An Adiabatic‐Expansion‐Induced Perturbation Study on Gas‐Aerosol Partitioning in Ambient Air ‐ Formation of NH4NO3 and Microdroplet Nitrogen Fixation (2)

1. Intruduction

Ammonia (NH₃) dominates the gas-phase alkaline species in the atmosphere and acts as the major contributor to neutralize atmospheric acids secondarily generated from natural and anthropogenic precursors [1,2,3]. Atmospheric NH₃ is widely recognized as being derived from various natural and anthropogenic sources in surface environments on the earth, including chemical fertilizers, livestock, natural soils and oceans, etc. [1,4,5,6]. However, satellite measurements have detected significant concentrations of NH₃ (above 15 pptv) in the upper troposphere and lower stratosphere (UTLS) over the subtropical regions of the southeastern Asian continent (20°-30°N, 70°-110°E) during the summer monsoon season, along with abundant neutralized NH₄NO₃ in the UTLS over remote oceanic zones, while the puzzle remains poorly understood [7,8,9]. In addition, whether oceans act as a net source or a net sink for NH₃ in remote, clean marine atmospheres is also a topic of ongoing debate [10,11,12,13]. Could these remote atmospheres harbor unrecognized sources of NH₃? This question, with its inherent complexity, may extend far beyond the conventional paradigm on NH₃ budgets in the atmosphere [14,15].

Excluding biogenic release of atmospheric NH₃, as well as the burning of fossil fuels and biomass, industrial NH₃ synthesis on Earth typically requires harsh operating conditions, such as high pressure, high temperature and highly efficient catalysts [1,5]. To circumvent these harsh conditions, there has been significant interest in the electrochemical reduction of NO and N₂ [15,16]. However, this process is severely restricted by the lack of efficient catalysts. Recently, Song et al. [14] reported a novel method for generating NH₃ under ambient conditions without the application of external electric potential or irradiation, i.e., spraying water microdroplets onto a magnetic iron oxide (Fe₃O₄) and Nafion-coated graphite mesh using compressed N₂ or air. The extremely strong perturbation signals of NH₄NO₃ reported in the companion paper raises two questions: 1) Does microdroplet-driven generation of NH₃ occur widely in cold, moist ambient air? and 2) Is the efficiency of microdroplet ammonia synthesis determined by the line speed of moving droplets, regardless of the presence of specific catalysts?

In this paper, we attempt to explore four key issues: 1) Whether the extremely strong perturbation signals of NH₄NO₃ resulted from the microdroplet redox generation reaction of NH₃ and HNO₃, followed by perturbation formation of NH₄NO₃? 2) Whether the microdroplet redox reaction of NH₃ and HNO₃ is independent of catalysts? 3) Whether the line speed of moving droplets determines the probability for occurring redox reactions of NH₃ and HNO₃, or the oxidation reactions of HNO₃ alone or organic nitrate? 4) What future studies are needed for microdroplet ammonia synthesis and the development of a new paradigm for atmospheric NH₃, particularly in marine atmospheres abundant in fast-moving microdroplets?

2. Experimental

This study includes data collected from coastal and marine atmospheres during nine distinct campaigns. Campaigns 1–7 were defined identically to those in the companion paper, i.e., four coastal campaigns performed in Qingdao, China, during the heating seasons in 2013, 2015 and 2022 (Campaigns 2 - 4), as well as the non-heating season in 2013 (Campaign 1), and three marine campaigns as Campaigns 5-7 over the marginal seas of China in November - December 2012, 2013 and July- August 2016 [17], respectively. Campaign 8 was performed at a coastal site <1 km away from that used for Campaign 4, with nine sets of samples collected during the summer of 2019 on 11, 20, 21 July, and 1, 4, 5, 6, 12 and 13 August. The chemical analysis was consistent with those detailed in Yu et al. [18], Xie et al. [19] and Hu et al. [20]. In addition, some Nano-MOUDI results from the spring and summer marine campaigns of 2015, as reported by Yu et al. [18], were also referenced for discussion. Campaign 9 was performed over the South China Sea (SCS, a tropical region) from 29 March to 2 May 2017 [21]. Eleven sets of Nano-MOUDI samples were collected during the campaign, with two sets intentionally gathered on 15 and 18 April to capture strong self-vessel combustion emissions. The remaining samples were collected on 30 March (en route to the South China Sea from the anchor port), 2, 7, 11, 21, 22, 23, 26, and 29 April. Note that 0.5%-1.0% high sulfur-content diesel was commonly used on research vessels at that time in China.

