The role of coarse aerosol particles as a sink of HNO₃ in wintertime pollution events in the Salt Lake Valley

Wintertime ammonium nitrate (NH₄NO₃) pollution events burden urban mountain basins around the globe. In the Salt Lake Valley of Utah in the United States, such pollution events are often driven by the formation of persistent cold-air pools (PCAPs) that trap emissions near the surface for several consecutive days. As a result, secondary pollutants including fine particulate matter less than 2.5 µm in diameter (PM_2.5), largely in the form of NH₄NO₃, build up during these events and lead to severe haze. As part of an extensive measurement campaign to understand the chemical processes underlying PM_2.5 formation, the 2017 Utah Winter Fine Particulate Study, water-soluble trace gases and PM_2.5 constituents were continuously monitored using the ambient ion monitoring ion chromatograph (AIM-IC) system at the University of Utah campus. Gas-phase NH₃, HNO₃, HCl, and SO₂ along with particulate NH${}_{\mathrm{4}}^{+}$, Na⁺, K⁺, Mg²⁺, Ca²⁺, NO${}_{\mathrm{3}}^{-}$, Cl⁻, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ were measured from 21 January to 21 February 2017. During the two PCAP events captured, the fine particulate matter was dominated by secondary NH₄NO₃. The comparison of total nitrate (HNO₃ + PM_2.5 NO${}_{\mathrm{3}}^{-}$) and total NH_x (NH₃ + PM_2.5 NH${}_{\mathrm{4}}^{+}$) showed NH_x was in excess during both pollution events. However, chemical composition analysis of the snowpack during the first PCAP event revealed that the total concentration of deposited NO${}_{\mathrm{3}}^{-}$ was nearly 3 times greater than that of deposited NH${}_{\mathrm{4}}^{+}$. Daily snow composition measurements showed a strong correlation between NO${}_{\mathrm{3}}^{-}$ and Ca²⁺ in the snowpack. The presence of non-volatile salts (Na⁺, Ca²⁺, and Mg²⁺), which are frequently associated with coarse-mode dust, was also detected in PM_2.5 by the AIM-IC during the two PCAP events, accounting for roughly 5 % of total mass loading. The presence of a significant particle mass and surface area in the coarse mode during the first PCAP event was indicated by size-resolved particle measurements from an aerodynamic particle sizer. Taken together, these observations imply that atmospheric measurements of the gas-phase and fine-mode particle nitrate may not represent the total burden of nitrate in the atmosphere, implying a potentially significant role for uptake by coarse-mode dust. Using the NO${}_{\mathrm{3}}^{-}$ : NH${}_{\mathrm{4}}^{+}$ ratio observed in the snowpack to estimate the proportion of atmospheric nitrate present in the coarse mode, we estimate that the amount of secondary NH₄NO₃ could double in the absence of the coarse-mode sink. The underestimation of total nitrate indicates an incomplete account of the total oxidant production during PCAP events. The ability of coarse particles to permanently remove HNO₃ and influence PM_2.5 formation is discussed using information about particle composition and size distribution.

1 Introduction

Episodes with high particulate matter (PM) pollution occur frequently in urban air basins across the globe during winter months when a stable boundary layer persists for multiple days, including in the mountain valleys in the western USA (Baasandorj et al., 2017; Bares et al., 2018; Green et al., 2015; Silcox et al., 2012; Whiteman et al., 2014), the Po Valley in Italy (Bernardoni et al., 2017; Vecchi et al., 2018), and the Sichuan (Tian et al., 2019) and Twin-Hu (Gao et al., 2019) basins in China. These mountain basins, including the Salt Lake Valley (SLV) in northern Utah, experience strong temperature inversions that develop into persistent cold-air pools (PCAPs), which suppress vertical mixing and trap emissions within a shallow boundary layer (Lareau et al., 2013; Whiteman et al., 2014). Under these conditions, mass loadings of fine particles smaller than 2.5 µm (PM_2.5) often reach values of 60–80 µg m⁻³ in the western USA, which are far above the US National Ambient Air Quality Standard (35 µg m⁻³, 24 h average) (Lareau et al., 2013; Silcox et al., 2012; Whiteman et al., 2014). Along with environmental impacts such as haze and localized deposition, mass loadings of PM_2.5 that exceed this standard have been associated with increased risk of mortality, especially cardiopulmonary or cardiovascular disease mortality (Pope et al., 2003, 2017). However, despite significant societal concerns about the impact on human health from these PM_2.5 pollution episodes and their common occurrence around the world, the major chemical processes that drive PM formation in these regions are still uncertain.

Figure 1 Processes influencing the concentrations of NH₃, NO_x, HNO₃, and NH₄NO₃. Pathways that occur during the day are shown in dashed grey arrows, while pathways that occur during the night are shown in black. The rate of uptake of HNO₃ to coarse particles, ≥ PM₁, composed of CaCO₃ and/or NaCl is unknown.

Several studies conducted in northern Utah have shown that PM_2.5 during these PCAP episodes is predominantly composed of ammonium nitrate (NH₄NO₃), accounting for roughly 60 %–80 % of the total dry particle mass (Baasandorj et al., 2017; Franchin et al., 2018; Hansen et al., 2013; Kelly et al., 2013; Kuprov et al., 2014; Long et al., 2002, 2003; Mangelson et al., 1997). NH₄NO₃ formation is thermodynamically favorable under conditions with low temperatures and high relative humidity based on equilibrium partitioning with gas-phase ammonia (NH₃) and nitric acid (HNO₃) (Mozurkewich, 1993; Nowak et al., 2010; Seinfeld and Pandis, 2006), illustrated in the right-hand portion of Fig. 1.

NH₃ emissions are often associated with agricultural activities and waste disposal; however, in recent years automotive emissions and industrial processes have become increasingly important sources in urban areas (Behera et al., 2013; Bishop et al., 2010, 2016; Livingston et al., 2009; Nowak et al., 2012; Roth et al., 2019; Suarez-Bertoa et al., 2014; Sun et al., 2017). In contrast, HNO₃ is formed in the atmosphere and is a major sink of nitrogen oxides, which are emitted primarily through fossil fuel combustion. The two dominant mechanisms highlighted in Fig. 1 that lead to HNO₃ occur either (1) from oxidation of NO₂ via reaction with OH (in orange), which is photochemically driven during the day, or (2) through heterogenous uptake of N₂O₅ (in black), which typically occurs at night. Gas-phase HNO₃ and NH₃ and particulate NH₄NO₃ are removed from the atmosphere through dry and wet deposition under typical atmospheric conditions. However, under PCAP conditions, the stably stratified boundary layer reduces convective mixing, which in turn allows pollutants to accumulate. In addition, HNO₃ can also be lost through heterogenous reaction with dust and sea salt components, such as CaCO₃ and NaCl (Beichert and Finlayson-Pitts, 1996; Dasgupta et al., 2007; Fenter et al., 1994; Liu et al., 2008a). When the loss of HNO₃ to deposition is suppressed under PCAP conditions, the loss to reactive uptake to airborne dust and sea salt can potentially become more important.

