Influence of COVID-19 lockdown on the variation of organic aerosols: Insight into its molecular composition and oxidative potential

Abstract

To prevent the transmission of the novel coronavirus disease 2019 (COVID-19), China adopted nationwide lockdown measures on January 25, 2020, leading to an evident diminution in the observed air pollutants. To investigate the influence of the lockdown on atmospheric chemistry, the specific molecular composition, oxidative potential of organic aerosols (OAs) in PM_2.5 were studied using a high-resolution orbitrap mass spectrometry at a typical coal-combustion city, Linfen, in the North China Plain (NCP). The major air pollutants including PM_2.5, PM₁₀, SO₂, NO₂, and CO were observed to be diminished by 28.6–45.4%, while O₃ was augmented by 52.5% during the lockdown compared to those before the lockdown. A significant decrease of oxygen-containing (CHO) compounds (24.7%) associated with anthropogenic acids was observed during the lockdown, implying a reduction in fossil fuel combustion. The coal-burning related sulfur-containing organosulfates (CHOS-) and nitrooxy-sulfates (CHONS-) have also shown attenuated in both their relative abundances and anthropogenic/biogenic ratios. Amine/amide-like CHON + components have decreased by 27.6%, while nitro/nitrooxy-containing CHON- compounds have only decreased by 7.1%. Multi-source nitrogen-containing (CHN) compounds have shown a moderate elimination of 24.0%, while the identified high-condensed azaarenes have fallen from 17.7% to 14.7%, implying a potential reduction in the health risk of OAs during quarantine. The measurement of OAs’ oxidative potential through dithiothreitol (DTT) assay has confirmed that as it had dropped from 0.88 nmol min⁻¹ m⁻³ to 0.80 nmol min⁻¹ m⁻³. High correlations were observed between the abundance of OA subgroups with the concentration of PM_2.5 after the execution of the lockdown, suggesting a potential elevation in the contribution of organic components to the total PM_2.5 level. Our study provides insightful compositional and health-related information in the variation of OAs during the lockdown period and attests to the validity of joint-control strategy in controlling the level and health risks of numerous atmospheric pollutants.

1. Introduction

After the outbreak of Coronavirus disease 2019 (COVID-19) in China, a series of lockdown measures were rapidly implemented by the government to prevent the transmission of the virus. These measures include compulsory isolation, closing nonessential public facilities, restricting transportation and contacts. As a result, anthropogenic activities like industrial production and vehicle emission were highly reduced, and the level of air pollutants was observed to plummet correspondingly (Chen et al., 2020b; Chu et al., 2021). These lockdown measures were firstly conducted on January 23 in Wuhan, Hubei province where this novel pneumonia-like disease was firstly reported (Zheng et al., 2020). On January 25, the Chinese government launched the highest level of public health emergency response in 31 provinces in mainland China (Tu et al., 2020). During this period, it is not only observed the decreases in the absolute concentration of ambient fine particulate matter (PM_2.5, particles with aerodynamic diameter ≤2.5 μm) and other air pollutants, their emission sources, chemical composition and related health effects could also be influenced. The COVID-19 epidemic has provided a valuable opportunity to investigate the significance of controlling societal production and human activities on the mitigation of air pollution.

Several studies have investigated the change in the chemical composition of PM_2.5 before and during the lockdown. Chu et al. (2021) have found a significant concurrent reduction in PM_2.5 (19% in China excluding Hubei) and NO₂ (30% in China excluding Hubei) levels during the epidemic and support that controlling the level of NO_x can further reduce PM_2.5 pollution in China. Cui et al. (2020) have investigated the composition and source of PM_2.5 at a rural site between Beijing and Tianjin, finding a considerable reduction from dust, coal combustion and industrial production. Chen et al. (2020a) have studied the variation of the major components of non-refractory PM_2.5 (NR-PM_2.5) and gaseous pollutants in urban Shanghai, finding a 60% and 45% concentration decrease of nitrate and ammonium respectively during the quarantine period. An intensive formation of secondary aerosol was observed due to the increased level of oxidant in Shanghai, leading to haze periods during the epidemic. Zheng et al. (2020) have assessed the change of chemical composition, source contribution of PM_2.5 in Wuhan, the first city that adopted the lockdown measures, revealing an overall decrease of PM_2.5 species ranging from 0.85 μg m⁻³ (chloride) to 9.86 μg m⁻³ (nitrate) mainly from the suspension of industrial activities. These studies mainly focused on the macroscopic level of chemical species (e.g., organic carbon, elemental carbon, trace element) or inorganic aerosols (e.g., nitrate, ammonium, chloride, sulfate). Characterization of the organic fraction (as known as organic aerosol, OA) of PM_2.5 at the molecular level, however, was rarely reported. Related health effects like oxidative potential (OP) were also ambiguous due to the research gap. To obtain a more specific and comprehensive understanding of the influence of lockdown measures on the control of atmospheric pollutants during the epidemic, an intensive study on the variation of OAs in their molecular composition and oxidative potential is necessary.