An Ambient Ion Monitor−Ion Chromatograph (AIM-IC, URG-9000D) was used for simultaneous sampling during Campaign 3 and part of Campaign 4, but not in the other campaigns. Moreover, the E-AIM model was employed to calculate the gas-aerosol partitioning of semi-volatile species [22] (http://www.aim.env.uea.ac.uk/aim/aim.php, last access: 20 January, 2025). For the cruise campaigns, meteorological data were obtained from an automatic meteorological station on the vessel. For the coastal campaigns, the meteorological data were download from the meteorological data were download from China Meteorological Data Service Centre (https://data.cma.cn/, last access: 25 January, 2025).

3. Materials and Methods

3.1. Perturbation Formation of Comparable NH₄NO₃ and Organic Nitrate in Campaigns 8 and 9

In Campaign 8, the strongest perturbation formation of NH₄NO₃ was detected at the size bin of 0.032-0.056 µm on 20 July 2019 (Figure 1a, b). The existence of higher NH₄NO₃ concentrations in atmospheric particles within the 0.032-0.056 µm size range, compared to the 0.056-0.10 µm size range with an identical dlogD_p value, was theoretically and practically impossible due to the size-dependent Kelvin effect and the universally observed larger surface area of atmospheric particles in the latter size range [3,23]. The Kelvin effect has also been reported to result in higher SO₄^2- concentrations than NO₃^- in atmospheric particles smaller than 100 nm [24], and did not support that over one order of magnitude higher concentration of NO₃^- (23 neq m^-3) than that of SO₄^2- (0.47 neq m^-3) observed at 0.032-0.056 µm size range (Figure 1a, c). Even surprisingly, the detected molar concentration of NH₄⁺ at that size bin was only accounted for approximately 1/4 of the corresponding NO₃^- concentration. This raised the possibility of perturbation formation of organic nitrate, which was commonly detected in ambient air [25,26], to account for the NO₃^- unexplained by NH₄⁺. Organic nitrate might have dissolved into the deionized water and been detected by the ion chromatography as NO₃^-. On the same day, strong perturbation formation of MSA^- was also detected, but not for SO₄^2- and DMAH⁺. On the remaining days of Campaign 8, substantially weak perturbation formation of particulate nitrate was mostly observed at the 0.018-0.032 µm size bin. Again, the detected amount of NH₄⁺ via perturbation formation was insufficient to explain the corresponding NO₃^- values.

In Campaign 9, the strongest perturbation formation of NH₄NO₃ was detected at the 0.032-0.056 µm size bin on 21 April 2017 (Figure 2a, b). However, similar to Campaign 8, the detected amount of NH₄⁺ (2.9 neq m^-3) in this size range accounted for only about 1/5 of the corresponding NO₃^-value (15.5 neq m^-3). This suggested that approximately 80% of the detected NO₃^- might have existed as organic nitrate. The same can be said to the perturbation formation of organic nitrate on 15 and 18 April when the self-vessel combustion plumes were deliberately observed. However, the perturbation signals were substantially weaker on these two days than those on 21 April. On the remaining days, the perturbation signals for NO₃^- were generally too weak to confirm, and the same was true for SO₄^2- and DMAH⁺. The lower occurrence frequency of perturbation formation of NH₄NO₃ and insufficient amount of NH₃ formed could be related to the reduced line speeds of air streams, as discussed in Section 3.4.