To formulate effective control strategies that reduce wintertime PM_2.5 in northern Utah, measurement campaigns have been conducted to improve our understanding of fine particulate formation during these PCAP episodes (Baasandorj et al., 2017; Bares et al., 2018; Hansen et al., 2010; Kelly et al., 2013; Kuprov et al., 2014; Lareau et al., 2013; Malek et al., 2006; Mangelson et al., 1997; Silcox et al., 2012; Whiteman et al., 2014). Several analyses of the PM build-up in cold-air pool events in the SLV have been published, all based on measurements from one or several ground sites (Baasandorj et al., 2017; Kuprov et al., 2014; Long et al., 2003; Silcox et al., 2012). Long-term PM_2.5 composition measurements in the SLV are made using filter samples (Kelly et al., 2013; Kuprov et al., 2014; Long et al., 2003; Mangelson et al., 1997; Silcox et al., 2012; Whiteman et al., 2014), with PM_2.5 speciation carried out on filter extracts in accordance with the US Environmental Protection Agency (EPA) Speciation Trends Network (STN) protocol. Kuprov et al. (2014) were the first to report the gas and PM_2.5 composition during pollution events in northern Utah using an ambient ion monitoring ion chromatograph (AIM-IC) system; however, this study only measured anionic species, providing HNO₃ and particle nitrate (pNO${}_{\mathrm{3}}^{-}$) concentrations at an hourly resolution. Ambient NH₃ concentrations were monitored using a chemiluminescence-based NH₃ analyzer, but particulate ammonium (pNH${}_{\mathrm{4}}^{+}$) had to be inferred based on the assumption that both pNO${}_{\mathrm{3}}^{-}$ and particulate sulfate (pSO${}_{\mathrm{4}}^{\mathrm{2}-}$) were in the form of their respective ammonium salts. Particle mass fractions of crustal and carbonaceous material in PM_2.5 were calculated on 24 h average measurements of Al, Si, and elemental and organic carbon in fine particles from the NCore program (US EPA, 2010).

Recent research in the SLV has focused on determining whether ammonium nitrate mass loading is most sensitive to reductions in NH₃ or HNO₃ to inform PM_2.5 reduction strategies (Kuprov et al., 2014). Baasandorj et al. (2017) recently proposed that PM_2.5 forms overnight in the upper layers of the PCAP and mixes down to the surface, enhancing the total PM_2.5 loading experienced in the SLV. This explanation of stratified particulate nitrate chemistry throughout PCAP events suggests a complex coupling of a HNO₃-limited surface layer with a NH₃-limited elevated layer (Baasandorj et al., 2017). Aircraft measurements during the 2017 Utah Winter Fine Particulate Study (UWFPS) provided the first detailed vertically resolved chemical composition of these PCAP episodes in the SLV. Analysis presented in McDuffie et al. (2019) confirms that the formation of NH₄NO₃ during these pollution events is largely limited by HNO₃ but with more frequent periods approaching NH₃-limited conditions in the upper boundary layer over time (McDuffie et al., 2019). Regimes of both NH₃-limited and HNO₃-limited conditions are also consistent with an aerosol thermodynamic model sensitivity study of the same data by Franchin et al. (2018), who found simulations of non-refractory PM₁ mass were sensitive to reductions in both total ammonium and total nitrate (Franchin et al., 2018). Box model studies of the odd oxygen budget during UWFPS by Womack et al. (2019) assumed that NH₄NO₃ formation during PCAP episodes is HNO₃-limited, and found full episodes tend to be NO_x-saturated. In conditions where pNO${}_{\mathrm{3}}^{-}$ production is HNO₃-limited, it is often assumed that reduction in NO_x would directly lead to reductions in HNO₃. However, Womack et al. (2019) showed the production of HNO₃, and therefore, NH₄NO₃ formation, is sensitive to changes in VOC concentrations emphasizing the complexity of chemical processes involved in PM formation under these conditions.

The balance of factors influencing HNO₃ formation are still being understood. Modelling results from both McDuffie et al. (2019) and Womack et al. (2019) predict that up to ∼ 50 % of HNO₃ production is from the heterogeneous N₂O₅ pathway in the residual layer. This is in line with ground-based measurements of N₂O₅ by Baasandorj et al. (2017) highlighting the importance of nighttime pathway (Baasandorj et al., 2017). Less attention has been paid to the possibility that alternate sinks for HNO₃ may also significantly limit the amount of nitrate that can contribute to PM_2.5 formation.

Here, results are presented from ground site observations in Salt Lake City from a moderately elevated site at the edge of the SLV during the UWFPS campaign in January and February 2017. Atmospheric concentrations of inorganic trace gases and water-soluble PM_2.5 constituents were measured by an AIM-IC, equipped with both cation and anion ion chromatographs. During PCAP episodes, we observed the formation and build-up of NH₄NO₃ over several days. Between the hours of 08:00 and 18:00 mountain standard time (MST) on several PCAP days, we also observed the presence of elevated levels of non-volatile cations in PM_2.5. To investigate the role of mineral–coarse-mode aerosols in limiting the availability of HNO₃ during these events, the final section presents estimates of HNO₃ lifetime against reactive uptake onto mineral–coarse-mode dust based on aerodynamic particle sizer (APS) data and estimates of the total nitrate budget using the chemical composition of the snowpack measured during UWFPS.

2 Methods

2.1 Site description

The SLV (1300 km²) is a mountain basin in northern Utah surrounded by steep mountain ranges with the Great Salt Lake sitting to the northwest. The Wasatch Mountains (peak at 3636 m a.s.l.) span the entire eastern border of the valley with the Oquirrh Mountains (peak at 3235 m a.s.l.) to the west. There is a narrow passage in the southern Traverse Range (1878 m a.s.l.) that creates a small opening into the Utah Valley. Salt Lake City sits at the base of the valley (1288 m a.s.l.), making a vast elevation difference between the metropolis and the surrounding mountains. Measurements were conducted from the rooftop of the William Brown Building 33 m above ground level at the University of Utah (UU) located 40^∘45^′58.7^′′ N, 111^∘50^′51.6^′′ W, in Salt Lake City. The UU site is located 155 m above the valley floor along the northeast sidewall of the SLV. The site is impacted by local traffic that enters the university on the north side of campus.

