults, respectively; BW is the average body weight of 15 and 70 kg for a child and adult, respectively; AT (day) is the average time for non-carcinogenic toxic risks as shown in Eq. (6),

 

Hazard Quotient (HQ) representing non-carcinogenic risk is determined using Eqs. (7) and (8):

  

where RfD (mg kg–1 day–1) is the reference dose. The RfD values are 3.01 × 104, 1.0 × 10–3, 2.86 × 105, 1.4 × 105, 2.06 × 102, 3.52 × 10–3, 7.0 × 10–3, and 3.01 × 101 for As, Cd, Cr, Mn, Pb, Ni, V, and Zn, respectively (Feng et al., 2016). The Hazard index (HI) can be obtained by summing up the HQ of single metal to estimate the total risk. If HI ≤ 1, there is no adverse effect of PM2.5 on human health, while there is if HI > 1.

For carcinogenic risk, it is determined by the following Eqs. (9) and (10):

  

where Ri is the carcinogenic risk caused by specific metal element i; LADD (mg kg–1 day–1) is the lifetime average daily dosage of single heavy metals exposure by inhalation which is described as Eq. (11),

 

where SFa (kg·day mg–1) is the slope factor with the values of 15.1, 6.4, 42, and 0.84 for As, Cd, Cr, and Ni, respectively (Wang, 2007); Rt is the total of carcinogenic risk exposure to humans. 


RESULTS AND DISCUSSION



PM2.5 Concentration Analysis

The daily average PM2.5 concentration of A1 (15.9 µg m–3) was higher than those of B1 (11.8 µg m–3) and C1 (10.3 µg m–3) on February 21st during the NDs as shown in Fig. 2(a), because site A is located at the downtown of Taitung City with many restaurants and traffics, while the flow of people and traffic is relatively fewer around sites B and C. Additionally, site A is situated at the downwind of site C on February 21st as the wind mainly blew northwesterly or northwest-westerly (see Fig. S1(a)). Thus, the average PM2.5 concentration of C1 was less influenced by the transport of PM2.5 from the downwind site A.


Fig. 2. Variation of PM2.5 concentrations (a) daily average concentration at three sampling sites during sampling period; (b) real-time concentrations of PM2.5 and PM10 concentrations during a firecracker-burning process.Fig. 2. Variation of PM2.5 concentrations (a) daily average concentration at three sampling sites during sampling period; (b) real-time concentrations of PM2.5 and PM10 concentrations during a firecracker-burning process.

The daily average PM2.5 concentrations of A2 (70.0 µg m–3) and B2 (30.0 µg m–3) on February 22nd, A3 (59.1 µg m–3) and B3 (47.6 µg m–3) on February 23rd were significantly higher than those on February 21st, when a vast number of firecrackers were burned along the pilgrimage activities on the EDs. Additionally, the daily average PM2.5 concentrations of A2 was much higher than that of B2, probably because site A was more adjacent to the intensive firecracker burning sites, while a relatively smaller amount of firecracker was burned around site B. The daily average PM2.5 concentrations of C2 (23.4 µg m–3) and C3 (17.7 µg m–3) were the lowest among three sampling sites, because on the one hand, the pilgrimage activities and the firecracker burning were banned at site C on February 22nd and 23rd; on the other hand, site C was still located upwind of site A when Taitung was dominated by northwest and west wind with low speed during the EDs (see Figs. S1(b)–S1(c)). On February 24th, after ending the pilgrimage activities, the average PM2.5 (24.0 µg m–3) of A4 was not influenced by firecracker burning anymore but was still higher than those of B4 (19.5 µg m–3) and C4 (13.8 µg m–3), which was as similar as the distribution character on February 21st.

