Size-resolved aerosol pH over Europe during summer

The dependence of aerosol acidity on particle size, location and altitude over Europe during a summertime period is investigated using the hybrid version of aerosol dynamics in the chemical transport model PMCAMx. The pH changes more with particle size in northern and southern Europe owing to the enhanced presence of non-volatile cations (Na, Ca, K, Mg) in the 15 larger particles. Differences of up to 1-4 pH units are predicted between suband super-micron particles, while the average pH of PM1-2.5 can be as much as 1 unit higher than that of PM1. Most aerosol water over continental Europe is associated with PM1, while PM2.5-5 and PM5-10 dominate the water content in the marine and coastal areas due to the relatively higher levels of hygroscopic sea salt. Particles of all sizes become increasingly acidic with altitude (0.5-2 units pH decrease 20 over 2.5 km) primarily because of the decrease in aerosol liquid water content (driven by humidity changes) with height. Inorganic nitrate is strongly affected by aerosol pH with the highest average nitrate levels predicted for the PM2.5-5 range and over locations where the pH exceeds 3. Dust tends to increase aerosol water levels, aerosol pH and nitrate concentrations for all particle sizes. This effect of dust is quite sensitive to its calcium content. The size-dependent pH differences carry 25 important implications for pH-sensitive processes in the aerosol.