2.2 Instrumentation

Hourly averages of PM_2.5 chemical composition and gas-phase precursors were measured using an online continuous AIM-IC instrument (Model 9000D, URG Corp, Chapel Hill, NC). Briefly, gases and particles are sampled from the same airflow by passing through a short inlet with elements for size selection (PM_2.5 impactor), gas collection (wet parallel plate denuder), and particle capture (saturation chamber) at a flow rate of 3 L min⁻¹. Aqueous solutions containing analytes originally in the gas and particle phases are separately transferred through 22 m inlet lines into 5 mL glass syringes that are subsequently analyzed by ion chromatographs (ICs) following hour-long collection periods. More in-depth details about the AIM-IC system, including important adaptations from the standard commercial system, can be found in Markovic et al. (2012). During the UWFPS campaign, the components of the inlet assembly were housed in a small weatherproof box 2 m above the roof of the UU site and the IC systems were housed in a laboratory below. Standard calibrations of the ICs were performed offline using mixed ion standard solutions before and after the measurement period. Charged species (Na⁺, NH${}_{\mathrm{4}}^{+}$, K⁺, Mg²⁺, Ca²⁺, NH₃, amines, Cl⁻, NO${}_{\mathrm{2}}^{-}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$) were measured using a Dionex ICS-2000 equipped with concentrator columns (TCC-ULP1 and TAC-ULP1), guard columns (CG17 and AG19), and 4 mm analytical columns (CS17 and AS19). Gradient elution methods, using methanesulfonic acid (MSA) and potassium hydroxide (KOH), were carried out using electrolytically regenerated suppressors to reduce the influence of the eluent in the conductivity detection. Carbonate salts are never quantified in the anion IC due to the natural presence of carbonate in the distilled deionized water reservoir that sustains the IC systems. Background measurements of the entire AIM-IC system were conducted by overflowing the sampling inlet using high-purity zero air to determine the method zero and detection limits of each species. The limits of detection (LODs), based on 3σ of the background, for trace gases were found to be 150, 20, 40, and 10 ppt for NH₃, HCl, HNO₃, and SO₂, respectively. The LODs for PM_2.5 chemical constituents were 0.2, 0.1, 0.1, 0.04, and 0.4 µg m⁻³, for pNH${}_{\mathrm{4}}^{+}$, pNa⁺, pK⁺, pMg²⁺, and pCa²⁺ and 0.03, 0.03, and 0.01 µg m⁻³ for pCl⁻, pNO${}_{\mathrm{3}}^{-}$, and pSO${}_{\mathrm{4}}^{\mathrm{2}-}$. Traces gases and PM_2.5 constituents measured as cations are reported from 21 January to 21 February 2017. Due to technical difficulties, anionic constituents measured from 21 January to 8 February were compromised. Therefore, PM_2.5 anionic components are only reported from 13 to 21 February and the trace gases measured as anions during the former period may be less reliable.

Continuous measurements of total PM_2.5 mass concentrations were also captured at the site using the Filter Dynamics Measurement System (FDMS) with a tapered element oscillating microbalance (TEOM) ambient particulate monitor (FDMS TEOM 1400ab, Thermo Fisher Scientific) provided by the Utah Division of Air Quality (UDAQ). Ambient air was sampled at ∼ 1.5 m above the roof through a ∼ 7 m long $\mathrm{1}/{\mathrm{4}}^{\prime \prime }$ o.d. PFA line with a volumetric flow rate of 20 L min⁻¹. A custom-made inertial PM impactor was connected to the entrance of the sampling inlet to remove coarse PM and water from the sample flow. A critical orifice was installed downstream of the PM impactor to restrict the sample flow, maintaining a pressure below 200 mbar. The air sample was sent through an additional filter to remove particles larger than 2.5 µm. An aerodynamic particle sizer (APS; TSI Inc, Model 3321) was operated from 26 January to 10 February to size and detect particles with an aerodynamic diameter of between 0.54 and 19.81 µm based on the time of flight of the particle between two (633 nm) He-Ne lasers. This instrument was attached to an aerosol inlet that included a 2.54 cm diameter stainless-steel tube connected to a stainless-steel rain cap with approximately a 10 cm diameter. Aerosol mass was calculated using the Aerosol Instrument Manager (AIM) software for the APS Model 3321. The particle density used for calculating the aerosol mass concentration was 1.0 g cm⁻³, and particles were assumed to be spherical. The APS was attached to this inlet system via a custom-made stainless-steel pickoff. To improve aerosol transmission through the inlet, an external pump was attached to the system. The pump pulled at approximately 7.2 L min⁻¹. The calculated transmission shows a 50 % cutoff at approximately 7 µm for this inlet system (Skiles et al., 2018).

2.3 Snow sampling and analysis

The inorganic chemical composition (Na⁺, NH${}_{\mathrm{4}}^{+}$, K⁺, Mg²⁺, Ca²⁺, NH₃, Cl⁻, NO${}_{\mathrm{2}}^{-}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$) of the snowpack was monitored throughout the campaign during periods of snow cover. The snowpack composition and depth were measured daily within a 2 m radius around the AIM-IC inlet atop UU using a clean graduated 5 cm diameter beveled snow corer. Samples were collected in triplicate along with a field blank to account for any contamination during sampling. Collected samples were melted in sealed containers. The resulting snowmelt solution was measured for total melt volume and pH (Hach sensION+ PH1 Portable pH Meter). Snowmelt samples were then filtered through 0.25 µm PTFE filters (Whatman, VWR International) before analysis utilizing ion chromatography.

2.4 Aerosol thermodynamic modelling using ISORROPIA

Thermodynamic modelling was used to examine the expected gas and particle partitioning of semi-volatile constituents based on the local meteorological conditions and the observed concentrations of gas and particle constituents. These models carry out bulk calculations to estimate the gas-phase and particle-phase composition of inorganic species, in which all particles are assumed to have the same chemical composition. To examine how the presence of non-volatile particle components observed during this period may affect nitrate partitioning, ISORROPIA v2.1 (Nenes et al., 1998; Fountoukis and Nenes, 2007) was used. Model calculations were run in forward mode, with observed total ammonia (NH₃ + pNH${}_{\mathrm{4}}^{+}$), particulate sulfate (pSO${}_{\mathrm{4}}^{\mathrm{2}-}$), total nitrate (HNO₃ + pNO${}_{\mathrm{3}}^{-}$), total chloride (HCl + pCl⁻), and non-volatile particulate species (pK⁺, pNa⁺, pCa²⁺, and pMg²⁺), along with ambient relative humidity and temperature, as inputs. Model runs were conducted using the metastable state option, which prevents solid formation by forcing all particulate components to remain in an aqueous state.

