Partitioning of Organonitrates in the Production of Secondary Organic Aerosols from α-Pinene Photo-Oxidation.

The chemical pathways for the production of secondary organic aerosols (SOA) are influenced by the concentration of nitrogen oxides (NOx), including the production of organonitrates (ON). Herein, a series of experiments conducted in an environmental chamber investigated the production and partitioning of total organonitrates from α-pinene photo-oxidation from <1 to 24 ppb NOx. Gas-phase and particle-phase organonitrates (gON and pON, respectively) were measured by laser-induced fluorescence (LIF). The composition of the particle phase and the particle mass concentration were simultaneously characterized by online aerosol mass spectrometry. The LIF and MS measurements of pON concentrations had a Pearson correlation coefficient of 0.91 from 0.3 to 1.1 μg m-3. For 1-6 ppb NOx, the yield of SOA particle mass concentration increased from 0.02 to 0.044 with NOx concentration. For >6 ppb NOx, the yield steadily dropped, reaching 0.034 at 24 ppb NOx. By comparison, the yield of pON steadily increased from 0.002 to 0.022 across the range of investigated NOx concentrations. The yield of gON likewise increased from 0.005 to 0.148. The gas-to-particle partitioning ratio (pON/(pON + gON)) depended strongly on the NOx concentration, changing from 0.27 to 0.13 as the NOx increased from <1 to 24 ppb. In the atmosphere, there is typically a cross-over point between clean and polluted conditions that strongly affects SOA production, and the results herein quantitatively identify 6 ppb NOx as that point for α-pinene photo-oxidation under these study conditions, including the production and partitioning of organonitrates. The trends in SOA yield and partitioning ratio as a function of NOx occur because of the changes in pON volatility.

Introduction

Secondary organic aerosols (SOA) particles represent a significant fraction of atmospheric particulate matter. They can influence climate radiative forcing[1,2] and adversely affect human health.[3] SOA particles are produced by the low-volatility products of atmospheric oxidation chemistry. Volatile organic compounds (VOCs) are oxidized by hydroxyl radicals (OH), nitrate radicals (NO3), and ozone (O3). The low-volatility products leave the gas phase and partition onto particles, possibly undergoing further heterogeneous reactions with other particle-phase organic molecules or gas-phase species.[4,5] Coupling the reaction cycle of VOCs oxidation with those of nitrogen oxides (NO = NO and NO2) can lead to the production of organonitrates (ON = ΣRO2NO2 + ΣRONO2). These reactions can be especially important for polluted atmospheric conditions. Ambiguity regarding the effects of anthropogenic emissions represented by NO concentration on aerosols formation, organicnitrates production, and particles composition contributes to the uncertainties of the effect of SOA particles on radiative forcing and human health.

The chemical mechanisms for the influence of NO on SOA production are complex, hampering efforts to simulate and control atmospheric particulate matter (PM), especially the human influence on it.[6] The chemical mechanism is initiated by a reaction between organic peroxy radicals (RO2) and NO, leading to the production of ON in the gas phase. These organonitrates, a gas-phase reservoir species for NO2, can be transported far from their source regions. The organonitrates can also partition from the gas phase into atmospheric PM. The production and growth rates of atmospheric SOA particles can be affected by the uptake of organonitrates.[6−12] In rural areas, highly functionalized pON contributed to about 20% of the increase in organic PM mass concentration during the night.[11] Under clean conditions, for a boreal forest in Finland, pON contributed about 35% of the organic PM mass concentration.[10] In the polluted southeastern USA, the daytime and nighttime contributions of pON to organic PM mass concentration were 3 and 8%, respectively.[6]

The production of organonitrates is one of the direct links for the observed effects of atmospheric NO concentrations on SOA particle mass concentrations and yields from gaseous precursors. Even so, the link is still not well quantified or understood.[4,13] In field studies, anthropogenic NO emissions enhanced the yield of SOA particle mass concentration from biogenic precursors in several different environments, such as in the rural environments of the Amazon forest when located downwind of pollution,[14,15] in biogenic–anthropogenic mixed air masses in western USA,[13] and during nighttime chemistry due to monoterpene oxidation by nitrate radicals in southeastern USA.[16] At other times, increases in NO concentration suppressed SOA mass concentration as well as particle formation from monoterpene species oxidation[17−20] and methylglyoxal[21] in laboratory experiments. A decrease of highly oxygenated multifunctional organic molecule (HOM) accretion products in the gas phase because of the elevated NO concentration can in some cases explain the observed decreases in SOA mass concentration.[20] These contrasting sets of studies on the effect of NO concentration on SOA particle mass yield highlight the need for greater understanding and quantitative measurements across the full range of atmospheric NO concentrations, including the link to organonitrates.