3.2. Whether Is Adiabatic Perturbation Alone Sufficient to Explain NO₃^- and NH₄⁺ Detected on the Last Three Stages of Nano-MOUDI Sampling in Cold Coastal Atmospheres?

To determine whether enhanced gas-aerosol partitioning alone is sufficient to generate the perturbation amount of NH₄NO₃ in the last three stages of Nano-MOUDI sampling in coastal atmospheres, additional simultaneous measurements of HNO₃ and NH₃ gases, along with their counterparts in PM_2.5, are required. Therefore, the perturbation formation of NH₄NO₃ during Campaign 3 were selected to examine the hypothesis (Figure 3). To increase the dataset for correlation analysis, two samples collected over the Yellow Sea and the Bohai Sea, taken two and a half months prior to Campaign 3, were also included, as their perturbation characteristics were more similar to those observed in Campaign 3 rather than in other marine campaigns. Strong perturbation formation of particulate NO₃^- was observed in 5 of 11 samples, with NO₃^-concentrations in PM_0.010-0.056 exceeding those in PM_0.056-1.0 by more than double (Figure 3a, b). The perturbation was, however, negligible in 2 out of 11 samples, with the remaining samples falling between these two extremes.

As shown in Figure 4 and analyzed in detail below, the perturbation caused by the thermodynamic equilibrium shift alone cannot explain most of higher concentrations of NO₃^- in PM_0.010-0.056 than the corresponding values in PM_2.5, as well as some of NH₄⁺ with higher concentrations in PM_0.010-0.056. When gas-aerosol thermodynamic equilibrium analyses were conducted using the particulate concentrations of chemical species measured by the AIM-IC in PM_2.5 and by the Nano-MOUDI in PM_0.010-3.2, respectively, along with gaseous species measured by the AIM-IC, the modeled results indicated that the gas-aerosol equilibrium was reasonably achieved (Figure 4a, b, d, e). The slightly over-predicted equilibrium concentrations of NO₃^- and under-predicted NH₄⁺ in PM_2.5, might be attributed to the exclusion of NH₄Cl and ammonium organics in PM_2.5 from the modeling. This argument was supported by the consistent equilibrium concentrations of NO₃^- and NH₄⁺ in PM_0.010-3.2 with the corresponding observed values, which were significantly higher than those in PM_2.5. This increase likely reduced the impact of the exclusion on the modeled equilibrium concentrations.

With the gas-aerosol thermodynamic equilibria to be achieved, adiabatic-expansion-induced perturbation would cause a shift in the equilibrium, leading to additional gaseous NH₃ and HNO₃ partitioning into the particle phase (Figure 4c, f). This was clearly demonstrated in Figure 4c, where the concentrations of (NH_3gas + NH₄⁺ in PM_2.5) measured by the AIM-IC were plotted against the corresponding NH₄⁺ in PM_0.010-3.2, with four data points aligning along the 1:1 line. On 14 and 15 December, 2015, two data points aligned with the 1:1 line when the concentrations of (HNO_3gas + NO₃^- in PM_2.5) measured by the AIM-IC were plotted against the corresponding NO₃^- in PM_0.010-3.2. On these two days, the data points in Figure 4c slightly scattered above the 1:1 line. However, 7 out of 9 data points in Figure 4f and 2 out of 9 data points in Figure 4c scattered far below the 1:1 line. In these cases, additional ultrafast oxidation and/or redox reactions definitely occurred, resulting in the formation of (HNO₃+NH₃) and organic nitrate in the last three-stage sampling.