3 Results and discussion

3.1 Trace gas and PM_2.5 composition in the Salt Lake Valley

The AIM-IC measurement period included two PCAP pollution episodes, 27 January to 3 February 2017 (episode 1) and 13 to 17 February 2017 (episode 2), defined by the meteorological conditions that stabilize the boundary layer and by the total observed PM_2.5 mass, separated by a relatively clean period from 5 to 12 February 2017. The PCAP episodes can be clearly seen in the build-up of PM_2.5 mass and the inorganic constituents measured by the AIM-IC as shown in Fig. 2. Pollution periods with a total 24 h average PM_2.5 mass > 17.5 µg m⁻³ are deemed PCAP periods, while periods with a 24 h average PM_2.5 mass ≤ 2 µg m⁻³ are considered clean periods, as previously defined by Whiteman et al. (2014). The pollution period from 27 January to 3 February 2017 (episode 1) was the most persistent and severe during the UWFPS measurement campaign.

Figure 2 (a) The time series of NH₃ and pNH${}_{\mathrm{4}}^{+}$ observed by the AIM-IC and total PM_2.5 mass loading (µg m⁻³) measured by the TEOM from 23 January to 19 February 2017. (b) The AIM-IC observations of inorganic gases HCl, HNO₃, and SO₂ and PM_2.5 components pCl⁻, pNO${}_{\mathrm{2}}^{-}$, pNO${}_{\mathrm{3}}^{-}$, and pSO${}_{\mathrm{4}}^{\mathrm{2}-}$ present in PM_2.5. (c) The AIM-IC observations of the non-volatile inorganic cations pNa⁺, pK⁺, pMg²⁺, and pCa²⁺ measured in PM_2.5 for the entire campaign period. The two PCAP pollution episodes captured during the campaign are emphasized by the dotted black line. Time is expressed in mountain standard time (MST).

Both pollution events are marked in Fig. 2, while the full suite of inorganic gases and PM_2.5 components was measured by the AIM-IC only from 8 to 19 February 2017. The increase in total PM_2.5 mass at the onset of episode 1 on 27 January coincides with a rise in pNH${}_{\mathrm{4}}^{+}$, seen in Fig. 2a, while there is an initial drop in ambient NH₃ concentration before it plateaus at a mixing ratio of ∼ 1 ppb as the PCAP progresses. The HNO₃ and pNO₃ data, depicted in Fig. 2b for the second pollution event, confirm the major water-soluble component of PM_2.5 in the SLV is NH₄NO₃, as previously reported (Hansen et al., 2010; Kelly et al., 2013; Kuprov et al., 2014; Long et al., 2003). Compared to the pCl⁻ and pSO${}_{\mathrm{4}}^{\mathrm{2}-}$ mass loadings measured during that event, Fig. 2b shows the total mass of pNO${}_{\mathrm{3}}^{-}$ is an order of magnitude larger. Particularly at midday, the majority of the PM_2.5 mass is NH₄NO₃. It is important to note the organic fraction of PM_2.5 was not measured by the AIM-IC. Based on TEOM total mass and AIM-IC-measured mass fractions, the AIM-IC captured 75 % of the total PM_2.5 mass.

The distinction between PCAP episodes 1 and 2 is important to note because the stability of the cold-air pool and its persistence influences the intensity of PM_2.5 build-up in the SLV. Episode 1 was the most severe, persisting for 9 d and allowing PM_2.5 daily averages to reach the highest observed through the measurement period, a maximum hourly average of 60 µg m⁻³. The episode occurred a few days after a storm deposited roughly 22 cm of snow across the SLV and Wasatch front. Ambient temperatures remained below freezing, between 263 and 273 K, sustaining the snowpack throughout the event. This in turn increased the surface albedo, limiting the strength of convection that can disrupt the stable boundary layer (Whiteman et al., 2014). Winds during this episode were light, averaging 1 m s⁻¹, with gusts under 7 m s⁻¹ during the most stable and calm cold pool period from 31 January to 2 February.

The second PCAP observed during the campaign was much shorter-lived with moderate cold pool conditions only present from 13 through 17 February. This episode was distinctly different than episode 1, having no snow cover and much warmer conditions, with temperatures ranging from 270 to 284 K. The winter snowpack in the surrounding mountain range showed signs of spring melt, which extended the surface area of exposed ground. Winds were slightly stronger, averaging 1.6 m s⁻¹ with stronger gusts averaging around 8 m s⁻¹ caused by stronger convection. The hourly PM_2.5 average at the UU site peaked at 28 µg m⁻³, and elevated PM levels only persisted for 2 to 3 full days, so the intensity and duration of episode 2 was about half that of episode 1.

The presence of non-volatile cations (pCa²⁺, pMg²⁺, pK⁺, and pNa⁺), seen in Fig. 2c, was also observed during PCAP events. Of all the non-volatile cations observed, only potassium increases with the same pattern as the increase in pNH${}_{\mathrm{4}}^{+}$ and total PM_2.5 mass measured by the TEOM. Potassium salts are often found in primary aerosols formed from biomass or biofuel burning (Pósfai et al., 2003; Rissler et al., 2006). Wood burning is still a commonly used source of heat during the winter months in the SLV and could be the source of pK⁺ observed during the severe PCAP episode 1. Aerosol mass spectrometer (AMS) data from Franchin et al. (2018) showed traces of levoglucosan, a known marker for wood combustion, during the pollution episode. However, levoglucosan was typically measured at close to background values. Overall, the data reported from AMS measurements indicated the organic material from wood combustion is not a dominant mass fraction on the regional scale (Franchin et al., 2018). Recent radiocarbon analysis from ground sites in the SLV indicates that fossil fuels were the dominant sources of carbonaceous aerosols during winter, contributing to 88 % (80 % to 98 %) of the black carbon in aerosols and 58 % (48 % to 69 %) of the organic carbon in aerosols (Mouteva et al., 2017). Similarly, the AIM-IC data also suggest that pK⁺ salts generally compose a small fraction of the total PM_2.5 mass.

