Ambient fine particulate matter and ozone pollution in China: synergy in anthropogenic emissions and atmospheric processes

This review is mainly based on two sets of data:

• (a)  
  The surface observation data of PM_2.5 and O₃ in China from 2013 to 2021

We obtained the surface observation data from the China National Environmental Monitoring Center network (https://air.cnemc.cn:18 007/). Data processing of PM_2.5 and O₃ concentration was conducted according to National Ambient Air Quality Standards (GB3095-2012) (available on www.mee.gov.cn/ywgz/fgbz/bz/bzwb/dqhjbh/dqhjzlbz/201 203/W020120410330232398521.pdf, in Chinese) and China's Technical Regulation for Ambient Air Quality Assessment (tentative) (HJ 633-2013) (available on www.mee.gov.cn/ywgz/fgbz/bz/bzwb/jcffbz/201 309/W020131105548549111863.pdf, in Chinese). It should be noted here the standard state data were used before 2019, and the actual state data were used starting in 2019 for the assessment of ambient air quality. The O₃ concentration depended significantly on this modification. We multiplied the O₃ observation data for 2019, 2020, and 2021 by 1.09 for consistent conversion (1.09 was calculated according to difference between the Bulletin of China Ecological Environment 2019 and Bulletin of China Ecological Environment 2018, available on www.mee.gov.cn/hjzl/sthjzk/zghjzkgb/, in Chinese).

• (b)  
  The anthropogenic air pollutants emissions inventory of China

We obtained anthropogenic air pollutants emissions data from five databases, including ABaCAS-EI (Zhao et al 2013, Wang et al 2014, Cai et al 2017, Zheng et al 2019; www.abacas-dss.com/abacas/), Community Emissions Data System (CEDS, O'Rourke et al 2021), Emissions Database for Global Atmospheric Research (EDGARv5.0, https://edgar.jrc.ec.europa.eu/overview.php?v=50_AP), Multi-resolution Emission Inventory for China (MEIC, Zheng et al 2018, 2021; http://meicmodel.org/?page_id=560), and Regional Emission inventory in Asia (REASv3.1, Kurokawa and Ohara 2020). ABaCAS-EI and MEIC also provided the emissions data of each province in China.

The literature cited in this study was obtained by searching in the Web of Science (www.webofscience.com/) database. The latest search date is 10 June 2022. The search terms are included in table S1 of the supplementary materials. We need to uncover suitable literature from many research domains because just a few papers have provided a thorough grasp of the synergy between PM_2.5 and O₃. When choosing the literature, we followed the following rules: (a) literature which receives a lot of citations or is extensively discussed; (b) literature which focuses on China or typical locations; and (c) literature that proposes novel phenomena and arguments for their explanations. Additionally, in order to track the most recent advancements in adjacent fields of research, we primarily focused on literature released after 2017 (152 references). Also a number of classic works of literature released before 2017 are included (18 references).

Based on air quality observation data in China, we analyzed spatial and temporal distribution of PM_2.5 and O₃ from 2013 to 2021. Annual evaluation values of PM_2.5 (average PM_2.5 concentration in one year) and O₃ (the 90th percentile of daily maximum 8-hour average O₃ in one year) in China and five key regions are shown in figure 1 (annual evaluation values of PM_2.5 and O₃ of all sites are shown in figure S1 in supplementary materials).

Zoom In Zoom Out Reset image size

Generally, the annual PM_2.5 concentration in China continuously decreased from 72 μg·m⁻³ to 30 μg·m⁻³ from 2013 to 2021. The annual PM_2.5 concentrations of the Jing-Jin-Ji (JJJ), the Yangtze River Delta (YRD) and the Pearl River Delta (PRD) regions decreased by 63 μg·m⁻³ (59%), 36 μg·m⁻³ (54%), and 26 μg·m⁻³ (55%), respectively. The annual PM_2.5 concentrations in Beijing, Shanghai and Guangzhou all met the national standard level 2 in 2021, according to the National Air Quality Standards (GB3095-2012). The spatial pattern of PM_2.5 is remained similar from 2013 to 2021, with higher PM_2.5 concentrations in eastern and northern China and lower PM_2.5 concentrations in western and southern China (Wang et al 2017a, Xu et al 2020). The North China Plain (NCP) is typically a polluted region in China. Seasonal variation of PM_2.5 has been described in many previous studies (Zhao et al 2016, Li et al 2017c, Chen et al 2018, Zhao et al 2018, Ye et al 2018b, Ma et al 2019b). As shown in figure 2(a), PM_2.5 is higher in winter and lower in summer. During 2013–2021, the averaged PM_2.5 concentration in winter was 2–3 times that in summer in China. In spring and autumn, the PM_2.5 concentration has different variation characteristics for different regions. In JJJ, the Fenwei Plain (FW) and the PRD, PM_2.5 in autumn is typically higher than that in spring, while the trends in the YRD and Sichuan Basin (SCB) are the opposite, as shown in figure S2 in the supplementary materials.