The results of continuous monitoring of PM2.5 and PM10 show that the near-ground mean concentrations of PM2.5 and PM10 reached up to 32 ± 10 µg m–3 and 102 ± 20 µg m–3, respectively, before one completed process of firecracker burning around site A. The higher concentrations of PM2.5 could be attributed to the heavier cooking fumes and traffic exhaust accompanying with the pilgrimage activities than the NDs on February 22nd (see Fig. 2(b)). When a large number of firecrackers was set off, the atmospheric concentrations of PM2.5 and PM10 rapidly increased to 613 µg m–3 and 2,049 µg m–3 in two minutes, and then slowly decreased to 22 µg m–3 and 93 µg m–3, respectively. The phenomenon inferred that the mean concentration of PM2.5 before and after the cracker burning contributed by the heavy traffic flow, and cooking fumes was much lower than the daily average PM2.5 of 70 µg m–3 at site A. Therefore, the daily average concentrations of PM2.5 at site A during the EDs were substantially attributed to the elevation caused by the continual high-concentration peak of PM2.5 during the firecracker burning.


Metallic Element Analysis

The mean metallic content in PM2.5 sampled at sites A, B, and C during the NDs accounted for 13.4 ± 2.4% of total PM2.5 concentration, while that at sites A, B, and C significantly increased to 22.8 ± 2.8% during the EDs due to firecracker burning. The mean concentration of each metallic element in PM2.5 of A2 and A3, denoted as (CiA2 + CiA3)/2, at site A during the EDs was much higher than that of A1 and A4, denoted as (CiA1 +CiA4)/2 (see Fig. 3(a)). The mean concentrations of K, Mg, Fe, and Al during the EDs were approximately 6-times higher than those during the NDs. Potassium is mainly originated from the oxidizing agents of firecrackers such as KNO3, KClO3, and KClO4, the oxidation of carbon and sulfur in powder which is a basic chemical reaction occurred when firecracker is exploded (Chen et al., 2016; Yang et al., 2018). Powdered Mg and Al are the common reducing reagents of firecrackers (Vecchi et al., 2008; Wang et al., 2013). Iron is often used as the propellant of firecrackers in the form of ferrocene (Han et al., 2019). The average concentrations of Ti, Mn, Cr, Ca, Na, V, Zn, Cd, and Cu at site A during the EDs were 3-times higher than those during the NDs. The rise of Ca, Na, Cu, Ti, Zn, and Cr concentrations could be attributed to the occasional burning of a few small fireworks displayed along with the pilgrimage since they are the main coloring agents of firework (Lin et al., 2016); the increase of Cd was majorly attributed to traffic exhausts (Pong et al., 2018). It’s noted that a few metallic elements as mentioned could be emitted by other sources. For instance, Iron and Titanium are the major crustal elements and thus could be released from the fugitive road dust during the EDs (Li et al., 2012, 2015); Sodium might be released from the sea salts during the cooking activities in dry and high-temperature condition such as meat roasting (Zhang et al., 2016); Zinc might also be released from the tire wear of vehicles (Lough et al., 2005; Hjortenkrans et al., 2007).


Fig. 3. Variation of daily metallic element concentrations at (a) site A; (b) site B; (c) site C; and (d) Ra for three sampling sites.Fig. 3. Variation of daily metallic element concentrations at (a) site A; (b) site B; (c) site C; and (d) Ra for three sampling sites.Fig. 3. Variation of daily metallic element concentrations at (a) site A; (b) site B; (c) site C; and (d) Ra for three sampling sites.Fig. 3. Variation of daily metallic element concentrations at (a) site A; (b) site B; (c) site C; and (d) Ra for three sampling sites.

The increments of metallic elements at sites B and C were lower than those at site A as shown in Figs. 3(b)–3(c). The concentrations of Al, Na, and Mg at site B raised by approximately 3-times while the concentrations of Cu, Zn, K, Ca, Cd, Cr, Ti, Fe, and Mn increased by about 2-times during the EDs. At site C, although the concentrations of Al and Mg increased by about 4-times during the EDs, most of the metallic elements such as Fe, Na, Cd, Cr, Cu, Mn, Ca, Zn, and K increased approximately by 2-times. Therefore, the temporal distribution of metallic elements was substantially influenced by the distance of the sampling site to the burning site. The shorter distance was between the sampling site and the burning site, the larger the increment of metallic elements was formed in the PM2.5. However, the concentration increment of Ni during the EDs was the smallest at each sampling site, and merely 1.5-times higher than those during the NDs; what’s more, their concentrations even decreased on February 23rd compared to that during the NDs, indicating that the concentration of Ni was less influenced by firecracker burning. It’s known that Ni mainly originated from coal or heavy oil-burning boilers (Kong et al., 2015; Yang et al., 2017), and thus the contribution from coal or heavy oil-burning boilers was much lower than the other pollution sources to the form of PM2.5 during the EDs.

