Decreasing Production and Potential Urban Explosion of Nighttime Nitrate Radicals amid Emission Reduction Efforts

Nighttime oxidation by nitrate (NO3) radicals has important ramifications on nocturnal aerosol formation and hence the climate and human health. Nitrate radicals are produced by the reaction of NO2 and O3. Despite large decreases in anthropogenic emissions of nitrogen oxides (NOx = NO + NO2), a previous study found significant increases in NO3 production (PNO3) from 2014 to 2019 in China, in contrast to decreasing trends in the U.S. and Europe. Using the summer observations from 2014 to 2022, we analyze the interannual variability of nocturnal PNO3 using a systematic framework, in which PNO3 is diagnosed as a function of odd oxygen (Ox = O3 + NO2) and the NO2/O3 ratio. We did not find an increase of PNO3 from 2014 to 2022 in China due to a continuous decrease in the NO2/O3 ratio, although PNO3 is modulated by the variation in Ox. Using in situ observations obtained in Beijing in 2007, we demonstrate the potential for an upsurge resembling an “explosion” in urban nighttime NO3 radicals amid emission reduction efforts.


INTRODUCTION
−4 While in situ measurements of NO 3 have been made for decades, 5 long-term records of nighttime NO 3 are not available due in part to the short lifetime and high variability of its concentrations, 6 making it difficult to assess the changes of nocturnal oxidation in the past due to anthropogenic emission changes in response to environmental regulations.However, the production rate of NO 3 , PNO 3 , near the surface can be diagnosed using observations of NO 2 and O 3 by air quality monitoring networks around the world since the reaction R1 is the main source of NO 3 in a polluted boundary layer.Given the short lifetime of NO 3 , PNO 3 can often be used to assess the nighttime NO 3 radical oxidation potential in summer. 1 However, the NO 3 concentration and its oxidation rate of volatile organic compounds (VOCs) also depend on two other losses.The reaction of NO 2 and NO 3 produces N 2 O 5 , which thermolyzes back into NO 2 and NO 3 with a lifetime of ∼1 min at 20 °C. 7Heterogenous reactions of N 2 O 5 that produce HNO 3 and ClNO 2 on aerosols remove NO 3 radicals from the atmosphere. 2 In addition, NO 3 is also removed by reaction with NO.We examine the potential impact of urban nighttime NO on NO 3 concentrations using in situ observations.

MATERIALS AND METHODS
Similar to the previous study, 1 we use the hourly observations of surface NO 2 and O 3 between 8 pm and 5 a.m.LT (at altitudes typically a few meters above the surface) by the China National Environmental Monitoring Center (CNEMC) network.To ensure data consistency, we only analyze summertime (June, July, and August) data from 2014 to 2022.Limiting the analysis to summer data ensures a more consistent physical and chemical environment.The ending time of nighttime analysis is changed from 6 am 1 to 5 am for the summer.We use only the data from stations that were continuously operational from 2014 to 2022.The observations from 640 surface stations are used in the analysis.Furthermore, we filter out the outlier data outside the range of , where Q 1 and Q 3 are the 25th and 75th quartiles, respectively, and k = 1.5. 82.65% of the measurement data are removed, although this filtering does not alter the analysis results.Lastly, to further ensure data consistency, we group the observations in China into 4 regions based on climate characteristics and topography (Figure 1).There are only 6 stations over the Tibetan Plateau region, and hence only data in the Northeast (NE), Southeast (SE), and Northwest (NW) regions are analyzed.
Chemical reaction rate constants depend on the atmospheric temperature and pressure.We use the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis version 5 (ERA5) hourly surface temperature and pressure at a resolution of 0.25°× 0.25°.Reaction rate constants are calculated using the kinetics data from the latest Jet Propulsion Laboratory (JPL) compilation. 7o add a point of reference for the pre-2014 period, we analyzed the observations from the Campaigns of Air quality Research in Beijing (CAREBeijing) in August 2007.Concentrations of O 3 , NO, and NO 2 were measured at an urban site located on a building rooftop (∼20 m above the ground level) on the campus of Peking University. 9,10The 1 minute observations were averaged hourly to be comparable to the CNEMC measurements.

