Characteristics and Source Apportionment of Volatile Organic Compounds in an Industrial Area at the Zhejiang–Shanghai Boundary, China

Abstract

As “fuel” for atmospheric photochemical reactions, volatile organic compounds (VOCs) play a key role in the secondary generation of ozone (O₃) and fine particulate matter (PM_2.5, an aerodynamic diameter ≤ 2.5 μm). To determine the characteristics of VOCs in a high-level ozone period, comprehensive monitoring of O₃ and its precursors (VOCs and NOx) was continuously conducted in an industrial area in Shanghai from 18 August to 30 September 2021. During the observation period, the average concentration of VOCs was 47.33 ppb, and alkanes (19.64 ppb) accounted for the highest proportion of TVOCs, followed by oxygenated volatile organic compounds (OVOCs) (13.61 ppb), alkenes (6.92 ppb), aromatics (4.65 ppb), halogenated hydrocarbons (1.60 ppb), and alkynes (0.91 ppb). Alkenes were the predominant components that contributed to the ozone formation potential (OFP), while aromatics such as xylene, toluene, and ethylbenzene contributed the most to the secondary organic aerosol production potential (SOAFP). During the study period, O₃, NOx, and VOCs showed significant diurnal variations. Industrial processes were the main source of VOCs, and the second largest source of VOCs was vehicle exhaust. While the largest contribution to OFP was from vehicle exhaust, the second largest contribution was from liquid petroleum gas (LPG). High potential source contribution function (PSCF) values were observed in western and southeastern areas near the sampling sites. The results of a health risk evaluation showed that the Hazard Index was less than 1 and there was no non-carcinogenic risk, but 1,3-butadiene, benzene, chloroform, 1,2-dibromoethane, and carbon tetrachloride pose a potential carcinogenic risk to the population.

1. Introduction

With the continuous increase in energy consumption resulting from the acceleration of urbanization and industrialization, air pollution, represented by ozone (O₃) and fine particulate matter (PM_2.5, an aerodynamic diameter ≤ 2.5 μm), has become an increasingly serious atmospheric problem [1,2,3]. Volatile organic compounds (VOCs), as “fuel” for atmospheric photochemical reactions, play a key role in the secondary generation of O₃ and PM_2.5 [4]. The chemical composition of VOCs in the ambient atmosphere is complex and mainly includes non-methane hydrocarbons (NMHCs), oxygenated volatile organic compounds (OVOCs), halogenated hydrocarbons, nitrogen-containing organic compounds, and sulfur-containing organic compounds. Carbonyl compounds, including aldehydes and ketones, have an important influence on the reactive radical cycle and secondary pollutant generation [5].

In recent years, high-level urban and regional O₃ pollution episodes have been frequent in China. As the primary precursor of ozone pollution, many studies on VOCs have been conducted to provide ideas for urban and regional O₃ prevention and control. Yu et al. [6] found that the proportion of vehicle emissions and liquid petroleum gas/natural gas (LPG/NG) contribution increased year by year, while the proportion of industrial and solvent sources showed a decreasing trend. Li et al. [7] found that VOCs in suburban areas are mainly from traffic in urban areas and that urban sites are very active in photochemical processes. The same study in Beijing showed a significant decrease in average VOC concentrations in 2018 compared to 2015 and 2016, with regional transport mainly from the south–southeast (SSE) and south–southwest (SSW) and a tendency toward a transitional state of O₃ sensitivity in the summer. Strong solar radiation, a high temperature, and low humidity exacerbate O₃ pollution, which is further exacerbated by regional transport from polluted areas in the south [8]. Tan and Wang [9] reviewed VOC characteristics during a Shanghai citywide closure period, showed that reductions in O₃ precursors and other pollutants did not prevent ground-level O₃ pollution during the 2022 closure period; high O₃ concentrations were primarily due to large reductions in nitrogen oxide (NOx) emissions, while reductions in volatile organic compound (VOC) concentrations could not overcome the titration effect of nitric oxide (NO). A study of air pollution in the Dongguan industrial area during the COVID-19 lockdown showed that the average concentrations of NOx and VOCs decreased compared to the pre-lockdown period. In contrast, the average ozone concentration increased slightly; the analysis concluded that the increase in ozone was the combined result of a large increase at night and a small decrease during the day [10]. VOCs may adversely affect public health (e.g., respiratory and cardiovascular systems) at certain concentrations. Previous studies found significant positive correlations between personal exposures and VOC concentrations and greater risk-of-hazard quotients and cancer in areas of high VOC exposure [11,12]. As a result, toxic VOCs like benzene are classified as hazardous air pollutants by the United States Environment Protection Agency (USEPA).

