The Impact of Photochemical Loss on the Source Apportionment of Ambient Volatile Organic Compounds (VOCs) and Their Ozone Formation Potential in the Fenwei Plain, Northern China

Abstract

The Fenwei Plain (FWP), one of China’s most polluted regions, has experienced severe ozone (O₃) pollution in recent years. Volatile organic compounds (VOCs), key O₃ precursors, undergo significant photochemical degradation, yet their loss and the implications for source apportionment and ozone formation potential (OFP) in this region remain unclear. This study conducted summertime VOC measurements in two industrial cities in the FWP, Hancheng (HC) and Xingping (XP), to quantify photochemical losses of VOCs and assessed their impact on source attribution and OFP with photochemical age-based parameterization methods. Significant VOC photochemical losses were observed, averaging 3.6 ppbv (7.1% of initial concentrations) in HC and 1.9 ppbv (5.6%) in XP, with alkenes experiencing the highest depletion (22–30%). Source apportionment based on both initial (corrected) and observed concentrations revealed that industrial sources (e.g., coking, coal washing, and rubber manufacturing) dominated ambient VOCs. Ignoring photochemical losses underestimated contributions from natural gas combustion and biogenic sources, while it overestimated the secondary source. OFP calculated with lost VOCs (OFP_loss) reached 34 ppbv in HC and 15 ppbv in XP, representing 20% and 25% of OFP based on observed concentrations, respectively, with reactive alkenes accounting for over 90% of OFPloss. The results highlight the importance of accounting for VOC photochemical losses for accurate source identification and developing effective O₃ control strategies in the FWP.

1. Introduction

Volatile organic compounds (VOCs), originated from both natural and anthropogenic emissions, are critical atmospheric pollutants with diverse chemical compositions, including non-methane hydrocarbons (NMHCs), oxygenated VOCs (OVOCs), and sulfur- and nitrogen-containing compounds [1]. Via complex reactions with atmospheric oxidants (e.g., OH, NO₃, O₃), VOCs contribute to the formation of secondary pollutants, including ozone (O₃) and secondary organic aerosol (SOA), exerting adverse effects on both air quality and human health. In addition, exposure to high concentrations of VOCs, O₃, and SOA can impair lung function and lead to cardiopulmonary diseases [2,3].

Home to over 55 million people, the Fenwei Plain (FWP) is one of China’s most polluted regions, and it has been included in the key governance areas by the Ministry of Ecology and Environment of China since 2018. The high levels of pollution in the FWP result from the combined effects of its unique topography, meteorological conditions, and intensive local emissions. Enclosed by mountains, the FWP features a basin-like terrain that leads to consistently low wind speeds across its urban valley areas, significantly limiting the dispersion of air pollutants [4]. Additionally, as a typical coal-resource-based region, coal accounts for approximately 90% of the energy consumption in the FWP, substantially higher than the national average of 60% [5]. The region is also heavily industrialized, with dominant sectors including coal coking and steel production [6].

While extensive studies have examined VOC characteristics in major economic regions in China, such as Beijing–Tianjin–Hebei (BTH), Yangtze River Delta (YRD), and Pearl River Delta (PRD) [7,8,9], research on VOC pollution in the FWP remains limited. Existing studies show much higher VOC levels in the FWP (50–60 ppbv) compared to cities like Beijing (29–44 ppbv), Shanghai (32–43 ppbv), and Guangzhou (33–35 ppbv) [10,11,12]. The primary sources of VOCs in the FWP are mainly local industries such as coke and rubber/plastic production, contributing significantly to OFP [5,13], notably differing from BTH and YRD, where VOCs are more closely linked to vehicular emissions and solvent usage [9,14].

Moreover, VOCs can be rapidly degraded by atmospheric oxidation, leading to significant difference between observed VOCs (VOCs_obs) and freshly emitted VOCs in the atmosphere [15]. As a result, without correcting photochemical losses, source apportionment and the OFP estimated based on VOCs_obs may be biased. Previous studies have used photochemical age-based parameterization methods to estimate initial VOC concentrations (VOCs_ini) by correcting for photochemical loss using species pairs with similar sources but different reactivities [16,17]. These studies suggest that ignoring photochemical losses can underestimate contributions from biogenic and solvent emissions and overestimate contributions from liquefied petroleum gas (LPG), vehicle emissions, and gasoline evaporation [18]. The OFP estimated from VOCs_ini is 2.2 times higher than that from VOCs_obs, mainly caused by photochemical consumption of alkenes [19]. However, studies in FWP mainly rely on VOCs_obs, which may be biased in source apportionment and OFP estimations, limiting the accuracy of current assessments and effectiveness of the mitigation strategies.

This study conducted summertime field observations of VOCs using online instruments in two FWP cities to characterize VOC pollution, source attribution, and its influence on ozone formation. By applying photochemical age-based parameterization methods, we estimated VOC photochemical loss and assessed its impact on VOC source apportionment and OFP.