Zoom In Zoom Out Reset image size

For O₃, as shown in figure 1, the 90th MDA8 O₃ in China, JJJ, YRD, PRD, and FW were the highest in 2019. The highest 90th MDA8 O₃ in SCB appeared in 2020. Compared with 2013, the 90th MDA8 O₃ in 2019 in JJJ, YRD, and PRD increased by 59 μg·m⁻³ (38%), 35 μg·m⁻³ (24%), and 37 μg·m⁻³ (24%), respectively. In terms of spatial distribution, eastern China is more polluted than western China, which is similar to the spatial distribution of PM_2.5. Due to higher volatile organic compounds (VOCs) emissions and stronger radiation, O₃ is higher in the warm season (Zhang et al 2015, Cheng et al 2017, Wang et al 2018, Yang et al 2020). Figure 2(b) shows the monthly averaged MDA8 O₃ concentrations of JJJ, YRD, PRD, FW and SCB in 2021. For JJJ and FW, the peak of monthly averaged MDA8 O₃ appears in June. For SCB, O₃ is highest in July or August. For YRD, O₃ is the highest in June or September. PRD has a distinct pattern, with the highest O₃ in September or October. Thus, the warm season of China is usually considered to be April to September or April to October.

Many studies have analyzed the correlation between PM_2.5 and O₃ by calculating the Pearson correlation coefficients based on monitoring data, as shown in table 1. From the spatial distribution, as pointed out in many studies, PM_2.5 and O₃ tend to be negatively correlated in North China and positively correlated in South China. In addition, Chen et al (2019a) found that PM_2.5 and O₃ in coastal areas tend to be positively correlated but negatively correlated in inland areas. In this study, we also analyzed the correlation between the daily PM_2.5 and MDA8 O₃ concentrations in JJJ, YRD, PRD, FW and SCB in 2021, as shown in figures S3 and S4 (remove seasonal cycle of PM_2.5 and O₃). It can be seen that the correlation between PM_2.5 and O₃ is more positive in PRD than in other regions, which is consistent with the conclusion above. From the temporal characterization, there is an obvious seasonal variation in that PM_2.5 and O₃ tend to be more positively correlated in the summer but tend to be more negatively correlated in the winter, as shown in both previous studies and our results. The daily variation shows that the peak value of the Pearson correlation coefficient appears at approximately 16:00 (Chen et al 2019a).

The secondary generation of PM_2.5 and O₃ is the key to explain the spatial-temporal variation of PM_2.5 and O₃ correlations. Qin et al (2022) associated atmospheric oxidizing capacity (AOC) with the correlation between PM_2.5 and O₃ in YRD. They found that the daily PM_2.5 and O₃ were highly correlated at moderate to high AOC levels. The promoted formation of secondary pollutants by AOC is the main reason (Liu et al 2020, Zhu et al 2020, Chen et al 2020c, Wang et al 2022a). We analyzed the correlation between daily O_x (O₃ + NO₂) and PM_2.5 (figure S5). The result showed that O_x and PM_2.5 is positively correlated in almost all key regions and seasons in 2021, indicating promoted production of secondary components of PM_2.5 by AOC. Chu et al (2020) suggested that the correlation between PM_2.5 and O₃ would become positive with the decrease of PM_2.5 in China due to the influence of PM_2.5 on solar radiation and heterogeneous reactions of HO₂. In summer or south China, where PM_2.5 concentration is relatively low, PM_2.5 has little impact on solar irradiation and HO₂ radicals, thus PM_2.5 and O₃ tend to have a positive correlation with secondary generation from NO_x and VOCs emissions. Chen et al (2019a) suggested similar causes for regional differences. Besides, they reported the climate and meteorological conditions would cause the regional differences by influencing the formation and removal of PM_2.5 and O₃. Lei et al (2022) revealed that the Spearman correlation coefficients between PM_2.5 and O₃ in key regions (JJJ, YRD, PRD, SCB and FW) were much higher in 2020 than in 2019. They emphasized the urgent demand for the co-control of PM_2.5 and O₃ as the primary emissions decreased steadily. We therefore discuss the synergy of PM_2.5 and O₃ as influenced by three factors: (a) the impact of emissions; (b) interactions between the two pollutants; (c) future challenges under climate change in the following sections.