In Shanghai, rapid urbanization and industrialization make VOC pollution more serious than ever before. Numerous studies have been conducted to analyze the characteristics of VOCs in Shanghai in recent years [13,14,15,16]. However, these studies have focused more on the overall characteristics of VOCs in Shanghai, and relatively few studies have been conducted in each district. Jinshan District, where many of Shanghai’s industrial parks are located, has attracted less attention. Therefore, it is necessary to measure and understand the characteristics of VOCs in Jinshan District.

In this study, 113 VOCs, O₃, and NOx at provincial boundary sites in Jinshan were monitored from August to September 2021 to carry out atmospheric VOC characteristics, diurnal variations, ozone and secondary organic aerosol formation potential determinations, and a source analysis. The results will be helpful to better understand ozone and secondary organic aerosol (SOA) pollution and the formation mechanism in this region, which can provide technical support for the development of regional air pollution control countermeasures.

2. Materials and Methods

2.1. Observation Overview

In this study, a GC3000-MS produced by Spectrum Technology was used to measure high-carbon VOC species from C₆ to C₁₀, and a GC3000-315L was used to determine low-carbon VOC species from C₂ to C₅. VOCs were monitored from 18 August to 30 September 2021 at provincial boundary station (121.27° N, 30.71° E) which is located in a large integrated industrial area in Jinshan Distinct, in the southwest of Shanghai (as shown Figure 1). There are multiple paint companies within a 3 km radius of the station which use large amounts of solvents in their production processes. There is a large chemical zone to the northeast of the site, with several Sinopec gas stations and petrochemical companies in the vicinity. In total, 113 VOCs, including 29 alkanes, 11 alkenes, 17 aromatic hydrocarbons, 35 halogenated hydrocarbons, 20 oxygenated organic compounds (OVOCs), and acetylene, were monitored online using a gas chromatography–mass selective detector/flame ionization detector (GC-MSD/FID, Puyu Technology Development Co., Ltd., Hangzhou, China). The time resolution was 1 h. The sampling system collects ambient air in two channels. The samples of the two channels were dehydrated and enriched by a low-temperature pretreatment system, and the gas samples were quickly transported to a capillary column for subsequent separation by direct thermal desorption at a high temperature. The separated volatile organic compounds were detected by MS and FlD detectors, and qualitative and quantitative analysis results were obtained through data processing. Details can be found in a previous study [17]. Synchronized observations of pollutants, O₃, NO₂ and meteorological parameters (temperature, humidity, wind speed, and direction) were carried out at the site. Statistical analyses were conducted using the Origin software. During the observation period, the average temperature was 27.2 °C, varying between 20.3 °C and 33 °C, the relative humidity ranged between 50% and 100% with an average of 94.3%, and the mean wind speed was 2.9 m/s during the observation period.

2.2. Ozone Formation Potential (OFP)

Various species of VOCs in the atmosphere have different concentration levels and reactive activities. The contribution to ozone formation varies greatly. Currently, the ozone formation potential (OFP) is commonly used to assess the potential contribution of VOCs to ozone formation. The calculation formula is as follows:

$${\mathrm{OFP}}_{\mathrm{i}}={\left[\mathrm{VOCs}\right]}_{\mathrm{i}}\times {\mathrm{MIR}}_{\mathrm{i}},$$

(1)

where OFP_i is the ozone formation potential of the VOC species i (μg/m³); [VOCs]_i is the average concentration of the VOCs species i (μg/m³); MIR is the maximum incremental reactivity of the VOC species i. MIR is referenced from Carter’s study [18].