2. Materials and Methods

2.1. Sampling Sites and Time Periods

The campaign was conducted at two sites: a suburban site in Longmen Town, Hancheng city (HC, 35°37′ N, 110°34′ E) from 12 August 2023 to 10 September 2023, and an urban site in Xingping city (XP, 34°17′ N, 108°27′ E) from 16 June 2023 to 23 August 2023. As shown in Figure 1, the HC site is located in a major industrial hub, approximately 23 km northeast of the city center. It is surrounded by industrial facilities, including coal washing plants, steel mills, and coking plants. Particularly, Longmen Town is home to the Longmen Metallurgical Industrial Park, one of the largest coke production bases in northern China, and one of the three major steel production bases in northwest China [20]. The XP site is also located in an industrial area, surrounded by rubber factories, fertilizer plants, and paper mills.

2.2. VOC Measurements and Meteorological Data

In HC, VOCs were continuously measured using the TH-300B online monitoring system at a time resolution of 1 h (Wuhan Tianhong, Wuhan, China). Briefly, the instrument integrates a carrier gas system, ultra-low-temperature pre-concentration electronic refrigeration, GC-FID/MS analysis, and automatic calibration system, featuring dual gas channels with a FID and MS. The FID detects C2-C5 hydrocarbons, while the MS detects C5-C12 hydrocarbons, halohydrocarbons, and OVOCs. The detection limit ranges from 0.007 to 1.035 ppbv. A total of 108 VOC species were measured, including 29 alkanes, 9 alkenes, acetylene, 33 halocarbons, 18 aromatics, 17 OVOCs, and carbon disulfide. Detailed information on the instrument is available in [21,22].

In XP, continuous online monitoring of VOCs was performed using an adsorption concentration (AC) online sampling system and GC-MS system (AC-GCMS 1000, Hexin, Guangzhou, China). The AC component enables online cyclic sampling (for 1 h) while performing ultra-low temperature dehydration and VOC enrichment. The GC-MS component achieves qualitative and quantitative analysis of the VOCs. A total of 102 VOC species were measured at 1-h resolution, including 28 alkanes, 11 alkenes, 34 halohydrocarbons, 17 aromatics, 11 OVOCs, and carbon disulfide. Details on the instrument are provided elsewhere [23,24].

Meteorological data, including wind speed, wind direction, atmospheric pressure, air temperature, and relative humidity, were measured at the Jinta Road station in HC and the Water Supply Company station in XP during the same time period.

2.3. Calculation of the Initial Concentrations of VOCs

The initial concentrations of VOCs (excluding isoprene and OVOCs) were calculated using the ratio method [25], as shown in Equations (1) and (2):

$${\left[{VOCs}_i\right]}_{obs}={\left[{VOCs}_i\right]}_{ini}\times exp\left(-k_i\left[OH\right]\Delta t\right)$$

(1)

$$\left[OH\right]\Delta t={\displaystyle \frac{1}{k_C-k_B}}\times (\ln ({{\displaystyle \frac{C}{B}})}_0-ln{\left({\displaystyle \frac{C}{B}}\right)}_t)$$

(2)

where i represents a specific VOCs, [VOCs_i]_obs and [VOCs_i]_ini are the observed and initial concentrations of VOCs_i; [OH] represents the OH radical concentration; Δt represents the photochemical age, and k_i is the reaction rate constant of VOCs_i with the OH radical. The calculation of [OH]Δt uses the evolution of the ratio between two primary hydrocarbon species (C and B in Equation (2)), where $({{\displaystyle \frac{C}{B}})}_0$ and ${\left({\displaystyle \frac{C}{B}}\right)}_t$ are the ratios of species C and B at initial time and the observed time, respectively; k_B and k_C are the reaction rate constants of species B and C with the OH radical, respectively. The basic assumption in Equation (2) is that B and C have similar origins but different rate constants with the OH radical. This study selected ethylbenzene and o-xylene as species C and B [26,27], and the OH radical reaction rate constant for o-xylene (13.6 × 10⁻¹² cm³ molecule⁻¹ s⁻¹) is 1.9 times that of ethylbenzene (7.0 × 10⁻¹² cm³ molecule⁻¹ s⁻¹). In this study, ethylbenzene and o-xylene concentrations demonstrated good correlation (r² = 0.95), indicating a common source (Figure S1). Based on the diurnal maximum of their ratio, $({{\displaystyle \frac{C}{B}})}_0$ was set to 1.23 for HC and 1.24 for XP. It should be noted that the selection of the initial ratio, $({{\displaystyle \frac{C}{B}})}_0$, based on diurnal maximum, assumes that VOCs do not undergo oxidation at night and the ratio remained unchanged during the transport. However, reactions with other atmospheric oxidants, e.g., O₃ and nitrate radical, as well as the influence of fresh emissions along the transport can introduce uncertainties. Nevertheless, this method still provides valuable insight into the photochemical loss of atmospheric VOCs and has been widely adopted in previous studies [28,29].

