Intermediate-Volatility Organic Compounds in Two Chinese Megacities: Sources and Implications for Ozone Formation

Abstract

]]­Intermediate-volatility organic compounds (IVOCs) are important precursors of ozone and secondary organic aerosol (SOA), yet their urban characteristics and impacts remain insufficiently understood. In this study, IVOCs were investigated during the summer of 2024 in Shenzhen and Beijing, representing southern and northern Chinese megacities, respectively. Beijing exhibited significantly higher average IVOCs concentrations (40.42 ± 30.11 μg/m³) than Shenzhen (6.22 ± 4.50 μg/m³), with residual UCM dominating in both cities. Source apportionment showed that diesel-related emissions were the primary IVOCs source in Shenzhen (44.4%), followed by biomass burning (29.8%), ship emissions (16.7%), and gasoline vehicles (9.1%). In contrast, Beijing’s IVOCs were mainly from the petrochemical industry (35.5%), gasoline vehicles (31.5%), and diesel vehicles (25.0%), with minor contributions from biomass/coal combustion (8.1%). PAHs, particularly fluorene, acenaphthylene, and acenaphthene, contributed substantially to IVOCs-related ozone formation potential (OFP). While VOCs accounted for the majority of OFP, IVOCs contributed 32.9% in Shenzhen and 22.1% in Beijing, underscoring their non-negligible role in urban ozone formation. These findings highlight the diverse sources and substantial photochemical reactivity of IVOCs in different urban settings. The results underscore the need to incorporate IVOCs, particularly PAHs, into air quality management and ozone mitigation strategies in similar rapidly urbanizing regions.

1. Introduction

Atmospheric volatile organic compounds (VOCs) play a crucial role in air quality and climate, influencing the formation of secondary pollutants such as ozone and secondary organic aerosol (SOA). , Among these compounds, intermediate-volatility organic compounds (IVOCs), defined by effective saturation concentrations (C*) ranging from 10³ to 10⁶ μg/m³ and roughly corresponding to the volatility range of C₁₂–C₂₂ straight-chain aliphatic hydrocarbons, have attracted increasing attention in recent years. − IVOCs exhibit high SOA formation potential and undergo rapid atmospheric oxidation, reacting efficiently with major oxidants including hydroxyl radicals, ozone, nitrate radicals, and chlorine radicals. , These reactions produce oxygenated products that readily partition into the particle phase, a process further enhanced by heterogeneous and aqueous-phase chemistry. Field studies indicate that gas-phase oxidation of VOCs and IVOCs contributes 70–90% of SOA, significantly influencing aerosol optical properties and regional radiative balance. ,

In recent years, stringent air pollution control measures have led to substantial improvements in air quality across China, particularly in reducing concentrations of fine particulate matter (PM_2.5). However, ozone pollution remains a persistent challenge. In 2024, the 90th percentile of the daily maximum 8 h average ozone across 339 cities ranged from 77 μg/m³ to 195 μg/m³, with a national average of 144 μg/m³, comparable to 2023 levels and representing a 5.8% increase since 2021. Ozone primarily forms through photochemical reactions between nitrogen oxides (NO _x ) and VOCs under sunlight, with IVOCs acting as important precursors. For example, evaporation tests using diesel fuel in a wind tunnel show that ozone formation from IVOC emissions can be comparable to that from VOCs after 3 days of atmospheric aging, and that reducing IVOC emissions results in significant reductions in ozone formation. IVOCs are ubiquitous in urban atmospheres, with observed concentrations typically ranging from 6.3 to 62.5 μg/m³. ,, Field observation studies indicate that IVOCs originate from a wide range of sources, encompassing fossil-fuel combustion, biomass burning, industrial activities, cooking emissions, and biogenic sources. ,− Their chemical complexity and diverse origins make it challenging to quantify their contributions to ozone and SOA formation. Understanding their role is therefore crucial for improving ozone pollution control in Chinese cities.

Shenzhen and Beijing are two megacities in southern China and northern China, respectively, experiencing different levels and types of air pollution. Shenzhen is a coastal city characterized by a subtropical monsoon climate, with prolonged and humid summers, and short, mild winters. The city has seen substantial air quality improvements in recent years, with annual average PM_2.5 concentrations declining to 17 μg/m³ in 2024, although ozone pollution remains a concern (137 μg/m³). Beijing is located at the northwestern margin of the North China Plain, approximately 150 km from the Bohai Sea to the southeast. It lies within the warm temperate semihumid zone and is characterized by a monsoon-influenced humid continental climate, featuring hot and humid summers and cold, dry winters. Historically affected by severe PM_2.5 pollution, the city’s annual average PM_2.5 decreased to 30.5 μg/m³ in 2024 under stringent control measures, yet summer ozone episodes continue to pose challenges. The high population density and intense economic activities in these megacities create complex and dynamic atmospheric environments, making them ideal locations for studying the sources and impacts of IVOCs.

In this study, we aim to characterize the sources and contributions of IVOCs to ozone formation in two Chinese megacities, Shenzhen and Beijing. Through field sampling and laboratory analyses, we quantified ambient concentrations of IVOCs and compared their pollution characteristics in these two megacities. In addition, we identified the dominant sources of IVOCs and explored their potential impacts on ozone formation. This study provides new insights into the source characteristics and photochemical impacts of IVOCs under summertime urban conditions and contributes to a better understanding of their potential role in urban ozone pollution.