Systematic Analysis Framework.
In polluted regions at night, O 3 is consumed to oxidize freshly emitted NO: We define nighttime odd oxygen as O x = O 3 + NO 2 . 11Figure 2 shows the summertime hourly averaged regional O x , O 3 , and NO 2 concentrations from 2014 to 2022.O x concentrations decreased at night largely following the decreases of O 3 since the variation of NO 2 was relatively small compared to those of O x and O 3 .The nighttime O 3 and NO 2 variation patterns in China are similar to those observed in the United States. 12e write the PNO 3 rate as a function of O x ,   The modulation effect by O x is particularly large from 2020 to 2022 in the SE as the increase of O x in these 3 years is faster than during any 3-year periods of 2014−2019 (Figure S2).However, the resulting increase in PNO 3 is slight (Figure S2) due to the decrease of the NO 2 /O 3 ratio from 2020 to 2022 (Figure 3).
Figure 4 shows that hourly average PNO 3 rates for the three regions decreased from 2014 to 2022 at night.The variation of PNO 3 at night is larger than in daytime in summer. 13The intranight decrease of PNO 3 by a factor of ∼2 was larger than those of O x .It was driven primarily by the decay of O x due to the quadratic dependence of PNO 3 on the O x concentrations (eq 1).Quantifying nighttime O x decreases is therefore necessary.
Figure 2 shows the nearly exponential decay of the O x concentrations at night in all three regions.The decay shapes were fairly consistent among different years, i.e., the seasonal mean difference at 8 pm LT was largely maintained until 5 am LT.To quantify the nighttime decay lifetime, we conducted least-squares regressions using x = x 8pm e −t/τ , where x is O x hourly concentration, t is the time lapse from 8 pm, and τ is the decay lifetime.The regressions have R 2 values in the range of 0.97−0.99.The lifetime of O x is ∼20 h in the NE and SE regions, but is higher at ∼30 h in the NW region (Figure 4), likely reflecting higher canopy dry deposition of O 3 and NO 2 in eastern China than the NW.The vegetation coverage of eastern China is much higher than the NW, 14,15 resulting in larger deposition of O 3 and NO 2 . 16,17Monthly mean midnight dry deposition velocities for O 3 and NO 2 , simulated using a regional chemical transport model, 12,18,19 are much higher in the NE and SE than the NW region (Figure S3).Model description and further discussion on the loss of O x due to dry deposition can be found in Supporting Information.
We apply least-squares regression analysis to compute interannual trends of O x lifetimes.The SE region is the only region with a statistically significant small decreasing trend,  −1.3% year −1 (p = 0.03) from 2014 to 2022.The other two regions also have negative trends, −0.28% year −1 (p = 0.63) for NE and −0.20% year −1 (p = 0.80) for NW.Moreover, taking into account significant seasonal variances that were larger than the trends, it is evident that nighttime O x lifetimes remained largely unchanged despite substantial reductions in emissions across the three regions from 2014 to 2022, leading to fairly consistent nighttime decay patterns of O x and PNO 3 among different years (Figures 2 and 3).
The finding of stable nighttime lifetimes of the O x from 2014 to 2022 is somewhat surprising considering the significant interannual variation of PNO 3 (Figure 3).Since the lifetime of NO 3 is short, PNO 3 equals the loss rate of NO 3 , which includes (1) heterogeneous reactions of N 2 O 5 on aerosols to produce HNO 3 and ClNO 2 , (2) reactions of NO 3 and VOCs, and (3) the reaction of NO 3 and NO.In the first reaction pathway, up to three O 3 molecules are lost for every N 2 O 5 molecule that is lost.In the second reaction pathway, up to two molecules of O 3 are lost for every NO 3 loss.The third reaction pathway did not affect O x .Assuming the regional effect of the third reaction pathway was insignificant, an increase of PNO 3 from 2014 to 2017−2018 (Figure 3) would imply a decrease of O x lifetime, and the decrease in PNO 3 from 2017 to 2018 to 2022 would have the opposite effect.However, the derived O x lifetimes in the NE and NW regions had a peak in 2018 when PNO 3 values were among the highest.The lack of correspondence between the O x lifetime and PNO 3 reflects in part the dominance of dry deposition of O 3 and NO 2 in determining the lifetime of the O x at night.
Another possible contributor is that the rate of N 2 O 5 heterogeneous reactions on aerosols decreased relative to the rate of the NO 3 oxidation of VOCs.The loss of O x through the latter is less than that through the former.The decrease in NO 2 concentrations (by 31−37%) from 2014 to 2022 (Figure 2) reduced the formation rate of N 2 O 5 considering that regional averaged PNO 3 rates decreased.Furthermore, the heterogeneous loss rate of N 2 O 5 depends on aerosol surface area.There was no regulatory monitoring of aerosol surface area.However, its qualitative trend can be estimated using PM2.5 concentrations by assuming that aerosol size distribution did not change significantly.Figure S5 shows that PM2.5 concentrations decreased by nearly a factor of 2 from 2014 to 2022 in the three regions.In addition, aerosol liquid water content also tends to decrease with PM2.5, 20 contributing to a decrease of aerosol surface area.The overall effect was a decrease of the aerosol N 2 O 5 reaction rate relative to the rate of NO 3 oxidation of VOCs, which tended to decrease O x loss and increase O x lifetime at night.
Temperature at night decreases significantly and reduces the PNO 3 rates.Figure S6 shows summertime hourly average temperatures from 2014 to 2022.Similar to that of O x , the interannual difference in average temperature is much less than the decrease in temperature from 8 pm to 5 am LT and the patterns of hourly temperature decrease are consistent among different years in the three regions.The resulting effect of interannual temperature variation on PNO 3 rates is small from 2014 to 2022, although lower temperature in the NW resulted in lower reaction rate constants for PNO 3 compared to those in the SE and NE regions (Figure S7).