In this study, we carried out a series of chamber experiments that ranged from simulating a pristine-like environment (i.e., NO < 1 ppb) to an industrial-like environment (i.e., NO of about 24 ppb) for α-pinene photo-oxidation. We used α-pinene to mimic the biogenic emissions of a monoterpene from a forested region, such as Amazonia,[22] where anthropogenic emissions from urban regions interact with biogenic VOCs.[23] The steady-state α-pinene concentration in the chamber of 5 ppb is similar to the real atmosphere in forested environments.[22] The mass concentrations of the produced SOA particles and the gas-phase and particle-phase organonitrates were measured across this range. SOA particle mass yield, the relative contribution of organonitrates to the particle mass yield, and the organonitrates partition fraction were calculated from the data sets. The connections among organonitrates, NO concentration, and SOA particle mass yield were considered.

Experimental Section

The experiments were carried out in the Harvard Environment Chamber (HEC).[24−28] Inside the chamber is present a PFA Teflon bag of 4.7 m3 in volume. Purified zero air, H2O2, α-pinene, and NO were continuously introduced for a total flow of 21.7 sLpm and a chamber mean residence time of 4.5 h. The chamber was operated as a continuously mixed flow reactor. Forty-six Sylvania 350 blacklights (40 W) of negligible emission below 310 nm provided ultraviolet irradiation. Experiments were conducted at ambient temperature (T varying between 24 and 28 °C), and the relative humidity was <10%. Our experiments used VOC, NO, and oxidant concentrations similar to those of the real atmosphere, albeit under dry conditions to remove any RH-dependent effects. Therefore, future experiments are necessary to evaluate the RH effects also.

For chamber inflows, α-pinene (C10H16; Aldrich, ≥98% purity) was injected at 40 ppb in zero air (Aadco 737-14A). Hydrogen peroxide (H2O2) (Sigma-Aldrich, 50 wt %) was injected at 4 ppm in zero air. Nitrogen oxide (NO; EPA protocol, 1 ppm) was injected at different concentrations among different experiments (Table ).

Table 1

Summary of Series of Experimentsa

experiment	initial NOx concentration (ppb)	steady-state NOx concentration (ppb)	steady-state O3 concentration (ppb)
background		0.26 ± 0.02	11.21 ± 1.60
industrial 1	1.61 ± 0.00	0.71 ± 0.03	18.40 ± 1.24
industrial 2	4.03 ± 0.02	1.47 ± 0.01	29.96 ± 6.00
industrial 3	5.92 ± 0.04	1.98 ± 0.06	32.19 ± 1.50
industrial 4	8.79 ± 0.01	2.69 ± 0.05	39.88 ± 0.61
industrial 5	12.17 ± 0.09	4.22 ± 0.05	47.55 ± 2.16
industrial 6	23.80 ± 0.03	7.94 ± 0.10	69.84 ± 2.55

Listed are the initial NO concentration measured before turning on the chamber lights and the NO and O3 concentrations for the chamber at steady state during the reaction.

Inside the chamber, photolysis produced hydroxyl radicals (OH) from H2O2. In turn, OH reacted with α-pinene (C10H16) to produce RO2 radicals, some of which subsequently reacted with NO. Many further reactions continued, a subset of which resulted in stable organonitrates.[29] The rate of decay of the α-pinene concentration from its steady-state value before turning on the lights to its value after turning on the lights (5 ppb) implied an OH concentration of approximately 2 × 106 molec cm3 (Figure S1). Ozone (O3) was also produced by NO2 photo-oxidation chemistry in the chamber (Table ). Using the MCM model, we estimated the α-pinene loss rate due to its reaction with O3 and OH; we found that, at the highest O3 concentrations reached during our experiments, the reaction of α-pinene + OH is 3 times higher than that of α-pinene + O3.