2.3. Secondary Organic Aerosol Formation Potential (SOAFP)

In this study, the contribution of various VOCs to the generation of secondary organic aerosols was assessed using the secondary organic aerosol formation potential (SOAFP) based on the Grosjean smoke box experiment with the following formula [19,20]:

$${\mathrm{SOAFP}}_{\mathrm{i}}={\left[\mathrm{VOCs}\right]}_{\mathrm{i}}\times {\mathrm{FAC}}_{\mathrm{i}},$$

(2)

where SOAFP_i is the secondary organic aerosol formation potential of the VOC species i (μg/m³); [VOCs]_i is the average concentration of the VOCs species i (μg/m³); and FAC_i is the aerosol generation coefficient for a species i. The values are taken with reference to Grosjean et al. [19,20].

2.4. Potential Source Contribution Function (PSCF)

The PSCF is an analysis method based on trajectory and pollutant concentration to identify potential pollution source areas [21,22,23]. Its analysis is run by TrajStat, which is a GIS-based software that uses different trajectory statistical analysis techniques (e.g., HYSPLIT) to identify potential sources of atmospheric pollutants [21,24]. It has been extensively used in previous studies [25]. The PSCF value can determine the proportion of pollution tracks in the grid. The PSCF is defined as

$${\mathrm{PSCF}}_{\mathrm{ij}}=\frac{{\mathrm{m}}_{\mathrm{ij}}}{{\mathrm{n}}_{\mathrm{ij}}},$$

(3)

where i is the latitude, j is the longitude, n_ij is the number of endpoints falling in the ij cell, and m_ij is the number of “polluted” trajectory falling in the cells. In this study, the average mixing ratio of VOCs is used as the pollution threshold. In order to reduce the influence of small n_ij values, it is necessary to multiply any weight function W (n_ij) by the PSCF value to better reflect the uncertainty in cells [26,27]. The weight function is defined as follows:

$${\mathrm{WPSCF}}_{\mathrm{ij}}=\frac{{\mathrm{m}}_{\mathrm{ij}}}{{\mathrm{n}}_{\mathrm{ij}}}\times {\mathrm{W}}_{\left({\mathrm{n}}_{\mathrm{ij}}\right)},$$

(4)

$${\mathrm{W}}_{\left({\mathrm{n}}_{\mathrm{ij}}\right)}=\left\{\begin{array}{cc}1.00,\hfill & \hfill 80<{\mathrm{n}}_{\mathrm{ij}}\\ 0.72,\hfill & \hfill 20<{\mathrm{n}}_{\mathrm{ij}}\le 80\\ 0.42,\hfill & \hfill 10<{\mathrm{n}}_{\mathrm{ij}}\le 20\\ 0.05,\hfill & \hfill {\mathrm{n}}_{\mathrm{ij}}\le 10\end{array}\right.,$$

(5)

2.5. Positive Matrix Factorization (PMF) Model

The source apportionment of VOCs was performed using the PMF model [23]. PMF is a receptor model for the source identification of atmospheric pollutants. PMF has been widely used for source analyses of VOCs, and its principle and applications have been detailed in many papers [28,29]. Here, for VOC species with concentrations equal to or less than the method detection limit (MDL), the uncertainty is set to 5/6 of the MDL (Equation (6)). And for species with concentrations higher than the MDL, the uncertainty is defined as in Equation (7).

$$\mathrm{Unc}=\frac{5}{6}\times \mathrm{MDL}({\mathrm{x}}_{\mathrm{i}}\le \mathrm{MDL}),$$

(6)

$$\mathrm{Unc}=\sqrt{{\left(\mathrm{Error}\mathrm{Fraction}\times {\mathrm{x}}_{\mathrm{i}}\right)}^2+{\left(0.5\times \mathrm{MDL}\right)}^2}\hspace{1em}({\mathrm{x}}_{\mathrm{i}}>\mathrm{MDL}),$$

(7)

where Unc is the uncertainty of a VOC species i, and x_i is the concentration of species i. In this study, the MDL is uniformly taken as 0.05. The method of calculating the Error Fraction is as follows: first, find out the standard deviation and mean value of x_i, and finally, express the Error Fraction as a relative standard deviation of x_i. In this study, a total of 46 representative VOCs with complete data were used for the source analysis.