Potential Explosion of Urban NO 3 Radicals.
Higher PNO 3 rates in the NE and NW than the SE region are due largely to regional differences in O x concentrations (Figure 3).The decrease in PNO 3 rates from 2014 to 2022, on the other hand, is due mostly to the decreasing NO 2 /O 3 ratio, reflecting much larger and more consistent NO 2 reductions compared to the variations of O x and O 3 in the three regions (Figure 2).We did not find evidence that nighttime variations of chemical species or temperature, which were much larger than their interannual variations, changed significantly from 2014 to 2022.The lifetimes of nighttime O x in the three regions did not show significant changes either (Figure 4).
While we find increases in PNO 3 rates from 2014 to 2017− 2018 in all three regions as Wang et al. 1 due to increased O x concentrations, the increases appeared to be transitory in nature due to persistent and rapid decreases in the NO 2 /O 3 ratio as NO x emissions were reduced.Some of the post-2020 NO x emissions decreases are COVID-related. 21Nonetheless, it is possible that the rates of nighttime PNO 3 increased prior to 2014.Sun   23 It is instructive to compare the 2014−2022 results for the NE region to the CAREBEIJING-2007 observations.The average PNO 3 rate at night during the CAREBEIJING-2007 campaign was lower than the summertime average in the NE region (Figure 3) mostly due to a lower average O x concentration.The geometric mean NO 2 /O 3 ratio during the campaign was 1.3.Considering that the measurement altitude of 20 m is above air quality monitoring stations, it is possible that this measurement may underestimate near-surface NO/ NO 2 ratios due to the conversion of NO to NO 2 during mixing from the surface to the measurement altitude.If this value was representative of the summer of 2008, the decrease in the NO 2 /O 3 ratio from 2008 to 2014−2002 would slightly increase PNO 3 until the NO 2 /O 3 ratio reached 1.After this point, the decrease in the NO 2 /O 3 ratio would start decreasing PNO 3 .However, this effect would not be as large as the increase in PNO 3 due to increasing O x (eq 1 and Figure 3).
Both historical observations of O 3 22,23 and CAREBEIJING-2007 measurements indicated that O x concentrations increased prior to 2013 as anthropogenic emissions of NO x and VOCs increased.The quadratic dependence of nighttime PNO 3 on O x implies that PNO 3 increased during this period.However, it does not imply that nighttime NO 3 concentrations increased, particularly in urban regions.CAREBEIJING-2007 1 minute measurements had 57% of nighttime NO concentrations >1 ppbv.At this level, the steady-state NO 3 radical concentration is negligible due to the rapid removal of NO 3 by NO, 7 NO concentrations were not reported by the CNEMC network.It is therefore difficult to quantify the effect of R3 in suppressing nighttime NO 3 radical concentrations in urban areas with high NO x emissions.Figure S8 shows that hourly average O 3 concentrations during the CAREBEIJING-2007 campaign stayed at ∼10 ppbv after midnight and were lower than hourly average NO 2 concentrations.The observed high NO 2 /O 3 ratios during CAREBEIJING are due to high concentrations of NO, which reacts with O 3 to produce Environmental Science & Technology NO 2 .We therefore use an indirect measure, the data fraction for NO 2 /O 3 > 1, to qualitatively examine this effect.Figure S9 shows that the data fractions for NO 2 /O 3 > 1 decreased from 32−39% in 2014 to 17−20% in 2022 in the three regions.In comparison, the corresponding fraction in CAREBEIJING-2007 data was 47%.
Figure S10 shows simulated hourly steady-state NO 3 radial concentrations as a function of NO concentration using hourly averaged values of the O x , O 3 , and NO 2 for the NE region from 2014 to 2002.The critical NO concentrations that suppress NO 3 radicals were at the level of 0.1−1 ppbv.Similar results were obtained for the SE and NW regions.As NO x emissions continuously decrease, it is plausible that in the future nighttime NO concentrations in certain urban regions may frequently decrease to levels below 0.1 ppbv.Under such circumstances, NO 3 concentrations will increase by orders of magnitude (Figure S7).
