2.1 Sampling site

An observation site (109^∘00^′7^′′ E, 34^∘13^′22^′′ N) managed by Xi'an Jiaotong University was used in this study (Fig. 1). All sampling equipment was deployed on the rooftop of a 15 m tall academic building. No obvious stationary pollution sources were found nearby, and the location can be considered as a typical urban location in Xi'an (Q. Zhang et al., 2015).

Figure 1Regional and local maps of the study area: (a) regional map showing the location of Xi'an and the surrounding geography; (b) local map of Xi'an showing the sampling site (blue dot), main roads (red lines), and secondary roads (blue lines).

2.2 Field sampling

A severe dust-to-haze episode was observed in Xi'an and the surrounding areas from 8 to 13 November 2016, and samples were continuously collected during this period to investigate the chemical compositions of both VOCs and fine PM. A total of 57 non-methane VOC species (i.e., C₂−C₁₂ saturated and unsaturated aliphatic and aromatic VOCs) were sampled hourly into offline multi-bed adsorbent tubes; the measured 57 VOCs were defined as VOC_PAMS. The loaded tubes were analyzed using a thermal desorption and gas chromatography/mass spectrometry (TD-GC/MS) method. In previous developmental work, humidity and temperature during sampling were found to significantly impact the analyses; for this study, all sample collections were made under optimized conditions (Ho et al., 2017, 2018). Sixteen airborne carbonyls (including mono- and dicarbonyls) were collected over diurnal cycles (i.e., 20:00–08:00 and 08:00–20:00 local time, LT) by 2,4-dinitrophenylhydrazine (DNPH) coated cartridges. Detailed sampling and analytical procedures for VOCs and carbonyls can be found in previous publications (Ho et al., 2017; Dai et al., 2012).

PM_2.5 filter samples were sampled with mini-volume samplers (model MiniVol, Airmetrics Co., Oregon, USA) by an overline flow rate of 5 L min⁻¹ (Cao et al., 2005). Fine PM was sampled with 47 mm quartz microfiber filters (Whatman QM-A, Maidstone, UK), and the filters were pre-heated at 900 ^∘C for 3 h before sampling. The loaded filters were transferred into clean polystyrene petri dishes and stored in a freezer.

2.3 Chemical analyses

Analytical procedures for VOC analysis have been described previously (Ho et al., 2017). In brief, the analytes in the adsorbent tubes were firstly desorbed in a thermal desorption unit (Series 2 UNITY-xr system with ULTRA-xr, Markes International Ltd., UK) coupled to a GC/MS (7890A/5977B, Agilent Technologies, Santa Clara, CA, USA). The loaded tubes were transferred into the TD unit and blown with ultra-high purity He gas. The targeted VOCs were desorbed at 330 ^∘C within 8 min and then refocused onto a cryogenic trap (U-T1703P-2S, Markes) at −15 ^∘C. The targeted VOCs were transferred to a cold GC capillary column head (Rtx^®-1, $\mathrm{105}\mathrm{m}\times \mathrm{0.25}\mathrm{mm}\times \mathrm{1}\mathrm{mm}$ film thickness, Restek Corporation, USA) at −45 ^∘C. The chromatographic condition can be found in our previous work (Ho et al., 2017).

For carbonyl compounds, the DNPH cartridges were firstly eluted with acetonitrile (HPLC/GCMS grade, J & K Scientific Ltd., Ontario, Canada) (Dai et al., 2012). The extracts were analyzed with a typical high-pressure liquid chromatography (HPLC) system (Series 1200; Agilent Technologies) equipped with photodiode array detector. The column was matched with a 4.6×250 mm Spheri-5 ODS 5 µm C-18 reversed-phase column (Perkin-Elmer Corp., Norwalk, CT) (Dai et al., 2012; Ho et al., 2011).

The particulate organic carbon (OC) and elementary carbon (EC) were analyzed with a DRI model 2001 carbon analyzer (Atmoslytic, Inc., Calabasas, CA, USA) (Chow et al., 2007, 1993). Anions (Cl⁻, ${\mathrm{NO}}_{\mathrm{3}}^{-}$, and ${\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$) and cations (Na⁺, ${\mathrm{NH}}_{\mathrm{4}}^{+}$, K⁺, Mg²⁺, and Ca²⁺) in particles were determined in aqueous extracts of the sample filters. Detailed extraction and analytical procedures were presented in a previous publication (Zhang et al., 2011). The abundances of 25 particulate elements (Na, Mg, Al, Si, S, Cl, K, Ca, Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, As, Se, Br, Sr, Ba, Pb, Ga, Zn) were measured by energy dispersive X-ray fluorescence (ED-XRF) spectrometry (Epsilon 4 ED-XRF, PANalytical B.V., the Netherlands). The X-ray source was matched with a metal-ceramic X-ray tube with an Rh and Ag anode, and the X-ray source was operated at a maximum current of 3 mA and a maximum accelerating voltage of 50 kV (maximum power 15 W).