2.6. Health Risk Assessment

A health risk evaluation quantifies the risk of pollutants causing harm to the health of the population, using the risk level as the evaluation index. Based on the carcinogenicity of the pollutants, which can be categorized into carcinogenic risk and non-carcinogenic risk, the carcinogenicity of VOCs is evaluated using the HI (Hazard Index) and Risk (Lifetime Cancer Risk) [30]. The calculation formulae are as follows:

$$\mathrm{EC}=\left(\mathrm{CA}\times \mathrm{ET}\times \mathrm{EF}\times \mathrm{ED}\right)/\mathrm{AT},$$

(8)

$$\mathrm{HQ}=\mathrm{EC}/\left(\mathrm{RfC}\times 1000\right),$$

(9)

$$\mathrm{HI}=\sum {\mathrm{HQ}}_{\mathrm{i}},$$

(10)

$$\mathrm{Risk}=\mathrm{EC}\times \mathrm{IUR},$$

(11)

where EC is the exposure concentration (μg/m³); CA is the concentration of the VOC species i (μg/m³); ET is the exposure time, 24 h/d; EF is the exposure frequency, 365 d/a; ED is the lifetime exposure time, 80 a; AT is the averaging time, 80 × 365 × 24 h; IUR is the risk per unit of inhalation (μg/m³); and RfC is the reference concentration (mg/m³); The values of IUR and RfC can be found in EPA’s Integrated Risk Information System (IRIS, http://www.epa.gov/iris, accessed on 10 December 2023). It is generally recognized that a non-carcinogenic risk exists when HI > 1 and a carcinogenic risk exists when Risk > 10⁻⁶ [31].

3. Results and Discussion

3.1. VOC Pollution Levels

3.1.1. General Characteristics Data Overview

Time series of meteorological parameters and pollutants during the observation period are shown in Figure 2. Based on the National Ambient Air Quality Standard (GB 3095-2012), a day with hourly ozone concentration exceeding the Grade II limit value of 200 μg/m³ is defined as an ozone-polluted day. According to the guideline, 18 August, 22 August, 26 August, 28 August, 1 September, 2 September, 6 September, 21 September, and 22 September were classified as polluted days. Notably, the hourly ozone concentration at 15:00 on September 1 reached the highest value of 291 μg/m³ during the period of pollution, and the concentrations at 14:00 and 17:00 on that day were the second and third highest. During the observation period, the mean VOC concentration was 47.33 ppb, which was higher than other studies in Nanjing (34.4 ppb) [32], Chongqing (24.0 ppb) [33], and Zhengzhou (28.8 ppb) [34]. The results of this study are comparable with results in Tokyo (43.3 ppb) [35]. The reasons for the differences in concentrations may be related to meteorological conditions, boundary layer height, and local control policies in different cities. In this study, VOC species are in the following order: alkanes > OVOCs > alkenes > aromatic hydrocarbons > halogenated hydrocarbons > acetylene. Their average concentrations were 19.64 ppb, 13.61 ppb, 6.92 ppb, 4.65 ppb, 1.60 ppb, and 0.91 ppb, respectively (the order of VOC species on polluted and non-polluted days was the same). This result was similar to previous study conducted in the Second Jinshan Industrial Area in Shanghai [36] In Shenzhen, the order is alkanes > OVOCs > halogenated hydrocarbons > aromatics > alkenes > acetylene > acetonitrile [33]. Among them, the concentration of alkanes is much higher than the concentrations of the other VOCs. It is noteworthy that on individual days, there is also a steep increase in the concentration of alkenes.

The diurnal variations in meteorological parameters for polluted and non-polluted days are illustrated in Figure 3. The temperature was always higher on polluted days than on non-polluted days; high temperatures favor photochemical reactions, and more ozone is produced. Similarly, the wind speed was always lower on polluted days than on non-polluted days, and a lower wind speed means more accumulation of pollutants. Except for the period from 19:00 to 05:00, the humidity was lower on polluted days than on non-polluted days, and low humidity weakens the wet removal of ozone, which also affects the concentration of ozone [37]. In summary, the meteorological conditions on polluted days were more favorable for the accumulation of pollutants and for the production and accumulation of ozone.

