Evaluation of the ISORROPIA-II model simulations and the influence of ALWC on volume growth factor (VGF) of aerosols
The ISORROPIA-II model was run in the "forward" mode using the hourly measured total (gas + particle) species concentrations instead of only particle-phase concentrations as input, as it is considered to be more accurate [20; 21]. As shown in Figure S1, the predicted PM1 NH4+ (r = 0.97), NO3- (r = 0.93), Cl- (r = 0.98), PM2.5 NH4+ (r = 0.87), NO3- (r = 0.98), Cl- (r = 0.99), and gas-phase NH3 (r = 0.93) showed very good correlation with the measured concentrations, confirming the reliability of the ISORROPIA-II model simulations at a range of temperatures (278–298 K).
The data points are color-coded by ambient temperatures to investigate how temperature variability can alter the ISORROPIA-II predicted gas and particle-phase concentrations. The pH and ALWC of PM1 and PM2.5 were estimated during 19 December 2017 to 10 February 2018. The predicted pH of PM1 varied between 2.2 to 5.6, and the mean PM1 pH (average ± SD) was 4.5 ± 0.5. The PM2.5 pH ranged from 2.5 to 6.5, with a mean value of 4.6 ± 0.5. The predicted pH of PM2.5 was similar to the measured PM2.5 pH of 4.6 ± 0.5 over Delhi, estimated during the Winter Fog EXperiment (WiFEX) campaign period of 2015-16, underscoring the reliability and accuracy of the ISORROPIA-II model simulation [18; 22].
As the high ammonium chloride concentration is a special characteristic of air pollution in India, it is important to explore its variability in prevailing meteorological conditions. To determine the impact of liquid water content on the chloride fraction and volume growth factor (VGF) of aerosols, here we investigate the growth of aerosols with increasing RH. The VGF is defined as the ratio of the volume of the wet particle to the corresponding volume of the dry particle [17; 23], and is calculated using the following equation:
$$VGF=\frac{\frac{\sum {m}_{i. MARGA}}{{\rho }_{i}}+(\frac{{ALWC}_{inorg}+{ALWC}_{org})}{{\rho }_{water}}}{\frac{\sum {m}_{i. MARGA}}{{\rho }_{i}}} \left(1\right)$$
where miMARGA is the mass concentration of species 'i' measured by MARGA-2S [24].
In this calculation, we have neglected the impact of organic fractions as studies have shown their volumetric growth is much less than water-soluble inorganic ions [25; 26; 27]. The major inorganic constituents of fine aerosols in IGP are chloride, sulfate, nitrate, and ammonium (CSNA), with their densities prescribed to be 1.52, 1.75, 1.75, and 1.75 g cm-3 respectively [28; 29].
Figures 1a and 1b show the VGF variability with RH for both PM1 and PM2.5. The sizes of the circles are scaled to the aerosol liquid water content (ALWC) and colored according to the chloride mass fractions (%) in PM1 and PM2.5 respectively. It can be seen that VGF for PM1 and PM2.5 increase exponentially with increasing RH and varies as VGF = exp (a * RH) + b, with high co-efficient of determination values for both PM1 (R2 = 0.94) and PM2.5 (R2 = 0.92) respectively. Below 70% RH, the mean PM1 VGF was 2.02 ± 0.77, which increased to 7.87 ± 4.58 when RH ≥ 70%. PM2.5 VGF was 2.08 ± 0.97 at RH ≤ 70% and increased to 7.38 ± 4.46 at RH ≥ 70%, causing a significant reduction in visibility and worsening air pollution. The steeper slope in VGF with increasing RH is attributed to the water uptake by hygroscopic constituents and the growth of aerosols due to enhanced multiphase reactions in highly humid conditions [30; 31; 32].
It can also be seen that aerosols with higher chloride mass fraction uptake more water than those with less chloride mass fraction under prevailing RH conditions. This is due to the co-condensation of HCl, NH3, and water, as the gas-phase HCl gets dissolved in aerosol water, dissociates, and then equilibrates with ammonia to form ammonium chloride, stabilizing chloride in the particle phase [33]. This particle-phase chloride can absorb even more water from the air, leading to enhanced growth of aerosols into fog droplets during winter and augmenting the particle mass loading [21]. These results demonstrate the role of ALWC in the growth of PM1 and PM2.5 aerosols, which needs to be thoroughly investigated for a deeper understanding of the complex thermodynamical control of high aerosol loading.
Sensitivity of gas-to-particle phase partitioning (ε) to pH and ALWC during winter
The average chloride partitioning ratio ε(Cl-) of PM1 and PM2.5 was 0.93 ± 0.09 and 0.96 ± 0.07 respectively (Table S1), implying the dominant presence of chloride in the particle phase during winter. The ε(Cl-) was 0.4 at RH ≤ 50%, which sharply increased to 0.95 at RH ≥ 80%, showing the enhanced phase-partitioning to the particulate chloride phase. The increased phase-partitioning of total chloride (HCl + Cl-) in highly humid conditions can further promote chloride formation caused by increased ALWC. The enhanced ALWC increases pH by dilution, further increasing total chloride partitioning and significantly increasing PM1 and PM2.5 chloride in a positive feedback loop [17; 34].