A large number of studies have been conducted to analyze the emission changes in China in recent years. Researchers have provided detailed reviews of both pollutants separated and sectors separated emissions (Wang et al 2014, Li et al 2017b, Li et al 2021a); thus, we briefly discuss the emissions change to provide support for later sections. Figure 3 shows the emission trends of SO₂, primary PM_2.5, NO_x , non-methane volatile organic compounds (NMVOC), and NH₃ from different studies and databases (Zhao et al 2013, Wang et al 2014, Crippa et al 2018, Zheng et al 2018, 2019, Kurokawa and Ohara 2020, O'Rourke et al 2021, Zheng et al 2021). The trends of SO₂ emissions and PM_2.5 emissions were similar. From 2000 to 2005, SO₂ emissions and PM_2.5 emissions increased by 32%–70% (EDGARv5.0, REASv3.1, CEDS) and 15%–21% (EDGARv5.0, REASv3.1), respectively, and then decreased from 2006 to 2010. Wide application of flue-gas desulfurization devices and end-of-pipe dust collectors in power plants is reported as one of the most effective measures (Lu et al 2010, Li et al 2017b). Small increases of SO₂ and PM_2.5 emissions appeared from 2011 to 2013. After 2013, SO₂ and PM_2.5 emissions decreased rapidly as a result of the implementation of the 'Air Pollution Prevention and Control Action Plan' (Cai et al 2017, Zheng et al 2019). Pollutant control strategies in industry, including strengthening industrial emission standards and removing pollutant emissions, were recognized as the main drivers of this reduction (Zheng et al 2018, Zhang et al 2019b). NO_x emissions increased by 80%–128% (EDGARv5.0, REASv3.1, CEDS) from 2000 to 2013. From 2013 to 2017, rapid implementation of an effective NO_x emissions control strategy (e.g. application of selective catalytic reduction (SCR) and selective non-catalytic reduction (SNCR) in power plants and industry) caused a 20%–37% (ABaCAS-EI, MEIC, CEDS) decrease of NO_x emissions in China. However, the emissions of NMVOC increased slowly and NH₃ emissions were stable from 2000 to 2017, due to the lack of effective control policies.

Zoom In Zoom Out Reset image size

Herein, we also analyzed the emissions variation in key regions from ABaCAS-EI, as shown in figure 4. Compared to 2007, the trends of SO₂ and primary PM_2.5 were consistent in different regions, decreasing by 73%–88% and 57%–73% in 2019, respectively. However, the emissions of NO_x were highest in 2013 in all 5 regions. Compared to 2013, the emissions of NO_x in JJJ, YRD, PRD, FW and SCB decreased by 42%, 41%, 27%, 52%, and 33%, respectively, in 2019. In contrast, the emissions of NMVOC varied less from 2013 to 2019, by −33%, −6%, −5%, −10% and +5% in JJJ, YRD, PRD, FW, and SCB, respectively. The emission of NH₃ was more stable from 2007 to 2019.

Zoom In Zoom Out Reset image size

As listed in table 2, the changes of anthropogenic emissions is the major driver of PM_2.5 change in recent years in China. Ding et al (2019b) revealed that the variation of anthropogenic emissions contributed 70.8%, 95.6% and 99.7% to the total decrease of annual PM_2.5 concentration in JJJ, Guangdong and Sichuan-Chongqing region from 2013 to 2017. Control measures on industry were evaluated as the largest source of PM_2.5 decline across China, which was consistent with the anthropogenic emissions change (Zheng et al 2019, Cheng et al 2019a, Zhang et al 2019b).