Isoprene primarily originates from biogenic emissions and peaks at midday when temperature and light intensity are strongest. The initial concentration of isoprene calculated using the ratio method would be significantly overestimated [30,31]. Therefore, the Methacrolein, one of the photochemical products of isoprene, was used to calculate the initial concentration of isoprene (R. 1–2 and Equations (3) and (4)) [32]:

Isoprene + OH → 0.63HCHO + 0.23 Methacrolein + 0.32 Methyl Vinyl Ketone

(R1)

Methacrolein + OH → products

(R2)

$${\displaystyle \frac{{\left[Methacrolein\right]}_t}{{\left[Isoprene\right]}_t}}={\displaystyle \frac{0.23k_1}{k_2-k_1}}(1-\mathit{exp}\left[\left(k_1-k_2\right)\left[OH\right]\Delta t\right])$$

(3)

$${\left[Isoprene\right]}_0={\left[Isoprene\right]}_t\times exp\left(k_1\left[OH\right]\Delta t\right)$$

(4)

where k₁ and k₂ are the rate constants for the reactions of isoprene and Methacrolein with OH, which are 1.00 × 10⁻¹⁰ cm³ molecule⁻¹ s⁻¹ and 3.3 × 10⁻¹¹ cm³ molecule⁻¹ s⁻¹, respectively [33]; [Methacrolein]_t and [Isoprene]_t are the observed concentrations at time t.

Furthermore, when calculating the initial mixing ratios of VOCs, we assumed that VOC losses mainly result from reactions with OH radicals, which are mainly generated from photolysis of O₃, HONO, HCHO, and H₂O₂ during daytime [34]. Therefore, initial concentrations of VOCs were calculated only for VOC observations in daytime (06:00–19:00 LT) [35]. Given that OVOCs have both primary emissions and secondary formation pathways and halohydrocarbons generally exhibit low reactivity and long atmospheric lifetimes, this study assumed the initial concentrations of these species were equivalent to their observed concentrations. Similar assumptions have also been adopted by Gao et al. [26]. Some studies attempted to estimate the photochemical loss of OVOCs using methods such as the isoprene loss reference approach or the traditional photochemical-age-based parameterization method (i.e., Equations (1) and (2)), but there still exists large uncertainties due to complex sources and formation pathways of OVOCs [36].

2.4. OFP Calculation

The OFP for each VOC species was calculated using maximum incremental reactivity (MIR) values as in Equations (5)–(7):

$${OFP}_{obs}={\sum }_i{\left[{VOCs}_i\right]}_{obs}\times {MIR}_i$$

(5)

$${OFP}_{ini}={\sum }_i{\left[{VOCs}_i\right]}_{ini}\times {MIR}_i$$

(6)

$${OFP}_{loss}={\sum }_i{(\left[{VOCs}_i\right]}_{ini}-{\left[{VOCs}_i\right]}_{obs})\times {MIR}_i$$

(7)

where [VOCs_i]_obs and [VOCs_i]_ini represent the observed and initial concentration of VOC species i, respectively. MIR_i is the MIR coefficient for species i in g O₃ g⁻¹ VOCs, with specific coefficients taken from Carter [37]. OFP_obs and OFP_ini represent the OFP based on the observed and initial concentrations of VOCs, respectively, and OFP_loss reflects the contribution of photochemical loss to OFP.

2.5. Source Apportionment by the Positive Matrix Factorization (PMF)

PMF 4.2 is employed in this study to analyze the sources of VOCs. The PMF model has been widely applied in the source apportionment of air pollution and detailed information about the model is provided by the PMF User Guide from the U.S. Environmental Protection Agency (www.epa.gov (accessed on 30 December 2024)). Briefly, the PMF model decomposes the sample matrix (X) into two matrices: the source contribution matrix (G) and the source profile matrix (F). The key sources of VOCs and their respective contributions are identified through the least squares method expressed as Equation (8):

$$X_{ij}={\sum }_{k=1}^p{G}_{ik}{F}_{kj}+E_{ij}$$

(8)

where X_ij represents the concentration of the jth component in the ith sample; G_ik is the contribution of the kth source in the ith sample; F_kj is the mass fraction of the jth component in the kth source; E_ij is the residual; and P is the number of sources. For a given value of p, G and F are obtained when the objective function Q approaches its minimum value, as illustrated by Equation (9):

$$Q={\sum }_{i=1}^n{\sum }_{j=1}^m[{\displaystyle \frac{E_{ij}}{U_{ij}}}{]}^2$$

(9)

where U_ij is the uncertainty of the jth component in the ith sample. The uncertainty calculation depends on a species’ concentrations relative to their detection limit (MDL) and is provided in Supplementary Material Table S1. Species that are below the MDL or have >50% invalid values were excluded.

In this study, factor numbers ranging from 3 to 9 were tested. The selection of the optimal number of factors was based on the variation rate of Q values with increasing factor numbers, the inter-profile correlations (all <0.5), and the physical interpretability of the resolved factors. Additionally, F_peak sensitivity analysis (step size of 0.2, ranging from −1 to 1) was conducted to further evaluate the solution stability. Finally, we chose 7 factors for HC and 5 factors for XP, based on the best fitness and interpretability.

