This work aims to elucidate the composition, sources, and health risks associated with ambient VOCs in a representative densely populated city (Nanjing, in the Yangtze River Delta region, China). Importantly, we quantify the source-specific health risks of VOCs by linking the source profiles resolved from positive matrix factorization (PMF) with their estimated carcinogenic and noncarcinogenic risks. Our findings are valuable for effective VOCs control and the reduction of their health hazards in the future.
2. Experimental Methods 2.1. Sampling Site, Instrumentation, and Chemical AnalysisThe sampling site was located in the Jiangning District of urban Nanjing, China (118.818161° E, 31.917282° N) (Figure 1a), and VOC samples were acquired in the open space on the rooftop of a 10-story building (~35 m above the ground). The site was close to the arterial roads in the west and north (~250 m away, with high traffic flows), and was surrounded by residential buildings, office buildings, industrial plants, and schools. This site thus represents a typical urban environment with VOCs released from multiple anthropogenic sources such as traffic, industrial, and residential activities. To better understand the impact of possible industrial emissions, we marked the locations of nearby factories in Figure 1b. These factories include electronics, machinery manufacturing and maintenance, plastic processing, printing, and painting, and the majority of them were located in the east/northeast of the site.The sampling period was from 11 October to 12 November 2020. The VOCs were sampled twice a day at ~9:00 a.m. and ~3:00 p.m. respectively, each lasting for one hour, by using SUMMA canisters (6.0 L, Entech Instruments Inc., Simi Vally, CA, USA), and a total of 60 valid samples were obtained. Before sampling, the canisters were cleaned by a canister cleaner (3100D, Entech Instruments Inc., Simi Vally, CA, USA) at least 3 times and then were pumped to vacuum (36]. In total, 108 compounds were measured, including 30 alkanes, 34 halogens, 17 aromatics, 13 alkenes, 12 oxygenated volatile organic compounds (OVOCs), 1 alkyne, and 1 other compound.The quality control and quality assurance of GC-MS mainly include the multipoint calibration (using PMAS and TO-15 mixed standard gases, Sigma-Aldrich, St. Louis, MO, USA), single-point calibration, and mass spectrometric tuning (using 4−bromofluorobenzene (BFB) 1 μL (50 ng)). The procedure of the multipoint calibration was as follows: the standard gas was diluted to a 10 nmol/mol working standard by using a dynamic dilution apparatus (4600D, Entech Instruments Inc., Simi Vally, CA, USA); a univariate linear regression was used to create the calibration curve by inputting results from six standards (20 mL, 50 mL, 100 mL, 200 mL, 400 mL, and 600 mL), and the correlation coefficients (r2) of the regressions for all compounds were assured to be >0.99, with a response factor relative standard deviation (RF RSD) of 6.1~26.8%. The detection limits of 108 VOCs ranged from 0.002 ppb to 0.050 ppb.
Concentrations of other pollutants including O3 (Model 49i, Thermo Fisher, Waltham, MA, USA), NOx (Model 42i, Thermo Fisher, Waltham, MA, USA ), CO (Model 48i, Thermo Fisher, Waltham, MA, USA), SO2 (Model 43i, Thermo Fisher, Waltham, MA, USA), and PM2.5 (XHQ500C, XianHe Co. Ltd., Shijiazhuang, China), and the meteorological data (XHPM200E, XianHe Co. Ltd., Shijiazhuang, China) including temperature, relative humidity, wind speed, wind direction, and PM2.5 were all obtained from the nearby national environmental monitoring station (~50 m away).