In this study, we analyzed surface observations from 640 stations in China to characterize regional near-surface PNO 3 changes from 2014 to 2022.It is worth noting that the differential optical absorption spectroscopy measurements by Yan et al. in Beijing showed increasing NO 3 radical concentrations with altitude due to in part to increasing NO 3 lifetime with altitude. 24Although not subject to regulatory monitoring, understanding the vertical distributions of O 3 , NO 2 , and other atmospheric species, along with their variations, is essential in gaining insights into nighttime oxidation processes.
3.4.Discussion.Satellite observations showed that NO x emissions started decreasing by 2011 in southern China. 25The year 2013 marked a significant turning point in emissions reduction policies in China, leading to substantial decreases in NO x emissions. 26Before 2013, the increased emissions led to O x increases in the summer, 22,23 likely leading to a PNO 3 increase since it is a quadratic function of O x .The sensitivity of PNO 3 to the NO 2 /O 3 ratio is significantly lower than the quadratic dependence on the O x (eq 1 and Figure 3).However, increased NO x emissions also led to higher NO concentrations in urban regions near emission sources.The lifetime of the NO could be fairly long when the O 3 was titrated through R1 by continuous emissions.Therefore, higher PNO 3 concentrations during this period did not necessarily imply higher NO 3 radical concentrations and nighttime oxidation in urban areas with high NO emissions (Figure S10).
After 2014, observations showed significant and consistent reductions of NO 2 and PM2.5 in China (Figures 2 and S1).Nighttime O x concentrations had transient increases peaking around 2017 and 2018 and then returned to and even went lower than the 2014 levels after 2020.The continuous decrease in the ratio of NO 2 /O 3 , on the other hand, contributed to a decrease in PNO 3 .This continuous decrease in PNO 3 will be modulated by O x variations due to meteorological conditions and O 3 photochemistry. 9,22If the modulation effect by O x in the future is neutral or negative as decreasing anthropogenic emissions cause O x to stay near the same level or start decreasing, a PNO 3 decreasing trend is expected.
The reduction in the NO 3 radical concentration following the decrease in PNO 3 is not always directly proportional.In regions with minimal impact from fresh NO emissions, the decrease in the level of NO 2 resulting from emission reduction (Figure 2) generally led to a decline in the formation rate of N 2 O 5 .Additionally, the decrease in PM2.5 (Figure S1) and, consequently, the reduction in aerosol surface area tended to slow down the loss of N 2 O 5 to HNO 3 and ClNO 3 on aerosols.Both factors would contribute to decreasing the loss of NO 3 by heterogeneous N 2 O 5 reactions on aerosols, resulting in an increase of NO 3 loss by its oxidation of VOCs for a given PNO 3 rate.The net effect would mitigate the potential decrease of the NO 3 oxidation of VOCs due to decreased PNO 3 .The opposite mitigation effect would likely occur during the phase of increasing anthropogenic emissions.
In urban regions significantly affected by fresh NO emissions, on the other hand, much more drastic nighttime NO 3 changes could be expected.If nighttime O 3 titration by NO emissions and a long NO lifetime occurred over a sufficiently long time and large area of an urban region during the phase of increasing emissions, an "explosion" of nighttime NO 3 radicals near the surface would likely occur during the period of emission reduction after 2013 when the NO lifetime shortened at night.As NO emissions continue to decrease, NO suppression of NO 3 radicals would continue to weaken.Significant nighttime nitrate radical concentrations are necessary conditions for the formation of HNO 3 from N 2 O 5 hydrolysis 2 and gaseous and particulate organic nitrates. 4herefore, in an urban area with high NO x emissions, an upsurge could exist from minimal nighttime inorganic and organic nitrate production to significant nighttime nitrate production as a result of efforts to reduce emissions.The analysis in this work targeted summer, but the probability for this mechanism to occur is higher in other seasons, particularly in winter, since the O 3 concentrations tend to be highest in summer.This mechanism could help explain some of the observed lower nitrate reductions compared to other secondary aerosol components in Beijing from 2011/2012 to 2017/2018.