For chamber outflows, the NO concentration was measured by a Thermo 42i-TL NO analyzer. The O3 concentration was measured by an ozone analyzer (model TMI-10; Teledyne Instruments). The NO2 concentration was measured by thermal dissociation laser-induced fluorescence (TD-LIF). The instrument has been described by Dari-Salisburgo et al.[30] and Di Carlo et al.[31] The same instrument also measured the concentration of gas-phase organonitrates (gON) and particle-phase organonitrates (pON), expressed as ON = ΣRO2NO2 + ΣRONO2, as well as speciated peroxy nitrates (ΣRO2NO2; gPN and pPN) and alkyl nitrates (ΣRONO2; gAN and pAN). The TD-LIF inlet had a bypass line for measuring total ON and a denuder line with activated carbon to remove gON. A three-way valve was used for automated switching (Figure S2). The measured removal efficiency by the denuder was 97% for control experiments, with NO2 at 8 ppb. The size distribution of the particle population in the chamber outflow was characterized by a scanning mobility particle sizer (SMPS; TSI, 3010). SMPS data also retrieved the total number, volume, and mass concentrations (Morg) with an aerosol density of 1.2 g/cm3. The particle population was also sampled by aerodyne high-resolution aerosol mass spectrometry (HR-AMS) following standard methods.[32,33] The data were processed by SQUIRREL and PIKA toolkits. The instrument provided the nitrate particle mass concentration. Moreover, the high-resolution data were used to identify organic nitrogen species, including CHN+, NO+, CHO1N+, and CHON+, for x as an integer greater than 1.

Results and Discussion

Production of Organonitrates in Different NOx Regimes

Figure shows the time series for the concentrations of NO and NO2, O3 and gON, gAN and gPN, and organic PM and pON for α-pinene photo-oxidation at 24 ppb of initial NO concentration in the chamber. When the lights are turned on, photo-oxidation begins. As a result, the NO concentration within 10 min decreases to that below the limit of detection. It is replaced by an increased NO2 concentration. Ozone is produced, and its concentration increases to a steady-state value with a time constant of 95 min. Along this same time constant, the NO2 concentration decreases from 19 ppb to its steady-state value of 8 ppb.

Figure 1

Time series for the concentrations of (a) NO and NO2, (b) O3 and gON, (c) gAN and gPN, (d) pAN and pPN, and (e) organic PM and pON for α-pinene photo-oxidation at an initial NO concentration of 24 ppb. Figure S3 shows similar plots for all NO concentrations listed in Table .

Organonitrates are produced, and they are measured in both the gas and particle phases. The concentration gON of gas-phase organonitrates increases from 0 to 4 ppb. gON speciates as 3 ppb gPN of peroxy nitrates and 1 ppb gAN of alkyl nitrates. The concentration pON of particle-phase ON increases to 1.5 ppb. In total, 8 ppb NO2, 4 ppb gON, and 1.3 ppb pON account for 13.3 ppb of the initial NO of 24 ppb. The remaining 10.7 ppb of NOx is lost by both O3 photochemical production and by NO Teflon wall loss, which has been previously reported.[34] The organic PM mass concentration measured by the SMPS increases once the UV lights are turned on, representing the production of SOAs. The speciated pAN and pPN reach 1.1 and 0.2 ppb, respectively, in the particle phase. The particle-phase concentration pON of ON measured by TD-LIF closely tracks the organic MS mass concentration measured by SMPS (i.e., orange points on blue line).

