Unequal toxic potencies of industry-emitted PM2.5
Exposure to different industry-emitted PM2.5 samples exhibits substantial discrepancies in toxic potencies, including intracellular OS and cytotoxicity (CT) potencies (Fig. 1 and Supplementary Fig. 2). EC1.5 (the concentration effectively inducing a 1.5-fold production of intracellular ROS generation relative to that of the control) and IC20 (the concentration effectively inducing a 20% inhibition relative to that of the control) values were determined to elucidate the endpoints of PM2.5-induced OS and CT potencies, respectively. The toxic potency increased with decreasing EC1.5 and IC20 values. ISI-emitted PM2.5 exhibited the highest OS potency (19.1 ± 10.9 μg/mL), which is nearly 5.0 and 5.4 times higher than that of power industry-emitted PM2.5 (94.9 ± 41.9 μg/mL) and cement industry-emitted PM2.5 (103.6 ± 46.3 μg/mL), respectively (Fig. 1a).23 The EC1.5 values for the PM2.5 originating from the three major polluting processes of basic oxygen furnace (BOF)-based crude steel production, i.e., the sinter ore (21.3 ± 11.2 μg/mL), pig iron (20.7 ± 12.1 μg/mL), and crude steel (14.7 ± 7.8 μg/mL) generation, are shown in Supplementary Fig. 2b, while the EC1.5 values for other ISI processes (including pelletizing, electric arc furnaces (EAF)-based steelmaking, and steel rolling) are shown in Supplementary Fig. 3. The three processes contributed nearly 90% to the nationwide ISI PM2.5 emissions in mainland China in 201931. The averaged OS potency of power industry-emitted PM2.5 resulted in very close EC1.5 values when the boilers were fed with different solid fuels, including coal (104 ± 44 μg/mL), solid waste (89.7 ± 48.1 μg/mL), and biomass (91.9 ± 35.7 μg/mL) (Supplementary Fig. 2c). Additionally, the EC1.5 values for the PM2.5 originating from the major production processes of cement plants were close, at 113.4 ± 53.0 μg/mL for the kiln head and 98.7 ± 39.9 μg/mL for the kiln tail (Supplementary Fig. 2d).
The distribution of the IC20 values further indicated similar substantial discrepancies among the three industrial sectors. The estimated CT potency of ISI-emitted PM2.5 (46.4 ± 28.2 μg/mL) was nearly 12.1 times that of power generation-emitted PM2.5 (561 ± 226 μg/mL) and 9.5 times that of cement production-emitted PM2.5 (442 ± 221 μg/mL) (Fig. 1b). The CT potency of ISI-emitted PM2.5 successively increased for the major production processes of sinter ore (60.0 ± 30.4 μg/mL), pig iron (43.4 ± 22.7 μg/mL), and crude steel (36.1 ± 26.4 μg/mL) (Supplementary Fig. 2f). The IC20 values for the other ISI processes are shown in Supplementary Fig. 3. The CT potency of power industry-emitted PM2.5 slightly varied among the different fuel-fired units, including coal- (604 ± 254 μg/mL), biomass- (434 ± 212 μg/mL), and solid waste-fired units (539 ± 137 μg/mL) (Supplementary Fig. 2g). In terms of clinker production, the CT potency of kiln tail emitted-PM2.5 (396 ± 214 μg/mL) was also close to that of kiln head-emitted PM2.5 (487 ± 221 μg/mL) (Supplementary Fig. 2h).
Estimation through laboratory and field studies of all the major processes under the same quality control conditions revealed that the PM2.5 emitted from the ISI is much more toxic than those of the power and cement industries, especially the PM2.5 emitted from blast furnace (BF)-based ironmaking and BOF-based steelmaking processes. A substantial inequality in the toxicological properties of the PM2.5 samples obtained from the three industrial sectors is uncovered according to the estimated OS and CT potencies. Source-specific PM2.5 comprises a complex mixture of inorganic and organic chemicals, some of which are potential contributors to PM2.5 related toxicity17,18. Hence, it is necessary to uncover and quantify the key chemical components that determine PM2.5 toxicological effects to better understand the causes of the abovementioned inequality in toxicological properties.