2. Methods

2.1. Field Sampling

The sampling site in Shenzhen was located on the rooftop platform of a fourth-story building on the campus of University Town of Shenzhen (longitude: 113.97°, latitude: 22.59°), surrounded by school and residential areas with no major industrial sources nearby. Sampling was conducted from July 4 to July 22 and from July 29 to August 2, 2024, covering 24 days (see Table S1). Sampling was interrupted for 6 days due to instrument malfunction and typhoon events. In total, 45 samples were collected during the campaign.

The sampling site in Beijing was located on the rooftop platform of a second-floor building on the campus of Tsinghua University (longitude: 116.33°, latitude: 40.00°). The surrounding area is primarily school and residential, with no significant industrial emissions. Sampling was conducted from July 12 to July 18, 2024, covering 7 days (see Table S1), during which 13 samples were collected.

Sampling was performed using a dual-channel sampler (flow rate range: 0.1–1 L/min, accuracy: ±5%) equipped with Tenax-TA adsorption tubes (Markes International, UK; adsorbent: 200 mg, length: 89 mm, outer diameter: 6.4 mm) and “PAH” adsorption tubes (Markes International, UK; adsorbent: 200 mg, length: 89 mm, outer diameter: 6.4 mm), operated in parallel. The “PAH” tubes were used to collect polycyclic aromatic hydrocarbons (PAHs) from ambient air, while the Tenax-TA tubes captured other gaseous IVOCs and VOCs, including aromatic hydrocarbons, halogenated hydrocarbons, and C8–C22 n-alkanes. PTFE filters (Markes International, UK; diameter: 5.1 mm) were installed before the sorption tube to remove particles from the air flow.

2.2. Sample Analysis

Samples were analyzed using a gas chromatography (GC)/mass spectrometry (MS) system (8890 GC/5977B MS, Agilent Technologies, USA) coupled with a thermal desorption system (TD100-xr, Markes International Ltd., UK). The desorption of the tubes occurred at 300 °C for 15 min under an inert helium flow of 20 mL/min to release the adsorbed analytes. The desorbed compounds were cryogenically focused on a cold trap at −20 °C and then rapidly heated to 300 °C, transferring the concentrated analytes onto the GC column for separation. The GC was equipped with an HP-5 capillary column (30 m × 0.25 mm × 0.25 μm, Agilent Technologies). The temperature program for GC was as follows: an initial temperature of 50 °C, held for 2 min; ramped at 20 °C/min to 150 °C, held for 2 min; then ramped at 10 °C/min to 310 °C, held for 25 min. The MS ion source was operated in electron impact (EI) mode at 70 eV. Full scan (SCAN) mode was run with a mass range of m/z 35 to 550.

A known amount of 1,4-dichlorobenzene-D4 was spiked into each tube prior to the thermal desorption to track the recovery of IVOCs during analysis. The identification of speciated IVOCs was based on their retention times and the NIST mass spectral library. The quantification of detectable species was performed using authentic standards (Details in Text S1 and Table S2). A total of 15 VOCs and 15 IVOCs were quantified in Shenzhen (Details in Table S3), and 14 VOCs and 14 IVOCs were quantified in Beijing (Details in Table S4).

Due to the chemical complexity of IVOCs, many components cannot be fully resolved using conventional GC. In addition to the speciated IVOCs that can be individually identified, the remaining unresolved compounds were collectively classified as an unresolved complex mixture IVOCs (UCM IVOCs). The total IVOC mass was calculated as the sum of the speciated IVOCs and the UCM fraction. A semiquantitative approach was applied following Zhao et al. (2014), in which compounds within similar volatility ranges in a mixed standard solution were used as surrogates to estimate the unresolved fraction. This method enables quantification of total IVOCs despite incomplete chromatographic resolution (see Text S2 for details).

Furthermore, the UCM fraction was partitioned into unspeciated branched alkanes (b-alkanes) and residual UCM based on characteristic ion signals (e.g., m/z 57) and retention time distributions. This classification allows partial chemical differentiation within the UCM by separating b-alkane-like compounds from the residual UCM (see Text S3 for details).

2.3. Quality Assurance and Quality Control

Before sampling, it was ensured that the actual sampling volume would not exceed the breakthrough volume of the thermal desorption tube. The desorption tube was conditioned at 300 °C for 2 h prior to use under a stream of purified nitrogen to remove potential contaminants. The cleaned tubes were analyzed to check if any IVOCs still existed and then stored in a dry, sealed bag and kept at 4 °C before sampling. The sampling flow rate was verified with a flow meter to ensure consistent and reliable sampling conditions before and after each sampling. After each sampling, the desorption tube was promptly sealed at both ends, wrapped in aluminum foil, and stored at −20 °C until analysis. Field blanks were obtained to expose the tube to ambient air with no active air flow for the entire sampling period. In total, three field blanks were collected for each campaign.

Prior to each GC–MS analysis sequence, a blank test was performed by injecting a blank sample using the activated thermal desorption tube to check for system contamination. In addition, the MS was calibrated using 4-bromofluorobenzene (25 mg/L) to ensure the correct ion abundance and mass spectral stability. Calibration curves for target compounds were established using mixed standards and internal standards, with R ² exceeding 0.99. No background contamination was detected throughout the entire process, from injection to pretreatment and analysis. Dynamic blank subtraction was applied to minimize background interference and improve semiquantitative accuracy.