Interestingly, when the UV lights are still off in the chamber at the beginning of the experiment, there is still a significant presence of ON in the gas phase, as seen in the plot for gON. This gON production can be explained due to the presence of O3 in the chamber that could partially oxidize α-pinene, thereby allowing RO2 production even in the dark (Figure S4). Moreover, at this time, the relative contribution of gAN to the total gON is increasingly important for higher initial NO concentrations (Figure S3). Without UV lights, NO is the dominant species in the chamber, and the peroxy radical produced by ozonolysis can preferentially react with NO to produce ΣRONO2 compounds. Once the lights are turned on, the gON species are photolyzed. The implication is that gON produced by α-pinene ozonolysis is less stable than the one produced by OH. At steady state, peroxy nitrates represent the dominant fraction of gON (i.e., 1 ppb gAN and 3 ppb gPN).

The gON species are believed to contribute to new particle production.[8]Figure highlights new particle production in the first few hours of the experiment during the transitory period in spin up to steady-state conditions in later hours. Panel a shows that the maximum in the peak particle number concentration of 5000 cm–3 is for the highest initial NO concentration and that the peak concentration steadily decreases for decreasing initial NO concentration, reaching 340 cm–3 for <1 ppb initial NO concentration. This trend is consistent with the importance of gON species in new particle production. With further time, there is less new particle production and more condensational growth, as reflected in the declining number concentration. Furthermore, because surface area scales with the square of diameter, whereas volume and mass scale with the cube of diameter, these distributions are less sensitive to initial new particle production and hence less sensitive to the initial NO concentration, as reflected in panels b, c, and d. The results for <1 and 1.5 ppb of initial NO concentration remain distinct in these panels, but they are convergent to first order for 4–24 ppb. To second order, there is a peak in volume and mass at 6 ppb (see also Figure and further related discussion there).

Figure 2

New particle formation events occurred after turning on the UV lights in the chamber, as a function of different initial NO. (a) Number concentration of particle population. (b) Surface area concentration of particle population. (c) Volume concentration. (d) Mass concentration. Initial NO concentrations are listed in the inset legend.

Figure 3

Panels (a) through (f) show the plots of the yield of gON, gPN, gAN, pON, pPN and pAN based on the mixing ratio of each measured by TD-LIF divided by the initial α-pinene mixing ratio. Panel (g) shows the plots of the particle mass yield given by the organic PM mass concentration Morg divided by the initial α-pinene mass concentration. Panel (h) shows the plots of both the particle N/C atomic ratio and the particle pON/Morg ratio. Panels (i) through (l) show the plots of particle fractions yields, based on the mass concentration measured by AMS, for CHN+, CHON+, NO+, and of total nitrogen species.

Gas- and Particle-phase Organonitrates as a Function of Initial NO Concentration

Figure shows the experimental results as a function of NO, after the chamber reached steady state at each NO concentration. Data have been collected on average for 2 days after the steady state was reached: the average, standard deviation, and propagation errors have been evaluated (shaded areas in Figures and 5, error bars in Figure ). The yield of gas-phase organonitrates gON steadily increases approximately linearly from 0 up to 0.15 for the progression from <1 to 24 ppb initial NO concentration (Figure , panels a–c). The slope is 0.006 yield ppb–1. Below an initial NO concentration of 6 ppb, the gON species are made up dominantly of gPN, whereas gAN are progressively more important above 6 ppb, reaching 40% ± 10% of the total speciation by 24 ppb. The yield of particle-phase organonitrates pON steadily increases approximately linearly from 0 up to 0.02 for the range of initial NO concentration (panels d–f). The slope is 0.0008 yield ppb–1. In an opposite behavior to that of the gas-phase composition, the pAN species dominate by a factor of 10 over the pPN species across the full range of studied initial NO concentrations.

Figure 5

Comparison between pON as a mixing ratio as measured by the TD-LIF instrument and as a mass concentration as measured by AMS.

Figure 4

Partition fraction of unspeciated (black) and speciated (pink and green) ON to the particle phase as a function of initial NO concentration.