In addition to the PM_2.5 concentrations, the chemical compositions and source apportionment of PM_2.5 changed significantly due to the changes of anthropogenic emissions. Dong et al (2020) and Cao et al (2022) stated that the local contribution to PM_2.5 of Beijing decreased largely for strict pollutant control measures from 2013 to 2017, reflecting the uneven emissions reduction in different cities. Chen et al (2019b) reported that compared with the pollution episode in March 2013, the relative contribution of coal combustion decreased by 29%, while the relative contribution of vehicle exhaust increased by 35% in the pollution episode in March 2018 in Beijing. They suggested that different emissions control measures on emission sources were the main cause. Li et al (2021c) also found an increased contribution of vehicles and a decreased contribution of coal combustion to PM_2.5 in Heze (a city in the central NCP) from 2017 to 2019.

In addition, more studies have focused on the chemical composition of PM_2.5. Sulfate decreased markedly due to the large reduction of SO₂ emissions in the past decade. For JJJ, YRD, PRD and SCB, the sulfate decreased by 51.1%, 40.8%, 37.5% and 36.2%, respectively, as reported by Cao et al (2021). However, nitrate showed different trends. Geng et al (2019) claimed that the proportion of nitrate in PM_2.5 increased by 5%, 7%, 4% and 3% in eastern China, JJJ, YRD and PRD from 2013 to 2017, respectively, even if NO_x emissions had been reduced. Explosive growth of nitrate during haze events was also observed (Kong et al 2018, Li et al 2018, 2019a, Zhou et al 2019, Xu et al 2019a, Fu et al 2020, Sun et al 2020, Zhang et al 2020). A simple explanation for the phenomenon was that a large reduction of SO₂ left more NH₃ available to form nitrate. However, some studies reported that the North China Plain was in an ammonia-rich atmosphere (Xu et al 2019b, Bhattarai et al 2020), contradicting the above explanation. The second explanation was that photochemical oxidants increased due to the reduction of NO_x emissions and constant VOCs emissions in NCP, so that most regions were VOC-limited in winter (Xing et al 2019, Lu et al 2019a, Fu et al 2020, Leung et al 2020), increasing the conversion efficiency of NO_x to HNO₃. Fu et al (2020) reported ∼30% increase of O₃ and OH and 38.7% increase of conversion efficiency of NO_x to HNO₃ from 2010 to 2017 in NCP. Leung et al (2020) showed that the above two explanations both accounted for high NO₃ ⁻ concentrations in eastern and central China. Another explanation was proposed by Zhai et al (2021). A large reduction of SO₂ emissions, smaller reduction of NO_x emissions (compared to SO₂ emissions), and stable NH₃ emissions were conducive for the transformation of gas-phase nitrate to particle-phase nitrate. By simulation with the Goddard Earth Observing System (GEOS)-Chem model, they found that the deposition of the total nitrate (gas + particle) decreased as the particulate fraction of total nitrate approached unity. As a result, nitrate sometimes showed a weak response to NO_x emissions reduction and an increase in haze events.

Unlike PM_2.5 which is influenced by a multitude of precursor emissions, the O₃ concentration is primarily affected by NO_x and VOCs emissions. However, the anthropogenic emissions driven O₃ change varied in different studies, as shown in table 2. Ding et al (2019a) showed that changes of anthropogenic emissions decreased O₃ in China and key regions from 2013 to 2017, while Liu and Wang (2020b) found that anthropogenic emissions increased O₃ in urban areas and decreased O₃ in rural areas. Lin et al (2021) projected that anthropogenic emissions changes caused O₃ to rise in eastern and central China, but meteorological conditions played the opposite role in the area from 2013 to 2020. Studies also showed that both radiative effect of particles and aerosol chemistry have significant effects on O₃ pollution (Li et al 2019b, Wang et al 2020a), which will be discussed in the next section. There are many factors that may cause different results, such as model selections, mechanism settings, amount and distribution of emissions. One of the most important factors is the definition of the temporal and spatial scope. Due to distinctive regional and temporal characteristics of O₃, same change of anthropogenic emissions may cause different variation of O₃ in different regions and times.