3. Results and Discussion

3.1. Characteristics of VOCs During the Observation Period

The time series of meteorological parameters and VOCs during the campaign are provided in Figure S2 in the Supplementary Material. The mean temperature, relative humidity, and wind speed over the whole campaign were 25 ± 4.0 °C (mean ± standard deviation), 65 ± 17% and 1.6 ± 0.7 m s⁻¹ at HC and 28 ± 4.6 °C, 64 ± 20% and 0.9 ± 0.6 m s ⁻¹ at XP, respectively. VOC concentrations at HC ranged from 18 to 130 ppbv, with an average of 47 ± 20 ppbv. VOC concentrations at XP were slightly lower, ranging from 11 to 76 ppbv, with an average of 32 ± 9.9 ppbv. As shown in the VOC windrose diagrams (Figure S3), VOC levels at HC were generally higher under northeasterly winds, averaging 54 ppbv. Even with wind speeds >3 m s⁻¹, concentrations reached 59 ppbv, contrary to the expectation that higher wind speeds facilitate pollutant dilution and dispersion. This may be due to the emission of VOCs from coal washing and coking plants located 5.5 km northeast of the HC site. At XP, high VOC concentrations were predominantly associated with northwesterly winds at speeds less than 2 m s⁻¹, averaging 36 ppbv, indicating possible influence from rubber and plastic factories within 0.55 km. When wind speeds exceeded 2 m s⁻¹, average VOC concentrations at XP were only 24 ppbv.

Figure S2 also shows the concentrations of grouped VOCs. Alkanes were the most abundant category at both sites, accounting for 31% of total VOCs (TVOC) at HC and 42% at XP. AT HC site, alkanes were most abundant, followed by aromatics (20%) and alkenes (20%), OVOCs (15%), halohydrocarbons (8.8%), alkynes (5.0%, solely acetylene), and carbon disulfide (0.51%). In contrast, XP site was dominated by alkanes (42%), followed by halohydrocarbons (25%) and OVOCs (19%), with much lower proportions of alkenes (9.7%) and aromatics (4.3%) compared to HC.

Within each VOC category, there were significant differences in the dominant VOC species. For example, for alkanes, ethane was the dominant species at HC (64%), while at XP multiple alkanes including 2,2-dimethylbutane (29%) contributed to alkanes. Ethylene (59%) and propylene (36%) together accounted for 95% of the alkenes at HC, while at XP, the main species were ethylene (26%) and 1-butene (23%). Aromatics at HC was dominated by benzene (67%), whereas toluene (22%) was the main species at XP. The top three species at HC were ethane (9.3 ± 4.4 ppbv, 20%), benzene (6.5 ± 7.0 ppbv, 14%), and ethylene (5.6 ± 5.5 ppbv, 12%), all closely associated with combustion sources [38]. At XP, the top species were acetone (4.0 ± 1.5 ppbv, 13%), 2,2-dimethylbutane (3.9 ± 1.7 ppbv, 12%), and freon−12 (3.8 ± 4.7 ppbv, 12%). Acetone may originate from industrial emissions and secondary formation [39,40], freon-12 is closely related to industrial sources [12], and 2,2-dimethylbutane likely comes from vehicle exhaust [41] and industrial raw material usage [42].

Figure 2 presents VOC concentrations and compositions in other regions of China in summer. Among FWP cities, VOC levels at HC ranked second to Xianyang and were comparable to Xi’an, while VOC concentration at XP is about half of Xianyang and is similar with Weinan, the least polluted area in FWP. Compared with other regional sites in China, HC’s VOC level was relatively high, second to Hong Kong (52 ppbv), possibly because that site was located in a roadside environment with heavy traffic, while XP’s VOC level was in the medium range.

In terms of composition, alkanes are the main VOC component (30–62%) in most regions (including HC and XP), except for Xianyang and Xi’an stations, where OVOCs were higher than alkanes (Xianyang: 53% vs. 29%; Xi’an: 37% vs. 27%), possibly related to higher photochemical reactions [52]. Although TVOC at HC was similar to Xi’an, contributions from alkenes and aromatics were far higher than Xi’an’s, while the proportion of OVOCs was relatively lower. Additionally, the proportion of alkene and aromatics at HC were significantly higher than other regions (20% vs. 3.5–15% for alkene, and 21% vs. 4.3–38% for aromatics).

In contrast, VOCs at XP had a relatively high content of halohydrocarbons, at the high end of reported values in the literature (25% vs. 3.7–25%), similar to Wuhan (25%). Given the high proportions of alkenes and aromatics in HC, together with the dominance of species such as ethylene and benzene, VOCs at this site are likely associated with coal combustion and related industrial activities [41,53]. A high proportion of halohydrocarbons was also observed in Wuhan, where vehicle emissions are a major VOC source. However, halohydrocarbons can originate from both industrial processes and long-lived background sources [12,54].

3.2. Photochemical Loss of VOCs

Table 1. Comparison of VOCs with photochemical loss (VOCs_loss) in FWP and other cities of China.

