Impact of quarantine measures on chemical compositions of PM2.5 during the COVID-19 epidemic in Shanghai, China

The COVID-19 epidemic broke out in Wuhan, Hubei in December 2019 and in January 2020 and was later transmitted to the entire country. Quarantine measures during Chinese New Year effectively alleviated the spread of the epidemic, but they simultaneously resulted in a decline in anthropogenic emissions from industry, transportation, and import and export of goods. Herein, we present the major chemical composition of non-refractory PM2.5 (NR-PM2.5) and the concentrations of gaseous pollutants in an urban site in Shanghai before and during the quarantine period of the COVID-19 epidemic, which was Jan. 8―23 and Jan. 24―Feb. 8, respectively. The observed results show that the reduction in PM2.5 can be mainly attributed to decreasing concentrations of nitrate and primary aerosols. Nitrate accounted for 37% of NR-PM2.5 before the quarantine period when there was no emission reduction. During the quarantine period, the nitrate concentration decreased by approximately 60%, which is attributed to a reduction in the NOx concentration. Ammonium, as the main balancing cation, showed an approximately 45% simultaneous decrease in concentration. The concentrations of chloride and hydrocarbon-like organic aerosols from primary emissions also declined due to limited human activities. By contrast, sulphate and oxygenated organic aerosols showed a slight decrease in concentration, with their contributions increasing to 27% and 18%, respectively, during the quarantine period, which resulted in two pollution episodes with PM2.5 exceeding 100 μg/m3. This study provides a better understanding of the impact of quarantine measures on variations of the PM2.5 concentration and chemical compositions. Atmospheric oxidation capacities based on the oxidant (Ox = O3 + NO2) and oxidation ratios have been discussed for elucidating the source and formation of haze in an environment with lower anthropogenic emissions. With increasing contribution of secondary aerosols, lower NOx and nitrate concentrations did not completely avoid haze in Shanghai during the epidemic.

The COVID-19 epidemic broke out in Wuhan, Hubei in December 2019 and in January 2020 and was later transmitted to the entire country. Quarantine measures during Chinese New Year effectively alleviated the spread of the epidemic, but they simultaneously resulted in a decline in anthropogenic emissions from industry, transportation, and import and export of goods. Herein, we present the major chemical composition of non-refractory PM 2.5 (NR-PM 2.5 ) and the concentrations of gaseous pollutants in an urban site in Shanghai before and during the quarantine period of the COVID-19 epidemic, which was Jan. 8-23 and Jan. 24-Feb. 8, respectively. The observed results show that the reduction in PM 2.5 can be mainly attributed to decreasing concentrations of nitrate and primary aerosols. Nitrate accounted for 37% of NR-PM 2.5 before the quarantine period when there was no emission reduction. During the quarantine period, the nitrate concentration decreased by approximately 60%, which is attributed to a reduction in the NO x concentration. Ammonium, as the main balancing cation, showed an approximately 45% simultaneous decrease in concentration. The concentrations of chloride and hydrocarbon-like organic aerosols from primary emissions also declined due to limited human activities. By contrast, sulphate and oxygenated organic aerosols showed a slight decrease in concentration, with their contributions increasing to 27% and 18%, respectively, during the quarantine period, which resulted in two pollution episodes with PM 2.5 exceeding 100 μg/m 3 . This study provides a better understanding of the impact of quarantine measures on variations of the PM 2.5 concentration and chemical compositions. Atmospheric oxidation capacities based on the oxidant (O x = O 3 + NO 2 ) and oxidation ratios have been discussed for elucidating the source and

Introduction
In December 2019 and in January 2020, a cluster of COVID-19 cases was reported in Wuhan, Hubei, China (Li et al., 2020). COVID-19 is confirmed to transmit between humans through respiratory droplets. Liu et al. (2020) recently found that RNA of COVID-19 was detectable in aerosols in two Wuhan hospitals. To slow down the spread of COVID-19, the Chinese government issued strict quarantine measures (lockdown) in Wuhan on Jan. 23, 2020 and enacted corresponding restrictive measures in different regions according to the level of the epidemic. These measures included reducing public transportation and closing schools and business centers. These quarantine measures effectively alleviated the spread of the COVID-19 epidemic in China  but also greatly suppressed anthropogenic emissions Zhang et al., 2020b). A study in Barcelona showed that quarantine measures led to an improvement in air quality (Tobías et al., 2020). However, emission reduction did not prevent the occurrence of severe haze in China, and a plausible explanation for this is derived from a model study that showed a large emission reduction in transportation and a slight reduction in industrial production are insufficient to prevent haze in China in the winter .