Herein, we mainly focused on the influence of anthropogenic emissions on the O₃ formation regime, which is closely associated with the control strategy, and has been considered in many studies as listed in table 3. As NO_x plays a dual role in O₃ formation, the O₃ formation regime is classified as VOC-limited regime (VOC-sensitive regime), transition regime and NO_x -limited regime (NO_x -sensitive regime) (Sillman 1995, Wang et al 2019b). Lu et al (2019a) reviewed the spatial-temporal characteristics of the O₃ formation regime across China. O₃ formation during high O₃ seasons in most urban areas was in the VOC-limited regime due to large amounts of anthropogenic emissions, while rural areas were in the NO_x -limited regime (Xing et al 2019, Yu et al 2020). In the VOC-limited regime, a reduction of NO_x emissions may lead to an increase of ozone. Thus, a rapid decrease of NO_x was identified as one of the reasons for ozone increases in urban areas in recent years and during COVID-19 (Sicard et al 2020, Xing et al 2020a, Chen et al 2020c, Huang et al 2021). With the change of anthropogenic NO_x and VOCs emissions, the formation mechanism of ozone changes dynamically, especially after 2013. In JJJ, the proportion of VOC-limited regime areas varied from 8.4%–44.2% to 5.6%–33.3% from 2014 to 2018 (Wei et al 2021b). Urban areas changed from VOC-limited regime to transitional regime, and rural areas were always in transitional regime or NO_x -limited regime (Cheng et al 2019b, Li et al 2021e). In YRD, the transitional regime accounted for more than half, and rural areas tend to be NO_x -limited (Wang et al 2019a). In PRD, more than half of urban areas belonged to the transitional regime, and most rural areas were NO_x -limited (Chen et al 2020d).

In terms of synergistic effects between PM_2.5 and ozone, joint control becomes possible. Based on empirical kinetic modeling approach (EKMA) curve, a reduction of VOCs emissions was proven to conduct under the current emission situation (Wang et al 2019a, Huang et al 2021). Studies conducted by Xiang et al (2020) and Ding et al (2021) revealed that earlier VOCs emission reductions were more cost-effective in reducing both PM_2.5 and O₃ in JJJ and surrounding regions. For further improvement of air quality in China, reduction of NO_x emission is still essential (Xing et al 2019, Li et al 2019c, Ding et al 2021). Based on the response surface model, Dong et al (2022) suggested that the emissions of NO_x , SO₂, NH₃, VOCs and primary PM_2.5 should be reduced by 50%, 26%,28%,28% and 55%, respectively, to achieve the National Ambient Air Quality Standard (NAAQS) in YRD region. Huang et al (2020) suggested that emissions of NO_x , VOCs and primary PM_2.5 should be reduced by 22%, 12% and 30%, respectively, to achieve NAAQS in PRD region.

Furthermore, many studies focused on control policies aimed at both carbon and air pollution reduction (Xing et al 2020b, Yang et al 2021a). Note here that there is a special issue for China referred to as achieving carbon neutrality by 2060. Several studies proposed a strategy to achieve carbon neutrality and evaluated the influence on air quality. Shi et al (2021) projected SO₂, NO_x , primary PM_2.5 and NMVOCs in China to reduce by 93%, 93%, 90% and 61% in 2060, respectively, compared to 2019, decreasing more than 70% PM_2.5 and 25% 90th MDA8 O₃. They expected PM_2.5 concentrations and O₃ concentrations in more than half of 337 cities to meet the World Health Organization (WHO) guidance (2006) in 2060. Cheng et al (2021) and Zhang et al (2021a) also highlighted that emission mitigations to achieve China's carbon neutrality goal could improve PM_2.5 exposure to below 10 μg·m⁻³ for the majority of China's population.

AOC refers to the ability of atmospheric chemical processes to oxidize primary pollutants. Oxidants (HO_x , NO₃ ⁻...) concentrations and reaction rates are always used to characterize AOC. AOC plays an important role in the formation of both PM_2.5 and O₃. As one of the important atmospheric oxidants, O₃ concentration is often positively correlated with AOC (Zhu et al 2020). In clear air, O₃ is generated by NO₂ photolysis, and consumed in the oxidation of NO, without the accumulation of O₃ in the process. However, when OH oxidizes VOCs and generates HO₂ or RO₂ radicals, they replace O₃ to oxidize NO, leading to an increase of the O₃ concentration. Another product of VOCs oxidized by OH is oxygenated compounds, leading to the formation of secondary organic aerosol (SOA). At the same time, the increase of atmospheric oxidants also accelerates the formation of secondary inorganic aerosol (SNA). SOA and SNA are reported to account for more than 50% of the PM_2.5 concentration in most cities in China (Yin et al 2020). Wang et al (2021b) analyzed simulation and observation data of oxidants in 2013 and 2020. The results showed that the level of AOC did not decrease significantly (Chen et al 2020c), and even increased slightly in NCP and PRD, causing an increase of secondary pollutants. The importance of AOC was again evident in studies on COVID-19. Zhu et al (2021b) revealed that enhanced AOC during COVID-19 contributed 40% and 53% of O₃ in NCP and YRD, respectively. Wang et al (2021d) reported that the changes of AOC and O₃ concentrations in YRD were consistent during the COVID-19 lockdown. Huang et al (2021) found that an increase of AOC due to large decreases of NO_x emissions promoted the formation of secondary particles. Moreover, Chen et al (2020a) revealed that the stratosphere-to-troposphere (STE) exchange of O₃ would increase PM_2.5 by 2.289 μg·m⁻³ in winter and 2.034 μg·m⁻³ in spring by enhancing AOC.