City	Period	VOCsloss (ppbv)	VOCsloss/VOCsini (%)					Top3 Photochemical Loss Species	References
			VOCs	Alkanes	Alkenes	Alkyne	Aromatics
HC	summer	3.6	7.1	1.0	22	0.85	7.0	propene, isoprene, ethylene	This study
XP	summer	1.9	5.6	2.7	30	/	13	isoprene, cis-2-pentene, 1-butene	This study
Guangzhou	spring	4.1	15	8.2	37	1.5	25	m,p-xylene, isoprene, ethylene	[35]
Guangzhou	summer	5.1	16	6.6	37	1.5	20	/	[18]
	Autumn	4.5	18	8.6	42	2.2	28	/
Beijing	summer	1.7	8.5	2.1	26	0.90	13	isoprene, ethylene, propene	[30]
Shanghai	annual	12.2	35	2.5	39	10	31	C8 aromatics, C2-C3 alkenes	[55]
Shanghai	spring and summer	11.7	45	32	66	/	42	trans-2-butene, trans-2-pentene, ethylene	[56]
Taiwan	summer	3.7	13	3.5	45	1.2	13	isoprene, 1,2,4-trimethylbenzene, toluene	[51]

Note: ‘/’indicates not mentioned in the article.

presents the photochemical loss concentrations of different VOC categories (VOCs_loss) and their ratios to VOCs_ini. The total photochemical loss concentration of VOCs during the observation period at HC was 3.6 ppbv, accounting for 7.1% of VOCs_ini. Among these, alkenes experienced the greatest photochemical loss (22% of VOCs_ini), followed by aromatics (7.0%), alkynes (0.85%), and alkanes (1.0%), which agrees with the fact that alkenes and aromatics react with OH radicals much faster compared to alkanes. The photochemical loss of VOCs at XP was similar to HC, accounting for 5.6% of VOCs_ini, attributed by alkenes (30% of VOCs_ini), aromatics (13%), and alkanes (2.7%).

Although alkenes showed the greatest photochemical loss in both HC and XP, there were marked differences in specific species (as shown in Figure S4). For instance, propylene, ethylene, and isoprene were the top three VOC species contributing to photochemical loss at HC, accounting for 66% of total consumption, while at XP, cis-2-butene, 1-butene, and isoprene accounted for 50% of the total photochemical loss.

Table 1 also shows the extent of photochemical loss for different VOC categories in other Chinese cities, ranging from 8.5% (Beijing summer, [30]) to 45% (Shanghai spring and summer, [56]) of initial VOC concentrations. The photochemical loss of VOCs and various categories in Beijing summer (8.5%, 2.1%, 26%, and 13% for TVOC, alkanes, alkenes, and aromatics, respectively) was close to the levels observed at HC and XP. The relatively low photochemical loss in the FWP during summer may be attributed to its unique topography and meteorological conditions, such as more dusty weather that reduces net radiation and suppresses photochemical reaction [57]. Additionally, source-specific VOC compositions may lead to varying reactivity under similar environmental conditions. For example, annual VOCs in Shanghai primarily come from vehicle emissions, with C₂–C₃ alkenes as the main alkene components and C₈ aromatics as the main aromatic components, resulting in similar photochemical losses between alkenes and aromatics [55]. In other cities where isoprene and C₆–C₇ aromatic concentrations are higher, the higher reactivity of isoprene relative to C₂–C₃ alkenes and the lower reactivity of C₆–C₇ aromatics compared to C₈ aromatics lead to significantly higher photochemical losses than aromatics [35,51]. In addition to the reactivity of species, chemical loss is also affected by environmental factors such as solar radiation, ambient temperature (T), and the OH concentrations. This partly explains the relatively higher chemical loss observed in Shanghai during the spring–summer period compared to the annual average [55,56]. Liu et al. [36] summarized previous studies on the chemical degradation of VOCs and found that losses in spring and summer were substantially higher than those in autumn and winter.

3.3. VOC Source Apportionment

Source apportionment analysis was conducted using observed and initial VOC concentrations, respectively, and the results are shown in Figure 3.

3.3.1. Source Identification Based on Observed Concentrations at HC

Seven factors were identified in HC with observed VOC concentrations (Figure 3a):

Factor 1 contained large amounts of OVOCs, with significant diurnal variations (higher at noon, lower at night), consistent with secondary formation processes of OVOCs, which are greatly influenced by temperature and solar radiation [58]. Therefore, this factor was identified as secondary source.

Factor 2 comprises C₅–C₁₂ alkanes, 1,3-dichlorobenzene, and carbon disulfide, likely related to coal washing processes. C₅–C₁₂ are major chemical components in raw coal [59]. To effectively remove chlorine, sulfur, and moisture from coal, the washing process volatilizes significant amounts of chlorides and carbon disulfide [60,61]. Considering multiple coal washing plants are located within 2.5 km of the sampling site, we identified this factor as coal washing process.

Factor 3 contained high proportions of alkanes (i.e., butane, ethane, propane, and higher carbon alkanes), accompanied by small amounts of alkenes (e.g., ethylene and 1,3-butadiene) and aromatics (e.g., toluene, ethylbenzene, and xylene). Among these, ethane, propane, iso/n-butane, and pentane are typical species emitted from vehicle exhaust [12], higher carbon alkanes are considered to be emitted from diesel vehicles [14], while alkenes and benzene characterize incomplete combustion [58]. Therefore, this factor was named as vehicle exhaust.