2.4 Quality control

The minimum detection limits (MDLs) of the VOCs were in the range of 0.003–0.042 ppbv with a 3 L sampling volume (Table S1 in the Supplement). The measurement precision at 2 ppbv was ≤5 % (Ho et al., 2017, 2018). Three field blank samples were collected within each sampling day, and they were analyzed using the same procedures as those for the ambient air samples. Most target compounds were not detected in the field blanks, and propylene, benzene, and toluene were below their MDLs (<0.23 g per tube and <10 % of the arithmetic mean of ambient samples). No breakthrough (∼0 %) was observed for VOC_PAMS except for C₂−C₃ hydrocarbons, which were <10 % when the air temperature was >30 ^∘C. The MDLs for the carbonyl target compounds were between 0.009 and 0.067 ppbv at a sampling volume of 3.6 m³. Negligible breakthrough (<5 %) was found under the sampling conditions and flow rates in the field.

3.1 Origins of ambient VOCs during dust and fine particle pollution events

In the present study, the mixing ratio of the sum of non-methane hydrocarbon was 36.0±15.7 ppbv, which was lower compared to those in Beijing and Guangzhou with values of 51.0 and 47.8 ppbv, respectively (Ho et al., 2004; Liu et al., 2008b). Similar levels of alkenes were seen in the cities of Beijing (9.4 ppbv) and Guangzhou (8.2 ppbv) compared to that in the present study (9.2 ppbv, Table S2) (Ho et al., 2004; Liu et al., 2008b). Unexpectedly, the levels of aromatics were slightly higher in Xi'an (10.3 ppbv) than in Beijing (9.6 ppbv) and 50 % higher than in Guangzhou (6.8 ppbv, Shao et al., 2009; Zou et al., 2015). In this respect, ethylene, ethane, toluene, isopentane, propane, n-butane, isobutane, propylene, n-pentane, and benzene were the 10 most abundant VOC_PAMS. The high fractions of these markers reflect strong emissions from traffic and coal combustion or from biomass burning (Liu et al., 2008a; Ho et al., 2009; Huang et al., 2015; Fan et al., 2014; Z. Zhang et al., 2015). Previous studies found higher contributions of non-fossil sources to carbonaceous aerosols in Xi'an than in Beijing (Ni et al., 2018). Generally, non-fossil emissions mainly originate from biomass burning (Ni et al., 2018), and the higher contribution of non-fossil sources to carbonaceous aerosols in Xi'an would indicate that remarkable biomass burning activities exist in Xi'an and the surrounding areas (Huang et al., 2014; Xu et al., 2016).

Receptor models and correlations between individual VOCs have been used for source assessments. In this study, a significant correlation (R²=0.62, p<0.05, slope of 1.59) was found for a least squares regression between toluene and benzene (Fig. S1 in the Supplement). The ratio of toluene to benzene (T∕B) has been shown to be different among combustion sources; for example, Liu et al. (2006) reported T∕B ratios of 1.5–2.0 in gasoline-related emissions collected in a tunnel. In contrast, T∕B ratios ranged from 0.23–0.68 and 0.13–0.71 for biomass burning and coal combustion, respectively (Z. Zhang et al., 2015). The T∕B ratios in our samples (R²=0.62, p<0.05, slope of 1.59) implied a strong impact from traffic on the ambient VOCs in Xi'an. Significant correlations (p<0.05) were observed among C₃−C₅ alkanes: between propane and n-butane (R²=0.75, slope =0.91), n-pentane and isopentane (R²=0.85, slope =0.35), and trans-2-butene and cis-2-butene (R²=0.99, slope =0.84) (Fig. S1). The observed ratio of propane to n-butane in Xi'an was 1∕1, which is close to that (1.36) observed in the tunnel study cited above (Liu et al., 2008a). High loadings of n-pentane and isopentane are indicative of unburned vehicular emissions, and Liu et al. (2008b) reported a ratio of isopentane / n-pentane of 3 in tunnel air, which is consistent with the slope of 2.85 found in the present study. The ratios of T∕B, trans-∕cis-2-butene, propane / n-butane, and n-pentane∕isopentane indicated that gasoline emissions were a dominant source of ambient VOCs. The source apportionment by PMF (positive matrix factorization) model and the detailed description of source apportionment will be carried out in the following section.