Introduction
Acidity is an aerosol property of central importance driving gas-particle partitioning and heterogeneous chemistry (Pye et al., 2019). pH affects the formation of semi-volatile particulate 30 matter and the nitrogen cycle by modulating HNO3/NO3and NH3/NH4 + gas-particle partitioning (Meskhidze et al., 2003;Guo et al., 2017;Nenes et al., 2019). Aerosol acidity can influence pH-https://doi.org/10.5194/acp-2019-1146 Preprint. Discussion started: 19 February 2020 c Author(s) 2020. CC BY 4.0 License. dependent heterogeneous atmospheric processes, like oxidation of SO2 to sulfate, formation of secondary organic aerosol and uptake of N2O5 on particles (Huang et al., 2011) and also influences aerosol hygroscopicity (Hu et al., 2014). Deposition of acidic particles causes damage on building 35 materials, forests, and aquatic ecosystems (Xue et al., 2011). Aerosol pH can change the solubility of metals, such as iron and copper, which have been linked to aerosol toxicity, and at the same time affects nutrient distributions with impacts on photosynthesis productivity and ocean oxygen levels (Meskhidze et al., 2003;Nenes et al., 2011). Adverse health outcomes have been linked to aerosol acidity, including respiratory diseases (Raizenne et al., 1996), oxidative stress (Fang et al., 2017) 40 and lung and laryngeal cancers (Hsu et al., 2008).
The nitrate partitioning to the aerosol phase is favored when pH exceeds a threshold value (between 1.5 and 3) that depends logarithmically on liquid water content and temperature (Meskhidze et al., 2003;Guo et al., 2016;Nenes et al., 2019). If aerosol pH is high enough (typically above 2.5 to 3), aerosol nitrate formation is favored, as most of the total nitrate formed 45 from NOx chemistry resides in the aerosol phase. For lower pH values (below 1.5 to 2), formation of aerosol nitrate is not favored and remains in the gas phase as HNO3. Between these pH value limits, a sensitivity window (of 1 to 1.5 pH units) exists in which nitrate can be found either as gas or as aerosol (Vasilakos et al., 2018;Nenes et al., 2019). Atmospheric aerosol has often pH values inside this sensitivity window, for which pH errors could translate to importance biases in aerosol 50 composition (Bougiatioti et al., 2016;Guo et al., 2015Guo et al., , 2017Vasilakos et al., 2018).
Aerosol acidity and partitioning of semi-volatile species, like nitrate, can be modulated by the presence of soluble inorganic cations of sea salt and mineral dust, such as Na + , K + , Ca 2+ and Mg 2+ (Vasilakos et al., 2018). These non-volatile cations (NVCs) tend to reside in the coarse mode of ambient aerosol (sea salt, dust), with much lower concentration in smaller particles (Seinfeld and 55 Pandis, 2006). Chemical transport models tend to overpredict aerosol inorganic nitrate levels in both US and Europe (Yu et al., 2005;Pye et al., 2009;Fountoukis et al., 2011;Tuccella et al., 2012;Heald et al., 2012;Walker et al., 2012;Im et al., 2015;Ciarelli et al., 2016;Zakoura and Pandis, 2018;Zakoura and Pandis, 2019). One of the reasons for these errors is that these models do not simulate properly the aerosol acidity introducing errors in gas-particle partitioning of semi-volatile 60 species, often affecting predictions of inorganic nitrate (Vasilakos et al., 2018).
The aerosol pH has been estimated combining field measurements and aerosol thermodynamic models. Katoshevski et al. (1999) estimated the aerosol pH in the marine boundary layer using the thermodynamic model ISORROPIA and found that it ranged between -0.5 to 9 for particle diameters smaller than 1 μm up to 10 μm. pH was estimated to be 0 to 2 for the accumulation mode and 2-5 for the coarse mode particles using aerosol and gas phase data collected over the Southern Ocean in combination with the EQUISOLV ΙΙ model (Fridlind and Jacobson, 2000). Keene et al. (2004) (Huang et al., 2011). Guo et al. (2015) estimated that PM1 particle pH varied from 0.5 to 2 in the summer and 1 to 3 in the winter in the Southeastern US. PM1 pH was estimated for the 75 northeastern US and its mean value was 0.77 (Guo et al., 2016). PM2.5 pH values of 0-2 were estimated combining ISORROPIA II and data collected at a rural southeastern US site during summer 2013 . Based on impactor measurements in Atlanta, GA during the spring of 2015, Fang et al. (2017) calculated a mean pH value of 3.5 for the coarse mode particles using the ISORROPIA II model. Guo et al. (2017) calculated PM1 and PM2.5 pH (equal to 1.9 and 80 2.7) from measurements during the CalNex study in combination with ISORROPIA II. An average PM1 pH equal to 2.2 was estimated in a rural southeastern US site using ISORROPIA II (Nah et al., 2018). Vasilakos et al. (2018) used the three-dimensional chemical transport model, CMAQ, along with ISORROPIA II, to predict the annual average PM2.5 pH over the Eastern US for 2001 and 2011 (pH equal to 1.6 and 2.5, respectively). Bougiatioti et al. (2016)  Most of the previous studies focused on the average pH of a particular size range neglecting 90 potential pH variation with particle diameter. There is evidence that pH may vary by as much as 6 units between particle diameters of 0.1 μm to 10 μm (Fang et al., 2017;Ding et al., 2019). The majority of previous work has focused on select locations in the US, Canada and Asia and there is still little information about Europe. Also, there is only one study that links aerosol acidity with altitude (Guo et al., 2016), indicating the need for further investigation.

95
The aim of our work is to investigate the size-dependent aerosol pH over Europe. owing to the large concentration of NH3, nitrate, sulfate and dust across all sizes. The role of dust, Ca 2+ and the variation of aerosol pH with altitude are analyzed in detail.