Air quality in Shanghai has experienced continuous, substantial improvement through the reduction of industrial emissions (Han et al., 2018). Further PM 2.5 reduction appears to be challenging, and severe haze still occasionally occurs in winter (Sun et al., 2019). Variation in the pollution levels during Chinese New Year (abbr. CNY, also known as Spring Festival) has attracted great attention because of the complex effect of anthropogenic activities on emissions. Around the CNY holiday, great amounts of migrant residents return to their hometowns to reunite with their families, which causes extremely high traffic flows. Meanwhile, anthropogenic emissions reduce to some extent because of the temporary and partial closure of factories. The concentration of pollutants usually drops during CNY and rapidly recovers during the post-CNY period (Huang et al., 2012a). Additionally, intensive fireworks activity can cause severe haze during CNY. Thus, a strict fireworks ban enacted in 2014 has helped to reduce fireworks emission during CNY .
Because of the joint effect of the quarantine measures and CNY holiday, emission reduction due to suppressed human activities lasted sufficiently longer in 2020, and its impact on the air quality was more distinct. In this study, we present the major chemical composition of PM 2.5 and the concentrations of gaseous pollutants in urban Shanghai from Jan. 8 to Feb. 9 in 2020. Chloride and hydrocarbon-like organic aerosols from primary emissions can be directly influenced by the suppression of human activities, while secondary aerosols including nitrate, sulphate, ammonium, and oxygenated organic aerosols are formed through gas-to-particle transformation, which is related to their corresponding precursors and the atmospheric oxidation capacity. We investigate variations in the PM 2.5 chemical composition before, during and after the CNY holiday, which can help to elucidate the impact of quarantine measures on air quality. Atmospheric oxidation capacities based on the oxidant (O x = O 3 + NO 2 ) and oxidation ratios have been discussed for elucidating the source and formation of haze in the environment in Shanghai with lower anthropogenic emissions.

Methods
A Time of Flight-Aerosol Chemical Speciation Monitor (Tof-ACSM, Aerodyne Inc.) was employed to measure the concentration of non-refractory PM 2.5 (NR-PM 2.5 ). The Tof-ACSM was placed on a 7-story building at the Jiangwan Campus of Fudan University (31.338°N, 121.511°E). The Tof-ACSM was equipped with a PM 2.5 aerodynamic lens and a standard vaporizer heated to 600°C (Xu et al., 2015). Since the Tof-ACSM was only able to measure NR components evaporable at 600°C, refractory components with higher melting points (CaSO 4 , CaCl 2 , MgCl 2 , NaCl, etc.) were not detectable. This might cause an underestimation of the concentration of sulphate and chloride that originated from soil dust and/or sea spay. The hourly concentrations of chloride (Chl), nitrate (NO 3 ), sulphate (SO 4 ), ammonium (NH 4 ), and oxygenated and hydrocarbon-like organic aerosol (OOA and HOA) were derived from the Tof-ACSM dataset (Sun et al., 2012) and used for further investigation. PM 2.5 , SO 2 , NO x , CO, and O 3 were monitored in an air quality station, Xinjiangwan station, which is 500 m away from the Tof-ACSM site. The site is located north of downtown Shanghai, on the border of the Yangpu District and Baoshan District, and close to the junction of the Huangpu River and Yangtze River (Fig. S1). Nitrogen and sulphur oxidation ratios were calculated from the molar ratios of the particulate concentration and the sum of gaseous and particulate concentrations: The study was divided into three phases ( Fig. 1) according to the level of quarantine measures in Shanghai. Phase 1 (P1) is from Jan. 8 to 23 when there were no quarantine measures. The lockdown in Wuhan and in other cities of Hubei started on Jan. 23 and the CNY holiday was extended to Feb. 2 in Shanghai. Phase 2 (P2) represents this period of the extended CNY holiday with strict quarantine measures. People in Shanghai gradually returned to normal life since Feb. 3. Because of the delayed return of migrant residents, the encouragement to work from home, and other quarantine measures, Phase 3 (P3), from Feb. 3 to 9, represents the partial resumption period. There were four typical PM 2.5 pollution episodes identified during the study (Fig. 1)-EP1 and EP2 for P1, Ep3 for P2, and EP4 for P3.