Several studies analyzed the nonlinear relationship between AOC or oxidants and precursors, and then proposed a control strategy for NO_x and VOCs to avoid the increase of secondary pollutants. On the basis of O₃, OH and NO₃ ⁻ isopleths, Kang et al (2021) found that NO_x emission reduction was beneficial for O₃ and PM_2.5 in summertime. Wang et al (2021b) found that NO_x emission reduction would decrease the AOC in summer, but increase the winter AOC in NCP and YRD. Both studies also proposed that the emission reduction of VOCs were beneficial for the further control of O₃ and PM_2.5. Furthermore, the effects of some other species on AOC have also been found. Ge et al (2019) showed that NH₃ could enhance AOC by accelerating the formation of gas-phase nitrous acid (HONO). Li et al (2021d) demonstrated that halogens increased PM_2.5 by 21% in northern China, and the enhancement of AOC was the main reason (Li et al 2020b).

Heterogeneous chemistry refers to reactions on the surface of aerosols, including reactions and uptake of gases. Thus, it can lead to changes in both the size and components of particles (Farmer et al 2015). For example, Tian et al (2021) reported that heterogeneous reactions (including the uptake of SO₂, NO₂, HNO₃ and H₂SO₄ on CaCO₃ and mineral oxides) contributed 20%–30% to sulfate formation over northern China during a dust pollution event. Wang et al (2017c) showed that heterogeneous reactions promote both fine mode and coarse mode nitrate and sulfate to be internally mixed with dust particles.

Heterogeneous reactions related to sulfate formation have been a concern since the end of the last century. Early studies revealed that SO₂ could be oxidized by oxidants (H₂O₂, OH, O₃...) and NO₂ on mineral dust (Zhang and Carmichael 1999, Ye et al 2018a), and by dissolved oxidants in the aqueous phase (Cheng et al 2016, Li et al 2017a). Recently, Gen et al (2019) proposed NO₂ and OH radicals produced from the photolysis of particle nitrate could oxidize aqueous-phase SO₂, which was a less explored pathway for sulfate formation. Song et al (2019) proposed that heterogeneous reactions of hydroxymethanesulfonate (HMS) might be important and could account for up to 1/3 of unexplained sulfate formation in an air quality model. In addition, many studies have focused on heterogeneous reactions of HONO. HONO, as an important source of OH radicals, is highly related to the secondary components of particles. Cui and Wang (2021) found that in polluted regions (JJJ, YRD, and PRD), the HONO concentration in winter was highest. However, the HONO concentration was typically underestimated in chemical transport models (CTM), particularly during haze events (Xue et al 2020). By considering heterogeneous reactions of HONO, the performance of HONO, OH, O₃, and secondary components in particles was improved significantly. Zhang et al (2019a) found that heterogeneous reactions (${\text{N}}{{\text{O}}_{\text{2}}}\mathop{-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!\longrightarrow} \limits^{{\text{aerosol/ground}}} {\text{HONO}})$ contributed approximately 85% of the daytime HONO concentration and improved the SOA concentration by more than 40% in Beijing. Fu et al (2019) reported that heterogeneous reactions of HONO on the surfaces ($2{\text{N}}{{\text{O}}_{\text{2}}} + {{\text{H}}_{\text{2}}}{\text{O}}\mathop{\longrightarrow} \limits^{{\text{surface}}} {\text{HONO + HN}}{{\text{O}}_{\text{3}}}$) were the major source (72%) of HONO concentration during a severe pollution in PRD. Compared to the simulation without any HONO sources, the daytime averaged O₃ and mean concentration of TNO₃ (HNO₃ and fine particle NO₃ ⁻) increased 70% and 67%, respectively. Zhang et al (2021b) revealed that heterogeneous reactions on ground contributed nearly 70% of total HONO production. In Beijing, heterogeneous reactions of HONO increased the OH concentration and SOA concentration by 98% and 50%, respectively. Heterogeneous reactions were proven to effectively improve the simulation of HONO and SOA, but there still exists a gap between simulations and observations, and more studies are needed.