In this study, 24 h of PM_2.5 samples were collected in a typical northern coal combustion city, Linfen, to study the influence of lockdown on the molecular composition and oxidative potential of OA. Two periods were selected in this study including before (January 1 to 24) and during the lockdown of the city (January 29 to February 9). High-resolution orbitrap mass spectrometry was used to characterize the specific molecular composition of OAs during these two periods. The oxidative potentials of the extracted OA components were ascertained by the dithiothreitol (DTT) assay. Correlation between the concentration of PM_2.5 and the molecular composition of OAs as well as its oxidative potential was also further discussed in this study to find out their potential relationships.

2. Materials and methods

2.1. Sample collection and pretreatment

PM_2.5 samples collection were conducted using quartz microfiber filters (QMA, 90 mm, Whatman International Ltd, UK) with a medium volume air sampler (ADS-2062E, AMAE (Shenzhen) Co., Ltd, China) for 24 h (∼140 m³ total volume) from January 1 to February 9, 2020. The sampling site was located at Mingjiang Town, Hongdong County, Linfen City, Shanxi Province (36°20′ N, 111°22′ E). The quartz filters used in this study were baked at 550 °C for 5 h in advance to eliminate possible contaminants. After sampling, all the filters were wrapped with aluminum foil and stored at a −80 °C freezer for further analysis.

The extraction procedure was performed with reference to earlier studies with modification (Wang et al., 2018). Briefly, a 4 cm² portion of the filter was cut into a tube, adding 2 mL acetonitrile-water (8/2, v/v) as the extraction solvent. Then the sample was extracted 3 times in an ultrasonic bath for 30 min. After pouring these extracts together, the mixture was filtered through a 0.22-μm Teflon syringe filter and blown with purified nitrogen gently to near dryness. Finally, the residual was redissolved in 1 mL acetonitrile-water (1/9, v/v) solvent for subsequent determination.

2.2. UHRMS analysis

Sample analysis was carried out with an ultrahigh-resolution mass spectrometer (Q Exactive Hybrid Quadrupole-Orbitrap Mass Spectrometer; Thermo Scientific, USA) coupled to an UHPLC system (Dionex UltiMate 3000, Thermo Scientific, USA). Analytes were separated using a Waters Acquity BEH C18 column (1.7 μm, 2.1 × 100 mm) with mobile phases consisting of (A) 2% acetonitrile and 0.04% formic acid in ultrapure water and (B) 2% ultrapure water in acetonitrile. Elution procedure was used in the gradient mode with a flow rate of 300 μL/min and the following gradients: start with 2% B for 1.5 min and firstly increased linearly to 20% B in 1 min and held for 5 min, then increased linearly to 98% B in 1 min and held for 2.5 min, back to 2% B in the next 5 min and hold for 1 min. The injection volume of each sample was 20 μL and blank filter samples were also analyzed for background subtraction.

The spray voltage of the ESI source equipped for Q-Exactive MS was −3.3 kV in the negative ion mode (ESI-) and 4.0 kV in the positive ion mode (ESI+). The sheath gas and auxiliary gas were 45 psi and 10 psi respectively. 320 °C capillary temperature and 55 S-lens RF level were adopted. The scanning range was set to m/z 80–800 with a resolution of 140,000 at m/z 200. The MS system was pre-calibrated using sodium acetate solution to ensure the mass accuracy of the measurements was less than 3 ppm.

2.3. Data processing

The obtained LC-MS spectrums were processed with Xcalibur software (V4.3.7, Thermo Scientific, USA) to subtract the blank sample for the elimination of background peaks from contaminants and other interferences. Then the data were processed by MZmine (V2.53) software, which enables several LC-MS data processing functions (Lin et al., 2012; Pluskal et al., 2010; Tao et al., 2014; Wang et al., 2017). The processing workflow and parameters were set with reference to previous studies and were described in the supporting information (Wang et al., 2017, 2020b). It should be noted that the compositional elemental of these formulas was constrained as C_1∼40H_0∼100O_0∼40N_0∼5S_0∼2 and that m/z tolerance was 3 ppm for both ESI modes. Several indicators were adopted to illustrate the characters of OA subgroups including double bond equivalent (DBE), modified aromaticity index (AI_mod), aromaticity equivalent (Xc) and Kendrick mass defect (KMD). Their specific calculation was presented in the supporting information.