To further investigate the characteristics and abundance of metallic elements in PM2.5 around the sampling sites, the relative abundance (Ra) of metallic elements were determined. It showed that the most significantly abundant metallic elements in PM2.5 with Ra > 2 at site A were K, Cd, Mg, Fe, and Al, confirming that they were mainly released from the firecracker burning (see Fig. 3(d)), while the Ra of other metallic elements such as Mn, Cr, Na, V, Zn, Ti, and Cu were higher than unity, indicating that the metallic elements released from firework display, traffic exhaust, kitchen fumes and etc. were enriched in PM2.5, but lower than those emitted from firecracker burning. Additionally, the Ra of Ni was lower than unity at site A, confirming that the influence of coal- or heavy oil-fired boilers on PM2.5 was much lower than other emission sources such as firecracker burning, firework display, traffic exhaust, cooking fume.

At site B, the Ras of Na, K, Al, and Mg were nearly equal to or slightly lower than unity, indicating that the concentrations of these metallic elements at site B were less affected by firecracker burning than those at site A. However, the Ras of Al and Mg at site C were much higher (Ra > 2), while those of Na, K, Zn, Fe, Ni, Ca, and Cd were slightly higher than unity. The results indicate that many metallic elements were also much abundant in PM2.5 at site C, inferring that the large proportion of PM2.5 at site C was influenced by the dispersion of PM2.5 from the intensive burning sites.


Water-soluble Ion (WSI) Analysis

The variation of WSI in PM2.5 sampled at sites A, B, and C is shown in Fig. 4. During the NDs, the mean concentrations of WSIs at sites A, B, and C were 6.2 ± 1.5, 5.8 ± 2.0, and 3.5 ± 0.6 µg m–3 which accounted for 31.2 ± 1.4%, 37.4 ± 0.3% and 28.9 ± 0.6% of the total mass of PM2.5, respectively. During the EDs, the mean concentrations of WSI at sites A and B significantly increased by 5.0- and 2.7-times and reached up to 31.2 ± 4.2 and 15.9 ± 5.3 µg m–3, respectively. The contents of WSI in PM2.5 sampled at sites A and B accounted for 48.2 ± 4.3% and 41.0 ± 2.1%, respectively, indicating that both the concentrations and contents of WSIs were promoted by the intensive firecrackers burning. While, the mean concentration of WSIs (7.5 ± 2.9 µg m–3) at site C during the EDs was approximately 2.2-times higher than that (3.5 ± 0.6 µg m–3) during the NDs, indicating that site C was far away from the firecracker burning site, however, the concentrations of WSIs in PM2.5 still increased, merely the increment was lower than those at sites A and B.


Fig. 4. Variation of daily WSI concentrations at three sampling sites.Fig. 4. Variation of daily WSI concentrations at three sampling sites.
 

The concentration variation of WSIs at site A is shown in Fig. 4. It shows that the mean concentrations of K+ and Cl- at site A during the EDs were 11.6- and 7.5-times higher than those during the NDs, respectively, resulting from the oxidation of charcoal in contact with KNO3, KClO4, and KClO3 during the firecracker burning as shown in Eq. (12):

 

The concentrations of SO42– at site A during the EDs were about 4.5-times higher than those during the NDs due to the oxidation of elemental sulfur powder by KNO3 as shown in Eq. (13):

 

The mean concentrations of NO3, Na+, Mg2+, and Ca2+ at site A during the EDs were 4.4-, 3.8-, 3.7-, and 3.1-times higher than those during the NDs, respectively. It was noted that the mean concentration of NH4+ during the EDs was also 2-times higher than that during the NDs, which was probably ascribed to the oxidation of elemental sulfur powder by NH4ClO4 as shown in Eq. (14):

 

In addition, NH4+ might be emitted from snack-street boiling because a small amount of ammonium chloride or ammonium phosphate as food ingredients are also used to increase the toughness of noodles (Li et al., 2014).