Results and discussion
The mean concentration of PM 2.5 during P1, P2, and P3 was 60.9 ± 42.4, 41.2 ± 26.3, and 34.0 ± 23.7 μg/m 3 , respectively (Fig. 3d). The PM 2.5 concentration decreased by 33% and 44% during P2 and P3, respectively. By contrast, the SO 2 and CO concentrations slightly decreased by 15% and 22%, respectively, during P3. CO is often used as a primary emission tracer because of its relatively long lifetime against oxidation by OH radicals (DeCarlo et al., 2010). Different degrees of reduction in the concentration of CO and PM 2.5 indicated that the declining pollution level cannot be solely explained by primary emissions. Secondary aerosols generated through gas-to-particle conversion are crucial to the pollution process. Sun et al. (2019) found that, in Shanghai, the chemical transformation of gaseous pollutants to secondary inorganic aerosols leads to the explosive growth of PM 2.5 in winter. The SO 2 concentration, in the range of 1.7-5.6 ppb during the study, is comparable with that in Europe (Querol et al., 2014). Strong emission reduction measures in the industrial and residential sectors have substantially lowered the ambient SO 2 concentration in China . The move-out and closure of some blast furnaces of Baosteel (the largest iron and steel producer in China) in the neighboring Baoshan District has notably improved local air quality since 2010 (Han et al., 2018). Huang et al. (2012a) observed obvious ascending trends of the SO 2 and CO concentration after the CNY holiday. The quarantine measures for the COVID-19 epidemic suppressed human activities. Thus, the expected increase in SO 2 and CO concentration was not observed, which was consistent with the delayed recovery of electricity demand (Fig. S4). The NO x concentration during P1 is~75% higher than that during P2 and P3. There were three durations with high NO x concentration during the study. The first two durations concurred with EP1 and EP2 during P1 and the third duration occurred between P2 and P3 (Fig. 1).
These NO x peaks all arose at nights with low windspeed (Jan. 12-14, Jan. 19-21, and Jan. 31-Feb. 5 as shown in Fig. S2), which might be attributable to the local emissions from the nearby industrial parks and ports, and from the heavy duty truck transportation (Fig. S1). The third high NO x duration, during Feb. 1 to 5, did not alter the descending trend of the NO x concentration. A top-down study showed that NO x emissions were significantly reduced during the CNY holiday and recovered with a slow rate afterwards (Zhang et al., 2020b). As shown in Fig. 1, NO 2 accounted for most of NO x , which results from the fast titration of NO by O 3 and slow photodissociation of NO 2 . To better elucidate the atmospheric oxidation capacity, the term oxidant (O x = NO 2 + O 3 ) is discussed. O x has been widely used to represent the real photochemical production of O 3 (Lu et al., 2010). The O 3 concentration showed an obvious increase with decreasing NO x concentration during P2 and P3. In comparison, O x concentrations showed minor differences, which indicated that the oxidation capacity during the study did not change significantly.