2.4. DTT assay

The oxidative potential of the extracted PM_2.5 samples was evaluated by the DTT assay in reference to the study of Ma et al. (2018) with a slight modification. Briefly, 20 μL of each extracted and filtered sample in the solution of acetonitrile-water (1/9, v/v) were dispensed into a 1.5 mL tube. 920 μL of potassium phosphate buffer (pH = 7.4) and 50 μL of 0.5 mM DTT were added into the tube successively. The tube was then placed into a shaker with the temperature-controlled at 37 °C (simulating the human environment), shaking for 90 min to ensure a better correlation between DTT consumption and reaction time. After spiking 100 μL of 1.0 mM 5,5′-dithiobis-2nitrobenzoic acid (DTNB) into the mixture, the residual DTT was quantified immediately by measuring its light absorbance at 412 nm using an ultraviolet–visible (UV–Vis) spectrophotometer (Hewlett, CA, USA). The DTT consumption of each extracted PM_2.5 sample was blank corrected and then normalized by the volume of air sampled (DTT_V, in units of nmol·min⁻¹ m⁻³) with the following equation:

$$DT{T}_V=\frac{R_{DTT}(\%)\times n_{DTT}\left(nmol\right)}{t\left(min\right)\times Airvolumn\left(m^{-3}\right)}$$ (1)

where $R_{DTT}(\%)$ is the consumption of DTT in percentage, $n_{DTT}\left(nmol\right)$ is the total amount of DTT that added, $t\left(min\right)$ is the reaction time, and $Airvolumn\left(m^{-3}\right)$ is the corresponding sampling volume of filters used.

2.5. Statistical analysis

The Pearson correlation analysis and the multiple linear regression model were carried out in the corresponding statistical analyses. A p value less than 0.05 was considered statistically significant. All the statistics were implemented using the Statistical Program for Social Sciences (SPSS, Version 11.0, IBM, SPSS Inc.).

3. Result and discussion

3.1. Changes in the level of major air pollutants

Similar to those in other cities in China, the air pollutants in Linfen also showed a downward trend during the lockdown. Fig. 1 showed the temporal variations in the concentration of major air pollutants and their variation rates. It is clear that the mean concentrations of PM_2.5, PM₁₀, NO₂, SO₂ and CO have decreased by 28.57%, 35.85%, 45.39%, 35.20%, 38.58% respectively, while that of O₃ increased by 52.49%. The amplitude of the variation was greater than the national mean change rates of PM_2.5 (−20.1%), PM₁₀ (−15.0%), O₃ (21.9%), NO₂ (−35.7%), SO₂ (−20.1%) and CO (−13.0%), indicating a strong influence of the lockdown measures on the alleviation of air pollution in Linfen (Zheng et al., 2020). Among all these 6 kinds of pollutants, it is only ozone (O₃) that showed a conversely upward trend after the implementation of quarantine measures. The ascent of O₃ could be attributed to the increase in its production rate, as the reduction of PM_2.5 could impair its role as a scavenger of hydroperoxyl and NO_x radicals that would otherwise produce ozone (Dang and Liao, 2019; Li et al., 2019a, 2019b). A similarly elevated ozone level was also observed in other cities and areas like Wuhan, Shanghai, Barcelona, Baghdad, and eastern India during the lockdown period (Chen et al., 2020a; Hashim et al., 2021; Sharma et al., 2020; Tobías et al., 2020; Zheng et al., 2020). A higher decrease rate of PM₁₀ than that of PM_2.5 was observed, indicating a strong reduction of primary aerosol compared to the secondary aerosol as the former showed a higher contribution to PM₁₀ than PM_2.5 (Seinfeld and Pandis, 2016). A significant decrease of CO (p < 0.01) also verified this result as CO is often used as a marker of primary emission considering its relatively long lifetime against oxidation by OH radicals (DeCarlo et al., 2010). A remarkable decrease of NO₂ (p < 0.001) was observed, with its mean concentration declining from 59.10 ± 14.36 to 32.27 ± 6.73 μg m⁻³, which may be related to the marked drop in traffic intensity (Wang et al., 2020a). A concurrent decrease in PM_2.5 and NO₂ during the Covid-19 epidemic was revealed regarding their temporal and spatial variations in China, implying the potential effectiveness of controlling NO_x in reducing PM_2.5 pollution (Chu et al., 2021). Similarly, the level of SO₂ also showed to decrease by 35.20% during the lockdown period. As an important indicator of coal combustion, the shrink of SO₂ may imply a potential reduction of coal-burning under the COVID-19 epidemic control measures. However, considering that coal combustion is the basic energy supply for electricity and central heating and contributed 57.7% to the total energy supply in China during 2019, the decrease rate of SO₂ was not as significant as CO and NO₂ (Statistics, 2020). A similar result was also observed in Wuhan during the covid-19 lockdown period, as the SO₂ decreased from 14–18 before the lockdown to 10–13 μg m⁻³ during the lockdown period (Shi and Brasseur, 2020).