In addition, heterogeneous reactions were proven to influence ozone concentration by impacting the production and loss of nitrogen oxides (NO₂, NO₃ and N₂O₅), hydrogen radicals (HO₂, OH, and H), ozone and halogen radicals (Jacob 2000), as shown in table 4. Li et al (2019b) projected that a 40% decrease of PM_2.5 was the most important driver of the O₃ increase in the North China Plain from 2013 to 2017, by slowing the uptake of HO₂. However, another study by Qu et al (2018) proved that effect of heterogeneous reactions of H_x O_y (OH, HO₂, H₂O₂) was less than that of heterogeneous reactions of nitrogen oxides and ozone. Based on field campaign measurements of radicals and trace gases, Tan et al (2020) conducted a sensitivity test of HO₂ uptake by using different uptake coefficients. The results showed that the performance of all radicals (OH, HO₂, and RO₂) was poor when using an uptake coefficient value of 0.2 (same as Li et al 2019b). When using an uptake coefficient of 0.08, the difference between the simulation and observation was much smaller, and the influence of aerosol uptake on HO₂ concentration was only 17%. Tan et al (2019) also found that the RO_x and HO₂ budgets of the PRD closed without aerosol characterization, indicating that heterogeneous reactions of HO₂ were not important. In summary, the impact of heterogeneous reactions on ozone is subjected to much debate. Both comprehensive field measurement and modeling studies are suggested.

Table 4. Heterogeneous reactions related to ozone concentration.

Reactions	Uptake coefficients	References
${\text{H}}{{\text{O}}_{\text{2}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!\longrightarrow} \limits^{{\text{aqueous aerosols}}} 0.5{{\text{H}}_{\text{2}}}{{\text{O}}_{\text{2}}}\left( {\text{g}} \right)$	0.2 (range 0.1–1)	Jacob (2000)
${\text{N}}{{\text{O}}_{\text{2}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!\longrightarrow} \limits^{{\text{aqueous aerosols}}} 0.5{\text{HONO}}\left( {\text{g}} \right){\text{ + 0}}{\text{.5HN}}{{\text{O}}_{\text{3}}}\left( {\text{g}} \right)$	10−4 (range 10−6–10−3)
${\text{N}}{{\text{O}}_{\text{3}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!\longrightarrow} \limits^{{\text{aqueous aerosols}}} {\text{HN}}{{\text{O}}_{\text{3}}}\left( {\text{g}} \right)$	10−3 (range 2 × 10−4–10−2)
${{\text{N}}_{\text{2}}}{{\text{O}}_{\text{5}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!-\!\!\!\!\!\longrightarrow} \limits^{{\text{aqueous aerosols}}} 2{\text{HN}}{{\text{O}}_{\text{3}}}\left( {\text{g}} \right)$	0.1 (range 0.01–1)
${\text{OH}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{OH}}\left( {{\text{ads}}} \right)$	Dust: 0.1 (range 0.01–0.2) Sea salt: 0.04 BC: 0.1	Deng et al (2010)
${\text{H}}{{\text{O}}_{\text{2}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{H}}{{\text{O}}_{\text{2}}}\left( {{\text{ads}}} \right)$	Dust: 0.1 (range 0.01–0.2) Sea salt: 0.01 BC: 0.1
${{\text{H}}_{\text{2}}}{{\text{O}}_{\text{2}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {{\text{H}}_{\text{2}}}{{\text{O}}_{\text{2}}}\left( {{\text{ads}}} \right)$	Dust: 10−4 (range 10−5-2 × 10−4) Sea salt: 10−4 BC: 10−4
${{\text{O}}_{\text{3}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {{\text{O}}_{\text{3}}}\left( {{\text{ads}}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {{\text{O}}_2}$	Dust: 10−4 (range 10−5–2 × 10−4) Sea salt: 5 × 10−5 BC: 4.5 × 10−5
${\text{N}}{{\text{O}}_{\text{2}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{N}}{{\text{O}}_{\text{2}}}\left( {{\text{ads}}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{NO}}_{\text{3}}^{-}$	Dust: 4.4 × 10−5 (range 10−5–10−4) Sea salt: 1.36 × 10−6 BC: 10−4
${\text{N}}{{\text{O}}_{\text{3}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{N}}{{\text{O}}_{\text{3}}}\left( {{\text{ads}}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{NO}}_{\text{3}}^{-}$	Dust: 0.1 (range 0.01–0.2) Sea salt: 0.046 BC: 0.5
${\text{HN}}{{\text{O}}_{\text{3}}}\left( {\text{g}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{HN}}{{\text{O}}_{\text{3}}}\left( {{\text{ads}}} \right)\mathop{-\!\!\!\!-\!\!\!\!\longrightarrow} \limits^{{\text{aerosols}}} {\text{NO}}_{\text{3}}^{-}$	Dust: 0.01 (range 0.001–0.1) Sea salt: 9 × 10−5 BC: 0.001