Substantial discrepancies among key toxic chemicals in PM2.5
The chemical composition, including carbonaceous species (i.e., organic matters and elemental carbon), water soluble inorganic ions, and elements, widely varied among the PM2.5 originating from the three industrial sources (Supplementary Fig. 4 and Supplementary Table 4). The role of EC was not taken into consideration due to its minor mass concentration in per unit PM2.5 mass and the relative lower toxic potency compared to PAHs and metals. Here, the targeted chemicals are focused on metals and PAHs, which are commonly regarded as key toxic components associated with the biological effects of PM2.517,18. Ten toxic metals (i.e., V, Cr, Mn, Fe, Ni, Cu, Zn, As, Cd, and Pb) were observed at substantially higher concentrations per unit mass of ISI-emitted PM2.5 (253 ± 106 mg/g), exceeding those per unit mass of power industry- (14.9 ± 4.5 mg/g) and the cement industry- (23.5 ± 7.8 mg/g) emitted PM2.5 by approximately 17 and 11 times, respectively (Fig. 2a). Fe dominated the mass concentrations, with relative concentrations ranging from 62.6–98.3%, 43.0–94.8%, and 80.4–94.1% in the selected metals within PM2.5 emitted from the ISI, power, and cement industries, respectively (Fig. 2a). The observation results regarding the variation in toxic metals are consistent with those obtained in our previous study in regard to flue gases emitted from industrial plants35. The concentrations of metals per unit mass of PM2.5 emissions grouped by the prevailing processes in iron and steel production, the fuel type in power-generating units, and the prevailing processes in clinker and cement production are detailed in Supplementary Note 7 and Supplementary Fig. 5a. The targeted toxic metals of concern predominately exist in the particulate phase, not the vapor phase, in typical industrial flue gases. This is mainly due to their substantially higher gasification temperatures36,37,38. Consequently, the condensation of these maters after being discharged from the stack can be disregarded or considered negligible.
Other key toxic chemical in PM2.5, i.e., 16 US Environmental Protection Agency (EPA) priority PAHs (i.e., Nap, Acy, Ace, Flu, Phe, Ant, Flt, Pyr, BaA, Chry, BbF, BkF, BaP, InP, DahA, and Bghip), were also revealed to substantially differ among the three industrial sectors studied (Fig. 2b). The content of these 16 PAHs (686± 775 μg/g) in mass-normalized ISI-emitted PM2.5 was one order of magnitude higher than that in mass-normalized power sector-emitted PM2.5 (27.7 ± 39.2 μg/g) and cement sector-emitted PM2.5 (14.6 ± 14.8 μg/g). Compared to the PAHs emitted from the power and cement industries, higher carcinogenic species (with toxic equivalent factor (TEF) equal to or greater than 0.1) were more abundant in ISI-emitted PAHs, accounting for 31.2–37.8% of the total mass of the 16 PAHs. In contrast, two- and three-ring PAHs (species with a lower toxicity) were the dominant species among the PAHs emitted from power and cement industries, owing to the higher combustion efficiency of their practical boilers than that of the boilers at ISI plants. The corresponding toxic equivalent benzo[a]pyrene values (BaPeq) of the 16 PAHs can directly reflect their potential toxic discrepancies (Supplementary Fig. 6). The BaPeq value per unit mass of ISI-emitted PM2.5 (44.5 ± 78.5 μg/g) was two orders of magnitude higher than that per unit mass of power industry-emitted PM2.5 (0.21 ± 0.23 μg/g) and cement industry-emitted PM2.5 (0.51 ± 0.68 μg/g). Thus, the PAHs originating from the iron and steel production posed a much higher carcinogenic risk than those originating from the other two industrial sectors. The concentrations of the 16 PAHs per unit of PM2.5 emitted from individual ironmaking and steelmaking processes, power generation units driven by different fuel types, and individual clinker- and cement-making processes are detailed in the Supplementary Note 7 and Supplementary Fig. 5b.