Model description
PMCAMx Karydis et al., 2010) is based on the CAMx air quality model (Environ, 2003) to simulate the processes of horizontal and vertical advection, horizontal and vertical diffusion, wet and dry deposition, gas-and aqueous-phase chemistry. A sectional approach 105 is used to dynamically track the evolution of the aerosol mass and composition distribution across 10 size sections covering a diameter range from 40 nm to 40 μm. The aerosol components modeled include sulfate, nitrate, ammonium, sodium, chloride, calcium, potassium, magnesium, other inert crustal material, elemental carbon, water, primary and secondary organic species. The gas-phase chemical mechanism used in this application is based on the SAPRC mechanism (Carter, 2000;110 Environ, 2003). The version of SAPRC mechanism used here includes 237 reactions of 91 gases, 18 radicals and 37 aerosol species. The thermodynamics of inorganic species was simulated using the ISORROPIA II model (Fountoukis and Nenes, 2007). Additional details regarding PMCAMx are provided in Fountoukis et al. (2011).
We use the hybrid approach to model inorganic aerosol mass transfer, where for particles 115 with dry diameters less than 1 μm, bulk equilibrium is assumed. For larger particles, the mass transfer to each size section is simulated using the Multicomponent Aerosol Dynamics Model (Pilinis et al., 2000). Trump et al. (2015) used the hybrid approach over Europe to improve the simulation of coarse particle chemistry. They found that PM1 nitrate overprediction in areas with high sea-salt levels was reduced with the hybrid approach due to the more accurate representation of 120 the interaction of nitric acid and ammonia with coarse mode sea salt. These interactions result in reduction of fine nitrate and increase of nitrate in the coarse mode. Given the importance of pH on the partitioning of nitrate, the hybrid approach is essential for capturing the size-resolved variability of pH.
pH is calculated in this work for particles smaller than 1 μm, 1-2.5 μm, 2.5-5 μm and 5-10 where [H + ] and [W] are the concentrations of particle hydronium ion and particle water in μg m -3 .

130
PMCAMx was applied over Europe, during the EUCAARI summer intensive campaign in May 2008 for which the model has been evaluated in previous work (Fountoukis et al, 2011). The domain covers a 5400×5832 km 2 region with 36×36 km grid resolution and 14 vertical layers extending up to 6 km. Inputs to the model include horizontal wind components, vertical diffusivity, temperature, pressure, water vapor, clouds and rainfall, all generated using the Weather Research and Forecast

135
(WRF) meteorological model (Skamarock et al., 2005). Anthropogenic gas-phase emissions include land emissions from the GEMS dataset (Visschedijk et al., 2007)  Three simulations were performed. The first was the "base case" simulation and included all emissions described above. Two other simulations are carried out and compared with the "base case" to understand how NVCs in dust affect water uptake and aerosol pH: one where dust lacks any 150 non-volatile soluble cations ("inert dust" simulation) and one where we neglect calcium (the major NVC in dust) from the "base case" simulation. Calcium is unique compared to the other NVCs in that it can react with sulfate ions and form insoluble CaSO4, which precipitates out of the aerosol aqueous phase and remains insoluble under subsaturated conditions -even for metastable aerosol (Fountoukis and Nenes, 2007). This unique interaction implies that Ca, if present in sufficient 155 amounts, can reduce aerosol sulfate and reduce acidity, but at the same time reduce hygroscopicity that promotes acidity-in a way that is not obvious by just comparing the base case simulation with the "inert dust" simulation. In all simulations, the total dust mass emissions were the same and only its assumed composition varied.

Size dependence of aerosol pH
The average ground level pH predictions for different size ranges are presented in Fig. 1. pH is 165 higher over the sea for all particle sizes compared to continental regions due to the presence of sea salt, lower NH3 and the systematically higher RH and liquid water contentall of which act to reduce aerosol acidity. The pH of marine aerosol increases with particle size, with the highest value equal to 4.5 for the 2.5-5 μm and 5-10 μm ranges, as sea salt is emitted mainly at the super-micron range and is the main aerosol component. Over the continental region, average PM1 pH ranges the average pH is in the 1.5-3 range for all particle sizes. The largest pH changes across size occur for regions where fine-mode aerosol acidity is dominated by the NH3-SO4 system (i.e., relatively lower NH3 levelsso that aerosol nitrate is low; Guo et al., 2018), and the largest sizes contain large amounts of NVCs from sea salt and dust. The pH of PM2.5 has often been the focus of previous measurement studies, due to the availability of the corresponding filter samples. However, the pH in the 1-2.5 range can be quite different from that in the sub-micrometer range (Fang et al., 2017;Ding et al., 2019). This difference may have important implications for aerosol toxicity, metal solubility, nitrate partitioning and other processes. The difference of average ground level aerosol pH predictions between PM1-2.5 and PM1 is shown in Fig. 2. The pH of these size ranges can differ up to 1 unit over the continental region.
This difference is even higher over the ocean (up to 1.4 units), owing to the effect of NVCs from 200 sea-salt levels.
Particle water concentrations for the different particle sizes are shown in Fig. 3. PM1 has the most water, compared to the other size fractions over the continental region. The coarse particles in PM2.5-5 and PM5-10 have the most water over sea owing to sea salt, which is found in higher levels in these particles, and exhibits the highest hygroscopicitycompared to all other inorganic salts found 205 in aerosol. Water levels for all particle sizes are higher at areas closer to the sea, owing to the relatively high dry aerosol mass concentration combined with the high RH typically associated with the marine environments; in just the 2.5-5 μm size range alone, water content exceeds 20 μg m -3 .