At site B, the mean concentrations of Cl, Na+, and K+ during the EDs were 5.6-, 3.7-, and 2.5-times higher than those during the NDs, respectively, and 2.9-times, 2.8-times, 2.2-times, and 2.1-times higher than those of Mg2+, NO3, Ca2+, and SO42– during the NDs, respectively (see Fig. 4). The results indicate that the increments of WSIs were much lower than those at site A. At site C, the increments of WSIs were far lower than those at sites A and B (see Fig. 4), revealing that the increments of WSIs differed significantly among sites A, B, and C, which were the same as metallic elements due to the variation of the distance between sampling site and burning site.

As shown in Table S1 the mass ratios of K/K+ of A1 and A4 were 1.6 and 1.2 during the NDs, respectively while those of A2 and A3 were 1.0 and 1.1 during the EDs, indicating that the large proportion of K existed in the form of ions, such as KCl, K2SO4, and KNO3, rather than those in the form of metallic element in PM2.5 after the massive explosion of firecrackers, which was in an agreement with Zhang et al.' (2018b) observation. The mass ratios of Mg/Mg2+ of A1 and A4 were 1.3 and 1.5, while those of A2 and A3 reached up to 3.4 and 2.9, respectively, indicating that most of Mg was not converted to Mg2+ after the firecracker explosion. The metallic element of Mg was probably originated from the residual Mg powder due to the incomplete explosive reaction of Mg alloy. In addition, the mass ratios of Cl/Na+ of A1 and A4 during the NDs nearly equal to unity, approaching to the equivalent ratio of sea salts, but increased to 1.5 and 1.6 of A2 and A3 during the EDs, respectively, resulting from the rapid release of Cl which then probably existed in the form of KCl after firecracker burning. Therefore, during the EDs, the metallic element of Mg, the ion of K+, and the ratio of Cl/Na+ are suitable to indicate firecracker burning.


Carbonaceous Content Analysis

During the NDs, the mean total carbon (TC) contents in PM2.5 at sites A, B, and C accounted for 11.9 ± 2.3%, 10.1 ± 1.4%, and 11.7 ± 0.5%, respectively. During the EDs, the mean TC content at site A increased to 22.5 ± 2.3% (see Fig. 5), indicating that the TC content in PM2.5 at site A was significantly elevated by the firecracker burning. Although the mean TC contents increased to 15.7 ± 0.7% and 16.2 ± 1.7% at sites B and C, they were less elevated than that at site A.


Fig. 5. Variation of daily carbonaceous concentrations at three sampling sites.Fig. 5. Variation of daily carbonaceous concentrations at three sampling sites.

During the EDs, the mean contents of OC and EC at site A were 11.6 ± 0.8 and 2.8 ± 0.5 µg m–3, respectively, and 6.7- and 3.9-times higher than OC (1.7 ± 0.8 µg m–3) and EC (0.7 ± 0.2 µg m–3) during the NDs. The mean contents of OC (4.7 ± 1.7 µg m–3) and EC (1.4 ± 0.5 µg m–3) at site B during the EDs were 4.15- and 2.98-times higher than OC (1.1 ± 0.5 µg m–3) and EC (0.5 ± 0.2 µg m–3) during the NDs, respectively. The mean contents of OC (2.5 ± 0.8 µg m–3) and EC (0.9 ± 0.2 µg m–3) at site C were 3.4- and 2.4-time higher than OC (0.7 ± 0.1 µg m–3) and EC (0.4 ± 0.2 µg m–3) at site C. The results confirmed that the distance between the sampling site and the burning site played an essential role in the spatial distribution of the contents of PM2.5 released from firecracker burning, namely, the longer the distance was, the lower the concentration increments of OC and EC in PM2.5 were.