The relative contributions of the targeted metals and PAHs to PM2.5-induced OS potency, quantified via the concentration addition reference model39, are shown in Supplementary Fig. 7. The PM2.5-induced OS potency contributed from metals were estimated basing on the tested chemicals performed in the highest valance due to the majority of them are usually in the form of the highest valence in PM2.540,41. The targeted metals contributed 84.6 ± 22.6%, 83.8 ± 9.8%, and 77.5 ± 8.4% to OS potency induced by the primary PM2.5 emitted from the ISI, power, and cement industries, respectively. While the identified PAHs accounted for minor fractions (2.7–3.6%) of the total PM2.5-induced OS potency among the three major sectors (Supplementary Fig. 7a). The results indicated that the PM2.5-induced OS potency could mainly be attributed to PM2.5-bound toxic metals. The higher concentration of toxic metals per unit mass of ISI-emitted PM2.5 led to a much higher ROS-inducing effect than that of the PM2.5 emitted from the power and cement industries. Thus, the unequal toxic potencies of industry-emitted PM2.5 could be attributed to the substantial discrepancies among toxic chemicals in PM2.5 emitted from the different industrial sectors (Supplementary Fig. 7b and c). The toxicity results observed from industrial PM are consistent with epidemiological studies, which indicate that transition metals are likely to be responsible for the adverse effects of PMs, including induce oxidative stress, and impairment in lung function42,43. Furthermore, multiple epidemiological evidence also shows the association between industrial PM emissions and adverse health outcomes44,45,46. On the global scale, industrial sector emissions roughly account for 11.7% of the PM2.5 disease burden44. Then the quantified toxic potencies of industries-emitted PM2.5 could be integrated with emission inventories to develop toxic potency-adjusted emission policies toward healthy air.
The iron and steel industry dominates PM2.5 toxic potency-adjusted emissions
The quantified toxic potencies were integrated into PM2.5 emission inventories as PM2.5-induced toxic potency-adjusted emissions of the three selected industrial sectors in mainland China from 2005–2019, as shown in Fig. 3. As the largest contributor to PM2.5-induced toxic potency-adjusted emissions, the ISI is grouped by its major processes, including sintering, ironmaking, BOF-based steelmaking, and other processes. There existed substantial disproportionalities between the PM2.5 emissions and PM2.5-induced toxic potency-adjusted emissions of the three industrial sectors. ISI-emitted PM2.5 accounted for only 29.0 ± 7.6% of nationwide PM2.5 emissions among the three sectors in 2005, but accounted for approximately 70.7 ± 32.9% of PM2.5-induced OS potency-adjusted emissions (Fig. 3a). While power and cement industries-emitted PM2.5 accounted for 71.0 ± 16.4 % of the total mass emissions, and only 29.3 ± 10.4% of the total PM2.5-induced OS potency-adjusted emissions. Owing to the drastic reduction in PM2.5 emitted from power industry under the enforcement of the toughest ULE standards since 201422, the relative fractional contribution of the ISI increased to 43.8 ± 12.0% of mass emissions and further contributed 81.8 ± 37.3% to the overall PM2.5-induced OS potency adjusted-emissions in 2019 (Fig. 3b).
The trends in total PM2.5-induced OS potency-adjusted emissions of the three industrial sectors between 2005 and 2019, as illustrated in Fig. 3c, substantially differed from the corresponding trends in PM2.5 emissions (Supplementary Fig. 8). The total PM2.5-induced OS potency-adjusted emissions fluctuated and eventually increased by approximately 12.2% between 2005 and 2012, although the total amount of nationwide PM2.5 emitted from the three sectors decreased by 20.0% during the same period. The increase in OS potency-adjusted emissions was mainly attributable to the ISI emissions increase, particularly the emissions originating from BOF-based steelmaking and sintering processes. Since the per unit mass of ISI-emitted PM2.5 exhibit much higher OS toxic potencies, the slight increase (~10.4%) in ISI-emitted PM2.5 resulted in an increase of ~12.2% in the total PM2.5-induced OS potency-adjusted emissions. The reduced OS potency attributed to the reduction in PM2.5 emissions in the power (629 kt) and cement (1462 kt) sectors was offset by the slight increase (716 kt) in the ISI-emitted PM2.5. The PM2.5-induced OS potency-adjusted emissions of the three sectors decreased by ~65.1% between 2012 and 2019. The decreases in the toxic potencies were mainly driven by the corresponding reductions in the ISI sector, with a contribution of ~86.8% to OS potency-adjusted emissions. While the reductions in the power generation and cement production sectors jointly contributed only 13.2% to the overall reduction in PM2.5-induced OS potency-adjusted emissions, although these two sectors jointly accounted for ~47.2% of the reduction in total PM2.5 mass emissions.