The alkali and alkaline metal components pCa²⁺, pMg²⁺ and pNa⁺ are commonly found in larger primary particles often emitted from the lofting of mineral dust and salts. These species are often found in mineral aerosols, which typically range from 0.1 to more than 10 µm, partially contributing to PM_2.5 (Zender, 2003). The coarse-mode fraction of these primary particles, with diameters greater than 2.5 µm, has sufficiently large settling velocities that the particles have much shorter lifetimes in the atmosphere compared to PM_2.5. If the non-volatile cations observed in PM_2.5 are present in the largest fine particles, then their lifetime in the atmosphere is considerably shorter than NH₄NO₃ found in the accumulation mode. This may be the explanation for the more pronounced diel pattern of pCa²⁺, pMg²⁺, and pNa⁺ during the PCAP periods. It is also possible that because vertical mixing is suppressed during PCAP episodes, the UU site is in contact with air from the mineral dust source region to a measurable degree only at midday. During the clean periods, the concentrations of these ions measured by the AIM-IC are very close to their respective detection limits. pNa⁺, pCa²⁺ and pMg²⁺ PM_2.5 components peak at midday during pollution events, suggesting similar and/or co-located sources. Concentrations of pMg²⁺, which is an effective tracer for mineral dust along with pCa²⁺ (Maxwell-Meier et al., 2004), are present throughout the entire measurement period, averaging 0.02 µg m⁻³ and reaching a maximum of 0.15 µg m⁻³. It is known that sea salt is the most common source of pNa⁺, so examining the Ca-to-Na ratios can suggest the predominant source. Particularly for the SLV region, the Great Salt Lake can also be a source of sea salt. In PCAP episode 1, Ca-to-Na ratios were below 1, suggesting the presence of sea salts. In contrast, in episode 2 ratios of Ca-to-Na were above 1, suggesting the sources of these particulate components differ between the two PCAP periods. Due to the heavy snowfall during episode 1, road salt was continuously applied to all major roads and walkways throughout the UU campus, which may act as a local source of Na salts during periods of heavy traffic. The lack of snow cover during the latter episode and higher temperatures may suggest some influence of mineral dust from the surrounding ground surface areas, where sediments and weathered minerals are often Ca-rich (Chesselet et al., 1972; Kassomenos et al., 2012; Kuo et al., 2005). Recent studies have found the Great Salt Lake is quickly receding, suggesting the exposed the lake bed could act as a source of dust and sea salt (Hahnenberger and Nicoll, 2012; Skiles et al., 2018; Wurtsbaugh et al., 2017). However, the bulk PM_2.5 chemical composition measured by the AIM-IC is not detailed enough to unambiguously identify the source of non-volatile cations observed during PCAP events.

3.2 Estimating HNO₃ loss to coarse-mode particles

Gas-phase HNO₃ exists in equilibrium with semi-volatile nitrate in PM_2.5, and the extent of gas-to-particle partitioning depends on temperature, relative humidity, and NH₃, via the formation of NH₄NO_3(s) in dry particles or its influence on particle acidity in deliquesced particles. At the same time, HNO₃ may undergo net uptake by primary particles when it reacts with their constituents, for example CaCO₃ or NaCl, to form non-volatile nitrate salts. The rate of HNO₃ loss can be impacted by the composition of the reactive salt and the reactive surface area available (Seinfeld and Pandis, 2006). During the two PCAP episodes, we observed elevated midday concentrations of pNa⁺, pCa²⁺, and pMg²⁺ in PM_2.5 using the AIM-IC.

3.2.1 Inferences from aerodynamic particle sizing

Coarse-mode particles (diameters > 2.5 µm) were not speciated during the UWFPS campaign. There are very few reports of coarse-mode nitrate in winter conditions that could offer an approximation of how much HNO₃ may have been taken up by coarse-mode particles in the SLV. The study of Hansen et al. (2010) is the only one commenting on coarse-mode nitrate in northern Utah region, in Lindon, ∼ 50 km located south of the SLV, during January to February 2007. The authors observed high mass loadings, up to 80 µg m⁻³ of coarse aerosols (PM₁₀–PM_2.5), during PCAP events, based on measurements from two GRIMM optical particle counters (Hansen et al., 2010). The authors did not report any quantification of the chemical components of the coarse aerosols collected by integrated filter samples, although they commented that the nitrate levels were relatively low.

Figure 3 (a, b) The total fine-mode (PM_2.5) and coarse-mode (> PM_2.5) surface area calculated in µm² cm⁻³ from APS data. (c) The particle size distribution as a function of time, 28 January to 4 February, with particle size displayed along the y axis and color contours representing mass that is normalized by size bin ($\mathrm{d}M/\mathrm{d}\log Dp$). The population of particles that the AIM-IC captures is below the labelled PM_2.5 line.

Figure 4 (a) Wind speed and temperature (^∘C) measured at UU. The coarse particle (PM₁₀–PM_2.5) mass loading in µg m⁻³ as a function of time, 28 January to 4 February 2017, at three separate ground-based sites: Rose Park (b) shown in green and Hawthorne Elementary (c) shown in black, which are EPA monitoring sites, and the UU site (d) in blue.

To obtain information about the coarse-mode loadings during PCAP events in UWFPS, APS data collected during the pollution episode from 28 January to 3 February are displayed in Fig. 3. The particle size bins are colored by mass normalized to the size bin ($\mathrm{d}M/\mathrm{d}\log Dp$), clearly showing elevated midday burdens of coarse-mode particles, which become more pronounced during the intensive cold-air pool period from 30 January to 2 February 2017. This is consistent with the midday increases in coarse-mode mass observed across the SLV, shown in Fig. 4. The coarse fraction of PM₁₀ measured at Hawthorne Elementary (the air quality regulatory site at the base of the SLV operated by UDAQ) shows a similar trend to the APS mass loading of PM₁₀–PM_2.5 observed at UU. In contrast, the Rose Park site (another UDAQ site), which is close to Great Salt Lake, exhibits moderate increases in coarse-mode mass compared to UU and Hawthorne. This difference is more evident during the calmest period of the pollution event, when the lowest wind speeds were observed, which suggests the sources of coarse particles are highly localized during PCAP events. An analysis of coarse particles by Li et al. (2013) found only moderate correlation between PM_10–2.5 mass loadings and wind speed at several sites across the USA (Li et al., 2013).

The time series of the total surface area of PM larger than 2.5 µm in µm² cm⁻³ (Fig. 3) shows the available coarse-mode surface area exceeds 1500 µm² cm⁻³ for several hours on each of the 3 most severe pollution days, 30 January to 1 February 2017. For comparison, the total surface area of the fine-mode particles is also shown, which is roughly 3 times the maximum surface area of the coarse fraction but with less diurnal variability. However, the fine-mode aerosol nitrate is assumed to be in equilibrium with gas-phase HNO₃, in contrast to the coarse-mode surface area, which could be viewed as a reactive sink allowing net uptake. The midday increases in coarse-particle surface area, therefore, could represent an important permanent sink for HNO₃ if the particles are composed of reactive salts such as CaCO₃ or NaCl.