The average ε(NO3-) of PM1 and PM2.5 was 0.83 ± 0.11 and 0.89 ± 0.08 respectively, showing the dominance of particle-phase nitrate over gas-phase HNO3 during winter. The high particle-phase nitrate concentrations significantly impact the total aerosol loading as the presence of more ammonium nitrate reduces the mutual deliquescence relative humidity (MDRH) of aerosols resulting in the formation of more secondary aerosols in polluted conditions [35; 36; 37]. As the efflorescence RH (ERH) of these aerosols are generally low and rarely reached by the ambient RH during winter, these species primarily remain in the aqueous or metastable phase [31]. The average ε(NH4+) for PM1 and PM2.5 was 0.42 ± 0.17 and 0.55 ± 0.15 respectively, attributed to the substantial excess ammonia left in the gas-phase. These phase-partitioning ratios in Delhi are in contrast to other regions, where ε(NO3-) was reported to be 0.26 ± 0.15 and 0.39 ± 0.16 in the USA, and ε(NH4+) was 0.2 ± 0.1 in China, and 0.55 ± 0.25 in the USA [38; 39]. The extremely high ALWC in ammonia-rich Delhi triggers the water uptake and plays a crucial role in gas-to-particle phase partitioning and developing high secondary inorganic aerosol (SIA) loading during winter.
We estimated the sensitivity of the phase partitioning of nitrate [ε(NO3-)], chloride [ε(Cl-)], and ammonium [ε(NH4+)] to pH, ALWC, and T [40; 41; 42] following the methodology given in Guo et al., 2015 as
$$\epsilon \left({NO}_{3}^{-}\right)=\frac{{H}_{{HNO}_{3}}^{*} R T {ALWC}_{i}\times 0.987 \times {10}^{-14}}{{\gamma }_{{H}^{+}}{\gamma }_{{NO}_{3}^{-}}{10}^{-pH}+ {H}_{{HNO}_{3}}^{*} R T {ALWC}_{i}\times 0.987 \times {10}^{-14}} \left(2\right)$$
$$\epsilon \left({Cl}^{-}\right)=\frac{{H}_{HCl}^{*} R T {ALWC}_{i}\times 0.987 \times {10}^{-14}}{{\gamma }_{{H}^{+}}{\gamma }_{{Cl}^{-}}{10}^{-pH}+ {H}_{HCl}^{*} R T {ALWC}_{i}\times 0.987 \times {10}^{-14}} \left(3\right)$$
and,
$$\epsilon \left({NH}_{4}^{+}\right)=\frac{\frac{{\gamma }_{{H}^{+}} {10}^{-pH}}{{\gamma }_{{NH}_{4}^{+}}} {H}_{{NH}_{3}}^{*} R T {ALWC}_{i}\times 0.987 \times {10}^{-14}}{1+\frac{{\gamma }_{{H}^{+}} {10}^{-pH}}{{\gamma }_{{NH}_{4}^{+}}} {H}_{{NH}_{3}}^{*} R T {ALWC}_{i}\times 0.987 \times {10}^{-14}} , \left(4\right)$$
where γ is the activity coefficient of protonated species in the aqueous medium, and ALWCi is the water associated with the inorganic constituents (µg m-3). H* is the equilibrium constant of HNO3, HCl, and NH3 adopted from [41] and [43] using molality-based units of mol2 kg-2 atm-1 [44; 45]. R is the universal gas constant (8.314 J K-1 mol-1), and the value 0.987 is for the transformation of 1 atm to 1 bar. The equations describe the HNO3-NO3-, NH3-NH4+, and HCl-Cl- partitioning, and the estimated values of ALWCi and T were used to evaluate the phase-partitioning of ε(NO3-), ε(Cl-), and ε(NH4+) at different pH regimes.
Figures 2a and 2b show the variability of gas-to-particle partitioning of chloride and nitrate with pH, ALWC, and T following equations 2–4. Three prominent zones are shown, in which ε(Cl-) and ε(NO3-) vary between the complete gas-phase (ε ~ 0%) to the complete particle-phase (ε ~ 100%). In the blue-color zone, ε(Cl-) and ε(NO3-) asymptotically approach 0, and the total species primarily remain in the gas-phase. In the gray-color zone, ε asymptotically approaches 1, and the whole species is in the particle phase, whereas in the yellow-color zone, ε varies between 0 and 1, and the species remain as a mixture of gas and particle-phase. A thermodynamical sweet spot, pH50 is defined, where ε(Cl-) and ε(NO3-) are 0.5, and total chloride and nitrate remain 50% in the gas-phase and 50% in the particle phase.