Based on the size distributions presented in Figure 3a, the maximum NO₃^- concentration via perturbation formation occurred across various size bins. By considering the sampling flow, the diameters (~200 μm) and numbers of nozzles (2940), the maximum line speed of air streams passing through the last three-stage nozzles can be roughly estimated to ~5*10² m s^-1. Giving this estimation, the observed perturbation formation of NO₃^- must have occurred at an ultrafast rate, within approximately 1*10^-4 s. However, the line speed of air streams should largely decrease after they left the nozzle outlet. The ultrafast redox and oxidation reactions of NH_3, HNO₃ and organic nitrate differ from those traditionally recognized reactions in atmospheric chemistry included in various air quality or climate modeling [3]. These reactions likely occurred via microdroplet chemistry when the microdroplets were sprayed at an ultrafast moving-speed [14,27,28]. While the contribution of vacuum ionization cannot be entirely ruled out [29], it is likely a minor factor due to the incomplete vacuum condition in the last three stages of MOUDI (~0.4 atm). In addition, NO, one of the gaseous pollutants primarily emitted from vehicle exhaust, industrial plants, or thermal power stations, can be converted to NH₃ using three-way catalysts equipped on vehicles [30]. The simultaneously observed concentrations of NO and NO₂, as shown in Figure S1, could be sufficient to generate NH_3, HNO₃ and organic nitrate via microdroplet chemistry, however, we did not measure microdroplets directly. Moreover, there were no significant correlations between the average ambient temperature or relative humidity and the perturbation-formed particulate nitrate (Figure S1). Note that the theory on spontaneous microdroplet redox chemistry has only recently become available [31]. Therefore, it is unsurprising that the unexpectedly high concentrations of NO₃^- and NH₄⁺ observed at the last three stages cannot be timely explained.

When the concentrations of NO₃^- in the 0.010-0.056 μm size bins were plotted against the corresponding NH₄⁺ values (Figure 3c), a good correlation was obtained, with a slope of 1.49. This slope also suggested that 1/3 of the observed NO₃^- via perturbation formation likely existed as organic nitrate. However, the ratio of N+2*S to NH₄⁺ in PM_0.056-1.0 was 0.95 (Figure 3d), implying that NO₃^- and SO₄^2- in this size range were primarily associated with NH₄⁺ and were almost completely neutralized.

In addition, some samples exhibited a distinct droplet mode of NO₃^- at 1-2 μm (Figure S2), conventionally attributed to fog-processing of aerosols [3]. This unique feature suggests the possibility that the microdroplet chemistry may have occurred in larger size bins for some samples of the Campaign 3. Based on the observational data, we artificially assumed that the microdroplet chemistry could cause 100% of the positive artifacts in the first four samples collected on 27-30 November, 2015. Under this assumption, we surprisingly gained the new good correlations as follows (Figure S3): [NH₄⁺]_PM2.5 = 0.92 * [NH₄⁺*]_PM0.056-3.2, R²=0.94, P<0.01; [SO₄^2-]_PM2.5 = 1.02 * [SO₄^2-*]_PM0.056-3.2, R²=0.97, P<0.01; [NO₃^-]_PM2.5 = 0.69 * [NO₃^-*]_PM0.056-3.2, R²=0.98, P<0.01. The tested results, however, require further validation in future studies.

3.3. Adiabatic Perturbation Superimposed Ultrafast Formation of Huge Amounts of NH₄NO₃ and Organic Nitrate at the Last Three Stages of Nano-MOUDI Sampling in Marine Atmospheres

In Campaign 5, strong perturbation formation of NH₄NO₃ was observed in 4 out of 14 samples (Figure 5a, b, c). The maximum perturbations occurred at size range of 0.032-0.056 µm and 0.018-0.032 µm, respectively. A strong correlation with a unity slope was found when the concentrations of NO₃^- in PM_0.010-0.056 were plotted against corresponding NH₄⁺ concentrations, indicating no detectable perturbation formation of organic nitrate. Although simultaneous measurements of NH_3gas and HNO_3gas were not available during this period, our observations from the marine atmosphere in April 2018 and December 2019 suggested that the adiabatic perturbation alone cannot account for the observed NH₄NO₃ formation_. Alternatively, ultrafast redox reactions of NH₃ and HNO₃ were required. During the 2012 cruise campaign, SO₄^2- and NO₃^- in PM_0.056-1.0 appeared to be completely neutralized by NH₄⁺ (Figure 5d).