Fig. 1.

Time series of the concentration of PM_2.5, PM₁₀, NO₂, SO₂, O₃ and CO in Linfen during the measurement periods (a) and their variation in absolute value and percentage (b). The light orange and light green shaded areas represent two periods: before and during the lockdown period. Different numbers of * represent different significant levels: *p < 0.05, **p < 0,01, ***p < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.2. Variation in the molecular composition of OAs

A total of 5338 and 4914 molecule formulas were obtained from samples before lockdown in positive and negative ESI respectively, while those numbers during lockdown were 5206 and 5028 (Table S1). The mean molecule weight (MW) showed a slight decrease during the lockdown period in both -ESI and +ESI. O/C and H/C ratios, combined with other indicators like DBE and AI_mod, were widely used to study the molecular compositional characters of OAs (Mo et al., 2018). In this study, the OA molecules showed a small increase in the level of O/C (0.41/0.40), H/C (1.23/1.21) ratios, and a decrease in the level of DBE (7.64/7.92) and AI_mod (0.39/0.40) during quarantine. This indicates that OA molecules tend to be more saturated and oxidated during the lockdown period, which may relate to the increase of the oxidation capacity (NO_x and O₃) with the rise of ozone during that period.

By classifying the obtained molecule formulas into different subgroups, detailed compositional information could be attained. Molecules collected from ESI+ were subdivided into five groups as CHO, CHON, CHONS, CHOS, CHN, while molecules obtained from ESI- fell into four classes as CHO, CHON, CHONS, CHOS (Figure S1). Due to the difference in their compositional functional groups, molecules showed different ionization tendencies under different ionization modes. Deprotonated compounds with carboxy or carbonyl groups usually showed high abundance in ESI-, while protonated compounds like amines were more sensitive in ESI+ (Lin et al., 2012). Specific molecular compositional information of each OA subgroup before and during lockdown periods was shown in Table 1 . As the most widely observed component in OAs, the CHO compounds comprise a substantial proportion before and during the lockdown periods. Former studies have found that most of the CHO compounds consisted of carboxylic acids, oxygenated polycyclic aromatic hydrocarbons containing deprotonatable carbonyl, carboxyl functional groups, which tend to occur preferentially in negative ionization mode (Lin et al., 2012, 2016). The CHON compounds should be discussed separately as most of the CHON + compounds (>64%) have an O/N ratio lower than 3, which were assigned to have reduced nitrogen-containing functional groups like amines, while 71% of the CHON + compounds showed an O/N ratio higher than 3, which may associate with nitro (-NO₂) or nitrooxy (-ONO₂) groups in these formulas (Mo et al., 2018; Wang et al., 2018). Most of the CHOS- (87%–92%) compounds and CHONS- (66–78%) compounds showed O/S ≥ 4 or O/N ≥ 4, indicating a potential sulfonic acid group (-OSO₃H) and/or nitric acid group (-ONO₂) (O'Brien et al., 2013). Considering that, these molecules could be potentially identified as organosulfates (OSs) or nitrooxy-organosulfates (Wang et al., 2017).

Table 1.

The number of formulas and the average values of molecular weight (MW), oxygen-to-carbon ratio (O/C), hydrogen-to-carbon ratio (H/C), double-bond-equivalent (DBE) and modified aromaticity index (AI_mod) of each subgroup in PM_2.5 before and during quarantine.

Period	Elemental composition	Number of formulas	MW	O/C	H/C	DBE	AImod
Before lockdown	CHO+	1537	288.72	0.30	1.07	8.89	0.42
	CHON+	2465	281.11	0.29	1.23	8.93	0.40
	CHONS+	1007	317.18	0.61	1.40	6.40	0.21
	CHOS+	171	210.50	0.24	1.22	6.01	0.34
	CHN+	357	199.70	0.00	1.22	7.42	0.52
	CHO-	1489	326.59	0.35	1.13	9.26	0.43
	CHON-	1371	296.49	0.34	1.04	9.46	0.63
	CHONS-	1307	345.78	0.67	1.45	6.74	0.24
	CHOS-	743	323.12	0.65	1.14	7.75	0.36
During lockdown	CHO+	1419	285.91	0.30	1.02	9.21	0.45
	CHON+	2428	283.03	0.29	1.22	8.71	0.41
	CHONS+	949	315.86	0.62	1.58	5.93	0.19
	CHOS+	187	213.32	0.24	1.24	6.15	0.34
	CHN+	384	192.82	0.00	1.23	7.00	0.51
	CHO-	1560	332.71	0.37	1.12	9.53	0.43
	CHON-	1541	299.91	0.35	1.06	9.33	0.62
	CHONS-	1373	348.61	0.67	1.45	6.14	0.24
	CHOS-	688	310.88	0.69	1.22	6.51	0.31