Factor 4 was characterized by high concentrations of acetylene and ethane. Acetylene is a tracer for combustion sources [6,62], while ethane is the main component of natural gas (NG) besides methane [63]. Thus, this source was identified as NG combustion.

Factor 5 was characterized by halohydrocarbons (e.g., carbon tetrachloride, methylenechloride, etc.) and aromatics (e.g., naphthalene, trimethylbenzene, etc.), accompanied by propylene and cis-2-butene. Halohydrocarbons are closely related to industrial processes [12], aromatics are widely used as solvents [64], and cis-2-pentene and propylene are important raw materials in industrial production [65,66]. Therefore, this factor was identified as an industrial source.

Factor 6 contained high concentrations of ethylene and benzene, plus small amounts of ethane, toluene, and 1,1-dichloroethane. According to Wu et al. [30,67], coal combustion produces relatively simple aromatics such as benzene and toluene. 1,2-dichloroethane is also considered another typical species produced by coal combustion [11,14], and ethylene and ethane are typical combustion products [21]. Therefore, this source was identified as industrial boiler.

Factor 7 was characterized by high contributions of 1,3-butadiene and propane, with small amounts of propylene, ethane, and ethylene. Among these, 1,3-butadiene, propane, and propylene are considered from coke oven chimneys [68,69], while the combustion process produces ethane and ethylene [21]. Additionally, several coking plants are within 5.5–11 km of the site, and there are also steel plants equipped with coking furnace facilities near the site. Therefore, this source was identified as coking flue gas.

3.3.2. Source Identification Based on Observed Concentrations at XP

Based on Obs-PMF, five factors were identified in XP (Figure 3b):

Factor 1 contained high proportions of 1,2-dichloroethane, 1,1-dichloroethane, and other halogenated species. Additionally, there were certain amounts of aromatics (such as benzene and toluene) and OVOCs, consistent with the characteristics of widely used or released halohydrocarbons, aromatics, and OVOCs in industrial production processes [58,62,70]. Among these, 1,2-dichloroethane is considered a typical tracer species for industrial chemical processes. Therefore, this factor was identified as industrial source.

Factor 2 was characterized by low-carbon alkanes (ethane and propane) and small amounts of alkenes (e.g., ethylene and propylene), along with some higher-carbon alkanes (e.g., hexane). These alkanes and alkenes are characteristic components of gasoline and diesel vehicle emissions [12,14]. Therefore, this factor was ascribed to vehicle exhaust.

Factor 3 contained high concentrations of C₆–C₈ alkanes (namely 2,2-dimethylbutane, 2,4-dimethylpentane, 2,3-dimethylbutane, 2-methylpentane, and octane) and acetone. These VOC components were found to emit during rubber product manufacturing processes [62,66]. Furthermore, the Shaanxi Rubber Machinery Factory was located within 500 m of the monitoring station, which also explains the high contribution of this source factor. Therefore, this source was identified as rubber industry.

Factor 4 mainly consisted of isopentane, n-pentane, and cyclopentane, with small amounts of methyl tert-butyl ether (MTBE). Pentane is a characteristic species of gasoline evaporation [44,65,66], and MTBE is a commonly used additive to improve the anti-knock performance of gasoline [50,58]. Since there are few combustion markers in this factor, it was identified as gasoline evaporation.

Factor 5 mainly contained isobutane, n-butane, and 1-butene, with 1,2-dichloroethane and aromatics also present. Among these, isobutane and n-butane were found to come from liquefied petroleum gas (LPG) use and combustion processes [62], and 1-butene is considered an important species from coal combustion emissions [66,71], and 1,2-dichloroethane and aromatics are also considered typical coal combustion products [11,72]. This factor was therefore identified as a combustion source.

Figure 3. Source profiles from Obs-PMF (observed VOC concentrations) and Ini-PMF (initial VOC concentrations): (a) HC; (b) XP.

Figure 3. Source profiles from Obs-PMF (observed VOC concentrations) and Ini-PMF (initial VOC concentrations): (a) HC; (b) XP.

3.3.3. VOC Source Contribution Based on Observed Concentrations

As in Figure 4, industrial boiler contributed the most to VOCs at HC (29.5%), followed by vehicle exhaust (14.4%), NG combustion (14.0%), industrial source (13.2%), secondary source (12.3%), coking flue gas (10.3%), and coal washing process (6.27%). In XP, the rubber industry was the main VOC source (42.4%), followed by combustion source (19.9%), industrial source (16.8%), vehicle exhaust (14.8%), and gasoline evaporation (6.17%). This indicates the absolute dominance of total industrial emission in the FWP, while the proportion of vehicle exhaust was relatively small.