Even stronger perturbation formation of NH₄NO₃ occurred during Campaign 6 (Figure 6a, b, c). Again, no detectable organic nitrate was found concurrently with the perturbation formation of NH₄NO₃ (Figure 6d). It is interesting that the perturbation formation of NH₄NO₃ was negligible on the first two days (6-7 November, 2013) but intensified in the following days, with the maximum perturbation occurring at size bins varying a lot. In 8 out of 15 samples in Campaign 6, the perturbation formation of NH₄NO₃ ≥ 140 µg m^-3 (with NO₃^- reaching 202±71 μg m^-3 and NH₄⁺ at 61±4 μg m^-3). The substantial amount of NH₄NO₃ and ~10^-4 s residence time of air streams definitely required ultrafast redox reactions of NH₃ and HNO₃ during sampling. It also indicated that such ultrafast redox reactions of NH₃ practically occurred under mild conditions without the need for additional electricity consumption. Unlike in Campaign 5, SO₄^2- and NO₃^- in PM_0.056-1.0 were likely incompletely neutralized by NH₄⁺ in Campaign 6 (Figure 6e).

3.4. Key Factors in Determining Ultrafast Formation of (HNO₃+NH₃) and Organic Nitrate: Evidences and Uncertainties

To comply with the Second Law of thermodynamics, the ultrafast formation of (HNO₃+NH₃) and organic nitrate must be driven either by catalytic reactions or by the highly efficient conversion of energy from microdroplet motion into the electrochemical energy required. To investigate the occurrence of catalytic reactions, we compared the chemical comparisons between the 8th-stage and 12th-stage micro-orifice nozzle plates. The internal and external surface chemical compositions of the 8th-stage (S8) and 12th-stage (S12) plates, corresponding to the size bins of 0.18-0.32 μm and 0.018-0.032 μm, respectively, are shown in Figure 7. Both plates were primarily composed of aluminum and oxygen, which together accounted for 61%-76% of the total weight on internal and external surfaces, following by sulfur contributing 5.8%±0.7% of the total weight. No significant differences in these major components were detected between the internal and external surfaces of each plates. However, more Fe and Ni were detected on 12th-stage plates, accounting for 6.2% and 3.0% of the total weight, respectively. The role of these minor components as potential catalysts needs further investigation. However, potential catalytic reactions cannot explain the ultrafast formation of (HNO₃+NH₃) and organic nitrate being overwhelmed in a certain stage, e.g., Figures 4a and 5b.

Considering the conversion of energy from microdroplet motion into the electrochemical energy required for microdroplet chemistry, the high line speed of microdroplets might increase the microdroplet surface electric voltage and subsequently enhance ultrafast reactions [33]. Various perturbation levels of NO₃^- in PM_0.010-0.056 under different MOUDI sampling flow rates were thus analyzed in Figure 8. The perturbation levels are thoroughly defined in the companion paper [17], i.e., negligible and unconfirmable perturbation formation for levels 0-1, moderate perturbation formation for level 2, strong and extremely strong perturbation formation for levels 3-4. The efficiency of (HNO₃+NH₃) and organic nitrate generation expectedly increased with increasing air stream line speed through the nozzles. Additionally, sea salt aerosols and lower ambient temperature may further enhance the ultrafast reactions, as presented below.