The SOA particle mass yield is plotted in panel g. It is calculated as the organic PM mass concentration Morg at steady state divided by the mass concentration of reacted α-pinene. The plot shows that this yield has a nonlinear dependence on the initial NO concentration. The maximum yield is 0.045 for 6 ppb NO. For NO < 1 ppb, the yield drops by 55% to 0.02. For 24 ppb NO, the yield drops by 25% to 0.035. Similar trends for a maximum yield at intermediate NO concentration are reported in chamber studies for toluene oxidation[35,36] and isoprene oxidation.[37,38] Just as in the atmosphere, in the chamber, there are two different limiting regimes for the chemistry at high and low NOx concentrations.[35−38] Qi et al. (2020) suggested that an increase in SOA yield could arise from higher OH concentrations produced by both the NO–HO cycle and O3 reactions. OH reactions promote the formation of low volatility products, increasing the SOA yield. On the other hand, for increasing the NO concentrations, more RO2 reacts with NO to produce products of greater volatility. Moreover, the OH + NO2 reaction is favored, reducing the concentration of OH. Therefore, the SOA yield decreases.[35] A similar trend of the two regimes has been reported by Liu et al. (2021).[36] In their study, for low NO2 concentrations, the increase in SOA yield was attributed to acid-catalyzed heterogeneous reactions; the production of low-volatility species could explain the decrease in SOA yield at higher NO2. Further studies correlated the nonlinearity in SOA yield with NO concentration to altered RO2 chemistry across the NO regimes.[37,38] For HO2-dominant conditions, RO2 reactions mainly produce organic peroxides and polyols. For greater NO concentrations, the RO2 chemistry becomes more complex, and the reactions with NO or NO2 become more significant.[38] For intermediate NO concentrations, 2-methylglyceric acid (2-MG) is produced, and there is extensive oligomerization; for these conditions, the highest SOA yields occur. For high NO concentrations, 2-MG production is less efficient in favor of the production of more volatile species, corresponding to a decrease in SOA yield.[38] The NO concentration of 6 ppb can be called the cross-over concentration from one chemical regime to another under our experimental conditions.

In conjunction with the changing SOA particle mass yield, further details on the changing nitrogen composition and the speciation of the particle-phase ON are plotted in panels h through l. The ratio pON/Morg (ppb/μg m–3) of panel h represents the relative concentration of ON in the particle. It steadily increases for increasing initial NO concentration, showing the growing importance of ON to particle composition. The important contribution of ON to SOA particle mass concentration has been previously reported.[8,39−42] Based on panels c, d, and e, the dominant species are alkyl rather than peroxy forms of ON. The increase of the nitrogen-containing compounds is also confirmed by particle-phase mass spectrometry. In panels i, j, and k, there is a steady increase in the mass concentrations of the CHN+, CHO1N+, and NO+ families from low to high NO (see also Figure S5). At 24 ppb, only the mass concentration of the NO+ family continues to increase, which is consistent with lower molecular weight species and hence lower SOA particle mass yield for higher NOx concentrations.[36,38] As we did not see any ammonium (NH4+) signal, the nitrogen-containing species, for example, CHN+, CHON+, CHON+, and NO+, should be neither from ammonia nitrates nor from amines.

The partition fraction of ON to the particle phase can be considered. Defined as pON/(pON + gON), the fraction is plotted in Figure as a function of initial NO concentration. The partition fraction decreases from 0.27 for NO < 1 ppb to 0.15 at the cross-over NO concentration of 6 ppb. It then remains approximately unchanged from 6 to 24 ppb. For NO concentrations lower than the cross-over concentration, the produced pON species have a lower volatility that favors partitioning to the particle phase. Conversely, the produced pON species have a higher volatility for greater NO concentrations. These implied trends in pON volatility are consistent with the decrease in SOA particle mass yield for NO concentrations greater than the cross-over concentration (Figure g). More specifically, the volatility of pON species increases with greater NO, as is consistent with lower molecular weights at higher NO reported in earlier studies.[36,38] In summary, the lower volatility of pON when NO < 6 ppb promotes the particle-phase partitioning and an increase in the SOA yield; on the contrary, when NO > 6 ppb, the formed pON shows a higher volatility, which should correspond to lower molecular weights and a decrease in SOA yield.