2.4. Positive Matrix Factorization Analysis

Positive matrix factorization (PMF) is a receptor modeling technique that decomposes the measured concentration matrix into factor contributions and profiles, allowing for the identification and quantification of potential emission sources. In this study, the EPA PMF 5.0 was employed to apportion the sources of atmospheric IVOCs. Prior to modeling, species were selected based on their relatively high concentrations and strong source-indicating characteristics. Concentration data below the method detection limit (MDL) were replaced with half of the MDL, while missing values were substituted with the median concentration of the corresponding species. Species with excessive zero values or poor data quality were excluded from the analysis. To achieve the uncertainty matrices, species below the detection limit were assigned a value of 5/6 of the MDL to represent their uncertainties. For species above the detection limit, their uncertainties were calculated as

$$\mathrm{U}\mathrm{n}\mathrm{c}=\sqrt{{(\mathrm{E}\mathrm{r}\mathrm{r}\mathrm{o}\mathrm{r}\mathrm{F}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\times c)}^2+{(0.5\times \mathrm{M}\mathrm{D}\mathrm{L})}^2}$$

where the Error Fraction was set to 10%. The signal-to-noise ratio (S/N) was used to evaluate data quality. The model was run with the number of factors ranging from 2 to 8. After a comprehensive evaluation of the model outputs, including Q/Q _expected, residuals distributions, and factor profiles, a four-factor solution was selected for both Shenzhen and Beijing (Details in Text S4). Additionally, the model input included data from identified VOCs, while molecular-level information for the remaining UCM was unavailable, potentially limiting the ability to fully resolve sources associated with this fraction. Nevertheless, previous studies have reported strong correlations between certain resolved IVOCs (e.g., branched alkanes) and the residual UCM, suggesting that they are coemitted from common primary sources. Therefore, our source apportionment can still provide meaningful insights into the likely sources of the residual UCM.

2.5. Estimation of Ozone Formation Potential

The ozone formation potential (OFP) of IVOCs was estimated using the maximum incremental reactivity (MIR) method, which quantifies the amount of ozone formed per unit mass of a precursor under idealized photochemical conditions. This approach enables the assessment of the relative contribution of different IVOCs to ozone production in the atmosphere. The OFP for each IVOC species was calculated using the following equation

$${\mathrm{O}\mathrm{F}\mathrm{P}}_i=C_i\times {\mathrm{M}\mathrm{I}\mathrm{R}}_i$$

where OFP _i is the ozone formation potential of compound i (μg/m³), C _i is the ambient concentration of IVOC species i (μg/m³), and MIR _i is the maximum incremental reactivity coefficient for species i (g O₃/g IVOC). The MIR coefficients employed in this study were obtained from Carter, 2010 For IVOCs lacking specific MIR values in the literature, surrogates with similar chemical structures and reactivities were selected to approximate their ozone-forming potential.

In this study, the average concentrations of each IVOC and VOC species in Shenzhen and Beijing were used to calculate their respective OFP values. It should be noted that MIR-based OFP depends on atmospheric photochemical regimes and is generally most applicable under high-NO _x or VOC-limited conditions, where ozone formation is more sensitive to incremental VOC additions. Previous studies have shown that summertime ozone formation in both Shenzhen and urban Beijing is typically VOC-limited or transitional due to the high-NO _x characteristics of Chinese megacities. − Therefore, the MIR method provides a useful metric for comparing the relative ozone formation potential of VOCs and IVOCs in these two cities, although the calculated OFP should be interpreted as a reactivity-based indicator rather than an absolute prediction of ambient ozone production. The total OFP for VOCs and IVOCs was obtained by summing the individual contributions of all measured species for each group. Detailed MIR and the corresponding OFP values for Shenzhen and Beijing are provided in Tables S5 and S6, respectively. Notably, the calculated OFP of IVOCs should be regarded as a lower-bound estimate, as MIR values cannot be assigned to the UCM fraction.

3. Results and Discussion

3.1. Ambient Concentration and Chemical Composition of IVOCs

The average IVOC concentration was 6.22 ± 4.50 μg/m³ in Shenzhen, with speciated IVOCs accounting for 41.8%, unspeciated b-alkanes contributing 2.6%, and the remaining being residual UCM (Figure ). A total of 15 IVOC species were identified, including 11 n-alkanes (C12–C22) and 4 PAHs (naphthalene, acenaphthylene, acenaphthene, and fluorene). The top five contributors were fluorene (9.5%), acenaphthene (6.2%), acenaphthylene (6.1%), n-pentadecane (3.8%), and n-hexadecane (3.5%). Comparatively, the average concentration of IVOCs in Beijing was 40.42 ± 30.11 μg/m³, significantly higher than that in Shenzhen. The speciated IVOCs and unspeciated b-alkanes contributed 6.3% and 1.9%, respectively, of total IVOCs. The identified 14 IVOC species consisted of 10 n-alkanes (C12–C21) and 4 PAHs (naphthalene, acenaphthylene, acenaphthene, and fluorene). The top contributors were n-tetradecane (1%), n-pentadecane (0.9%), acenaphthene (0.9%), and n-heptadecane (0.6%). The ambient IVOCs were dominated by residual UCM in both Shenzhen (55.5%) and Beijing (91.8%).

1.

Chemical composition of IVOCs observed in (a) Shenzhen and (b) Beijing.