The mean mass ratio of OC/EC during the EDs was 4.2 ± 1.0 and significantly higher than that (2.4 ± 0.4) during the NDs at site A. The OC/EC ratios during the EDs approached 3.1–5.7 for firecracker burning observed by Feng et al. (2016), but much lower than 7.3 for wood burning (Choosong et al., 2010), suggesting that the firecracker burning as the primary PM2.5 source had played an essential role in the form of organic carbon during the Lantern Festival. The OC/EC ratio of 3.2 ± 0.2 during the EDs was also higher than that (2.3 ± 0.1) during the NDs at site B. It shows that the concentrations of OC and EC increased significantly, particularly for OC which increased much higher than that of EC. It was probably that the burning of organic matter in firecracker such as phenolic resin, polyvinyl alcohol, polyoxyethylene would release a large amount of OC to the atmosphere through the explosion. The OC/EC ratio (2.7 ± 0.3) at site C during the EDs was much lower than those at sites A and B, which could be attributed to that site C was far away from the burning site.

It is noteworthy that the OC/EC ratio at site A was much higher than that at site B, indicating that the carbonaceous species in PM2.5 at site A was not only influenced by PM2.5 released from firecrackers burning, but also by PM2.5 with higher ratio of OC/EC emitted from other anthropogenic activities. It is known that the OC/EC ratio of PM2.5 released from various cooking activities such as meat roasting, cafeteria frying, and snack-street boiling, ranged from 8.8 to 55 since a large number of OC are released through oxidation, decarboxylation, and cyclization reactions from the raw oil and meat components (Li et al., 2014). Compared to site B, the OC/EC ratios of PM2.5 at site A increased dramatically due to potential proliferated cooking activities during the EDs.


Analysis of Backward Trajectory

The analysis of backward trajectory is illustrated in Figs. S2(a)–2(d). It showed that on February 21st the air mass originating from Shandong Peninsula of mainland China successively passed through the Yellow Sea, the East China Sea, and the east coast of Taiwan Island before arriving at Taitung City. On February 22nd, the long-range transported air mass mainly came from the Yellow Sea and passed through the East China Sea before entering Taitung City, while a small amount of air mass originating from the north of Philippines, passed through the Bashi Channel and the Taiwan Strait, and landed at the north of Taiwan Island before entering Taitung City (see Fig. S2(b)). Since the air mass entering Hualien City, bounded on the north of Taitung City, originated from the same area and had the similar transport path as that entering Taitung City on February 22nd ((see Fig. S2(c)), the mean PM2.5 concentration of Hualien City was 17.3 ± 4.2 µg m–3 for comparison, indicating that no episode occurred in Hualien City, and thus the air mass transported from the same area as that of Hualien City contributed little to the PM2.5 pollution of Taitung City on February 22nd as well.

On February 23rd, the air mass originating from the Yellow Sea landed in the northern part of Taiwan through the East China Sea, and successively passed through Ilan, Hualien cities along the east side of the Central Range, and finally reached Taitung City (see Fig. S2(d)). Thus, the air mass might carry upwind PM2.5 from the north to Taitung City. The daily mean PM2.5 concentration of Hualien was 11.7 ± 3.7 µg m–3 on February 23rd, suggesting that no PM2.5 episode occurred in the upwind of Taitung City. Therefore, the outside air mass had little influence on the PM2.5 episode of Taitung City on February 23rd. On February 24th, the air masses mainly originated from the west coast of the Korean Peninsula, which had also little impact on the concentration of PM2.5 in Taitung as well, since it passed through the Yellow Sea and the East China Sea before entering Taitung City (see Fig. S2(e)). Therefore, the PM2.5 episode was mainly dominated by local sources in Taitung City instead of the air masses from outside region.