A lower limit to the lifetime of HNO₃ with respect to uptake by the coarse aerosol surface can be calculated to infer the potential importance of mineral–coarse aerosol particles during these heavy pollution events. The approximate first-order removal rate of HNO₃ due to the heterogeneous reaction with coarse aerosols, k_het, depends on its average molecular speed, c; the surface area density of coarse aerosols, S_A; and the uptake coefficient, γ, given in the following expression (Ammann et al., 2013; Crowley et al., 2010; Kolb et al., 2010; Tang et al., 2017):

$$\begin{array}{}\text{(1)}& k_{\mathrm{het}}=\mathrm{0.25}\cdot c\cdot S_{\mathrm{A}}\cdot \mathit{\gamma }.\end{array}$$

The variation in mineralogy of coarse-mode particles can influence the efficiency in HNO₃ reactive uptake via the uptake coefficient, γ. This is represented in the range of reported uptake coefficients of HNO₃ to CaCO₃ (γ=0.07 at RH = 50 %) and NaCl (γ=0.11 at RH = 55 %) (Liu et al., 2008a; Saul et al., 2006). The average RH at peak coarse-mode loading was 54 %; therefore, the reported uptake coefficients close to this RH value were used in our calculations. Literature reports have shown RH significantly enhances the reaction probability of HNO₃ uptake with mineral dust and sea salts (Liu et al., 2008a; Saul et al., 2006; Vlasenko et al., 2006). The detailed kinetics of HNO₃ reacting with CaCO₃ particles in a flow reactor by Liu et al. (2008a) found low uptake values of γ≤0.003 at 10 % RH, while at 80 % RH, γ can be ≥ 0.21. This agrees with reports from Vlasenko et al. (2006), in which HNO₃ uptake on dust aerosols containing ∼ 5 % CaCO₃ was also enhanced by an increase in RH. Saul et al. (2006) also found RH affects the uptake of HNO₃ onto pure NaCl showing a maximum γ of 0.12 at 50 % RH, while at 85 % RH, γ is 0.05. They also observed an increase in reactivity with the presence of MgCl₂ in NaCl powder. The greater reactivity of MgCl₂ is consistent with the hygroscopic character of magnesium salts that facilitate the increase in adsorbed water at the particle surface, which can in turn increase the effective uptake of HNO₃.

Figure 5 Predicted HNO₃ loss rate to coarse-particle surface area concentration measured by the APS from 28 January to 4 February 2017.

Because pNa⁺ and pCa²⁺ were the dominant non-volatile species that showed common behavior with the coarse-mode mass loading, the rates of uptake with respect to NaCl and CaCO₃ were calculated to approximate how quickly HNO₃ might be sequestered. Literature values for the uptake coefficients of HNO₃ on NaCl (0.11 at 55 % RH) and CaCO₃ (0.07 at 50 % RH) were used to estimate a lifetime against reactive uptake based on the measured coarse-mode surface area (Fenter et al., 1994; Liu et al., 2008a; Vlasenko et al., 2006). The time series of k_het from 27 January to 4 February 2017 is presented in Fig. 5. The effective lifetime of HNO₃ against reactive uptake to CaCO₃ during episode 1 ranges from 2 to 90 min with the shortest lifetimes during periods of peak coarse-mode loading. The lifetime against NaCl (0.11) ranges from 1.3 to 57 min. These lifetime calculations assume that the entire surface is reactive and do not account for any mass transfer limitations that may occur on the particle surface. Therefore, the time range is an estimate of the upper limit to the rate of HNO₃ uptake, or how quickly the reaction could potentially occur.

Under high-RH conditions, which often occur during PCAP events, it is also possible coarse particles may be deliquesced, allowing exchange between surface and bulk ions within a particle. This would result in an increased loss of nitrate if uptake were not limited to the surface. If the total mass of coarse particles measured during episode 1 were assumed to be composed of Ca(NO₃)₂, in which carbonate has been completely displaced, that would amount to an average of 6.2 µg m⁻³ of HNO₃ sequestered with a maximum of 32 µg m⁻³ during midday. This assumes HNO₃ can react with the bulk particle components, thereby representing an upper estimate of HNO₃ mass loss to PM₁₀. However, due to the size cutoff of the APS, with significant transmission losses above 10 µm, it is possible that the mass of coarse-mode particles is underestimated under some conditions.

The calculations of the removal rate do not include changes in reactive uptake associated with the changes in hygroscopicity of the resulting nitrate salts formed. The presence of calcium or magnesium nitrate salts can enhance the absorbance of water, which has been shown to increase the relative uptake of HNO₃ because it is no longer limited to the particle surface, so the estimation of HNO₃ loss can also be highly variable (Beichert and Finlayson-Pitts, 1996; Goodman et al., 2000). This is dependent on the deliquescence relative humidity (DRH) of the nitrate salt formed, in which DRH of Ca(NO₃)₂ is 10 % at 298 K (Liu et al., 2008b; Sullivan et al., 2009a, b) and NaNO₃ is 81 % at 273 K (Seinfeld and Pandis, 2006). The DRH is also temperature-dependent and tends to increase with decreasing temperature. The temperatures during episode 1 were consistently below 273 K; therefore, it is assumed DRH would be higher. The inference from the heterogeneous uptake calculation suggests that the loss of HNO₃ can be very rapid. The question of how this loss in HNO₃ to coarse aerosols could be shifting the gas-to-particle equilibrium that exists between HNO₃, NH₃, and NH₄NO₃ is still unknown. Therefore, further investigation of time-resolved coarse-aerosol chemical composition, in addition to PM_2.5 composition, is needed to better quantify how much coarse aerosols may be limiting HNO₃ availability during these PCAP pollution events and elucidate their effect on NH₄NO₃ formation.

Studies from coastal sites provide evidence that the uptake of nitrate to coarse-model aerosols can be substantial. For example, single particle measurements of sea salt particles using an aerosol time-of-flight mass spectrometer (ATOFMS) in California found that the chloride mole fraction could decrease from 0.3 to 0, while the nitrate mole fraction could increase from 0 to 0.5 following exposure of sea salt to urban air pollution (Gard et al., 1998). In coastal Florida, measurements using an online ion chromatographic technique showed that deliquesced sea salt particles could have more than 50 % of the chloride content replaced by nitrate (Dasgupta et al., 2007). Measurements from a size-resolved integrated sampler showed that nitrate peaked in the coarse mode, where its concentration was close to that of chloride (on a molar basis) and half that of sodium. Size-resolved particle composition measurements near the coast in British Columbia during the summer showed that during the day, nitrate was only present in coarse-mode particles (corresponding to a deficiency in chloride), whereas during the night, nitrate could also be measured in accumulation mode particles (Anlauf et al., 2006). In addition, Lee et al. (2018) identify that coarse-mode nitrate particles, formed from acid displacement, are more important in national park areas in Arizona and Tennessee. Measurements were taken during spring and summer, respectively, so did not have competing NH₄NO₃ formation; however, they do highlight the fact coarse particle nitrate extends into the PM_2.5 regime and not all nitrate in this regime is associated with NH₄.

3.2.2 Evidence from snow analysis

During the campaign period, a snow event occurred prior to the first PCAP episode (21 January), in which the snow remained on the surface until 3 February. This provided a snowpack that was subjected to the first pollution event, providing a record of chemical composition changes caused by PCAP exposure. A snow event that happened later in the season occurred during clean conditions (23 February to 2 March 2017), when no PCAP was observed, thus providing a representative example of chemical composition changes in snow not associated with PCAP exposure.