The temporal changes in PM2.5-induced CT potency-adjusted emissions in the three industrial sectors exhibited a similar tendency to those in OS potency-adjusted emissions, as shown in Supplementary Fig. 9. Substantial reductions in the PM2.5 mass-based emissions of the power industry, which could be attributed to the implementation of the toughest-ever ULE standards based on the PM mass concentration, do not necessarily suggest an equivalent alleviation of PM2.5 health effects, as evidenced, for example, by the toxic potency (including OS and CT potencies)-adjusted emissions in the present study. In contrast, the slight increases or decreases in ISI-emitted PM2.5 substantially dominated the variations in PM2.5 toxic potency-adjusted emissions among the three sectors. The disproportionate variation in nationwide PM2.5 toxic potency-adjusted emissions was mainly due to the unequal source-specific toxic potencies at equal PM2.5 mass concentrations. Furthermore, regional discrepancies in the industrial structure among the three sectors could lead to spatial variations in PM2.5 emissions and toxic potency-adjusted PM2.5 exposure risks nationwide.
Priority control in the iron and steel industry can result in a cost-effective reduction in the population-weighted PM2.5 exposure risk
Two reduction routes were developed to quantify the reduction effects on mass-based industrial PM2.5 emissions and toxic potency-adjusted industrial PM2.5 emissions (including the estimation of OS and CT potencies) if all facilities in power, cement, and iron and steel industries meet the ULE levels, respectively (Fig. 4a). The implementation of ULE standards in the power industry (1000 kt) could yield a nearly 1.5- and 1.6-fold higher reduction in nationwide mass-based PM2.5 emissions than that obtained via the implementation of ULE standards in the ISI (621 kt) and cement industry (653 kt), respectively. However, the implementation of ULE standards in the ISI could lead to the largest reduction in PM2.5-induced toxic potency-adjusted emissions (including the estimation of OS and CT potencies) among the three industrial sectors owing to the unequal OS and CT potency at an equal PM2.5 mass, respectively. Spatial distributions of the ambient PM2.5 concentration contributed by primary PM2.5 originating from the three sectors before and after ULE standards implementation in each sector are shown in Supplementary Fig. 10. The nationwide primary PM2.5 concentrations originating from the three industries if all existing units in the power, cement, and iron and steel industries, respectively, satisfied ULE standards in 2019 are shown in Supplementary Fig. 11. The nationwide annual mean population-weighted PM2.5 exposure could decrease by 37.6%, and 26.0%, and 36.4% if all facilities in power, cement, and ISI, respectively, met ULE standards in 2019 (Supplementary Table 5). The toxic potency-adjusted exposure risk (TPAE) metric was further developed to estimate a health-relevant risk index for population exposure to industrial emitted PM2.5, including estimation of PM2.5-induced OS potency- and CT potency-adjusted exposure risks (TPAEOS and TPAECT). The cited toxic equivalent values of PM2.5-induced OS and CT potencies are provided in Supplementary Table 6.
Figure 4b−d show the spatial distributions of the corresponding reduction in the total TPAEOS value attributed to meeting ULE standards in the power, cement, and ISI in 2019, respectively. The decrease in the power and cement industries only accounted for a minor faction of the total reduction in the nationwide TPAEOS value if all their units achieved ULE standards in 2019 (Fig. 4b and c). The highest reductions in power industry- and cement industry-related TPAEOS could be occurred in north and central China, respectively, which were consistent with the density of power and cement plants. While the substantial reduction in the nationwide TPAEOS value could be obtained if all ISI facilities satisfied ULE standards in 2019 (Fig. 4d). Much higher reductions in ISI-associated TPAEOS could be observed on the northern and eastern China, especially in densely populated areas such as Hebei and Jiangsu provinces. Since these populated areas concentrated with a large portion of ironmaking and steelmaking plants, and majority of these plants could attain ULE levels due to the priority implementation of ULE standards and a range of advanced technologies23. TPAEOS values attributed to each industry source with scenarios before and after ULE standards introduction are shown in Supplementary Fig. 12. The implementation of ULE standards in the power industry yielded the largest reduction in nationwide PM2.5 emissions (Fig. 4a) and PM2.5 concentration (Supplementary Figs. 10 and 11) among the three industrial sources. However, the implementation of ULE standards in the ISI sector could dominate the reduction in the nationwide TPAEOS value among the three sectors if the unequal OS potencies of source-specific PM2.5 were included (Supplementary Table 4).