Fig. 2(a) showed the temporal change of the abundance for each subgroup and their proportion in percentage. An evident decreasing tendency for each subgroup was identified during the lockdown period, which is consistent with the variation of PM_2.5. Fig. 2(b) showed the variation of each subgroup in absolute abundance and percentage. It is clear that all the subgroups showed a declining tendency in both ESI+ and ESI- after the execution of the lockdown measures. However, the changing rates for each subgroup vary from −7.09% to −31.95%. As a basic component of the oxygenated organic aerosol, CHO compounds have a mean decreasing rate of 24.73%. Among these oxygenated molecules, organic acid was a good indicator reflecting the source of secondary organic aerosols (SOAs) (Chan et al., 2011; Nozière et al., 2010). By identifying the potential composition, organic acids could be used to illustrate the source of SOAs from anthropogenic or biogenic origins. Figure S2 showed the relative intensity (RI) of identified organic acids derived from the anthropogenic/biogenic source and the ratios of adipic acid to azelaic acid (C₆/C₉) and phthalic acid to azelaic acid (Ph/C₉) respectively. These ratios have been widely adopted to reflect the relative contributions of anthropogenic versus biogenic SOAs (Hansen et al., 2014; Yu et al., 2021). By comparison, it is clear that both the abundance of identified acids and the anthropogenic/biogenic source ratios decreased during the lockdown period, indicating the reduction of acidic anthropogenic source SOAs. Meng et al. (2021) have investigated the impact of lockdown on the level and source of oxalic acid, revealing a 29% decrease of biomass burning related oxalic acid during the lockdown.

Fig. 2.

Time series of the abundance of different molecular compositional subgroups obtained from both positive and negative ESI mode (a) and their variation in absolute value and percentage (b). The pie charts show the relative abundance of different subgroups before and during lockdown periods. The light orange and light green shaded areas represent two periods: before the lockdown period and during the lockdown period. Different numbers of * represent different significant levels: *p < 0.05, **p < 0,01, ***p < 0.001. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

The CHON + compounds may represent a large proportion of oxidized amines/amides with relatively higher AI_mod (or DBE) value and lower O/N ratio (<3) (Wang et al., 2017), while the CHON- molecules were mainly organonitrates with higher O/N (≥3) (Wu et al., 2019). A significant decrease of CHON + compounds (27.57%, p < 0.01) may imply a large reduction of amine/amide-like substances, which could be associated with the reduction of direct emissions of industrial and tobacco smoke (Ge et al., 2011). Different from CHON + compounds and other subgroups, CHON- molecules showed a lower decreasing rate of only 7.09%, making its relative proportion increase by 3.4% (pie chart in Fig. 2 (a)). This may be related to the weaker nitrate radical chemistry caused by the reduction of NO_x, and a similar change was also observed during the 2014 Asia-Pacific Economic Cooperation (APEC) summit period (Mo et al., 2018).

The CHOS (organosulfates) and CHONS (nitrooxy-organosulfates) molecules were tended to be ionized in negative ESI, leading to a lower percentage in ESI+ (∼10%) and a high percentage in ESI- (∼40%) among all the molecules. Earlier studies have proven that they are formed from acid-catalyzed and/or NO₃-initialed oxidation reactions that are related to the combustion of fossil fuel (especially coal) or biogenetic sources (Cai et al., 2020). As a result, the remarkable decrease of NO₂ and SO₂ (Fig. 1) may be highly correlated to the reduction of OSs and nitrooxy-OSs, as both of these components were originated from the combustion of fossil fuel (Mead et al., 2015). Figure S5 showed the relative intensity of identified 19 organosulfates from anthropogenic (A) and biogenic (B) sources as well as the ratio of the total intensity between these sources. Both the intensity of anthropogenic OSs/nitrooxy OSs and the A/B ratio decreased in the lockdown period, suggesting a diminution of sulfur-containing OAs caused by the reduction of human activities. Similarly, the abundance of organosulfates was also observed to decline during the APEC period, with a series of emission control measures and a significant decline in sulfur dioxide (Mo et al., 2018).

The potential identification of CHN compounds shown in Figure S4 indicate that they were mostly azaarenes (e.g., quinoline and its homologs) or aromatic amines (e.g., C₉H₁₃N, dimethyl toluidine) originated from the combustion of biomass/fossil fuels (e.g., coal, oil) that take place during household heating, cooking, municipal waste incineration, transportation, industrial/power production, farming, and natural fires (Bandowe et al., 2016; Kuang et al., 2018; Wang et al., 2018). Compared to other species, CHN compounds showed a moderate-level decrease of 23.95% during the lockdown, making its relative proportion increase from 15.7% to 16.1%. This could be partially due to the combined effect of the above-mentioned sources. While some of these human activities could decrease dramatically during the lockdown, others, such as home cooking and household heating, might increase.