The time series of the chemical composition of NR-PM 2.5 are presented in Fig. 2. Nitrate was the dominant contributor to NR-PM 2.5 during P1, and the nitrate concentration decreased by 55% and 60% during P2 and P3, respectively (Fig. 3e). Since NOR exhibited a slight decrease during the study (Fig. 3f), the descending NO x concentration appeared to be the main cause for nitrate reduction during P2 and P3. Ammonium, as the main balancing cation, showed an approximately 45% simultaneous decrease in concentration during P2 and P3 in response to nitrate reduction. HOAs are mainly emitted from gasoline and diesel combustion sources (Sun et al., 2012). The HOA concentration exhibited a descending trend (Fig. 3e) because of a significant reduction in transportation during the quarantine period . Biomass burning can cause pollution events during the harvest season in Yangtze River Delta Du et al., 2011). Residential heating by biomass and coal is not a common activity in Shanghai in winter and primary emission from the power sector is constrained due to the implementation of "ultralow emission" standard . The PMF analysis on the mass spectra does not show a significant contribution from the biomass burning and coal combustion to the PM 2.5 concentration during the campaign. Primary aerosols from fireworks used to be a dominant pollution source during CNY in Shanghai, but a strict fireworks ban enacted in 2014 has effectively reduced PM 2.5 during CNY . Primary aerosols from fireworks were not observed in this study.
The decreased nitrate and primary aerosols substantially reduced the PM 2.5 concentration during P2 and P3. Meanwhile, the concentrations of sulphate and OOA barely decreased, which inhibited further PM 2.5 reduction. PM 2.5 concentrations exceeding 100 μg/m 3 were observed during EP3 and EP4 (Fig. 1). EP1 and EP2 showed similar chemical proportions of NR-PM 2.5 with nitrate as the dominant contributor. During EP3 and EP4, the proportions of sulphate and OOA increased to 22.7% and 21.9%, respectively, to their maxima, while the proportion of nitrate decreased to the minimum of 28.6%. The variation of the sulphate concentration matched well with that of SOR (Fig. 3f). To further elucidate the formation mechanism of sulphate and nitrate, diurnal courses of NOR and SOR are presented in Fig. S5. NOR showed bimodal mode peaking at noon and midnight, which inferred that both photochemical and nocturnal heterogeneous reactions contributed to nitrate production. A recent study showed that photochemistry in winter can   SO 4 , and NH 4 represent chloride, nitrate, sulphate, and ammonium, respectively. OOA and HOA represents oxygenated and hydrocarbon organic aerosol, which are derived from the Tof-ACSM dataset based on PMF analysis. Four typical PM 2.5 pollution episodes are presented with corresponding double-headed arrows. be as active as during the summer smog and facilitates nitrate production in China by the extremely efficient radical cycling process including volatile organic compounds (VOCs) and NO x (Lu et al., 2019). In addition, low NO and sufficient O 3 during the night (Fig. 1) enabled nocturnal nitrate formation through the hydrolysis of N 2 O 5 , which has been observed in earlier study performed in Shanghai (Pathak et al., 2011). By contrast, SO 2 peaked and SOR bottomed at noon, which made it difficult to distinct the contribution of photochemical and heterogeneous reactions to sulphate production. In the photochemical reaction system, SO 2 and NO x competes to react with OH radicals to form sulphate and nitrate, respectively. When NO x reduced and O x (oxidation capacity) kept constant during P2 and P3, SO 2 could participate more in photochemical reactions and form sulphate. Under low emission scenario during quarantine, the proportion of sulphate in PM 2.5 obviously increased during P2 and P3, especially when PM 2.5 concentration was low (Fig. 4). It should be noted that SOR and NOR rose simultaneously at the initial stage of EP1 and EP2, but SOR and sulphate concentration rose ahead of the occurrence of EP3 and EP4 (Fig. S5). Sulphate are hydrophilic and can promote heterogeneous reactions under humid conditions (Fu and Chen, 2017;Wang et al., 2020b). Thus, sulphate might play a more important role at the initial stage of EP3 and EP4 during quarantine period. OOA, as a surrogate of secondary organic aerosol, was correlated to nitrate and sulphate during the study (Fig. S5). The slopes between OOA and nitrate concentrations varied among different phases, and there was no clear difference in the slopes between OOA and sulphate concentrations for different phases. The formation of OOA is more complicated, but there were no collocated measurements in our study to analyze its formation mechanism. An earlier study in Shanghai with a high-resolution aerosol mass spectrometry (HR-AMS) identified semi-and low-volatility OOA (Huang et al., 2012b), which correlated with nitrate and sulphate, respectively, and the photochemical oxidation processes from HOA to SV-OOA to LV-OOA was implied as the cause of their temporal variation.