PMF Analysis

The PMF results of PM2.5 source apportionment during the NDs showed that the emission sources of PM2.5 included four main categories as shown in Fig. 6(a) and Table S1. In the Category I, Zn, EC, and OC had the highest loading, while Fe and Ca had moderately higher loading. Zinc is widely reported to be released by the wear of tires as mentioned, while EC and OC are primarily emitted by diesel and gasoline engine exhausts (Watson et al., 2001; Cheng et al., 2015; Corbin et al., 2018; Li et al., 2018). Metallic elements such as Fe, Ca, Mn, Cd, and Cu could be released from the combustion of heavy oil (Li et al., 2013b). Additionally, Fe and Ca could also be emitted by brake wear (Li et al., 2013b). Thus, such a source profile was mainly attributed to the vehicular exhausts, and the determined mass of PM2.5 was 3.2 µg m–3, accounting for approximately 28.3% of the total mass of 11.3 µg m–3 analyzed by PMF. In Category II, the mean concentration of chemical species was 3.1 µg m–3, accounting for 27.4% of the total mass. Ni, SO42–, Mg2+, Na+, Cl, and NO3 were the dominant chemical species in Category II. In addition to Ni, which is mainly released by local coal-burning boilers as mentioned, V is also a metallic indicator of coal burning. SO42– and Mg2+ are the essential chemical composition of coal-burning boilers as well (Sharp et al., 2013; Pei et al., 2016). Thus, the sources of these chemical species were attributed to coal-burning boilers. In Category III, the highest loading was identified as NO3, SO42–, Ca2+, and K+, and followed by NH4+, Mg, Ti, and Al, which were characterized as road dusts (Li et al., 2012; Liu et al., 2019). The mean concentration (2.6 µg m–3) of PM2.5 from the road dusts accounted for 23.0% of the total mass of PM2.5. The chemical species with a total mass concentration of 2.4 µg m–3 in Category IV accounted for 21.2% of the total mass of PM2.5. Mg2+, Ca2+, and OC had the highest loading among these species, while Na+, Cl, K+, and EC had higher loading. The composition highly matched the chemical feature of kitchen fumes (Li et al., 2015).


Fig. 6. Source profile and the loadings of chemical species during (a) the NDs and (b) the EDs by PFM analysis.Fig. 6. Source profile and the loadings of chemical species during (a) the NDs and (b) the EDs by PFM analysis.Fig. 6. Source profile and the loadings of chemical species during (a) the NDs and (b) the EDs by PFM analysis.

A total of five source profile categories representing main PM2.5 sources during the EDs is shown in Fig. 6(b). In the Category I, many chemical species including Cl, NO3, SO42–, Na+, K+, Mg2+, Ca2+, Ca, Fe, Mg, K, Na, OC, EC, Mn, Mg, and Al had high loadings, with a total mass of 20.1 µg m–3 accounting for 53.6% of the total average PM2.5 concentration of 37.5 µg m–3 during the EDs, and thus, the chemical feature of the Category I could be attributed to the firecracker burning. What’s more, the firecracker burning contributed approximately 76.7% of the increment of the average PM2.5 concentrations (26.2 µg m–3) between the NDs and the EDs. Therefore, firecracker burning played a vital role in the formation of PM2.5 during the EDs. The pattern of the category II exhibited high loadings of OC, K+, Mg2+, and K, which were attributed to the feature of cooking fumes. Although the contribution percentage (16.0%) of the total mass of PM2.5 was lower than that during the NDs, the concentration released from cooking fumes increased from 3.3 µg m–3 (NDs) to 6.0 µg m–3 (EDs) due to the dramatic increase of food stalls during the festival. In Category III, Zn, EC, and Ti had the highest loading among the chemical species, which was identified as the emission of vehicular exhausts, accounting for 14.6% of the total mass of PM2.5. The mean concentration of 5.5 µg m–3 contributed by vehicles exhaust was slightly higher than that (4.5 µg m–3) during the NDs due to the increase in the traffic flow of motor vehicles along with the pilgrimage activities. NH4+, Ca2+, and Mn showed the highest loading in Category IV, while Ni and V showed the highest loading in Category V, which originated from the fugitive road dust (8.1%) and local coal-fired boilers (6.1%), respectively. Correspondingly, their mean concentrations of 3.4 µg m–3 and 2.5 µg m–3 were lower than those during the NDs, indicating that the contributions of road dust and local coal-fired boilers to PM2.5 decreased significantly during the EDs. To sum up, a small portion (13.6%) of the increment was contributed by cooking fume and vehicles exhaust, while the contribution to the increment was nearly zero by local coal-fired boilers. 