A PMF model was used to identify the major pollution sources; the data input to the model was the mixing ratios and uncertainties in the mixing ratios of the VOCs for all valid samples collected during the study. Five sources were identified (Fig. S2), and the detailed process of source apportionment is given in the supplement. Biomass burning and gasoline exhaust were the two most significant pollution sources, contributing 25 % and 18 %, respectively. The combustion of LPG and CNG (25 %), diesel exhaust (15 %), and coal combustion (17 %) were also found to be important sources of ambient VOCs (Fig. S2). Biomass is commonly used for heating and cooking in rural areas of the basin in winter due to its low cost compared to natural gas and electricity. Consistent with our results, previous studies have found a high contribution of biomass burning and gasoline exhaust to the organic aerosol in the Guanzhong basin (Cao et al., 2005).

Clear air conditions occurred at the beginning of the sampling period, but severe dust and fine particle pollution events were observed afterward. The high dust event was defined by the loading of particulate matter ≤10 µm with an aerodynamic diameter (PM₁₀) between 300 and 500 µg m⁻³, and these conditions occurred from 12:00 LT on 9 November to 13:00 LT on 10 November. The abatement of dust before the fine particle pollution event is referred to as the transition period (i.e., PM₁₀<300 µg m⁻³ and PM_2.5<100 µg m⁻³). The loading of PM_2.5 subsequently increased, and heavy fine particle pollution (PM_2.5>100 µg m⁻³) occurred after 18:00 LT on 11 November.

Ratios of individual VOCs can be used to identify the origins of the compounds and to study atmospheric aging processes due to the special composition of VOCs in a typical source and the different lifetimes of VOCs species (Xue et al., 2017; Z. Zhang et al., 2015). In addition, influences from meteorological variation and atmospheric transport also need to be considered when the potential sources of the compounds in ambient air are characterized. To investigate the impacts of air mass transport on VOC concentrations, we calculated air mass back trajectories using the NOAA HYSPLIT model for the dust event (Fig. S4a) and for the fine particle pollution episode (Fig. S4b) (Stein et al., 2015; Rolph et al., 2017). The trajectories were calculated at an arrival height of 500 m above ground level at the observation site. In view of the short atmospheric lifetimes of VOCs (for example, isoprene ∼1.4 h; propylene ∼5.3 h; toluene 2.1 d) (Atkinson and Arey, 2003), 24 h back trajectories were used for this assessment.

Clear different air mass back trajectories and VOC ratios were observed between dust pollution and haze pollution periods. From 9 to 10 November (in the dust pollution period), the air mass reaching Xi'an passed over areas to the west of the city (i.e., Gansu province and the Ningxia Hui Autonomous Region) through long-range transport; after 11 November (formation of haze), the transport of air mass was mainly limited to areas around southern Xi'an. Differences in the chemical compositions of ambient VOCs in the dusty versus in the hazy events can clearly be seen (Fig. 2) in the ratios of toluene to benzene (T∕B) (toluene: K_OH $\mathrm{5.96}\times {\mathrm{10}}^{-\mathrm{12}}$ cm³ molecule⁻¹ s⁻¹; benzene: K_OH $\mathrm{1.22}\times {\mathrm{10}}^{-\mathrm{12}}$ cm³ molecule⁻¹ s⁻¹) and m- and p-xylene to ethylbenzene (X∕E) (m-xylene: K_OH $\mathrm{2.30}\times {\mathrm{10}}^{-\mathrm{11}}$ cm³ molecule⁻¹ s⁻¹; p-xylene: K_OH $\mathrm{1.43}\times {\mathrm{10}}^{-\mathrm{11}}$ cm³ molecule⁻¹ s⁻¹; ethylbenzene: K_OH $\mathrm{7.00}\times {\mathrm{10}}^{-\mathrm{12}}$ cm³ molecule⁻¹ s⁻¹). During the clear and dust periods, the T∕B and X∕E ratios varied significantly with time of day; that is, the highest values for T∕B (4.5–9.0) and X∕E (0.98–1.05) were seen during rush hour (07:00–09:00 and 17:00–19:00 LT), while the lowest values (0.50–1.95 for T∕B and 0.89–0.96 for X∕E) occurred in the early afternoon (i.e., 14:00–15:00 LT). The timings of the high T∕B and X∕E ratios suggest that fresh emissions from local traffic were the major source of the ambient VOCs, and this implies that long-range transport did not have a strong impact on the ambient VOCs during the clear or dust parts of the study (Ho et al., 2004; Liu et al., 2008a). During the transitional and fine PM pollution periods, both T∕B and X∕E varied but at relatively lower values compared to the earlier parts of the study (T∕B: 3.33±1.97, 2.21±0.86, 1.91±0.74, and 2.01±0.56 in clear, dust, transitional, and fine particle pollution periods, respectively; X∕E: 1.00±0.05, 1.05±0.12, 0.93±0.17, and 0.95±0.13 in clear, dust, transitional, and fine particle pollution periods, respectively). These synchronous lower values of T∕B and X∕B in transitional and fine particle pollution periods were indicative of aged air masses (Z. Zhang et al., 2015; Xue et al., 2017; Warneke et al., 2013).