The nationwide total population-weighted TPAEOS in the three sectors was dominated by the ISI (83%), owing to the higher OS potency per unit mass of ISI-emitted PM2.5. The implementation of ULE standards in the ISI sector could generate 5.4 ± 4.0 and 8.2 ± 6.6 times higher reduction in the nationwide population-weighted TPAEOS than that obtained via the implementation of ULE standards in the power and cement industry sectors in 2019, respectively (Supplementary Table 7). The reduction in PM2.5-induced CT potency-adjusted emissions and the reduction effect on the nationwide TPAECT as well as population-weighted TPAECT exhibited similar tendencies to that of TPAEOS in mainland China (Fig.4a and e−g, Supplementary Fig. 13 and Supplementary Table 7). The results indicate that implementation of ULE standards in the ISI sector can achieve the largest reduction effect in both nationwide TPAEOS and TPAECT in the three industrial sectors. Furthermore, the marginal reduction costs for PMs emitted from industries can be integrated into the analysis when installation and operating costs of dedusting systems to meet ULE standards are considered.
Figure 5 shows the marginal reduction costs for per unit of industry-emitted PM2.5 mass and per unit of toxic potency-adjusted industry emitted-PM2.5 mass under the scenario with all facilities in each industrial sector meeting ULE standards across the nation in 2019. The marginal abatement cost per ton of the PM2.5 mass of the ISI sector was nearly 73% and 142% that per ton of the PM2.5 emitted from the power and cement industries (Fig. 5a), respectively, while the abatement cost for per unit of OS potency-adjusted mass of ISI-emitted PM2.5 was only 14% and 25% that for per unit of OS potency-adjusted mass of power- and cement-emitted PM2.5 (Fig. 5b), respectively. The marginal cost for abating per unit mass of the CT potency-adjusted ISI-emitted PM2.5 is only 12% and 15% that for per unit mass of CT potency-adjusted power- and cement-emitted PM2.5 (Supplementary Fig. 14), respectively. The cost estimates indicated that reductions in the emission sources of higher toxic PM2.5 could greatly benefit the decrease in PM2.5-induced OS potency- and CT potency-adjusted mass emissions instead of the reduction in the largest PM2.5 emission source. Thus, we suggest that priority control in the ISI could achieve a much more cost-effective reduction in the adverse impacts associated with exposure to PM2.5 originating from the three industrial sources.
Discussion and policy implication
The unequal toxic potencies of primary PM2.5 emitted from the different industrial sources, including the ISI, power, and cement industries, were elucidated via real-world measurements and laboratory analysis under the same quality control conditions (including field and procedure blank correction). Oxidative stress is established as one of the important mediators leading to adverse health outcomes of air pollution. Therefore, oxidative potential is a good and robust starting point to compare the mass-based and health-oriented frameworks in both technical and economic considerations16. Furthermore, increasing number of epidemiological studies have found that OP was more strongly associated with cardiorespiratory disease than PM2.5 mass47,48, indicating that OP may be a more health relevant health metric than PM2.5 mass. However, this health-relevant metric is still absent in current control policies based on pollutant mass concentrations worldwide3,16. As the conventional top polluting industry ranked by the total PM mass emissions, the toughest control policies were primarily enforced at power plants. However, the reduction in the nationwide TPAE value attributed to the implementation of ULE control policies at power plants accounted for only 6.6–13.1% of total TPAE reduction among the three industrial sectors across mainland China, owing to the low toxic potency per unit mass of power industry-emitted PM2.5. In contrast, the decrease in the ISI after the implementation of ULE standards at all facilities dominated the nationwide reduction in the total TPAE value (76.2–86.7%) among the three sectors in 2019 across mainland China. Furthermore, the marginal abatement cost for the mitigation of per unit mass of toxic potency-adjusted ISI-emitted PM2.5 is only 12–25% that for the mitigation of per unit mass of toxic potency-adjusted power industry-emitted PM2.5 under the scenario whereby all ISI facilities in mainland China reach satisfied standards in 2019 (Fig. 5). Therefore, the ISI should be prioritized among the three sectors, given its cost-effective role in reducing adverse impacts associated with exposure to industry emitted-PM2.5. We highlight that it is urgent to shift from the mass-based control strategy to the health-oriented control policies via integrating unequal toxicities of industrial particulate emissions.