3.1.2. Analysis of OFP and SOAFP for VOCs

The concentration, OFP, and SOAFP of the top 15 VOCs during the observation period are shown in Figure 4. The sum of the concentration of the top 15 VOCs was 35.16 ppb, accounting for 74.2% of the total VOCs. Isopentane, propane, propionaldehyde, n-butane, and ethylene are the top five in order of concentration, which indicates that these substances contribute a high proportion of the local VOC concentration. In Figure 4b, the total OFP of the 15 VOCs was 291 μg/m³. Among these VOCs, OVOCs such as propionaldehyde, acetone, and acetaldehyde are not only associated with primary emissions but also with secondary photochemical generation. Alkenes such as ethylene and propylene dominate the ozone generation potential due to their own higher concentrations and high MIR values. This indicates that it is necessary to pay attention not only to VOCs with high concentration values but also to those VOCs with low concentration values but a strong influence on air pollution. Ethylene is widely used in industrial development [38], and its concentration was high because the observation site is located in an industrial area. Isoprene is an indicator of biogenic emissions and is emitted from many plants [39,40]. The C2-C5 hydrocarbon species are the main components of fossil fuels [41] due to the high volume of traffic and the presence of a large number of factories with strong industrial activity in the vicinity.

The largest contribution to the SOAFP was m/p-xylene (19.72 μg/m³, 26.6%), followed by toluene (18.77 μg/m³, 25.3%) and ethylbenzene (10.90 μg/m³, 14.7%). Among the 15 VOCs that contribute the most to the SOAFP, 11 of them are aromatic hydrocarbons, accounting for the largest proportion (over 90%), which is consistent with the research results of Li et al. [42]. Most of the aromatic hydrocarbons, such as toluene, ethylbenzene, and xylene, come from the use of coating solvents, so strengthening control over the coating solvent industry is essential for reducing the local SOA concentration.

3.2. Diurnal Variations

The diurnal variation characteristics of different air pollutants are shown in Figure 5. The ozone concentration was lower and decreased faster on polluted days than on non-polluted days before 08:00. Meteorological conditions on polluted days were more conducive to the accumulation of pollutants than on non-polluted days, resulting in a stronger titration effect. Ozone concentration increased at a faster rate after 08:00, reaching a maximum concentration (205.46 μg/m³) at 15:00. Higher temperatures, lower wind speeds, and lower humidity on pollution days made ozone generation and accumulation easier. After 15:00, ozone concentrations started to decrease; this occurred more rapidly on pollution days than on non-pollution days due to the fact that more VOCs accumulated on polluted days and depleted ozone more rapidly.

The diurnal concentration of nitrogen dioxide was characterized by a “double peak and a valley”, maintaining a high concentration during the night, decreasing around 02:00 and reaching a minimum around 04:00 due to the titration effect in the early morning. Due to boundary layer uplift and the depletion of photochemical reactions [43], the concentration reached its first peak at 07:00 and then started to decrease rapidly. On polluted days, there were occasional increases in NO₂ concentrations, suggesting an additional source of NO₂ production during this period. The concentration dropped to a minimum at 15:00, when the ozone concentration also reached a maximum. Thereafter, the concentration of nitrogen dioxide rose again due to the beginning of the evening rush hour, increased traffic flow, and reduced light radiation, resulting in weaker photochemical reactions. Finally, the concentrations equalized at night.

The diurnal variation in ethylene concentration also roughly showed the characteristics of “two peaks and one valley”. The concentration started to rise around 05:00, reached a peak, and then was consumed by the reaction; the concentration started to fall and then rose again in the evening peak. Compared with non-polluted days, the second peak was reached later on polluted days, and the concentration tended to then stabilize.

For isoprene, the polluted days displayed a similar pattern to the non-polluted days, both of which were single-peaked. The concentration was the lowest at night and started to increase after early morning, peaked around 15:00, and then started to decrease. The concentrations on polluted days were slightly higher than on non-polluted days because the high temperature makes plants emit more isoprene [44], and the low wind speed also facilitates the accumulation of isoprene.

Toluene showed the same diurnal pattern on polluted and non-polluted days, with the highest maximum concentration on polluted days peaking earlier than on non-polluted days. The concentration of toluene gradually increased at night and peaked around 08:00, after which the concentration started to decrease at a faster rate. After reaching a minimum around 15:00, the concentration remained at a relatively stable level.