Temporal evolution of pH 210
To study the temporal evolution of pH, eight sites ( Fig. S1) with different characteristics were selected based on their different type, location and dust/sea salt levels (Table S1). Iza, in Ukraine, has the lowest PM1 pH of all examined locations with a value equal to 0.25 (Fig. S2). The pH is predicted to vary between -0.5 and 1.4 (on an hourly basis) in this region of Eastern Europe that is characterized by high sulfate levels. Mace Head, on the other hand, has the highest average PM1 pH 215 (1.7) with values ranging between 0.9 and 2.3. Finokalia has the most variable PM1 pH with a range covering 4 units. This site is affected by both relatively dry air masses with continental aerosol characteristics (low pH) and by air masses with relatively high sea-salt and dust levels as well as biomass burning influences (higher pH; Bougiatioti et al., 2016). The distribution of pH values of the 1-2.5 μm diameter particles moves to higher values (less acidic particles) for all sites compared 220 to the sub-micrometer particles. The pH values of all sites for the 2.5-5 μm range are similar to those in the 5-10 μm range and higher than the fine aerosol pH values.
The pH diurnal profiles for Cabauw, Melpitz, Paris, Finokalia are shown in Fig. 4. These sites were selected based on their different type, location and dust/sea salt levels (Table S1)and because ambient pH data is available for both of them (Pye et al., 2019). pH follows the same trend Paris), and then decreasing values during the day reaching a minimum in the afternoon. The pH diurnal profile is different in Finokalia, since pH has its peak (up to 3.6) at noon (between 15-16 230 UTC time) and then starts to decrease. These variations are caused by a variety of factors including the relative humidity (that is higher during the early morning, leading to higher liquid water content and higher pH), the temperature (which tends to evaporate nitrate) and mixing height variation (which in turn tends to affect precursor concentrations).

pH variation with height
All the results presented so far are for the ground level (lowest 50 m). The predicted aerosol water content for all size ranges decreases with altitude (Fig. S3). This is mainly due to the decrease of the relative humidity and aerosol concentrations with altitude (Mishra et al., 2015;Wang et al., 2018).
As height increases, pH values for all particle sizes decrease, due to the reduction of aerosol water 240 per unit mass of dry aerosol, with height ( Fig. 5) which is exclusively an effect of relative humidity decrease. A secondary effect is that the lower concentration of aerosol tends to drive partitioning of semi-volatile species (nitrate, ammonium) to the gas phase (Nenes et al., 2019). As a result, particles of all sizes that are acidic at ground level become more acidic when they move higher in the atmosphere.

245
For PM1, in the less acidic areas over Europe the pH decreases from 2-2.5 near the ground to around 1.5-2 at 2.5 km altitude. For PM1-2.5, the reduction is even larger, since the pH values decrease from 3-3.5 at the ground to 1.5-2.2 at 2.5 km (the larger drop in pH is a result of the evaporation of nitrate aerosol and the decrease of liquid water content). Similar decreases of 1-1.5 pH units are predicted for the coarse particles in the first 2.5 km of the atmosphere in areas over 250 land. The predicted decrease in aerosol water content and pH for the super-micrometer particles is even more pronounced in the marine atmosphere and coastal areas due to the high levels of sea-salt near the ground. This reduction of pH with altitude is smaller for the PM1 size range in the marine atmosphere.