Figure 2Variations in the ratios of indicator volatile organic compound low loading (VOC) species (toluene/benzene (T∕B) and m-,p-xylene/ethylbenzene (X∕E)) and fine particle loadings during the study period.

Download

Variations in the air mass transport pathway and T∕B or X∕E ratios during different sampling periods (clear, dust, transitional, fine particle pollution) confirmed that ambient VOCs were fresh in the clear and dust periods but relatively aged during the transitional and fine particle pollution periods (Z. Zhang et al., 2015; Xue et al., 2017; Warneke et al., 2013). This indicates that the long-range transport of air mass had a relatively weak influence on the ambient VOCs even during the high dust period. Otherwise, the composition of ambient VOCs should be relatively aged due to the long exposure time with dust transport. Indeed, emissions from local vehicular exhausts and biomass burning in Xi'an and the surrounding areas were the main contributors to ambient VOCs throughout our study.

3.2 Transformation of VOCs between dust and fine particle events

With the shading of dust, levels of ambient VOCs decreased with time, and the low concentrations (8.3–33.9 ppbv) were observed from 13:00 LT on 10 November to 01:00 LT on 11 November (Fig. 3). During the fine particle pollution period (12–13 November), the ∑VOC_PAMS increased, reaching an average of 38.0 ppbv in the last 24 h, compared to 19.0 ppbv in the transitional period and 21.5 ppbv in the first 12 h of the fine particle pollution episode (Fig. 3). This buildup of VOCs can be explained by weak dispersion and relatively shallow boundary layers (400–1000 m) during the event (Fig. 3). In addition, during this transition period, much lower ratios of T∕B and X∕E were observed in comparison with those in other periods (as mentioned in Sect. 3.1.2). We propose the possibility that windblown dust, which includes sustainable TiO₂, can influence the atmospheric photochemistry of VOCs, which would accelerate the oxidation of ambient VOCs (Chu et al., 2019; Nie et al., 2014).

Figure 3Temporal variations in volatile organic compound (VOC) concentrations and particle levels during the sampling period (9–13 November 2016).

Download

Changes in the emission sources and their strengths, physical dispersion, regional transport, and the aging of air masses could all affect VOC levels and composition (Xue et al., 2013, 2017). As a result, to evaluate the aging of ambient VOCs in different periods, the impact of dust on the transformation of ambient VOCs, and the relative processes, the mentioned factors should be fully considered.

To evaluate the impact of source types on the variation of VOCs in the dust-to-haze episode, the diurnal variation of VOCs was depicted. During the clear and dust periods – and similar to the trends in T∕B and X∕E ratios – peaks in ∑VOC_PAMS were seen from 17:00 to 20:00 and 09:00 to 12:00 LT (Fig. 3), which highlighted the impacts of local traffic emissions (Liu et al., 2008a; Huang et al., 2015). 1,3-Butadiene is often used as a marker of gasoline-powered motor vehicles (Huang et al., 2015), while ethane is a key chemical marker for biomass and coal combustion (Liu et al., 2008a). Time series plots of 1,3-butadiene and ethane (Fig. S5) show that peaks in 1,3-butadiene mostly occurred during rush hour, while higher concentrations of ethane were seen during the night. These results support the conclusion that there were strong impacts from gasoline-powered motor vehicles in the daytime and from biomass burning or coal combustion for heating at night. In addition, winter heating activities was relatively active because of low temperatures during the transitional period, and this limited the possibility of reduced emission amounts. Hence, the variations of source strength were eliminated as being a major factor which caused the extremely low concentration and relative aged composition of ambient VOCs.