We compared the observed summer IVOCs concentrations in Shenzhen and Beijing with those reported in previous studies (Figure ). It should be noted that comparisons across cities focus on total IVOC concentrations rather than on individual species. The summer IVOCs concentration in Shenzhen is comparable to that observed in Pasadena, California (6.3 μg/m³) and the average concentration recorded at Yangshan Port, Shanghai, across summer and winter (5.1 μg/m³). However, it is significantly lower than the levels reported in urban Guangzhou during autumn (25.0 μg/m³) and in Wangdu, Hebei, during winter (40.8 μg/m³). Among the studies summarized in Figure , the summer IVOCs concentration observed in Beijing in this work is the highest, highlighting the relatively elevated IVOC levels during the sampling period. Both Shenzhen and Guangzhou are located in the Pearl River Delta, where pollution levels are generally higher in autumn than in summer. Thus, the disparity between the IVOCs concentration observed in Shenzhen during summer and that reported by Fang et al. (2022) for Guangzhou in autumn is likely driven by seasonal variations. In addition, as shown in Figure , sites with higher total IVOCs concentrations also exhibit a larger fraction of residual UCM relative to speciated IVOCs. This observation suggests that increased IVOCs abundance is accompanied by a higher contribution from UCM, reflecting a more chemically complex composition in more polluted environments. Such chemically complex UCM may undergo multigeneration oxidation and continuously produce lower-volatility products during atmospheric aging, thereby contributing substantially to SOA formation. ,

2.

Comparison of the concentrations and chemical composition of IVOCs between this study and previous studies.

As shown in Figure , the highest IVOCs concentration in Shenzhen during summer was observed in Bin17, followed by Bin16 and Bin14. In contrast, IVOCs in Beijing were predominantly concentrated in higher volatility ranges, with Bin12 exhibiting the highest concentration, followed by Bin19 and Bin13. Previous studies have shown that emissions from gasoline and diesel vehicles are primarily concentrated in the Bin12 and Bin13 ranges. , In this study, the average IVOCs concentration in the Bin12–13 range during summer in Shenzhen was 1.20 μg/m³, accounting for 19.3% of the total IVOCs concentration, whereas it reached 24.94 μg/m³ in Beijing, accounting for 52.6% of total IVOCs. These findings suggest that vehicle exhaust emissions are a major contributor to IVOCs in Beijing’s ambient air. In addition, IVOCs concentrated in the relatively volatile Bin12–13 range are more likely to remain in the gas phase and participate in rapid atmospheric oxidation, thereby enhancing their contribution to photochemical ozone formation. In contrast, lower-volatility IVOCs enriched in higher bins generally exhibit stronger gas-particle partitioning tendencies and higher SOA yields. ,

3.

Average concentrations of IVOCs quantified in each bin for (a) Shenzhen and (b) Beijing.

Since the launch of the 14th Five-Year Plan, Shenzhen has actively promoted the transition to new energy vehicles (NEVs) through initiatives such as the New Energy Vehicle Industry Development Plan issued by the General Office of the State Council and the Shenzhen New Energy Vehicle Promotion and Application Work Plan (2021–2025). By 2024, the penetration rate of NEVs in Shenzhen had reached over 70%, ranking second globally. This high adoption rate has led to relatively low IVOCs concentrations in the Bin12–13 range, contributing to the overall lower IVOCs levels in the city. In contrast, the significantly higher proportion of IVOCs in the Bin12–13 range in Beijingaccounting for nearly half of the total concentrationfurther reinforces the dominant role of vehicle exhaust emissions in shaping Beijing’s IVOCs profile. This conclusion aligns with the findings of Zhang et al. (2023). In addition, meteorological conditions likely also contributed to the observed intercity differences in IVOCs concentrations. During the sampling period, Shenzhen experienced higher average wind speeds (4.33 m/s) than Beijing (2.61 m/s), indicating more favorable atmospheric dispersion conditions. Although summertime planetary boundary layer heights (PBLH) in Beijing are generally higher than those in Shenzhen due to stronger daytime surface heating and more developed convective mixing under continental climatic conditions, Shenzhen is strongly influenced by coastal meteorology, including sea-land breeze circulation, high humidity, and urban heat island effects, which can constrain PBLH variability while simultaneously enhancing horizontal air mass exchange and pollutant dilution. , Therefore, the stronger ventilation conditions in Shenzhen likely contributed to the comparatively lower IVOCs concentrations observed during the sampling period. Regional transport may also have contributed to the elevated IVOCs levels observed in Beijing. Located within the North China Plain, Beijing is frequently influenced by air masses transported from surrounding industrialized regions, which can enhance the accumulation of combustion-related pollutants. , In contrast, Shenzhen is more frequently affected by marine air masses from the South China Sea, which generally exhibit lower anthropogenic pollutant burdens and stronger dilution effects. , Therefore, regional transport and large-scale atmospheric circulation patterns likely played an important role in shaping the intercity differences in IVOCs concentrations.