The purple band shows the variability in ε due to variation in ALWC at a constant pH, indicating the significance of ALWC in modulating the particle-phase loading. From Fig. 2a, it can be seen that ε(NO3-) can vary between 70–100% at pH ≈ 4, due to variability in ALWC. The red line is the sigmoidal curve (S) fitting, which depicts the sensitivity of ε to pH and ALWC over a location. The estimated average pH of PM1 and PM2.5 of 4.49 and 4.58, respectively (Section 3.1), falls on the flat side of the S curves in the blue-color zone. Here, chloride and nitrate remain almost exclusively in the particle-phase, and ammonia remains primarily in the gas-phase. The possibility that emerges as the pathway to modulate high aerosol concentrations is reducing precursors like HCl and HNO3, which is feasible as it would not adversely affect agricultural productivity and potentially impact the ecosystem.
Aerosol mass loading sensitivity to HCl, HNO3, and NH3 perturbations in the IGP
We explored a thermodynamically consistent mathematical framework to reduce ammonium chloride and ammonium nitrate concentrations in the IGP. Following equations 5–13, and using pH and ALWC as coordinates, we defined six "sensitivity regimes", where aerosols are sensitive to HCl, HNO3, and NH3 perturbations. As chloride dominates the inorganic mass fraction of fine aerosol in the IGP, we specifically investigate the aerosol mass sensitivity to HCl emissions to define an "HCl sensitive regime". This is significantly different from studies conducted in China, USA, and Europe where researchers have investigated aerosol sensitivity to only HNO3 and NH3 and did not investigate the "HCl sensitive regime" [4; 5].
The "sensitivity regimes" are shown in different colors in Fig. 3, where 1100 hourly observational data points are plotted to check the instantaneous response of aerosol loading to HCl, HNO3, and NH3 variability. Figures 3a and 3b show PM1 and PM2.5 aerosols to remain in the off-white shaded region, where aerosols respond proportionally to changes in the HCl and HNO3 emissions but tend to be insensitive to NH3 emissions. The off-white region is defined as an "HCl and HNO3 sensitive regime", indicating that HCl and HNO3 reduction would be the most effective pathway in controlling aerosol pollution over IGP. The blue-shaded region is defined as an "HCl, HNO3, and NH3 sensitive regime", where aerosols are sensitive to HCl, HNO3, and NH3. It can be seen that very few observational data points fall in this regime, in contrast to the USA, where most of the aerosols fall in this regime [4]. During winter, ammonia concentration is much higher than HCl and HNO3 in IGP, but fine aerosols are not sensitive to NH3 variations. Instead, HCl and HNO3 are by far the limiting factors in aerosol loading, which should be controlled by controlling the major HCl and NOx emissions over IGP.
In Figs. 4a and 4b, all the data points are color-coded with PM1 and PM2.5 concentrations, respectively, demonstrating that higher aerosol loading is often associated with higher ALWC. Interestingly, the ALWC usually ranges between tens to hundreds of micrograms per cubic meter in climatic regions like China and USA [4; 5], but over IGP the ALWC is an order of magnitude higher, sometimes reaching ~ 2000–2400 µg m-3 for PM1 and PM2.5. This high ALWC significantly influences the loading of SIA and causes a reduction in visibility as evidenced by Henry's law, which shows that particles with high ALWC would take up more gaseous pollutants, and the equilibrium would lead to an increase in their water content and the formation of more secondary aerosols like ammonium chloride and ammonium nitrate [46; 47; 48].
In Fig. 5, we have schematically shown the physical, chemical, and thermodynamical processes influencing aerosol loading. It shows the dominant impact of multiphase and heterogeneous chemical processes on the aerosol growth processes in the wintertime polluted atmosphere in the IGP. From a thermodynamic perspective, it can be seen that acidic gases are first absorbed on the surface layer of aerosols, and heterogeneous reactions rapidly occur at the surface resulting in the rapid increase in secondary aerosol mass concentrations. The newly formed particle mass then gets dispersed through the liquid phase in high ALWC, where multiphase reactions govern. The higher surface-area-to-volume ratio of PM1 than PM2.5 suggests the heterogeneous chemistry in PM1 to be more crucial than PM2.5.
Recent studies conducted in Europe, the USA, and China suggested that ammonia reduction is more cost-effective than NOx reduction and would be the most effective pathway to reduce aerosol loading [5; 39; 49]. But in this study, we argue that the sensitivity and effectiveness of the adopted reduction mechanisms are more crucial than the cost-effectiveness. If aerosols are not sensitive to the reductions of a specific precursor, then the cost-effectiveness would not assist in developing an effective mitigation policy. The results that we present illustrate the importance of using thermodynamically consistent sensitivity analysis to effectively address the particulate matter pollution mitigation problem in the Indian region.