Besides revealing the macroscopic change of chemical species in OAs, understanding their specific variation in molecule level was also crucial. Fig. 3 showed the Van Krevelen (VK) diagram of OA molecules that were lost/decreased (b) after the implementation of the lockdown measures. By constructing the VK diagram with the H/C and O/C of each molecule, one can get the categorical distribution of each subgroup. In ESI-, most of the eliminated molecules were sulfur-containing CHOS and CHONS compounds, accounting for 65% of the total number of molecules. As described above, these molecules were mainly composed of OSs or nitrooxy-OSs that were highly related to the combustion of fossil fuel. The vast reduction of these compounds, therefore, may imply a potential deduction of fossil fuel combustion. Besides that, these S-containing molecules were mainly distributed in the range of O/C 0.25–0.75, H/C 0.3–1.2, similar to the range that was recognized as the unique coal-combustion-produced molecules located in the area of O/C 0–0.6, H/C 0.3–1.0 (Tang et al., 2020). These results have further verified that the lost molecules obtained from ESI- were mainly caused by the reduction of coal combustion. An extensive loss of CHONS compounds (27.7%) was also noticed in ESI+, showing a broad distribution compared to other species. Considering the remarkable decrease of their DBE and AI_mod level (Table 1), these molecules could be attributed to aromatics-containing reduced sulfur that is originated from fossil fuel combustion as reported in Shanghai and Wilmington (Mead et al., 2015; Wang et al., 2017). The decreased/lost CHON + molecules mainly concentrated in the low-oxidated area with an O/C less than 0.5, indicating that they were CHON + compounds containing more condensed nitrogen atoms, which could be produced from a minor pyrolytic and oxidative processing (e.g., smoldering burning) of N-heterocycle compounds (e.g., imidazole) (Lin et al., 2012). The decrease of these molecules implied a potential diminution of biomass burning during the quarantine period. Similar to CHON + compounds, most of the disappeared CHO + compounds (∼79%) also showed an O/C less than 0.5 and could be considered as low oxidated aliphatic/aromatic compounds (Wang et al., 2018). Wang et al. (2017) have investigated the molecular composition of OAs in Shanghai and suggested that the CHO + molecules were likely to be associated with the direct emission of biomass burning. A reduction of these compounds may suggest the mitigation of biomass burning during quarantine. As there is no oxygen atom in the CHN compounds, the variation of CHN molecules was shown separately in Figure S5. Compared to those molecules before lockdown, which shows a relatively higher DBE (7.42), the CHN compounds during lockdown were observed to be more saturated (DBE = 7.00). The fraction of naphthalene/anthracene/pyrene-core structural aromatic CHN molecules has decreased from 56.6% to 42.9%. CHN compounds with 12–23 DBE and 15–29 carbon atoms were regarded as highly condensed azaarenes molecules (above the dashed line in Figure S5), which were reported to have strong mutagenesis and potentially carcinogenic (Wang et al., 2018), having decreased from 17.7% to 14.7% after the implement of lockdown measures. As the markers of coal combustion, the decrease of condensed azaarenes verified the reduction of coal-burning and may further indicate a potential decline in the health risk of PM_2.5 during the quarantine period (Alves et al., 2017; Bandowe et al., 2016).

Fig. 3.

Van Krevelen diagram of OA molecules obtained from ESI- (left) and ESI+ (right) that were lost/decreased after the implementation of lockdown measures. Different subgroups were shown by different colors and shapes. The pie chart represents the number proportion of each subgroup. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.3. Oxidative potential and correlations

To better understand the change in the health risk of OAs, together with the relationships between that with their molecular composition, the oxidative potential (OP) of the extracted components were also evaluated. The OP represents the ability of particular matters to generate reactive oxygen species (ROS), which has been suggested as a key mechanism underlying the toxic effects of exposure to OAs (Delfino et al., 2011; Esposito et al., 2014). A variety of methods were developed to evaluate the OP of particular matters, and the DTT assay was most adopted due to its high sensitivity to a variety of PM components (e.g., quinones, transition metals and HULICs) (Bates et al., 2019; Charrier and Anastasio, 2012; Verma et al., 2015). In this study, the DTT assay was used to evaluate the OP of OAs before and during lockdown periods in order to investigate the correlation of that with the molecular compositional subgroups of OAs.