3.2. Temporal Variations of IVOCs

Figure illustrates the temporal variations of total IVOCs concentrations alongside two key air pollutants, PM_2.5 and 8 h O₃, during the observation period in Shenzhen and Beijing. In Shenzhen, air quality was generally classified as “excellent,” with peak PM_2.5 and 8 h O₃ levels occurring primarily between July 7–10 and July 29 to August 2. During July 7–10, IVOCs concentrations exhibited an inverse relationship with PM_2.5 and 8 h O₃ trends, whereas during other periods, the three pollutants followed generally consistent patterns. In contrast, Beijing’s air quality during the observation period was rated as “moderate”, with IVOCs concentrations largely mirroring the variation trends of PM_2.5 and 8 h O₃. Peak PM_2.5 levels were recorded on July 14, while the highest 8 h O₃ concentration occurred on July 13. Although these peak pollution events did not coincide with the highest IVOCs concentrations, IVOCs levels also showed notable increases during this period. Overall, IVOCs exhibited weak correlations with PM_2.5 (R ² = 0.07 in Shenzhen, 0.0002 in Beijing) and O₃ (R ² = 0.04 in Shenzhen, 0.17 in Beijing). The weak correlations between IVOCs and O₃/PM_2.5 indicate that ambient IVOC concentrations alone cannot explain the variability of secondary pollution. This is expected because O₃ formation is governed by nonlinear photochemical processes involving VOCs, IVOCs, NO _x , radical chemistry, solar radiation, temperature, boundary layer dynamics, and regional transport. Similarly, PM_2.5 is influenced by primary emissions, secondary aerosol formation, gas–particle partitioning, atmospheric oxidation, and removal processes. Therefore, the weak correlations do not necessarily imply that IVOCs have a negligible role in O₃ or PM_2.5 formation; rather, they suggest that simple linear correlation analysis is insufficient to quantify their impacts. In this study, the correlation results are interpreted only as exploratory evidence of possible associations.

4.

Temporal variations of IVOCs with PM_2.5 and 8h-O₃ in (a) Shenzhen and (b) Beijing.

To further investigate IVOCs variations, we analyzed the observation period by differentiating between weekdays and weekends. In Shenzhen, the average IVOCs concentration on weekdays was 6.23 μg/m³, compared to 6.19 μg/m³ on weekends, indicating only a minor difference. This minimal weekday-weekend difference may be attributed to the high adoption rate of NEVs, which reduces the impact of fluctuations in vehicular activity on ambient IVOCs levels. In contrast, in Beijing, the average IVOCs concentration was significantly higher on weekends (53.59 ± 28.71 μg/m³) than on weekdays (44.68 ± 23.57 μg/m³), suggesting a more pronounced variation. In contrast, NEV penetration is relatively low in Beijing; according to the Beijing Municipal Bureau of Statistics (2025), only ∼12% of private passenger vehicles are NEVs, indicating that gasoline and diesel vehicles remain the dominant emission sources. In addition, Wang et al. (2022) reported that Beijing’s driving restriction policy, prohibiting vehicles with certain license plate numbers during weekday peak hours, reduces the probability of driving by ∼7.9–8.1% on weekdays. As this policy is not implemented on weekends, vehicle activity increases, which may contribute to the observed weekday-weekend differences in IVOCs concentrations.

3.3. Source Apportionment of IVOCs

For the source apportionment of IVOCs in Shenzhen (Figure ), Factor 1 was characterized by high contributions from n-dodecane, n-tridecane, and n-hexadecane, and accounted for nearly all of the observed n-C12. Compared to diesel and heavy fuel oil, IVOCs from gasoline combustion exhibit higher volatility. Previous studies have shown that gasoline vehicle emissions are typically enriched in relatively volatile low-carbon alkanes, particularly n-C12 and n-C13, whereas diesel emissions contain larger fractions of long-chain and lower-volatility hydrocarbons associated with incomplete combustion and lubricant oil emissions. ,, Fang et al. (2022) further reported that the gasoline-related factor in southern coastal Chinese cities exhibited extremely high contributions of n-C12 and n-C13 (about 60–75%), which is highly consistent with the species composition of Factor 1. Therefore, Factor 1 is tentatively attributed to gasoline vehicle emissions. Factor 2 was dominated by n-heptadecane to n-eicosane, along with acenaphthene and fluorene. Previous studies have shown that diesel exhaust contains substantial amounts of medium-molecular-weight n-alkanes, primarily emitted by on-road sources (e.g., heavy-duty trucks) and nonroad sources (e.g., heavy construction machinery). − IVOC dominates diesel-related I/SVOC emissions, with n-alkanes peaking at n-heptadecane (C17), and trucks have been identified as the dominant contributor to road-transport-related IVOC emissions. , Additionally, acenaphthene and fluorene are PAHs that can originate from crude oil and refined petroleum products (e.g., kerosene, diesel, lubricating oil) under moderate combustion temperatures (100–300 °C). The species composition of Factor 2 closely matches diesel-related emissions. Factor 3 was distinguished by a high proportion of PAHs, particularly naphthalene and acenaphthylene, whose contributions were significantly higher than those of n-alkanes. Combustion of coal and biomass, including wood and crop residues, is known to emit considerable amounts of naphthalene. Previous studies have further shown that biomass burning emissions are typically enriched in low-molecular-weight PAHs, especially naphthalene and acenaphthylene. , Furthermore, Fang et al. (2021) observed that IVOCs from corn combustion contain a significantly higher proportion of naphthalene compared to alkanes. Given the limited contribution of coal combustion to air pollution in Shenzhen and the species composition of Factor 3, we identify it as a biomass burning emission source. Factor 4 was dominated by n-heneicosane and n-docosane. The volatility of IVOCs from fossil fuel combustion is influenced by the volatility of the fuel itself. , Ships typically use heavy fuel oil, which has much lower volatility than gasoline and diesel. Xiao et al. (2024) demonstrated that heavy-fuel-oil-fueled vessels exhibit distinct I/SVOC profiles characterized by elevated PAHs and high-molecular-weight n-alkanes. Higher-molecular-weight n-alkanes, such as n-eicosane to n-docosane, are relatively more abundant in ship exhaust. The species composition of Factor 4 aligns with these characteristics, identifying it as a ship emission source.

5.