The variation of physical dispersion was also eliminated. With the shading of dust transport, shallow boundary layers were observed in the transitional period. For the clear and dust transport period, the boundary layer between 08:00 to 14:00 LT was relatively high (1150–1500 m). In contrast, the boundary layer height decreased sharply to <800 m on 11 November in the transitional period. This limited the possibility that diffusion caused the sharp decrease in ambient VOCs in the transitional period.

The significant impact of air mass input was eliminated. Input of air mass would certainly cause the variations in the composition and loading of VOCs (Xue et al., 2014). In the present study, the long-range transport of air masses had limited impacts on the characteristic of ambient VOCs during the sampling period. In another aspect, relatively active VOCs would be firstly degraded; hence, the composition of ambient VOCs would be aged with long-range transport (Ho et al., 2009; Xue et al., 2017). In the present study, as mentioned above, the composition of ambient VOCs was relatively fresh under the long-range transport of air masses (during dust transport). In contrast, the composition of ambient VOC was relatively aged in the transitional period, and the air mass in this period was limited within Xi'an and the surrounding area. This phenomenon indicated that regional transport cannot be a major factor inducing the relatively aged composition and excess low loading of the ambient VOCs in the transitional period.

Figure 4Temporal variations in trans-/cis-2-butene ratios in the dust, transitional, and fine particle pollution periods.

Download

Synchronous changes in the VOC isomerides were found in the windblown dust-to-haze episode, which supplied the evidence of the accelerated photochemistry reactions. In the present study, we found a fast decrease in trans-/cis-2-butene ratios during dust transporting, which confirmed the accelerated photochemical reactions of ambient VOCs (Fig. 4). Trans-2-butene and cis-2-butene are two isomerides that are mostly emitted from the same sources (Fang et al., 2017; Y. Zhang et al., 2015). Trans-2-butene has a higher photochemical reaction rate with OH radicals in the atmosphere (k_OH $\mathrm{6.40}\times {\mathrm{10}}^{-\mathrm{11}}$ cm³ molecule⁻¹ s⁻¹) than cis-2-butene (k_OH $\mathrm{5.64}\times {\mathrm{10}}^{-\mathrm{11}}$ cm³ molecule⁻¹ s⁻¹) (Perring et al., 2013); hence, trans-/cis-2-butene ratios would decrease with the photochemical reactions (Y. Zhang et al., 2015). Firstly, relatively higher trans-/cis-2-butene ratios were observed during rush hour (evening rush hour 17:00–20:00 LT; morning rush hour 07:00–10:00 LT) (Fig. 4), which indicated fresh emissions from local traffic activities (Y. Zhang et al., 2015). In addition, a sharp decrease in trans-/cis-2-butene ratios was observed from the later half of the windblown dust period to the end of the transitional period (Fig. 4). The quick shrinking of trans-2-butene compared to cis-2-butene in the dust pollution period indicated that the oxidation of ambient VOCs was accelerated in the period with high loading of the suspended dust particles (Y. Zhang et al., 2015).

A significant increase in particulate active metals was found in the dust pollution period, which further verified the promotion of dust on the heterogeneous reactions. Previous studies found that mineral dust can affect the chemistry of the atmosphere by scavenging gaseous compounds (Zhang et al., 2000; Chen et al., 2012); it can also promote heterogeneous reactions of atmospheric substances, including VOCs, because the particle surfaces can provide sites for photo-catalytic reactions (Cwiertny et al., 2008; Ndour et al., 2009). In the present study, iron (Fe) and titanium (Ti) contents of the particulate increased significantly during the period with dust transport (Fig. 5). In detail, the content of Fe increased from 19.3 µg m⁻³ on clear days to 40.8 µg m⁻³ on dust pollution days, and the content of Ti increased from 0.92 to 2.98 µg m⁻³. Hence, the huge increase in Ti and Fe concentrations in the particulate phase during the period of dust pollution could possibly promote the solid–gas photochemical reaction of the ambient VOCs, which would reasonably explain the relative low level and aged composition of ambient VOCs in this period (Chu et al., 2019; He et al., 2014; Song et al., 2019).