1) The efficiency of NH₄NO₃ perturbation generation via microdroplet chemistry was found to be dependent on the air stream line speed through the nozzles, which was determined by the MOUDI sampling flow rate (Figure 7 and Figure S4). For example, strong to extremely strong perturbation generation was observed during the 2015 coastal campaign and the 2012 and 2013 marine campaigns at a MOUDI (Model 122) flow rate of 29.4 L min^-1, corresponding to an air stream nozzle-outlet line speed of at least 5*10² m s^-1. In contrast, no microdroplet chemistry was detected at a sampling flow rate of 10 L min^-1 (using a different MOUDI, Model 125) during the November 2013 coastal campaign. However, the perturbation generation of organic nitrate was likely less influenced by the air stream line speed, as indicated by the presence of NO₃^- perturbation generation during the 2019 coastal campaign.

2) The presence of ultrafine sea salt aerosol significantly enhanced the efficiency of perturbation generation for (HNO₃+NH₃) and organic nitrate. As illustrated in Figure 8, the frequency of NO₃^- perturbation generation was higher in marine atmospheres than in coastal atmospheres. Specifically, 29 out of 60 samples collected from marine atmospheres were classified in Levels 2-4, whereas only 11 out of 64 samples from coastal atmospheres reached these levels. Additionally, the intensity of perturbation was greater in marine atmospheres, with 12 out of 15 samples falling into Levels 3-4 during the 2013 marine campaign, compared to only 3 out of 9 samples during the 2015 coastal campaign.

3) Higher ambient temperature likely disfavored the perturbation generation of (HNO₃+NH₃) and organic nitrate, as evidenced by the absence of NH₄NO₃ perturbation generation during the May 2013 coastal campaign and the July 2016 marine campaign. However, it was also possible that higher ambient temperatures disfavored gas-aerosol partitioning. In such cases, the microdroplet-generated (HNO₃+NH₃) and organic nitrate gases may have passed directly through the nozzles rather than being captured on the filters.

Overall, an innovative and controllable experiment system is necessary to examine the uncertainties. For example, to study the impact of air stream line speed on the efficiency of perturbation generation for (HNO₃+NH₃) and organic nitrate, water microdroplet could be sprayed into a NO₂/N₂-filled (or NO/NO₂/N₂-filled) chamber through a nozzle at varying air stream line speeds, with temperature and pressure controlled. Ionic concentrations would be dynamically monitored to observe changes. Additionally, spraying NaCl-containing microdroplet and adjusting the system’s temperature would enable further investigation into the effects of sea salt aerosol and ambient temperature on the efficiency of perturbation generation. Although our current experiments lacked the high temporal resolution to detect dynamic changes in microdroplets, this limitation may not pose a significant issue for the broader research community.

4. Implication

This study reports unique findings on the ultrafast redox reactions of generating substantial amounts of (HNO₃+NH₃) or organic nitrate in adiabatic-expanded ambient air within a time scale of approximately 10^-4 s. Such ultrafast and highly efficient redox reactions of (HNO₃+NH₃) or organic nitrate have definitely not been previously recognized in the atmospheric chemistry community, and are not currently included in air quality and climate models. However, these reactions may occur due to microdroplet chemistry under ultrafast moving speeds of microdroplets.

There remain numerous uncertainties regarding these new findings, including: 1) Does low ambient temperature enhance the redox reaction of (HNO₃+NH₃) or organic nitrate? 2) What factors control the ratio of (HNO₃+NH₃) to organic nitrate in the ultrafast redox reactions in addition to line speeds of micro-droplets? 3) To what extent does sea salt aerosol significantly increase the efficiencies of redox reactions via microdroplet generation? 4) Does NH₃ generation occur from N₂ or NO? 5) Is this process critical for marine nitrogen cycling, particularly giving the presence of microdroplets during typhoon periods or strong cumulus convection? This may also apply to upper atmosphere. 6) Is the NH₃ emitted from on-road vehicles generated by ultrafast-moving microdroplets? 7) Does microdroplet chemistry influence the formation of secondary species in freshly emitted combustion plumes within the initial seconds?