Source profiles of various IVOCs emission sources identified by PMF in Shenzhen. Abbreviations: Nap (naphthalene), Ace-1 (acenaphthylene), Ace-2 (acenaphthene), and Flu (fluorene).

Factor 1 in Beijing exhibits a species composition similar to that in Shenzhen (Figure ), with n-dodecane and n-tridecane as the dominant contributors, and is attributed to gasoline vehicle emission. Factor 2 was characterized by a high contribution of acenaphthylene and n-alkanes ranging from n-tetradecane to n-nonadecane. Similar to the diesel-related factor identified in Shenzhen, the predominance of medium-molecular-weight n-alkanes suggests strong contributions from diesel-related emissions. Factor 3 was dominated by naphthalene, a compound commonly associated with biomass burning, liquid fossil fuel combustion, and coal combustion. , Compared with Shenzhen, the coexistence of PAHs and combustion-related species in Beijing indicates a mixed biomass/coal combustion source, reflecting the more complex combustion emissions in northern Chinese cities. Factor 4 was distinguished by high contributions of n-heneicosane, acenaphthene, and fluorene, along with notable amounts of n-heptadecane to n-eicosane. This profile closely resembles IVOCs emissions from heavy fuel oil combustion. Given Beijing’s industrial landscape, Factor 4 is most likely linked to the petrochemical industry, which extensively uses heavy fuel oil for power generation, refinery processes, and chemical production. Additionally, PAHs are key byproducts of crude oil and refined fuel combustion. Therefore, Factor 4 was identified as a petrochemical industry emission source. As noted above, the source apportionment in both Shenzhen and Beijing is based primarily on the relative contributions and composition patterns of the measured IVOCs species.

6.

Source profiles of various IVOCs emission sources identified by PMF in Beijing. Abbreviations: Nap (naphthalene), Ace-1 (acenaphthylene), Ace-2 (acenaphthene), and Flu (fluorene).

Diesel-related emissions are the dominant source of ambient IVOCs in Shenzhen, contributing 44.4% of the total, followed by biomass burning at 29.8% (Figure a). Ship and gasoline vehicle emissions account for 16.7% and 9.1%, respectively. These findings are consistent with Fang et al. (2021), who reported that diesel emissions were the largest IVOCs source in urban Guangzhou, contributing 46.6% of the total. Emission inventory studies further support this result, indicating that diesel vehicles in China emit substantially higher levels of IVOCs than gasoline vehicles and ships, , highlighting the prominent role of diesel combustion in urban IVOCs pollution. Recent studies in the Pearl River Delta (PRD) and other coastal cities have similarly identified diesel-related emissions and ship activities as major contributors to urban VOCs/IVOCs, indicating that the diesel-dominant source structure observed in Shenzhen is regionally representative of highly urbanized coastal megacities. , Ships also represent an important source of IVOCs. As a major port city with the third-largest cargo throughput globally, Shenzhen hosts multiple port zones with intensive maritime activity. Pollutant emissions during port entry, exit, and berthing are substantially higher than during open-sea navigation, making ship-related IVOCs a notable contributor to local air pollution. Although the International Maritime Organization (IMO) mandated in 2020 that ships operating in China’s jurisdictional waters use fuel with sulfur content not exceeding 0.50% m/m, sulfur levels remain a critical factor influencing both the emission intensity and chemical composition of IVOCs. , For instance, ships using low-sulfur fuel emit approximately 2.4 times more IVOCs than those using high-sulfur fuel. As regulations on land-based pollution sources continue to tighten, the relative impact of ship emissions on Shenzhen’s air quality may become increasingly important under future emission-control scenarios. Furthermore, since the 13th Five-Year Plan, Shenzhen has actively implemented the national new energy vehicle (NEV) strategy. The city now leads the country in NEV adoption, with a penetration rate exceeding 70%. This widespread transition has significantly reduced the number of traditional fuel-powered vehicles, resulting in gasoline vehicle emissions accounting for only 9.1% of total IVOCs. Han et al.(2024) suggest that large-scale electrification of urban transportation can substantially reduce traffic-related VOCs emissions, particularly gasoline-related species, which likely contributes to the relatively low gasoline contribution observed in Shenzhen compared with northern cities that still maintain higher proportions of conventional fuel vehicles.

7.

Source apportionment results of IVOCs in (a) Shenzhen and (b) Beijing.

The source apportionment results indicate that petrochemical industry emissions are the dominant source of ambient IVOCs in Beijing, contributing 35.5% of the total, followed by gasoline vehicle emissions at 31.5% and diesel-related emissions at 25.0%. In contrast, biomass/coal combustion contributes relatively little, accounting for only 8.1%. These findings are consistent with previous studies on VOCs sources in Beijing during summer, which identified vehicle and industrial emissions as the major contributors to urban VOCs pollution. − Recent studies in Beijing and the Beijing-Tianjin-Hebei (BTH) region have consistently identified petrochemical industry emissions and vehicle exhaust as dominant contributors to urban IVOCs and SOA formation. As one of China’s major petrochemical industry clusters, the BTH region hosts numerous large-scale petrochemical enterprises, whose low-volatility raw materials and intermediate products substantially enhance ambient IVOCs levels. Vehicle emissions also represent a significant source of IVOCs in Beijing. Although the IVOCs emission factor for diesel vehicles is approximately nine times higher than that for gasoline vehicles, , the substantially larger number of gasoline vehicles in the city results in a slightly higher contribution from gasoline vehicle emissions during the summer observation period. It is also important to note that sampling for this study was conducted during the nonheating season, which likely explains the relatively low contribution (8.1%) from biomass/coal combustion. Previous wintertime studies in North China have often reported substantially higher contributions from coal and biomass combustion due to residential heating demand and stagnant meteorological conditions, indicating strong seasonal variability in IVOCs source structures across northern China.