The implementation of stricter control measures targeting ISI-emitted PM2.5 could be a big challenge in practice because iron and steel production involves complex manufacturing processes and multiple emission sources23, including the major processes of sintering, ironmaking, and BOF-based steelmaking. Due to their highest PM2.5 emissions and highest PM2.5-induced toxic potency values, these major processes represent important targets for further adverse impacts reduction. Additionally, compared to the effects of conventional BOF-based steelmaking processes, EAF-based steelmaking processes could alleviate PM2.5-induced toxic potencies, as indicated by our limited EAF-emitted PM2.5 samples (Supplementary Fig. 3). The obtained toxicity results expand the previous recognition of EAFs as an advanced technology option for the mitigation of ISI-emitted air pollutants31. The shift from BOFs to EAFs is recommended over the next decades to achieve carbon neutrality and reduce ISI-emitted air pollutants, especially in high income regions. The uncertainties in the toxicity of EAF-emitted PM2.5 may hinder the evaluation of the effectiveness of current EAF technology in addressing the health effects of PM2.5 pollution. Hence, future efforts are warranted to quantify and clarify the toxic potency of EAF-emitted PM2.5. The obtained understanding may offer important insights to policymakers formulating targeted control policies to achieve air quality, climate, and health cobenefits in the ISI.
Several limitations and uncertainties in field investigations and estimations exist in this study. This study focused on the filterable PM (refers to particles that are directly emitted as a solid or liquid at stack conditions), while condensable PM (the material in vapor phase at the stack but condenses to form solid or liquid PM immediately after discharge from the stack into the ambient air) originating from industrial sectors, as well as source-specific secondary and fugitive aerosols emission from manufacturing iron/steel and cement processes, were not considered in this study, which could result in underestimation of the adverse impacts associated with exposure to specific source-emitted aerosols. The observed toxic potencies based on in vitro cell tests are not strictly consistent with the actual PM2.5 exposure environment in the human body, and thus, toxic estimation methods limit the understanding of real-world human exposure scenarios. In future, it is necessary to investigate the potential toxicities of industrial condensable PMs and explore the quantitative connection between their toxic potencies and the corresponding chemical composition. Furthermore, quantifying the role of condensable PMs in adjusting the toxicity of filterable PMs is crucial for a comprehensively understanding of the adverse impacts originating from industry sources. Additionally, the industry source-emitted secondary and fugitive aerosols and their induced toxic potencies are encouraged to be quantified to comprehensively estimate the adverse impacts associated with exposure to industry-emitted PM2.5. Wet plumes, formed when saturated wet flue gas is discharged from stack and enters to the atmosphere, are frequently observed in industrial plants49. However, the toxicity of wet plumes remains unclear, emphasizing the need for systematic studies to address these issues in the future. Toxicological studies (in vitro and in vivo tests) are urgently needed to quantify the interactive effects of individual chemical, such as sulfate and transition metals, and to clarify and validate the associations by integrating with epidemiological studies. In addition, more realistic toxicological exposure studies, including human organoid-based in vitro model, are required to elucidate the multiple health-relevant end points associated with exposure to industry-emitted PM2.5. In vivo and epidemiological studies are strongly encouraged to clarify and validate the relationship between exposure to industry-emitted PM2.5 and adverse health outcomes. On the other hand, corresponding ambient air PM samples may be collected in future studies for better risk assessment for large populations. The mixture effects of ambient PM2.5, particularly in industrial areas, should be determined to better characterize and to dissect the quantitative contribution of toxic components and finally elucidate contributions of hazardous sources to mixture effects. Multilevel evidence could shed light on reducing the uncertainty in the estimation of the health effects of industry PM2.5 pollution and the formulation of tailored industry source control policies for health protection.
Despite the above limitations, our analysis could offer a perspective to policymakers, thereby emphasizing that a focus shift from emission mass-oriented policies to health-oriented control policies could help facilitate the most cost-effective reduction in the health effects of industry PM2.5 pollution. The results may provide guidance for low and medium income regions suffering from similarly severe air pollution driven by multiple industrial sources that are seeking economic approaches for the mitigating of the health effects of industry-related emissions, such as India and Brazil50. Since the industry structure, manufacturing processes, and energy consumption exhibit large discrepancies among different regions worldwide, future work to quantify and rank the toxicity of individual industry source-emitted PM2.5 (including condensable PM2.5) are strongly recommended for further quantitative cost-benefit analysis and finally to formulate cost-effective control strategies to effectively mitigate source-specific PM2.5 effects on public health.