Figure 5Composition of selected metallic elements in the PM_2.5 samples.

Download

Figure 6Variations in (a) the mixing ratios of 17 carbonyl compounds and acetone to methylglyoxal (A∕M) ratios in the gas phase, (b) particulate carbon fractions, and (c) particulate water-soluble ions during the study period.

Download

3.3 Variation of carbonyl compounds between dust and fine particle pollution periods: further formation of oxygenated VOCs with aging of primary VOCs

The aging of primary VOCs and the formation of carbonyl compounds were observed synchronously as the fine particle pollution event developed (Figs. 3, 6a). As discussed above, relatively low T∕B and X∕E ratios were observed during the transitional and fine PM periods after the dust event (Sect. 3.2). In our study, the carbonyl levels increased after the clear and dust periods, and the highest levels were seen during the fine particle pollution event (Fig. 6a). Carbonyl compounds are produced from the primary sources and form through secondary processes (Dai et al., 2012; Duan et al., 2012). We found higher carbonyl concentrations during daytime than at night (Fig. 6a). This is consistent with previous studies in Xi'an (Dai et al., 2012), which confirmed the secondary formation of carbonyl compounds under sunlight illumination.

Methylglyoxal is generally considered to be a secondary species, while acetone is mainly from primary emissions; the ratio of acetone to methylglyoxal (A∕M) has been used as an indicator of air mass aging (Dai et al., 2012; Liu et al., 2006). In the present study, A∕M ranged from 12 to 14 during the clear period and the first half of the dust period but then dropped sharply and stayed between 6 and 9 during the later parts of the dust pollution period, the transitional period, and the high PM event (Fig. 6a). Increases in the abundance of carbonyl compounds and lower A∕M ratios suggested relatively stronger aging of the air masses, which is further evidence of the fast degradation of VOCs in the late half of the windblown dust event, and the primary VOCs were oxidized and served as precursors of SOA. In consequence, composition of particles changed with the oxidation of ambient VOCs across the sampling periods.

3.4 Variations of PM_2.5 chemical composition during dust and fine PM pollution periods

Significant variations of water-soluble inorganic ions, OC, and EC were observed diurnally and between dust and fine particle pollution events (Fig. 6b, c). For instance, the concentrations of ${\mathrm{NO}}_{\mathrm{3}}^{-}$ were relatively high in the daytime, while K⁺ and Cl⁻ were more abundant at night. The diurnal cycles can be explained by the formation of secondary particles through photochemical processes during the daytime and by the impacts from biomass and coal burning for heating at night (Dai et al., 2012; Zhang et al., 2018; Cong et al., 2015). The concentrations of Ca²⁺, Mg²⁺, and Na⁺, which are typically associated with dust in inland areas (Wu et al., 2011), increased sharply during the dust period and then declined rapidly afterwards.

As discussed, the apparent contribution of VOCs to the formation of SOAs increased when the dusty conditions transitioned into a fine particle pollution event. Temporal changes in the chemical composition of PM_2.5 are consistent with this suggestion. During the fine particle pollution period, the concentrations of secondary ions, particularly ${\mathrm{NO}}_{\mathrm{3}}^{-}$, increased as the haze event developed. A similar trend was seen for OC (Fig. 6b), and the content of particulate OC increased from 11.1 µg m⁻³ in the dust event period to 47.1 µg m⁻³ in the haze period. In another aspect, the ratio of OC∕EC increased from 1.3 to 4.9 in the dust-to-haze episode. The previous studies on the characterization of particles from traffic emissions reported OC∕EC ratios in the range of 0.28 to 0.92 in diesel vehicles, and the OC∕EC ratios were reported as >2 in gasoline vehicles (Cadle et al., 1999; Huang et al., 2006). In addition, the OC∕EC ratios were reported in the range of 0.9 to 1.6 in the urban region of the city of Guangzhou (Tao et al., 2019). In the present study, the consistent increase in OC∕EC would prove the formation of SOAs in the dust-to-haze episode. Combined with the findings regarding the composition of VOCs and PM_2.5, these results indicate that the reactions of VOCs led to the formation of SOAs, and in so doing they contributed to the fine particle pollution.