A comparison between Shenzhen and Beijing highlights pronounced regional heterogeneity in urban IVOCs sources across China. Shenzhen is characterized by diesel- and ship-related emissions due to intensive port activities and high freight transportation demand, whereas Beijing is more strongly influenced by petrochemical industry and gasoline vehicle emissions associated with dense urban traffic and industrial activities. In addition, Shenzhen’s high NEV penetration likely suppresses gasoline-related IVOCs emissions compared with northern cities. These intercity differences demonstrate that urban IVOCs compositions are jointly shaped by industrial structure, transportation patterns, energy use, and emission-control policies, highlighting substantial regional variability in IVOCs sources and photochemical characteristics.

3.4. Estimation of OFP

Figure presents a comparative analysis of the concentrations and OFP of IVOCs and VOCs in Shenzhen and Beijing. In Shenzhen, the total OFP of identified VOCs and IVOCs is 15.41 μg/m³, with VOCs contributing 67.06% and IVOCs 32.94%. Among the top ten contributors to OFP, six compoundstoluene, o-xylene, p-xylene, styrene, ethylbenzene, and n-nonaneare classified as VOCs, while fluorene, acenaphthylene, acenaphthene, and n-pentadecane are IVOCs. Notably, fluorene, acenaphthylene, and acenaphthene exhibit OFP values of 1.95 μg/m³, 1.28 μg/m³, and 1.27 μg/m³, respectively. These three compounds are all PAHs, highlighting the significant role of PAHs in the contribution of IVOCs to ozone formation. In Beijing, the total OFP of identified VOCs and IVOCs is higher than in Shenzhen, at 19.23 μg/m³, with VOCs contributing 77.86% and IVOCs 22.14%. The top ten OFP-contributing species include the same five VOCs observed in Shenzhentoluene, o-xylene, p-xylene, styrene, and ethylbenzenealong with n-nonane and n-octane. As in Shenzhen, fluorene, acenaphthylene, and acenaphthene are the dominant IVOC contributors to OFP, with values of 1.43 μg/m³, 0.86 μg/m³, and 0.62 μg/m³, respectively.

8.

Concentration (inner circle) and ozone formation potential (OFP; outer circle) of individual IVOCs species in (a) Shenzhen and (b) Beijing. For comparison, the concentration and the OFP of identified VOCs are also shown.

Overall, VOCs contribute more substantially to ozone formation than IVOCs in both cities. However, the role of IVOCs, particularly PAHs, in ozone formation remains non-negligible. These findings underscore the need for future air pollution studies and control strategies to incorporate the effects of IVOCs, especially those with high reactivity such as PAHs, in order to develop more comprehensive and effective approaches to ozone pollution mitigation.

3.5. Uncertainty Analysis

Compared to Shenzhen, the sampling duration and sample size in Beijing are more limited and thus require careful consideration. As noted above, the sampling period fell within a typical summer meteorological window; each sample was time-integrated, which smooths out short-term fluctuations. The IVOC species inventories largely overlap between the two cities, and the marker species patterns of corresponding source factors are qualitatively consistent. The cross-city consistency of the source profiles demonstrates their physical stability. Moreover, previous studies have demonstrated that meaningful and stable source profiles can be resolved even from relatively small data sets. For example, Feng et al. (2023) systematically investigated the influence of sample size on PCA-MLR, PMF, and FA-NNC models using data sets as small as 20 samples and found that, although source contributions may vary, the extracted source profiles remained generally stable and comparable to those obtained using larger data sets. In addition to the physical consistency, the robustness of the PMF solutions was further assessed using BS and DISP diagnostics to evaluate factor reproducibility and rotational ambiguity.

BS analysis was conducted to evaluate solution stability and the influence of individual samples. For Shenzhen, 100 BS runs were performed using block resampling to preserve temporal autocorrelation, with a correlation threshold of R = 0.6 for factor matching. The resulting mapping rates were 93% (Factor 1), 99% (Factor 2), 92% (Factor 3), and 98% (Factor 4), all exceeding the recommended 80% criterion, indicating strong robustness and no dominance by a small subset of samples. For Beijing, BS analysis (100 runs, R = 0.6) yielded mapping rates of 84% (Factor 1), 97% (Factor 2), 85% (Factor 3), and 99% (Factor 4), all above the 80% threshold, suggesting generally stable solutions despite the smaller data set. DISP analysis was applied to assess rotational ambiguity. For Shenzhen, no factor swaps occurred at the strictest constraint (dQ _max = 4), and the maximum decrease in Q was 0.000, confirming that the solution lies at a stable minimum with negligible rotational ambiguity. Factor swaps appeared only under relaxed constraints (dQ _max = 8, 15, 25), as expected, indicating that the solution is well-defined. Similarly, for Beijing, the maximum Q decrease was 0.000, with no swaps were observed at dQ _max = 4 or 8, further indicating a stable solution. Swaps began to appear only at dQ _max = 15 and increased at dQ _max = 25, reflecting typical behavior under relaxed constraints. Notably, Factor 4 remained the most robust factor throughout DISP analysis, consistent with its high BS mapping rate.