Effect of aerosol pH on inorganic nitrate
The highest average nitrate levels (0.7 μg m -3 ) are predicted for the 2.5-5 μm size range (Fig. S4) for the whole domain. This size range is characterized by the highest average pH (with a value of 2.63).
Nitrate partitioning to the aerosol phase is favored when the aerosol pH is higher than 2.5 (Guo et al., 2016;Vasilakos et al., 2018). At the same time, the mass transfer of the produced nitric acid in 260 the gas phase is faster for the particles in the 2.5-5 μm range compared to those in the PM5-10 range also contributing to higher concentration. Finally, the removal of the larger particles from the atmosphere is faster adding one more reason for the maximum of the nitrate size distribution.
The size-dependent average nitrate diurnal profiles for Cabauw, Melpitz, Paris, and Finokalia, are shown in Fig. 6. In Cabauw, predicted total nitrate levels start to increase early in the 265 morning, have their peak at noon or afternoon and decrease during the afternoon and early evening.
Most of this variation is due to the formation of ammonium nitrate in the PM1 size range. The increase in PM1 nitrate is accompanied by an increase in the ammonium levels ( Fig. S5) in this area that is characterized by high ammonia concentrations. The morning increase in the PM5-10 nitrate levels is due to the formation of sodium nitrate and calcium nitrate during daytime. In Cabauw, all 270 super-micron particles have pH above 2.5, favoring the partitioning of nitrate to the aerosol phase forming also ammonium nitrate. In Melpitz, the behavior of nitrate is quite different than in Cabauw due to the differences in the pH behavior (Fig. 4). PM1 nitrate peaks early in the morning with peaks in coarse nitrate a few hours later. The peak in fine nitrate is at the same period as the pH in this range, while for the coarse particles it is a few hours later due to the delays in mass transfer to the 275 larger particles. Nitrate levels in all size ranges are predicted to decrease during the afternoon with nitrate reaching a minimum in the late afternoon. The behavior of nitrate in the fine and coarse particles in Paris is quite similar as in Melpitz reflecting the similarity in the behavior of pH. Nitrate in all size sections peaks in the early morning and has a minimum in the afternoon. The main difference in this case is that there is more nitrate in the coarse particles due to the higher predicted 280 levels of dust. In Finokalia, the predicted nitrate increases gradually in all sizes during the morning, reaches its maximum values in the afternoon and then gradually decreases.

Effect of dust on particle pH
The impact of the NVCs from dust on pH can be quantified comparing the results of the simulation 285 in which the dust was assumed to be inert with the base case simulation. Aerosol water levels are higher in all particle sizes for the base case simulation (Fig. S6) compared to the inert dust simulation, as result of the water uptake associated with the NVCs. Dust is predicted to cause an increase of 1.2-2 μg m -3 in aerosol water concentration even for the submicrometer particles over Europe with the highest changes in the northern areas. The water increases for PM1-2.5 due to dust 290 varies from 1 to 2.5 μg m -3 over continental region. The effect, as expected, is higher for PM2.5-5 reaching up to 3 μg m -3 in areas like the Po Valley in Italy and even higher for PM5-10 ranging between 1 and 6 μg m -3 . The highest differences in aerosol water levels between the two simulations are predicted for areas that combine relatively high values of RH and relatively high values of dust ( Fig. S7) during the simulated period.

295
The predicted aerosol pH is lower in all particle sizes for the inert dust case compared to the base case simulation (Fig. 7). The soluble NVCs in dust tend to increase pH, as due to their lack of volatility they irreversibly neutralize bisulfate ions that are generated by the NH3/NH4 + equilibrium, and therefore elevate aerosol pH. NVCs also elevate aerosol water in a way that leads to pH increase, directly through their hygroscopicity and indirectly, through promoting the condensation of 300 aerosol nitrate (and its associated water content; Guo et al., 2018). PM1 is also affected, even though it contains small amounts of dust, as its pH increase by approximately by 0. The effect of dust on pH and aerosol water is reflected on the predicted aerosol nitrate.