Chemical analysis of the snowpack exposed to PCAP episode 1 reflects a combination of cloud water composition and the atmospheric gas and particle composition of species scavenged by snowfall or later deposited to the snow, including particles larger than the PM_2.5 cutoff monitored by the AIM-IC. Figure 6 shows a scatter plot of the NH${}_{\mathrm{4}}^{+}$ versus NO${}_{\mathrm{3}}^{-}$ measured in the snowpack, where each data point is representative of the averaged chemical composition of the entire snowpack (sampling area 0.008 m²) on a single day, and in the atmosphere by the AIM-IC (gas phase + PM_2.5), where PM_2.5 NO${}_{\mathrm{3}}^{-}$ is assumed to be equivalent to 90 % of PM_2.5 NH${}_{\mathrm{4}}^{+}$ on a molar basis, and by a Twin Otter aircraft (gas phase + PM₁). Aerosol particle measurements from the aircraft do not include non-volatile cations and their associated nitrates, which the AMS does not measure (Franchin et al., 2018). Gas-phase HNO₃ measurements collected on the aircraft are from the University of Washington high-resolution time-of-flight chemical ionization mass spectrometer (HR-ToF-CIMS) instrument operated similarly to the method described in Lee et al. (2018). Gas-phase NH₃ measurements from the Twin Otter were taken using a quantum cascade tunable infrared differential absorption spectrometer (QC-TILDAS) instrument (Aerodyne Research Inc, MA, USA) and are described in Moravek et al. (2019). Based on aircraft data (Franchin et al., 2018) and AIM-IC measurements at the ground site, the sum of nitrate in the gas phase and particle phases, or total NO₃ (TNO₃) per cubic metre of ambient air, is less than the sum of ammonia in the gas phase and particle phases, total NH_x per cubic metre of ambient air. Therefore, we might expect that there would be more ammonium depositing into the snow compared to nitrate. However, in Fig. 6, it is clear that the deposition of NH_x is less than half that of total NO₃ on a molar basis for the highest deposition amounts. The average TNO₃ : NH_x molar ratio of 3 : 1 measured in the snowpack is significantly different compared to what is observed in the atmospheric measurements. This implies that there could be another source of nitrate being deposited to the snow that is not reflected in the atmospheric measurements of the gas-phase and fine-mode particles. Therefore, there is a missing portion of the overall nitrate budget that is connected to how much NH₄NO₃ can be formed.

Figure 6 Total NO₃ (pNO${}_{\mathrm{3}}^{-}$ + HNO₃) with respect to total NH_x measured and predicted from AIM-IC data (grey circles), aircraft measurements (open blue circles) per cubic metre of ambient air and snow (red triangles) within 0.008 m² sampling area from 21 January to 3 February 2017.

Further investigation into the composition of the snowpack reveals there is a strong correlation between the amount of NO${}_{\mathrm{3}}^{-}$ and the amount of Ca²⁺ (r²=0.836), seen in Fig. 7. Both Na⁺ and Ca²⁺ are also found in the snowpack in higher abundances than NH${}_{\mathrm{4}}^{+}$, despite being present at much lower levels in PM_2.5. The larger concentrations of pNa⁺ relative to pCa²⁺ seen in the AIM-IC PM_2.5 data are consistent with the larger concentrations of Na⁺ measured in the snowpack relative to Ca²⁺ concentrations.

Figure 7 Daily total micromoles of NH₄, Ca, and Na vs. NO₃ within the snow from 21 January to 3 February 2017, leading into the most severe PCAP period from 27 January to 4 February 2017.

This suggests the coarse-mode aerosols during this PCAP event were Na-rich. Due to the large excess of micromoles of Na⁺ when compared to NO${}_{\mathrm{3}}^{-}$, the amount of Cl⁻ was examined to determine whether Na-rich particles came in the form of NaCl salt. The snowpack showed the highest concentrations of Cl⁻ out of all of the inorganic anions measured, over 10 times more than NO${}_{\mathrm{3}}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ combined. The total water-soluble ion balance measured in the snowpack shows a slight excess of anions, partially due to the high Cl⁻ content. However, the concentrations of Cl⁻ measured in the snow were greater than what can be accounted for by observed Na⁺. The low concentrations of Mg²⁺ in the snowpack had no noticeable correlation with Cl⁻. A closer look at the daily inorganic composition changes and decrease in snowpack height reveals Ca²⁺ and NO${}_{\mathrm{3}}^{-}$ exhibit similar trends throughout the lifetime of the snowpack. This is also true for changes in K⁺ and Cl⁻ total snow column concentrations.

The proportions of non-volatile cations, where K⁺ > Na⁺ > Ca²⁺ > Mg²⁺, found in the snow exhibit a similar distribution to that observed in PM_2.5 composition. We infer that our measurements of these constituents in PM_2.5 also reflect a larger unmeasured contribution of these non-volatile cations in the coarse mode. In addition, the nitrate present in the snow appears to have a strong correlation with Ca²⁺, suggesting there could be a significant amount of particle nitrate in the coarse mode that is not accounted for in PM_2.5 measurements. This underestimation of particle nitrate during peak coarse-mode periods that occur during the day could lead to an underestimation in the production rate of nitrate due to photochemistry. This could impact estimates of the sensitivity of PM_2.5 to NH₃ and NO_x emissions reductions.

3.2.3 ISORROPIA analysis

The comparison of ISORROPIA model runs of predicted NH₄NO₃ (µg m⁻³) with and without non-volatile components measured by the AIM-IC are depicted in Fig. 8 through pollution episode 2 from 13 to 19 February 2017 when both anion and cation gas and particle data were available. ISORROPIA calculates the equilibrium concentrations of the NH${}_{\mathrm{4}}^{+}$–SO${}_{\mathrm{4}}^{\mathrm{2}-}$–NO${}_{\mathrm{3}}^{-}$–Na⁺–Cl⁻–K⁺–Mg²⁺–Ca²⁺–H₂O system. Overall, the difference in model outputs for these two input conditions demonstrates how influential the presence of non-volatiles (pNa⁺, pCa²⁺, pMg²⁺, and pK⁺) is on the thermodynamic equilibrium between HNO₃ and its particulate counterpart pNO${}_{\mathrm{3}}^{-}$.

Figure 8 Predicted concentrations (ISORROPIA) with respected to measured (AIM-IC) ambient concentrations of NH₄NO₃ in µg m⁻³ through the second pollution event from 13 to 19 February 2017. The model was run in three separate conditions: (1) with (red cross) and (2) without (open black circle) the addition of pK⁺, pCa²⁺, and pMg²⁺ and (3) with non-volatiles and the increase in total nitrate based on the ratio of NH${}_{\mathrm{4}}^{+}$ to NO${}_{\mathrm{3}}^{-}$ measured in the snow.