Similar VOC contributions were reported in other FWP cities [20], where coking process (27%) and solvent usage (20%) were major VOC sources in Weinan, and solvent usage and industrial sources contributing to 61% of VOCs in Xianyang, whereas vehicle exhaust contributing 24% [43]. This contrasts with results from the three mega-cities in China where vehicle exhaust is the main contributor to VOCs. In Beijing, diesel vehicle exhaust, gasoline vehicle exhaust, and industrial emissions were the main VOC sources, accounting for 20%, 15%, and 25%, respectively [73]. In Shanghai, VOCs mainly came from vehicle exhaust (33%) and combustion source (25%) [74]. In Guangzhou, vehicle exhaust (34%) and petrochemical emissions (26%) were the major contributors [10].

3.3.4. Source Apportionment Based on Initial Concentrations

Considering the impact of photochemical loss on VOC components from different sources, we also conducted PMF source apportionment based on initial VOC concentrations (Ini-PMF). The number of factors and source profile characteristics identified by Ini-PMF were generally similar to the Obs-PMF, except for factor 1 in HC (see Figure 3a). Factor 1 in Obs-PMF is attributed to the secondary source, while factor 1 in Ini-PMF was dominated by isoprene, accompanied by various OVOCs, and was therefore identified as biogenic source [19,35].

Figure 4 presents the source contributions of Ini-PMF. The initial VOCs in HC came from industrial boiler (26.9%), NG combustion (21.9%), vehicle exhaust (15.2%), industrial source (11.8%), coal washing process (10.5%), and biogenic source (2.74%). Compared to Obs-PMF, the contribution of NG combustion increased significantly (7.9%), mainly because it contains more highly reactive alkenes and aromatics. VOC contributions from the coal washing process, vehicle exhaust, and coking flue gas increased only slightly (0.26–4.5%), while industrial boiler and industrial source decreased slightly (1.4–2.6%). Notably, biogenic source contribution (2.7%) was only resolved in Ini-PMF, which underwent rapid photochemical reactions [75,76], and could not be identified in the Obs-PMF. It is also worth noting that in Ini-PMF, OVOCs representing secondary sources were allocated to industrial and biogenic sources. Bias could also be introduced by the simplified assumption that OVOC concentrations stay stable. Additionally, OVOCs are formed through complex atmospheric chemical reactions, often involving multiple precursors like isoprene and anthropogenic VOCs. These compounds can undergo various transformations, resulting in a wide range of oxidized products. As a result, their chemical composition often overlaps with that of primary sources, such as industrial or biogenic emissions. This makes it difficult for PMF models to clearly distinguish between primary and secondary sources, leading to potential misallocation of secondary OVOCs to industrial or biogenic categories in the analysis.

At XP site, the rubber industry accounted for 44% of TVOC, followed by combustion source (22%), industrial source (15%), vehicle exhaust (13%), and gasoline evaporation (5.5%), which were similar (variations within ±3%) to the contributions in Obs-PMF.

Previous studies have also indicated that biogenic emissions, solvent usage, and industrial sources tend to be underestimated, consistent with present results in FWP. Conversely, vehicle exhaust contributions are generally overestimated, agreeing with observations at XP but differing from HC, likely due to the redistribution of propane in source apportionment [30,35,77].

3.4. Contribution of Photochemical Loss to OFP

Based on observed VOC concentrations, the observed OFP (OFP_obs) at HC (170 ppbv) was significantly higher than that at XP (59 ppbv). In terms of composition, alkenes were the dominant contributors at both sites (HC: 57%; XP: 50%). However, the ranking of the remaining components differed: at HC, OVOCs (23%) were the second-largest contributors, followed by aromatics (14%), alkanes (4.8%), and alkynes (1.4%), while at XP, alkanes (26%) ranked second, followed by aromatics (13%) and OVOCs (11%). This difference was mainly attributed to the much higher concentrations of high-MIR-valued VOC species in HC compared to XP. At HC, the top three species, ethylene, propylene, and propionaldehyde contributed to 72% of the total OFP_obs, whereas in XP, the top three species, 1-butene (7.2 ppbv, 12%), ethylene (6.8 ppbv, 12%), and trans-2-pentene (4.5 ppbv, 8.0%) contributed to 32% of the OFP_obs.

This is consistent with regions where alkenes are the main OFP components [78,79,80]. Combined with Obs-PMF source apportionment results, OFP_obs in HC mainly came from secondary sources (60%), followed by the coal washing process (20%), industrial sources (9.8%), and coking flue gas (6.6%). For XP, industrial sources contributed the most, at 45%, followed by combustion sources (26%), the rubber industry (22%), and vehicle exhaust (5.9%). Compared to other regions, OFP sources based on observed concentrations showed that vehicle exhaust and industrial sources are major contributors in Wuhan [81]. In Hong Kong, LPG and solvent usage were identified as key OFP contributors [48]. Gasoline emissions and liquid gasoline evaporation contributed the most to OFP in Beijing [82].