1
study, Wang et al. analyzed the trend of nocturnal PNO 3 in the warm season of China from 2014 to 2019 in comparison to those in the United States and Europe. 1 They found a significant increasing trend of nocturnal PNO 3 in China in contrast to the decreasing trends in the United States and Europe and further suggested that nocturnal oxidation is becoming more important in China.In this study, we analyze the variation of nighttime PNO 3 in China from 2014 to 2022 using a systematic framework to separate the effects of odd oxygen (O x = O 3 + NO 2 ) and the ratio of NO 2 /O 3 .

Figure 1 .
Figure 1.CNEMC sites (black dots) with continuous measurements from 2014 to 2022 in three analysis regions.

Figure 2 .
Figure 2. Summertime hourly averaged regional O x , O 3 , and NO 2 concentrations at 8 pm to 5 am LT from 2014 to 2022 in the NE, SE, and NW regions.The hourly data points are plotted in the middle of the hour.The corresponding standard deviation distributions among the observation sites are shown in Figure S1.

Figure 3 .
Figure 3. Regional averaged summer PNO 3 as functions of the geometric mean of NO 2 /O 3 and average O x from 2014 to 2022 in the NE (squares), SE (circles), and NW (triangles) regions.The background color image was computed using eq 1 at a temperature of 20 °C.PNO 3 was computed using hourly CNEMC measurements and ERA5 temperature.Therefore, regional nighttime average PNO 3 values do not necessarily match those in the background color image.Averaged PNO 3 for the CAREBEIJING-2007 campaign is denoted by a diamond.The geometric mean of NO 2 /O 3 is used since it is in log scale.

Figure 4 .
Figure 4. Same as Figure 2 but for hourly average PNO 3 rate and the corresponding O x lifetime.The corresponding standard deviations of PNO 3 among the observation sites are shown in Figure S4.
et al. showed an increase of about 9.6 ppbv of maximum daily 8 h average O 3 concentrations in July and August from 2003 to 2015. 22They attributed about half of the O 3 increases to increased emissions of the NO x and VOCs.Although the interannual variability of O 3 between 2003 and 2015 was not studied, it is likely that nighttime PNO 3 increased from 2003 to 2015 given the quadratic dependence of PNO 3 on O x .Furthermore, Ding et al. showed increasing lower-tropospheric O 3 concentrations in Beijing and the North China Plain region from 1995 to 2005.

27 ■ ASSOCIATED CONTENT * sı Supporting Information The
Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.est.3c09259.Model computed nocturnal dry deposition velocity distributions for O 3 and NO 2 (Text S1); standard deviations of O x , O 3 , and NO 2 , yearly summertime averages of PNO 3 and O x , model computed monthly mean dry deposition velocities for O 3 and NO 2 , standard deviations of PNO 3 , summertime hourly average PM2.5 concentrations, summertime hourly average ERA5 temperature, summertime hourly average reaction rate constant of R1, hourly average concentrations of O x , O 3 , and NO 2 during the CAREBEIJING-2007 campaign, data fractions of nighttime hourly NO 2 /O 3 ratio >1 from 2014 to 2022, and model calculated hourly steady-state NO 3 radical mixing ratio as a function of NO (Figures S1−S10).(PDF) Yuhang Wang − School of Earth and Atmospheric Sciences, Georgia Institute of Technology, Atlanta, Georgia 30332, United States; orcid.org/0000-0002-7290-2551;Email: yuhang.wang@eas.gatech.eduEnvironmental Science & Technology