Table 2 summarized the Pearson correlation coefficients of the abundance for each subgroup with the concentration of gaseous pollutants and DTT. It could be seen that all the species showed an amplified correlation to the content of PM_2.5 during the lockdown period in comparison with that before the lockdown. Considering the enhanced secondary formation of OAs associated with the increased O₃ level, this result may imply a potential increase in the contribution of organic components to the total concentration of PM_2.5. A similar increased share of organic carbon was also observed in Wuhan and Shanghai (Chen et al., 2020a; Gu et al., 2021; Zheng et al., 2020). Among all these organic subgroups, significant positive correlations were found between CHONS- (p = 0.700) and the concentration of PM_2.5 as well as between CHOS- (p = 0.825, r < 0.05) and PM_2.5 during quarantine, which was much higher than that before the quarantine. As a tracer of secondary formatted OAs, OSs and nitrooxy-OSs originated from the oxidation of biogenic VOCs (e.g., isoprene and α-/β-pinene) as well as anthropogenic PAHs (Ning et al., 2019). The increment in their correlation coefficients may imply an elevated contribution of SOAs to the total concentration of PM_2.5, due to the enhanced atmospheric oxidizing capacity. High associations of the level of ozone with CHONS- (p = 0.847, r < 0.05) and with CHOS- (p = 0.935, r < 0.01) also verified the boosted OSs and nitrooxy-OSs levels due to the increase of O₃ during the lockdown. A former study also signified the enhanced photochemical formation of secondary SOAs during the COVID-19 lockdown in northern China (Meng et al., 2021). As mentioned before, PM₁₀ and CO were more sensitive to the POAs (DeCarlo et al., 2010; Seinfeld and Pandis, 2016). A relatively higher association of the organic species with the level of PM₁₀ and CO than that with PM_2.5 may indicate a higher level of direct emission caused by human activities before lockdown.

Table 2.

The Pearson correlation coefficients between the abundance of each subgroup with the daily concentration of gaseous pollutants and DTT activity. Different numbers of * represent different significant levels: *p < 0.05, **p < 0,01, ***p < 0.001.

		CHO+	CHON+	CHONS+	CHOS+	CHN+	CHO-	CHON-	CHONS-	CHOS-
PM2.5	Before lockdown	0.325	−0.006	0.368	−0.534*	−0.292	0.131	0.217	0.453	0.568*
	During lockdown	0.388	0.637	0.510	0.009	0.330	0.432	0.371	0.700	0.825*
	Total	0.420	0.217	0.420	−0.399	−0.071	0.288	0.259	0.551**	0.609**
PM10	Before lockdown	0.329	0.028	0.364	−0.514*	−0.265	0.159	0.243	0.456	0.541*
	During lockdown	0.373	0.644	0.474	0.043	0.358	0.413	0.383	0.672	0.779
	Total	0.451*	0.286	0.416	−0.343	−0.004	0.344	0.281	0.580**	0.591**
O3	Before lockdown	0.132	−0.123	0.018	0.244	−0.067	0.195	0.353	0.044	0.283
	During lockdown	0.772	0.921**	0.860*	0.427	0.727	0.805	0.668	0.935**	0.845*
	Total	0.081	−0.115	0.111	0.138	−0.056	0.107	0.342	−0.004	0.172
NO2	Before lockdown	0.148	0.231	0.216	−0.393	0.121	0.014	−0.047	0.032	−0.195
	During lockdown	0.505	0.549	0.664	0.131	0.301	0.640	0.509	0.584	0.603
	Total	0.421	0.504	0.323	−0.124	0.348	0.365	0.092	0.375	0.073
SO2	Before lockdown	0.464	0.321	0.515*	−0.393	0.192	0.296	0.398	0.241	0.211
	During lockdown	0.463	0.342	0.485	0.464	0.403	0.557	0.474	0.237	0.009
	Total	0.553**	0.452*	0.537*	−0.207	0.340	0.460*	0.422	0.377	0.271
CO	Before lockdown	0.547*	0.161	0.674**	−0.711**	−0.090	0.228	0.263	0.356	0.401
	During lockdown	−0.016	−0.070	0.153	−0.766	−0.498	0.078	0.017	0.179	0.570
	Total	0.577**	0.372	0.594**	−0.453*	0.098	0.403	0.251	0.507*	0.4712*
DTT	Before lockdown	−0.300	0.046	−0.349	0.405	0.053	0.153	0.226	0.313	0.286
	During lockdown	0.950***	0.855*	0.795	0.501	0.858*	0.903*	0.921**	0.744	0.486
	Total	0.140	0.295	−0.017	0.394	0.293	0.394	0.407	0.435*	0.313