3.3. Source Analysis

3.3.1. Identification of PMF Factors

Variation in the mixing ratio of VOCs during the sampling period may be due to changes in the contribution of emission sources. Ratios of specific VOCs are commonly used to identify emission sources. In this study, the reaction rates of n-butane and isobutane with OH radicals are similar, while the difference in their ratios is less than 10%, which indicates that they are from different sources. Butane is usually from natural gas, LPG, vehicle emissions, and biomass combustion, and the isobutane/n-butane ratios of different sources are different (e.g., 0.2–0.3 for vehicles, 0.46 for LPG, and 0.6–1.0 for NG) [45,46]. In this study, n-butane and isobutane showed a highly significant correlation (R² = 0.882 > 0.6), indicating that the sources of these two species are similar. The ratio of n-butane/isobutane was 0.54, indicating that the source of n-butane and isobutane emissions was LPG.

N-pentane and isopentane have similar atmospheric lifetimes; thus, the ratio of n-pentane to isopentane is widely used to determine the effects of vehicle emissions and combustion emissions, with values varying with sources (e.g., 0.56–0.80 for coal combustion, 1.5–3.0 for liquid gasoline, and 2.2–3.8 for vehicle exhaust) [36,47]. In this study, n-pentane and isopentane did not show a more significant correlation (R² = 0.218 < 0.6), indicating that the two substances were from different sources.

The emission sources of m/p-xylene and ethylbenzene are similar, but the former is three times more reactive to OH radicals than the latter [48,49]. Therefore, a lower X/E ratio usually indicates more pronounced aging of the gas mass, that is to say, a greater influence of external transport. In this study, the correlations between m,p-xylene and ethylbenzene on non-polluted days (R² = 0.8046) and on polluted days (R² = 0.7886) were significant. In particular, the X/E ratio was 2.44 on a polluted day compared to 3.41 on a non-polluted day. This indicates that the external transport influence was greater on the polluted day compared to the non-polluted day.

Compared to specific VOC proportion analyses, a PMF analysis allows for the quantitative determination of VOC source contributions [50]. In this study, six major factors were identified, including solvent usage, surface coatings, industrial processes, vehicle exhaust, combustion emissions, and liquefied petroleum gas (LPG) usage (as shown in Figure 6).

Factor 1 was mainly characterized by a few short-chain alkanes, and n-hexane, methylcyclopentane, 2-methylpentane-3-methylpentane, and 2,3-dimethylbutane, which are C6-C8 alkanes. Aromatic hydrocarbons such as toluene, xylene, and ethylbenzene made high contributions. The proportions of halogenated hydrocarbons such as carbon tetrachloride and dichloromethane were high. All of these substances are important organic solvents and typical tracers of solvent usage [51,52]. Therefore, factor 1 was highly related to solvent usage.

Factor 2 contained C6-C8 alkanes and a high fraction of aromatic hydrocarbons. Paints usually contain these components [51,53]. Among them, 1,2,3-trimethylbenzene (22.9%), 1,2,4-trimethylbenzene (21.6%) and 1,3,5-trimethylbenzene (21.6%) contributed more. Trimethylbenzene is widely used in coatings [50]. Therefore, factor 2 was identified as surface coatings.

Factor 3 was rich in aromatic and halogenated hydrocarbons. There were large contributions to benzene (83.8%) and toluene (23.6%), which are mainly used in industrial solvent production [54]. Chlorotrifluoromethane, methylene chloride, and 1,2-dichloropropane were shown in previous studies to be associated with industrial processes [26,36]. Ethylene, which is the blood of industrial development [38,55], is also present. Factor 3 also contained long-chain alkanes, of which n-undecane is widely used in industry [32,39]. Thus, we considered this factor an industrial process.

Factor 4 consisted of a high percentage of short-chain alkanes, alkenes, and a small amount of aromatic hydrocarbons. Among them, ethane (69.1%), propane (74.8%), n-butane (37.8%), isobutane (52.6%), n-pentane (76.8%), isopentane (93.6%), ethane (60.8%), and acetylene (73.5%) were particularly prominent. All of these substances are important tracers of automobile exhaust gases [22,51,54,56]. Therefore, factor 4 was defined as vehicle exhaust.