Previous studies have shown the inclusion of non-volatile cations in thermodynamic models can more accurately reproduce gas-particle partitioning observations by correctly reflecting the ion balance and the ammonium : sulfate ratio (Guo et al., 2018). With the addition of pCl⁻, pNa⁺, pCa²⁺, pMg²⁺, and pK⁺ in ISORROPIA model runs, the model consistently underpredicts the concentrations of NH₄NO₃ observed. This underestimation is expected because the implicit assumption in the ISORROPIA run is that all the non-volatile cations are associated with particle nitrate, which is unlikely to be the case. The model also does not accommodate the presence of carbonate, nor does the AIM-IC measure particulate carbonate, which can also associate with non-volatile cations. Therefore, the model runs represent an upper limit of how much nitrate could be permanently trapped in the particulate phase. The model outputs imply that some non-zero fraction of the nitrate in the gas + PM_2.5 system is associated with non-volatile cations. Based on the strong relationship ISORROPIA predicts between non-volatile cations and particle nitrate, the next step is to determine how sensitive NH₄NO₃ formation is to this portion of missing nitrate. Figure 8 demonstrates how much more NH₄NO₃ could be formed if HNO₃ was not sequestered by the coarse particles, depicted in blue squares. Based on measured gas-phase NH₃ and pNH${}_{\mathrm{4}}^{+}$, it is assumed there is a very little coarse-mode NH${}_{\mathrm{4}}^{+}$ in the atmosphere. To estimate the missing nitrate budget associated with coarse particles, assuming the total concentration of snow nitrate is derived from coarse particles is unrealistic. Therefore, with ambient NH_x measurements as a point of reference, the total NH${}_{\mathrm{4}}^{+}$ in the snow was used to estimate the fraction of snow nitrate potentially from coarse particles, in the absence of any other direct coarse-mode measurements. The total nitrate input into ISORROPIA was calculated using the NO${}_{\mathrm{3}}^{-}$-to-NH${}_{\mathrm{4}}^{+}$ ratio observed in the snowpack during the first PCAP event. This serves as an upper limit of the missing nitrate budget and the impact this HNO₃ loss has on NH₄NO₃ formation. As seen in Fig. 6, the TNO${}_{\mathrm{3}}^{-}$ concentration was on average 2.5 times greater than the TNH${}_{\mathrm{4}}^{+}$ concentration measured in the snowpack. This ratio is strictly for the concentrations measured during the PCAP event. When averaging TNO${}_{\mathrm{3}}^{-}$ : TNH${}_{\mathrm{4}}^{+}$ molar ratios for the full lifetime of the snowpack, the average ratio is slightly greater (3). This average ratio estimates the relative amount of TNO${}_{\mathrm{3}}^{-}$ that may come from coarse particles. This does not account for HNO₃ dry deposition loss; therefore, the ratio of NO${}_{\mathrm{3}}^{-}$ to NH${}_{\mathrm{4}}^{+}$ is an overestimate. However, with the combination of stable PCAP conditions, in which the vertical mixing is suppressed, and relatively low ambient concentrations of HNO₃ (< 1 ppb), the dry deposition velocities would be relatively small. Therefore, this overestimate is within reason. This average molar ratio of 2.5 measured during the pollution event was applied to the total NH_x measured by the AIM-IC and used as the total nitrate input into ISORROPIA along with pCl⁻, pNa⁺, pCa²⁺, pMg²⁺, and pK⁺ concentrations observed by the AIM-IC. Overall, the model predicts greater amounts of NH₄NO₃ and, therefore, PM_2.5 will form. Based on the upper-limit estimates of the missing nitrate, the increased amounts of NH₄NO₃ modelled also represent an upper limit of how much more PM could potentially form. When observed NH₄NO₃ concentrations are above 10 µg m⁻³, the model predicts roughly 2 times more NH₄NO₃ mass will form when nitrate lost to coarse particles is accounted for. The additional modelled pNO₃ mass is on average 6.8 µg m⁻³ with a maximum of 18 µg m⁻³, which is within the upper-limit range of total nitrate loss predicted when assuming uptake to the entire bulk of the aerosol particle. This simple analysis also shows how strongly HNO₃-limited the PCAP event in the SLV was, when additional nitrate led to increases in NH₄NO₃. This suggests the coarse particles have the potential to control NH₄NO₃ formation indirectly by acting as a permanent sink for HNO₃, thereby reducing the amount of available HNO₃ to form PM_2.5. Therefore, future campaign studies should also include coarse particle speciation, especially in systems involving semi-volatile nitrate salts.

4 Conclusion

The AIM-IC measurements confirmed the dominance of NH₄NO₃ in wintertime PM_2.5 pollution events in the SLV. The detection of non-volatile cations (pNa⁺, pK⁺, pCa²⁺, and pMg²⁺) in PM_2.5 by the AIM-IC suggests the presence of mineral dust and/or salt during PCAP pollution events, which can potentially impact the availability of HNO₃. This is due to the reactive uptake of HNO₃ onto NaCl- and CaCO₃-containing particles, which can be predominantly found in the coarse mode. This can introduce a permanent sink for HNO₃. Due to the lack of coarse-mode speciation, auxiliary data were used to estimate the amount of HNO₃ potentially lost to reactive uptake. The APS surface area data were used to calculate a lifetime for HNO₃ with respect to uptake that was on the order of minutes, depending on the uptake coefficient used, which is influenced by particle composition (Fenter et al., 1994; Liu et al., 2008b; Vlasenko et al., 2006). The potential presence of coarse-mode nitrate was also supported by the strong correlation between total Ca²⁺ and total NO${}_{\mathrm{3}}^{-}$ observed in the snowpack. Total NO${}_{\mathrm{3}}^{-}$ in the snowpack was 2.8 times larger than the total NH${}_{\mathrm{4}}^{+}$, implying there is additional NO${}_{\mathrm{3}}^{-}$ that is not observed by PM_2.5 pNO${}_{\mathrm{3}}^{-}$ and gas-phase HNO₃ measurements alone.

The equilibrium dynamics were explored by thermodynamic modelling of trace gas and PM_2.5 composition, during episode 2. Despite the discrepancies between observed and modelled NH₄NO₃ concentrations during PCAP episode 2, the inclusion of coarse nitrate, from snowpack estimates, into the model shows that coarse particles could be limiting a significant portion of the HNO₃ being generated, preventing its reaction with NH₃ to form NH₄NO₃. The extent to which coarse-mode aerosols are limiting PM_2.5 formation remains unanswered and warrants further investigation of coarse-mode composition and the role these particles play in the atmospheric chemistry during pollution events.