With respect to speciation, Figure a–fshow that peroxy nitrates dominate in the gas phase, whereas alkyl nitrates make more important contribution to the particle phase, indicating that the partitioning of alkyl nitrates from the gas phase into the particle phase is heavily favored. The partition fraction for pAN/(pAN + gAN) has a similar trend to that of pON, decreasing from about 0.7 to values lower than 0.3 at a cross-over NO concentration of 6 ppb. In contrast, the partition fraction for pPN/(pPN + gPN) is stable at <0.1 across the range of NO concentrations. This behavior confirms that the partitioning in the particle phase of the alkyl nitrates is highly favored compared to the peroxy nitrates, even as the gas-phase availability is significantly higher for the latter (Figure b,c). The pAN species produced for NO concentration <6 ppb have a lower volatility that increases as a function of the NO concentration. This result also suggests that the pAN species could have a shifting chemical composition toward decreasing molecular weights with increasing initial NO concentrations.

The TD-LIF and AMS measurements of ON can be compared. TD-LIF reports in units of ON molecular mixing ratio (ppb) at 1 atm, whereas AMS reports in units of organonitrates mass concentration (μg m–3). For the organonitrates concentrations derived from the AMS measurements, fragments of the CHN+, CHO1N+, CHON+, and NO+ families were used. Figure shows the plot of each as a function of increasing initial NO concentration. The overlay shows a similar monotonic relative increase. In a 1:1 plot (not shown), the Pearson coefficient of correlation R is 0.9. There are important differences, however, between the two plots. The two quantities are related by the ideal gas law and an average molecular weight, and for <1 ppb NO, the implied molecular weight is 130 g mol–1 to match the TD-LIF and AMS measurements. For 24 ppb NO, the implied molecular weight is 40 g mol–1. The molecular weight thus appears to decrease with increasing NO concentration. Figure S5 shows that, as NO increases, the molecules containing more oxygens decrease in favor of NO+. This result suggests that there is production of less-oxidized particle-phase material as a function of NO. Moreover, as shown in Figure h, the N/C ratio increases even as the particle number concentration remains almost stable from 12 to 24 ppb of NO (not shown). This behavior could arise because (1) the same number of nitrogen-bearing molecules are produced but of higher N/C stoichiometry or (2) more nitrogen-bearing molecules are produced but of the same N/C stoichiometry. This topic needs to be further investigated. These implied molecular weights can be somewhat explained by the recognition that AMS typically underestimates the actual ON mass concentration,[43,44] which would have the effect of underestimation of about 21%, up to 42% when ON dominates the organic aerosols.[44] Moreover, TDLIF has been found to overestimate by a factor 5 the AMS and CIMS measurements of organonitrates.[6] This aspect is still under investigation.

Atmospheric Implications

Chamber experiments were carried out to understand and quantify the partitioning of ON in the production of SOA from α-pinene photo-oxidation. The studied conditions ranged from simulated preindustrial to polluted environments with respect to initial NO concentrations. Organonitrates were produced in both the gas and particle phases, with maximum yields of 0.15 and 0.02 at the highest studied NO concentration of 24 ppb. Between the two contrasting preindustrial and polluted environments, the slopes of the yields of gas- and particle-phase ON were 0.006 and 0.0008 of yield per ppb, respectively, with respect to the initial NO concentration. The gas-to-particle partition fraction of ON decreased from 0.27 to 0.13, passing from the pre-industrial to the polluted environment. For this latter environment, the partition fraction did not change further. Thus, the implication is that species of different vapor pressures were produced in the two different low- and high-NO regimes, reflecting different dominant pathways in the chemical mechanisms. Peroxy nitrates dominated in the gas phase, whereas alkyl nitrates had a more important contribution to the particle phase, indicating that the partitioning of alkyl nitrates from the gas phase into the particle phase was heavily favored. The SOA particle mass yield also had a local maximum at the cross-over NO concentration between the two chemical regimes. The results imply that for environments in which biogenic emissions and low NO concentrations hold, the production and partitioning of ON to the particle phase are favored in a relative sense based on gON availability compared to that at higher NO concentrations. The combined set of results is consistent with the production of lower molecular weights ON in the high-NO regime. The higher NO concentrations also affect the vapor pressures and the chemical composition of pON, impacting the SOA particle composition and production. Future experiments are necessary to measure the α-pinene oxidation products to better describe the chemical regimes behind our findings.