Overall, the BS and DISP results consistently demonstrate that the four-factor solutions for Shenzhen and Beijing are stable, robust, and physically interpretable, with minimal rotational ambiguity.

A quantitative assessment of OFP uncertainty is essential for robust interpretation of the results. The total uncertainty in OFP arises mainly from two sources. The first is the uncertainty in VOC and IVOC concentration measurements, which includes the analytical precision of the GC–MS/FID system, calibration curve errors, and the reproducibility of standard gases. In this study, the relative uncertainty of individual VOC concentrations, estimated from repeated measurements, ranges from 0.27 to 3.39. The second source of uncertainty comes from the MIR coefficients, which are reported to vary between 10% and 30%. A representative value of 20% was adopted in this study. The OFP uncertainty for each species was estimated using the root-sum-of-squares method, assuming independent contributions from concentration measurements and MIR values. The results show that the relative uncertainty of OFP for individual IVOC species varies widely, from approximately 33% for n-pentadecane to over 340% for naphthalene. Despite the relatively large uncertainties of some species, the total OFP, which accounts for the combined contributions of all measured VOCs and IVOCs, exhibits a lower relative uncertainty of 20–25%. These results indicate that, while caution is warranted when interpreting species-level OFP, the aggregated OFP estimates are comparatively more robust and suitable for evaluating overall ozone formation potential.

It should be noted that the calculated IVOCs-related OFP may be systematically underestimated because the current OFP estimation only includes resolved species with available MIR coefficients, whereas residual UCM accounted for 55.5% and 91.8% of total IVOCs in Shenzhen and Beijing, respectively. Previous studies have shown that UCM IVOCs emitted from fossil-fuel combustion, vehicle exhaust, and petrochemical industries contain large amounts of reactive branched alkanes, cyclic hydrocarbons, and substituted aromatics, which generally exhibit considerable atmospheric oxidation reactivity and SOA/O₃ formation potential. Therefore, the exclusion of these residual UCM may lead to a systematic underestimation of the contribution of IVOCs to ozone formation. Compared with Shenzhen, Beijing exhibited a substantially higher residual UCM fraction, while its IVOCs were mainly associated with vehicle and petrochemical emissions, which are typically characterized by greater chemical complexity and higher UCM contributions. As a result, the underestimation of OFP is likely more pronounced in Beijing. These findings suggest that although VOCs dominated the calculated OFP in this study, the actual contribution of IVOCs to ozone formation in UCM-rich urban atmospheres may be substantially higher than currently estimated.

4. Conclusions

This study presents a systematic intercity comparison of ambient IVOCs in two major Chinese megacities, Shenzhen and Beijing, based on field observations. By integrating source apportionment with OFP analysis, we provide quantitative insights into IVOC concentrations, source characteristics, and their potential atmospheric implications in contrasting urban environments.

Substantial differences in IVOC levels and source structures were observed between the two cities. The mean IVOC concentration in Beijing (40.42 μg/m³) was approximately seven times higher than that in Shenzhen (6.22 μg/m³), indicating marked contrasts in emission intensity and composition. These differences are closely associated with variations in urban development patterns and energy use. The widespread adoption of NEVs in Shenzhen likely contributes to the relatively low transportation-related IVOCs contribution, whereas the higher reliance on conventional fuel vehicles in Beijing is associated with elevated concentrations and greater temporal variability. Source apportionment indicates that diesel-related emissions dominate IVOCs in Shenzhen (44.4%), followed by biomass burning (29.8%), ship emissions (16.7%), and gasoline vehicles (9.1%). In contrast, IVOCs in Beijing are primarily attributed to petrochemical industry emissions (35.5%), gasoline vehicles (31.5%), and diesel-related sources (25.0%). These results highlight substantial regional variability in IVOCs source structures between coastal and inland megacities. IVOCs accounted for 32.9% of the total OFP in Shenzhen and 22.1% in Beijing, underscoring their significant role as ozone precursors. Notably, fluorene, acenaphthylene, and acenaphthene exhibited relatively high reactivity and disproportionately large contributions to OFP, suggesting that tPAHs may warrant further attention in future studies of urban ozone formation.

Overall, our findings demonstrate that IVOCs pollution levels and source profiles are strongly influenced by urban structure and energy systems. Given the short-term summertime observations at single urban sites, the findings should be interpreted as indicative rather than fully representative of annual urban IVOCs characteristics across China. Nevertheless, this study provides useful insights into the potential atmospheric importance of IVOCs in urban environments and highlights the need for further long-term and multiseason observations in Chinese megacities.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c05752.

• Detailed experimental procedures and calibration methods (Text S1); determination methods for total IVOCs and volatility binning (Text S2); quantification approach for unresolved complex mixture (UCM) IVOCs (Text S3); PMF diagnostics and factor-selection procedures (Text S4); sampling campaign details (Table S1); qualitative and quantitative ions of target compounds (Table S2); summary statistics of VOCs and IVOCs in Shenzhen and Beijing (Tables S3–S4); maximum incremental reactivity (MIR) coefficients and ozone formation potentials (OFP) (Tables S5–S6); and references cited in the Supporting Information (PDF)

Guanqun Tang: Investigation, Validation, Writing - original draft; Chudi Wang: Data curation, Investigation, Validation, Visualization; Mingze Song: Data curation; Lina Chen: Data curation; Yanzhi Wang: Data curation; Haiyan Li: Conceptualization, Methodology, Visualization, Writing - original draft, Writing - Reviewing & Editing.

The authors declare no competing financial interest.