Factor 5 comprised ethane, propane, ethylene, acetylene, and benzene. Acetylene is a typical tracer of combustion emissions [50]. The characteristic combustion products of coal are propane, ethylene, acetylene, and benzene [54]. Ethane is also associated with the use of natural gas [57]. Therefore, factor 5 was apportioned as combustion emissions.

Alkanes (propane, n-butane, and isobutane) and alkenes (ethylene, propylene, and butene) were identified as factor 6, and the use of LPG may release large amounts of short linear alkanes [3]. Combining the previous ratio of n-butane and isobutane, this factor was thus proposed to be LPG.

Based on the results of the source apportionment, the proportion of each VOC source and its OFP were calculated during the observation period (as shown in Figure 7). The largest source of VOCs was industrial processes (36.62%), which is due to the fact that the monitoring site is located near an industrial park and is therefore most influenced by industrial activities. The second source was vehicle exhaust (31.56%), which is related to the high traffic volume in the vicinity. The contribution of LPG usage, which is used more frequently by the locals due to its popularity, was third (15.45%). In contrast, the contributions from surface coatings (3.91%), combustion emissions (8.11%), and solvent usage (4.35%) were relatively small. While in Zhengzhou [58], vehicle exhaust provided a greater contribution (28%), and industrial processes had a lower proportion (28%). Compared to Beijing, the contribution of vehicle exhaust (39.95%) is significant, with natural gas/LPG (22.04%) making the second largest contribution and industrial processes (20.64%) making a relatively low contribution [8]. In terms of OFP, vehicle exhaust made the largest contribution (38.2%), followed by LPG (24%), while industrial processes rank third with a large concentration contribution (17.3%). On the contrary, combustion emissions (10.2%), surface coatings (7.1%), and solvent usage (3.2%) contributed less. Compared to the results of the Zhengzhou study, the contributions of vehicle emissions and industrial processes were significant, but the contribution of solvent usage in this study was not as significant as in Zhengzhou. Due to the high source and OFP contributions from vehicle emissions, reducing vehicle exhaust would be a very effective measure.

3.3.2. The PSCF Results

In addition to the direct impact of local pollution sources, regional transport can also have an impact on VOC pollution. Areas with high PSCF levels indicate high-potential regional transport sources [51]. The results of the PSCF analysis are shown in Figure 8. The highest PSCF values for O₃ in August were mainly distributed in the west and south of the observation site; the highest PSCF values for NO₂ were mainly distributed in the west and southeast, and the TVOCs with the highest PSCF values were mainly distributed in the west. In September, the highest PSCF values for O₃ were mainly distributed in the north and southeast, the highest PSCF values for NO₂ were mainly distributed in the northwest and southeast, and the highest PSCF values for TVOCs were mainly distributed in the southeast. As a result, the potential source areas of pollutants in August were primarily in the west of the site, and the potential source areas in September were the southeast of the site. Again, the role of local pollution sources cannot be ignored.

3.4. Health Risk Assessment

Table 1. Carcinogenic risk assessment of 11 VOCs.

VOC Species	IUR	Risk
1,3-Butadiene	3.00 × 10−5	1.26 × 10−5
Benzene	7.80 × 10−6	2.44 × 10−5
Vinylchloride	8.80 × 10−6	9.73 × 10−7
1,4-Dioxane	5.00 × 10−6	4.32 × 10−7
Trichloroethene	4.10 × 10−6	2.91 × 10−7
Dichloromethane	1.00 × 10−8	2.73 × 10−9
Chloroform	2.30 × 10−5	5.88 × 10−6
Tetrachloroethylene	2.60 × 10−7	3.27 × 10−8
1,2-Dibromoethane	6.00 × 10−4	5.07 × 10−6
Carbon tetrachloride	6.00 × 10−6	4.41 × 10−6
Bromoform	1.10 × 10−6	1.40 × 10−8

and

Table 2. Non-carcinogenic risk assessment of 26 VOCs.