Risk Assessment

The hazard quotients (HQs) of heavy metals in PM2.5 for children and adults at sites A, B, and C on February 22nd were compared and summarized in Table 1. The HItotal of heavy metals for children and adults at each site was lower than the accepted risk threshold of unity, indicating that the heavy metals in PM2.5 were unlikely to cause noncarcinogenic risk on the spectators in the vicinity of firecracker burning site during the EDs. The HItotal of heavy metals at site A for children (3.70 × 10–4) and adults (2.10 × 10–4) was higher than those of 3.20 × 10–4 and 1.83 × 10–4 at site B, respectively. While the HItotal for children (1.84 × 10–4) and adults (1.04 × 10–4) at site C were the lowest at the three sites since site C was far away from the firecracker burning site. These results indicate that the HItotal of heavy metals varied significantly with the downwind distance to the firecracker burning site. Additionally, the HItotal of heavy metals for children was generally higher than that for adults, indicating that children were more susceptible to the health hazards of firecracker burning. It was noted that the highest risk level for these heavy metals was attributed to Cr ranging from 0.93 × 10–4 to 3.60 × 10–4, which was mainly released by the occasional burning of small fireworks, namely, the HItotal of heavy metals was not directly associated with the firecracker burning. However, a large amount of SO2, NOx, CO, and other organic matter released from the firecracker burning are harmful to humans’ health, which needs further investigation (Greven et al., 2019).


Table 1. Health risks of heavy metals in PM2.5 sampled at three sampling sites.

The comparison of the carcinogenic risk values (Rt) of heavy metals at each site showed that, during the EDs, the Rt values at sites A and B were 1.21 × 10–6 and 1.06 × 10–6, respectively, which were higher than the carcinogenic risk threshold of 1.00 × 10–6. Therefore, a possible carcinogenic risk to the spectators on sites A and B can not be neglected due to the firecracker burning. The Rt value at site C was 0.58 × 10–6, which was lower than the threshold, indicating that the carcinogenic risk to the residents in the vicinity of site C was lower than those at sites A and B due to its long downwind distance to the firecracker burning site. Besides, the largest Ri was contributed by Cr (0.58 × 10–6–1.21 × 10–6) as well. It is noted that the cancer risk of PAHs in PM2.5 should not be neglected (Fan et al., 2017). Nevertheless, the cancer risk of heavy metals in PM2.5 estimated via the chronic daily intake is usually far higher than those of PAHs (Christopher et al., 2018). Thus, this study did not investigate the health risk of PAHs.


CONCLUSIONS


During the EDs, the downwind distance to the burning site presented a significant impact on the spatial distribution of the chemical composition of PM2.5. The increments of both the contents and concentrations of metallic elements, WSI, carbonaceous species at site A were significantly higher than those at sites B and C, which were located far away from the firecracker burning site. The main sources of PM2.5 during the EDs were the firecracker burning, kitchen fumes, and vehicles exhaust, while the increment contributed by local coal-fired boilers was far lower than those during the NDs. During the EDs, the metallic elements in PM2.5 had low noncarcinogenic risk to residents living in the vicinities of the sampling sites. However, the carcinogenic risk was highly influenced by the downwind distance to the burning site. The carcinogenic risk at the site close to the firecracker burning site was higher than the carcinogenic risk threshold, while that far away from the burning site was lower than the carcinogenic risk threshold.


ACKNOWLEDGEMENT


This study was performed under the auspices of the Ministry of Science and Technology, Taiwan, under the contract number of NSC105-EPA-F-009-002 and National Natural Science Foundation of China, under the contract number of 21477042. The authors are grateful to their financial supports in order to accomplish this study.



Share this article with your colleagues 

 

Subscribe to our Newsletter 

Aerosol and Air Quality Research has published over 2,000 peer-reviewed articles. Enter your email address to receive latest updates and research articles to your inbox every second week.

5.9
2020CiteScore
 
 
81st percentile
Powered by
Scopus

2020 Impact Factor: 3.063
5-Year Impact Factor: 2.857

Aerosol and Air Quality Research (AAQR) is an independently-run non-profit journal that promotes submissions of high-quality research and strives to be one of the leading aerosol and air quality open-access journals in the world. We use cookies on this website to personalize content to improve your user experience and analyze our traffic. By using this site you agree to its use of cookies.