2.2. Data Analysis 2.2.1. Source ApportionmentPMF is a powerful and widely used tool to resolve distinct sources of pollutants based on measured data without prior information regarding the source profiles [37,38,39]. This study used the US EPA (Environmental Protection Agency) PMF 5.0 software kit [40,41,42,43] for source apportionment of VOCs. The PMF algorithm decomposes the measured data matrix into the product of two matrices (factor profiles and time series of the factors’ concentrations) plus a residual matrix. One advantage of PMF is that it carefully weighs the measurement uncertainties, making its solution robust, meaningful, and reliable. In this work, the uncertainty of a measured VOC was calculated by Equation (1):Unc=Error Fraction×C2+0.5×MDL2
(1)
where Error Fraction is chosen as 0.1 here, C is the concentration of a VOC (in ppb), and MDL is the method detection limit of that VOC. Measured data below the MDL were replaced by 1/2 MDL and corresponding uncertainties were set to 5/6 MDL. Some missing data were substituted by the geometric mean of its neighborhood measured values with an uncertainty of 4 times the uncertainty of the measured values. The data were then classified into three categories according to their signal-to-noise (uncertainty) (S/N) ratios. The data points with S/N ratio 2 were regarded as “strong” and were directly used in PMF analysis.In addition, we also took into account the species which are recognized as specific source markers. After the pretreatment, 35 VOCs were chosen, including 11 alkanes, 4 alkenes, 1 alkyne, 5 halogens, 8 aromatics, and 6 OVOCs. The PMF solutions were considerably evaluated by exploring different numbers of factors, rotational ambiguity, and bootstrapping (100 runs) for an estimation of the uncertainty of the solution, following the standard protocol in the EPA PMF 5.0 operation manual. Finally, we selected the 5-factor solution as the best result (see details in Section 3.3.2). 2.2.2. Calculation of Ozone Formation Potential (OFP) and Secondary Organic Aerosol Formation Potential (SOAFP)VOC reactivity is critical to the formation of ozone and SOA. A number of studies focused on the mechanisms of ozone formation [44,45] and proposed different approaches to quantify the OFPs of VOCs [46,47,48]. A widely used and simplified method is described by Equation (2) [49]: where OFPi (in μg/m3) refers to the OFP of VOCi; VOCi (in μg/m3) refers to the measured concentration of VOCi, and MIRi is the value of maximum increment reactivity of VOCi. In this study, the VOCs’ MIR values were adopted from those documented in Carter [50], and are available for most VOCs (93 out of 108).Similar to OFP, SOAFP quantifies the ability of a VOC to generate SOA [51,52,53], which is shown in the following equation (details in Hui et al. [54]): where SOAFPi (μg/m3) is the SOAFP of VOCi, and SOAPi refers to the coefficient of VOCi to form SOA. The SOAPi values used here were from Derwent et al. [55]. The values are only available for 31 VOCs. Although the SOA formation depends on various environmental factors, the use of SOAPi allows reasonable estimations of contributions of individual precursors to SOA formation and demonstrates the relative importance of these precursors [56]. 2.2.3. Health Risk AssessmentWe performed a quantitative health risk assessment on the VOCs by using the method recommended by the US EPA and the Exposure Factor Handbook of Chinese Population (US EPA, 1989; Ministry of Ecology and Environment of China, 2013), which have been widely used previously [57,58,59]. There are two health risk indicators: Lifetime carcinogenic risk (LCR) and noncarcinogenic risk (NCR). The calculation equations for a certain VOC are listed below:Here, EC is the exposure concentration (in μg/m3); CA is the ambient (measured) concentration (in μg/m3); ET is the exposure time (hours/day); EF is the exposure frequency (days/year); ED is the years of exposure (years), and AT is the average time (hours). In this study, EF is 365 d/year from the US EPA Integrated Risk Information System (IRIS), ET is 3.7 h/day, ED is 74.8 years, and AT is therefore 74.8 × 365 × 24 h, all from the Exposure Factor Handbook of Chinese Population (Adult) based on Ministry of Ecology and Environment of China. IUR is the inhalation unit risk (in m3/μg), RfC is the reference concentration (in mg/m3). The IUR and RfC values of different species were obtained from the risk assessment information system (RAIS) developed by the University of Tennessee. The values are compiled in Table S1 in the Supplementary Materials. We were able to calculate LCR for 16 VOCs and NCR for 39 VOCs. 4. ConclusionsIn this study, ambient concentrations of 108 VOCs were determined in urban Nanjing during the autumn of 2020. The mean TVOC concentration was 29.04 ± 14.89 ppb, which was relatively low compared with measurement results in other cities. The average concentration was much higher in the morning than in the afternoon (35.84 ppb vs. 22.24 ppb). Alkanes (36.9%), OVOCs (19.9%), and halogens (19.1%) were the three major VOCs types. The VOCs/NOx (ppbC/ppbv) ratio was on average 3.31 in the morning and 4.11 in the afternoon, demonstrating that ozone control is VOC-limited throughout the day. In contrast, aromatics became the most important VOC group in OFP (41.9%), as well as in SOAFP (94.3%), strongly suggesting that a preference for aromatics control can benefit both ozone and PM2.5 reductions.
Diagnostic ratios of E/X, I/N, and T/B all point to the large influence of traffic on VOCs in this site. Further PMF analysis did separate five sources with traffic as the largest contributor (29.2%); solvent use (22.2%), biogenic source (20.8%), oil/gas evaporation (14.1%), and industry (13.7%) were the other four sources. Traffic and solvent use, however, were less important, and biogenic source instead became the largest contributor (33.3%) in the afternoon. Moreover, we calculated the LCR and NCR of measured VOCs and found that the overall health risks were low, except for a few compounds including acrolein, benzene, 1,2-dichloroethane, and 1,2-dibromoethane. The PMF-MLR analysis successfully apportioned the TVOC NCR to individual sources. It is interesting to find that a biogenic source rather than traffic became the most important source to the TVOCs’ NCR and its contribution to the afternoon samples dominated over the sum of all other sources. In summary, our findings reveal the importance of controlling aromatics as well as traffic/industrial emissions to the coordinated reduction of PM2.5 and O3. In addition, we want to highlight that biogenic emissions should be paid attention to in the future when considering the direct health risks of VOCs.
留言 (0)