Our earlier study showed that both regional transport (windspeed N2 m/s) and stagnant weather (windspeed b2 m/s) can lead to winter haze in Shanghai (Sun et al., 2019). Descending windspeed (from 3 m/ s to 0 m/s) and reducing trajectory path occurred during the pollution evolution of EP1 and EP2 (Figs. S2 & S6), which indicated regional transport initiated EP1 and EP2 and local formation maintained the pollution events until the improvement of the meteorological parameters. This is in line with a recent study indicating that surface particle pollutant can deteriorate the boundary-layer meteorological factors and lead to PM 2.5 explosive growth in Beijing (Zhong et al., 2018). As illustrated in Figs. S2 & S6, EP3 was characterized with fast windspeed and long trajectory path, and EP4 was the opposite, which indicated that EP3 and EP4 was related to regional transport and local formation, respectively.

Conclusion
Reduction in anthropogenic emissions during the quarantine period of the COVID-19 epidemic manifested in an obvious reduction in the concentrations of PM 2.5 , SO 2 , NO x , and CO. In contrast to previous studies on the festival effect, where the pollutant concentration rapidly recovers during the post-CNY period, PM 2.5 and gaseous pollutants did not show inclining trends because of the suppression of human activities during this study. O 3 concentration did increase during the quarantine period when NO x dropped, and the oxidant (the sum of O 3 and Fig. 3. Average mass spectra and corresponding chemical proportions of NR-PM 2.5 during the three phases (a-c). The mean concentration with one standard deviation of (d) PM 2.5 , SO 2 , NO x , CO, O x , (e) chloride (Chl), nitrate (NO 3 ), sulphate (SO 4 ), ammonium (NH 4 ), oxygenated and hydrocarbon-like organic aerosols (OOA and HOA), and (f) nitrogen and sulphur oxidation ratios (NOR and SOR) during the three phases. Note that some species are presented with different units and are multiplied by 10 for clearance, which are marked on their labels.
NO 2 ) did not change significantly before and during the quarantine period, which indicates a constant atmospheric oxidation capacity.
With descending SO 2 emission, nitrate displaced sulphate to be the major PM 2.5 component in Shanghai (Sun et al., 2019). Before the quarantine period, nitrate was the dominant NR-PM 2.5 component. A substantial decrease in the NO x concentration reduced the nitrate concentration and subsequently the PM 2.5 concentration. Hence, NO x reduction can be implemented as an effective measure for PM 2.5 control. However, a slight decrease in the concentration of sulphate and OOA inhibited further PM 2.5 reduction. Two pollution episodes with PM 2.5 concentration exceeding 100 μg/m 3 occurred during the quarantine period. Active oxidation is the hypothesized cause for fast secondary aerosol formation. Air quality has been notably improved in Shanghai by emission reduction measures (Han et al., 2018) and firework bans . Since the demand for economic development has increased in China, relatively high concentrations of gaseous precursors will be a huge challenge in the future. Our study shows that fast chemical transformation of gaseous pollutants to secondary aerosols could lead to air pollution even if anthropogenic activities were suppressed to a great extent. In addition to emission reduction measures, understanding the atmospheric oxidation process with respect to secondary aerosol formation should be given more attention.

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.