Similar to the interaction between the species abundance and the concentration of PM_2.5, an increased relevance between these OA subgroups in PM_2.5 with its oxidative potential could also be ascertained from Table 2, highlighting their determining role in the oxidative potential of PM_2.5 during the lockdown. It is usually accepted that the DTT assay is highly sensitive to both organic species and transition metals in PM_2.5 (Fang et al., 2016). It is proposed that the reduction of human activity could mitigate the level of metal elements in the atmospheric environment, thus emphasizing the organic carbon-related ROS production during the lockdown period (Cui et al., 2020). As shown in Table 2, among the organic species, it is molecules appertained to CHO+ (r = 0.950, p < 0.001) and CHON- (r = 0.921, p < 0.01) groups that showed the two highest correlations with the DTT activities. Fig. 4 showed the temporal variation in the abundance of these two subgroups, DTT activities, and their linear correlations. A significant linear association was found between the abundance of CHO+ and CHOS- with the level of DTT activity during the lockdown, suggesting that these two components may play a vital role in the formation of the oxidative potential of PM_2.5 in that period. Many previous studies also found high correlations between the organic species of PM_2.5, especially water-soluble organic carbon (WSOC), with its oxidative potential. Nevertheless, few of them have clarified the specific chemical species that were dominant in the ROS production (Chen et al., 2019; Li et al., 2019c). Ma et al. (2018) have identified the individual species of water-soluble HULIS associated with the oxidative potential of PM_2.5 and found a series of aromatic acids (CHO) and nitrophenol analogs (CHON). The results of this study provide further support that directly emitted CHO molecules and nitrooxy functional CHON compounds were highly related to the oxidative potential of PM_2.5. It should be noted that the results may be limited due to the small number of samples collected during the lockdown. In-depth research with an increased sample should be conducted in the future to further investigate the correlation between the atmospheric species and their oxidative potential during the lockdown period.

Fig. 4.

Time series of the DTT activities, abundances of CHO+ and CHON- of OA samples and their linear correlation during lockdown period.

4. Conclusions

Different from previous studies that focused mainly on the macroscopic level of chemical species, this research investigated the specific molecular composition of OAs and the relation between these species and the OP of PM_2.5 before and during the COVID-19 lockdown period. The results showed a relatively higher reduction of atmospheric pollutants including PM_2.5, PM₁₀, SO₂, NO₂, and CO in Linfen, China compared to the national mean level during the lockdown period. Similar to other cities and regions, the level of O₃ in Linfen increased by 52.5% synchronously, which in turn enhanced the atmospheric oxidizing capacity. All the subgroups in OAs showed a decline in their abundance ranging from −7.09% to −32.59% during the lockdown period. Sulfur-containing CHOS and CHONS compounds showed the highest decreasing rate of 27.85%, indicating a vast reduction of fossil fuel combustion caused by the lockdown measures. Compared to the large decrease of oxygen amine CHON + compounds (27.57%), nitro (-NO₂) or nitrooxy (-ONO₂) functional CHON- compounds have shown a relatively lower decrease rate of 7.09%, which may be associated with the weaker nitrate radical chemistry caused by the diminution of NO_x. As high-condensed azaarenes and aromatic amines take a vital position in the CHN compounds, a relative decrease of 23.95% may imply the decline in the health risk of PM_2.5. Change in the oxidative potential of OAs has verified the result as it has declined by 8.76% during the lockdown period. Besides that, a high correlation was observed between the abundance of CHO+ (r = 0.950, p < 0.001) and the level of DTT activities during the lockdown, so was CHOS- (r = 0.921, p < 0.001). The likelihood that these components become the predominant contributor to the oxidative potential of PM_2.5 in the quarantine period is thus rather high. The overall correlation between secondary-formatted OSs and the concentration of PM_2.5 as well as that between nitrooxy-OSs and PM_2.5 have shown to increase after the implementation of the lockdown measures, while the correlation between these organic species and the direct emitted PM₁₀ and CO has decreased, suggesting an enhanced secondary formation of OAs to the total concentration of PM_2.5 during the lockdown period. The above observations manifest the validity of the joint-control strategy during the COVID-19 lockdown period in controlling the anthropogenetic emission thus reducing the level and health risks of numerous atmospheric pollutants. Enhanced secondary formation of OAs associated with the high atmospheric oxidation capacity should be taken into consideration in the future to further reduce the pollution level.

Credit author statement

Wei Wang: Writing original draft, Data curation, Formal analysis, Methodology. Yanhao Zhang: Data curation, Writing-review & editing, Methodology. Guodong Cao: Methodology, Resources. Yuanyuan Song: Resources. Jing Zhang: Resources. Ruijin Li: Resources. Lifang Zhao: Resources. Chuan Dong: Resources. Zongwei Cai: Supervision, Manuscript editing and revising, Resources, Project administration.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.