Aerosol was proven to have direct effects on radiative forcing by absorbing or scattering radiation and indirect effects by participating in cloud formation (Bond et al 2013, Hong et al 2017, Liu and Matsui 2021). Gao et al (2022) reported anthropogenic aerosols decreased shortwave radiative force at the bottom of the atmosphere by 0.45–140.00 Wm⁻² in Asia. Thus, the radiative effect can influence ozone formation via its impact on thermal and photolysis reactions. Previous studies revealed that aerosol direct effects lead to a lower photolysis rate and less ozone. (Xing et al 2017, Mukherjee et al 2020, Qu et al 2021, Zhao et al 2021). Mukherjee et al (2020) reported that the direct radiative effects of black carbon reduced more than 30% ozone concentrations by changing the photolysis rate in South Asia during a heavy dust and biomass burning event. Besides, aerosol direct effects can change meteorological factors, such as the temperature, wind speed, humidity, planetary boundary layer height, by influencing the energy balance in atmosphere (Li et al 2017d, Wu et al 2020, Wang et al 2022b). Li et al (2022) estimated that aerosol radiative effect decreased 2 m air temperature and planetary boundary layer height by 0.4 °C–1.8 °C and 30.9–183.6 m, respectively, during severe winter haze events in the North China Plain. Impact of aerosol radiative effects on meteorological factors is proven to influence O₃ concentration. Xing et al (2017) found that changes of atmospheric dynamics caused by aerosol radiative effect decreased the daily maximum 1 h O₃ in January by up to 39 μg·m⁻³ but increased the peak ozone in July by up to 4 μg m⁻³ in China. Moreover, Qu et al (2021) found that aerosol direct effects also influenced ozone concentration through the vertical transmission of water vapor. They projected that the annual ozone concentration decreased by 14.9%, 8.7%, 4.3% and 7.1% in JJJ, YRD, PRD and SCB, respectively, due to the combined impact of aerosol direct effects.

Regarding the impact of aerosol radiative effects on ozone, some studies have pointed out that rapid aerosol reduction in China may tend to enhance ozone production. According to Zhao et al (2021), from 2005 to 2019, the photolysis rate of nitrogen dioxide in Beijing increased by 5 × 10⁻⁵ s⁻¹ per year, while the annual aerosol effect decreased in this period. Wang et al (2020a) got the similar conclusion that the decrease of particles enhanced solar radiation and thus led to an increase in the photolysis rate of nitrogen dioxide of 3.6 ± 0.8% per year. However, whether the aerosol radiative effects play an important role in ozone variation is contentious. Ma et al (2021) showed that aerosol radiative effects contributed 23% of the summertime ozone change in the NCP from 2013 to 2019. Zhu et al (2021a) proved aerosol radiative effects contributed to a 7.8% ozone rise over the NCP during COVID-19. However, Li et al (2019b), Liu and Wang (2020b), and Li et al (2021f) all showed that the influence of aerosol radiative effects on ozone was weak, generally less than 1 ppb MDA8 O₃ from 2013 to 2017 across China. The varied mechanism of aerosol radiative effects in different methods and models was an important reason for the difference between studies, which should be set more carefully and accurately.