The impact of photochemical loss of VOCs on O₃ formation (OFP_loss) for HC and XP were 34 ppbv and 15 ppbv, accounting for 20% and 25% of OFP_obs, respectively. Regarding specific VOC components, the photochemical loss of alkenes contributed the most to OFP_loss at both sites (HC: 90%; XP: 88%), followed by aromatics (HC: 9.8%; XP: 8.9%), alkanes (HC: 0.44%; XP: 3.0%), and alkynes (HC: 0.060%). From the perspective of individual species (Figure 5), the top 10 reactive VOC species at both sites accounted for over 90% of OFP_loss, essentially reflecting the potential impact of local VOC photochemical loss on O₃ pollution. At HC, the top ten contributors to OFP_loss included six alkenes (propylene, isoprene, ethylene, cis-2-butene, 1,3-butadiene, and cis-2-pentene) and four aromatics (1,2,4-trimethylbenzene, m/p-xylene, 1,3,5-trimethylbenzene, and toluene), accounting for 96% of total OFP_loss. At XP, the top 10 species (isoprene, cis-2-pentene, 1-butene, 1,3-butadiene, 1-hexene, ethylene, 1,2,4-trimethylbenzene, cis-2-butene, 1,3,5-trimethylbenzene, and m/p-xylene) contributed 92% to total OFP_loss. These results are similar to those shown in Tianjin, where alkenes contributed over 90% of OFP_loss [83], while alkene contributions in Beijing, Shanghai, and Chengdu ranged from 40% to 82% [26,31,84].

Compared with OFP_obs (Figure 6), OFP_ini significantly underestimated the secondary source contribution (by 60%), while industrial source and coal washing process contributions increased significantly (by 39% and 17%), primarily due to the substantial contribution from OVOCs. Coking flue gas, vehicle exhaust, NG combustion, and industrial boiler showed little change (<6%), possibly due to their low MIR value components such as alkanes (e.g., ethane) and aromatics (e.g., benzene). At XP, the changes in In-OFP contributions were relatively minor (0.34–4.7%), with the rubber industry and combustion source increasing by 3.2% and 3.8%, respectively. Industrial, vehicle exhaust, and gasoline evaporation decreased by 4.7%, 2.0%, and 0.34%, respectively. Similar studies have also shown that calculations based on observed concentrations overestimate the contributions of vehicle exhaust, diesel and gasoline evaporation, and solvent usage to OFP, while underestimating the contributions of biogenic source, industrial source, etc., to OFP [35,51,83].

The VOCs_loss is significantly higher in the afternoon, typically between 12:00 and 15:00 local time when the solar radiation is strongest and temperatures are highest. OFP_loss also shows a clear diurnal trend. Taking HC as an example, OFP_loss reaches its maximum at 15:00, while it peaks at 14:00 at XP. These periods coincide with the most active reactions of O₃ precursors and O₃ formation.

The O₃ pollution control policy should prioritize time-specific emission reductions to key alkene-dominated sources during photochemically active period from noon to afternoon. Photochemical loss correction needs to be integrated into source apportionment analysis for a better source allocation and strength quantification.

4. Conclusions

This study investigated photochemical losses of VOCs and their impacts on source apportionment and ozone formation potential (OFP) in two industrial cities in Fenwei Plain. During the observation period in 2023, we observed average VOC concentrations of 47 ± 20 ppbv in Hancheng (HC) and 32 ± 9.9 ppbv in Xingping (XP) but with distinct compositional profiles. The dominant components in HC were alkanes (31%), aromatics (20%), and alkenes (20%), whereas in XP, the major components were alkanes (42%), halohydrocarbons (25%), and OVOCs (19%).

Traditional source apportionment (Obs-PMF) analysis identified industrial activities—such as boilers (30% in HC) and rubber manufacturing (42% in XP)—as major VOC contributors, contrasting with other Chinese regions where vehicular emissions typically play a major role. This highlights the unique industrial footprint of the FWP. Accounting for photochemical losses of VOCs led to significant discrepancies in original concentrations and sources. Initial VOC concentrations were 51 ppbv at HC and 34 ppbv at XP, respectively, with alkenes experiencing the largest photochemical loss. After correcting photochemical losses, NG combustion contribution increased by 7.9%, coal washing process, vehicle exhaust, and coking flue gas increased by 0.25–4.5%, and a previously obscured biogenic source (2.7%) emerged in HC. In XP, less pronounced changes were observed, with combustion and rubber industry contributions changing marginally (±3%).

The study also estimated the contributions of VOCs to OFP in FWP. OFP_loss concentrations were 34 ppbv at HC and 15 ppbv at XP, with alkenes contributing the most to OFP_loss (HC: 90%; XP: 88%). Revised OFP calculations (OFP_ini) found that secondary sources were overestimated by the original calculation (OFP_obs); industrial sources and coal washing in HC were underestimated by 39% and 17%, respectively. In XP, the contribution to OFP from rubber industry and combustion sources increased by 3–4%. These findings demonstrate the vital importance of considering VOC photochemical losses for accurate source identification and for designing O₃ mitigation policies in industrial regions of FWP. Note that photochemical loss correction is not applied to OVOCs, which is assumed to be stable in this study. This simplification introduces uncertainties in OVOC source analysis and OFP estimation. Future work should focus on the impact of OVOC photochemical degradation on source attribution and ozone formation.