VOC Species	Rfc	HQ
m&p-Xylenes	1.00 × 10−1	4.20 × 10−2
o-Xylene	1.00 × 10−1	2.15 × 10−2
Toluene	5.00 × 100	6.95 × 10−4
Ethylbenzene	1.00 × 100	2.02 × 10−3
1,3-Butadiene	2.00 × 10−3	2.09 × 10−1
Benzene	3.00 × 10−2	1.04 × 10−1
MTBE	3.00 × 100	7.57 × 10−4
MMA	7.00 × 10−1	1.27 × 10−4
Tetrahydrofuran	2.00 × 100	1.46 × 10−4
n-Hexane	7.00 × 10−1	1.35 × 10−3
Styrene	1.00 × 100	5.38 × 10−4
Cyclohexane	6.00 × 100	8.66 × 10−5
Naphthalene	3.00 × 10−3	6.16 × 10−2
2-Hexanone	3.00 × 10−2	6.53 × 10−3
Vinylchloride	1.00 × 10−1	1.11 × 10−3
1,4-Dioxane	3.00 × 10−2	2.88 × 10−3
1,2-Dichloropropane	4.00 × 10−3	7.30 × 10−2
Trichloroethylene	2.00 × 10−3	3.55 × 10−2
1,1-Dichloroethylene	2.00 × 10−1	3.33 × 10−5
Dichloromethane	6.00 × 10−1	4.56 × 10−4
Methyl bromide	5.00 × 10−3	4.48 × 10−2
Perchloroethylene	4.00 × 10−2	3.14 × 10−3
1,4-Dicl-benzene	8.00 × 10−1	6.24 × 10−6
1,2-Dibromoethane	9.00 × 10−3	9.39 × 10−4
Carbon tetrachloride	1.00 × 10−1	7.35 × 10−3
Methyl isobutyl ketone	3.00 × 100	5.00 × 10−5
HI		6.20 × 10−1

list the calculated non-carcinogenic and carcinogenic risk values for the VOC species available in the EPA Integrated Risk Information System (IRIS) library. According to the risk assessment, an HI value less than 1 indicates that there was no non-carcinogenic risk. In this study, the carcinogenic risk values of 1,3-Butadiene, Benzene, Chloroform, 1,2-Dibromoethane, and Carbon tetrachloride were higher than 10⁻⁶_, indicating that these substances have a potential carcinogenic risk. Among them, benzene had the highest lifetime cancer risk value of 7.80 × 10⁻⁶, which was higher than the risk of 2.30 × 10⁻⁶ obtained in a study conducted in Beijing [12]. It was also higher than the carcinogenic risk of 2.18 × 10⁻⁵ in Tianjin [59]. This result may be attributed to the fact that there are a large number of paint solvent companies around this site and the petrochemical industry is booming, both of which are important benzene sources. Therefore, benzene needs to be given sufficient attention to reduce the risk of cancer.

4. Conclusions

In this study, 113 VOCs and ozone were measured in an industrial area in Jinshan District, Shanghai, for 43 days. Days on which the hourly ozone concentration exceeded 200 μg/m³ were classified as polluted days. The highest hourly ozone concentration reached 291 μg/m³ on a polluted day. The average concentration of VOCs was 47.33 ppb during the observation period. The ranking of VOC concentrations was alkanes > OVOCs > alkenes > aromatics > halogenated hydrocarbons > acetylene. Alkenes were the predominant components that contributed to the OFP, while the greatest contributor to the SOAFP were aromatics. The diurnal variation in VOCs showed two peaks and one valley, indicating that typical VOCs had an accumulation phenomenon in the early morning. Six sources were identified by PMF, including combustion emissions, industrial processes, LPG, solvent usage, vehicle exhaust, and surface coatings. Among them, the big contributors are industrial processes (36.62%), vehicle exhaust (31.56%), and LPG (15.54%). And among the contributions of the OFP, vehicle exhaust is the largest (38.2%), followed by LPG (24%), and industrial process is in third place (17.3%). Therefore, vehicle emissions make an important contribution to the formation of local ozone and should be emphasized for control. The results of a PSCF analysis showed that the potential sources of pollutants were mainly in the west in August and in the southeast in September. In addition to focusing on regional linkages of the surrounding potential source cities, local sources should also be emphasized. The observed hazard indices for VOCs were less than 1, with no non-carcinogenic risk, but 1,3-butadiene, benzene, chloroform, 1,2-dibromoethane, and carbon tetrachloride were higher than 10⁻⁶, with a potential carcinogenic risk to the population. It is necessary to take adequate measures to control their emissions.