305
Nitrate in all particle sizes decreases when dust is assumed to be inert (Fig. S8), as the corresponding pH reduction (for cases when pH < 2.5) does not favor the partitioning of nitrate to the aerosol phase. The effect of dust on submicrometer nitrate is negligible in most areas, but there is still an effect in the Netherlands and the surrounding areas (Fig. S8). The dust is predicted to cause average increases of PM1-2.5, PM2.5-5 and PM5-10 nitrate up to 0.5 μg m -3 , 1.3 μg m -3 and 1.2 μg 310 m -3 , respectively in parts of northern Europe with higher dust levels and also Italy. This nonlinear impact of relatively minor amounts of NVCs from dust occurs because relatively small changes in aerosol pH, when occurring in the "pH sensitivity window" of nitrate partitioning can lead to large responses in nitrate uptake (Vasilakos et al., 2018).

The role of calcium
Predicted aerosol water levels decrease in the absence of calcium compared to the base case simulation (Fig. S9). The increase of aerosol water concentration caused by the calcium ranges between 0.8-1 μg m -3 for PM1, 0.8-1.1 μg m -3 for PM1-2.5, 0.8-1.2 μg m -3 for PM2.5-5 and 0.7-1.3 μg m -3 for PM5-10 over continental Europe. The effect is more significant in the coarse particles where 320 most of the calcium is found. Considering the possible effects calcium can have on soluble sulfate and water uptake, the simulations suggests that the primary effect of calcium is through its action as a soluble ion. If calcium is neglected in the simulation, aerosol pH decreases for all particle sizes compared to the base case simulation (Fig. 8) highest pH differences are predicted for the coarse particles, consistent with that the coarse particles are richest in calcium.

Conclusions
The size-dependent aerosol pH was simulated over Europe during an early summer period. We find 330 that fine mode aerosol is persistently more acidic than coarse mode particles. The size-dependence of pH is strongest in northern and southern Europe, where the difference can be as large as 4 units between submicron and 10 μm particles. This difference is reduced over continental regions, but can still be as large as 1 pH unit between PM1 and PM1-2.5. PM1 has the most water over continental areas, while PM2.5-5 and PM5-10 have the most water in the marine and coastal areas.

335
Particles of all sizes become increasingly acidic with altitude owing to the reduction of aerosol water levels with height and volatilization of particulate ammonium and nitrate due to dilution. The highest pH decrease between the ground and 2.5 km altitude is 0.5-1 units for PM1, 1.5-2 units for PM1-2.5 and PM2.5-5, 1.3 units for PM5-10. The largest drop in pH is observed for the PM1-2.5 fraction because it coincides with where aerosol nitrate resides mosthence its evaporation 340 with altitude tends to have a larger impact on pH than reductions of liquid water from the RH effect alone.
The nitrate concentration tends to peak a few hours later than the pH in all examined sites due to the time required for the production of nitric acid and its partitioning to the aerosol phase. If aerosol pH becomes low enough to impede fine mode nitrate formation, its preferential 345 condensation to larger sizes tends to increase the pH difference across size. The highest average nitrate levels over Europe are predicted for the 2.5-5 μm range for which the average pH is equal to 2.6 during the simulated period.
Dust causes increases of the aerosol water levels in all particle sizes. The increase in water levels ranges from 1 to 6 μg m -3 with the highest change for PM5-10 in parts of northern Europe with 350 relative high concentrations of dust. Dust also causes an increase in aerosol pH for all particle sizes with higher effects in the coarse particles. This effect can be more than 1 pH unit. This increase in pH is accompanied by increases in aerosol nitrate, which can be as large as 2.5 μg m -3 . This effect of dust is mainly due to its calcium content, suggesting the importance of simulating accurately not only the dust concentration but also the calcium levels.

355
This study clearly shows that aerosol acidity and liquid water content changes considerably across size, location, time and height over Europe. These changes will impact aerosol formation and its response to emissions controls, solubility of aerosol trace metals and deposition. With this