the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 06 May 2026
Phase-State and Humidity Trend Controls on Wintertime Nitrate Aerosol Formation
Abstract. Particulate nitrate is a dominant component of winter haze in East Asian megacities, yet its real-world formation mechanisms remain incompletely understood. We integrate high time-resolution aerosol composition measurements, explainable machine learning (ML), and conventional analyses to disentangle key drivers of wintertime nitrate production. While NO₂ availability is the primary control, our results reveal critical but underrepresented processes: (1) persistence of nitrate formation during late-morning relative humidity (RH) decline, sustained by metastable semi-liquid particles with retained liquid water that facilitate continuous gas-to-particle partitioning of photochemically produced HNO₃, and (2) temperature threshold effects, where subfreezing conditions suppress further nitrate formation primarily due to thermodynamic precursor saturation, compounded by potential diffusion limitations in highly viscous or solid phases. Contrary to common assumptions, boundary layer height contributes minimally to peak nitrate events. These findings demonstrate the need for air quality models to incorporate RH trends, aerosol phase transitions, and temperature-dependent reactivity to accurately predict nitrate episodes. The mechanistic framework presented here is transferable to other urban environments affected by secondary inorganic aerosols and offers new leverage points for mitigation strategies.
- Preprint
(1529 KB) - Metadata XML
-
Supplement
(1956 KB) - BibTeX
- EndNote
Status: open (until 17 Jun 2026)
- RC1: 'Comment on egusphere-2026-1510', Anonymous Referee #2, 12 May 2026 reply
-
RC2: 'Comment on egusphere-2026-1510', Anonymous Referee #1, 21 May 2026
reply
This work investigated the driving factors of nitrate in PM1 using ML approach. In fact, numerous of studies have revealed the nitrate formation mechanism using thermodynamic analysis, box model analysis, as well as ML. This work is not novel enough for understanding the nitrate formation in East Asia, and the measurement dataset is limited extent. I recommend to reconsider this manuscript if the following comments are addressed.
- From the perspective of nitrate formation mechanisms, nitrate formation is not solely related to NOx. In fact, nitrate formation results from the combined action of multiple reactions. Each of these reactions is influenced by numerous factors, which is why the factors affecting nitrate formation are highly complex. The parameters considered in this manscirpt are far from sufficient to assess the primary drivers of nitrate formation. For example, simulation results based on the box model suggest that VOCs are also important precursors for nitrate formation, but they were not considered in this study.
- Why were these specific parameters included in the ML model? Could the author provide the training and testing results for the machine learning model, as well as the results of the model’s transferability validation? How were the results for RH increase and decrease in Figure 3 derived? Were they obtained through a sensitivity analysis of the machine learning model?
- The author uses TNO₃ = pNO3+ HNO3to describe total nitrate. This relationship is not accurate. This is because nitrate undergoes gas-particle partitioning, and this process is dynamic, potentially influenced by a variety of factors such as RH, temperature, and the chemical composition of particulate matter. Furthermore, these factors also affect the pH of the ALWC and particulate matter. The paper does not discuss changes in pH or how they affect TNO3. In addition, nitrate formation will also enhance ALWC, while the authors did not consider this effect.
- Line 169: How do the authors calculate NOR?
- Line 173-174: This is a general conclusion, which have been widely proved by other studies.
- Line 179-184: Conclusions here seems to be based on speculation. Throughout the text, there are too many such speculative conclusions; the author needs to provide more evidence.
- Line 265-267: The author points out here that changes in RH are sufficient to cause nitrate deliquescence. Is this narrow change in RH sufficient to cause nitrate deliquescence? Furthermore, the author assumes here that sulfates and nitrates are externally mixed, which is not actually the case. Internally mixed particles account for a significant proportion. For internally mixed particles, their deliquescence behavior differs from that of pure sulfates or nitrates, and they do not have a precise deliquescence point.
Citation: https://doi.org/10.5194/egusphere-2026-1510-RC2 -
RC3: 'Comment on egusphere-2026-1510', Anonymous Referee #3, 26 May 2026
reply
The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-1510/egusphere-2026-1510-RC3-supplement.pdf
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 101 | 18 | 7 | 126 | 10 | 4 | 4 |
- HTML: 101
- PDF: 18
- XML: 7
- Total: 126
- Supplement: 10
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Many Asian regions have high concentrations of fine particle nitrate in cool seasons. There is significant interest in understanding the sources and processes that affect these aerosols to provide insights on how to lower PM2.5. The goal of this paper is to “investigate the atmospheric drivers of particulate nitrate formation in wintertime Seoul”. The approach focusses on measurements between January 17 and February 16, 2018, in Seoul, Korea, a period that includes episodes of high PM1 concentration (>40 ug/m3). ML was used to interpret the measurement data. Aerosol liquid water content (ALWC) was predicted by ISORROPIA in reverse mode.
Measurement data included:
AMS PM1 data, including nitrate (pNO3).
BC was reported but did not see info on the measurement method.
Gas phase O3 and NO2
Various meteorological data (T, RH, wind speed/direction, solar strength).
There are major limitations in assessing drivers of PM1 nitrate given this data set due to it’s limited extent. This is especially true for attempting to link sources and processes (NOx emissions and subsequent chemistry) to PM1 nitrate concentration since there are many processes involved and species in addition to nitrate that can also be formed, such as various organo-nitrates (ie, gas phase PAN and related products, and particle organonitrates), and nitric acid. The ML analysis may identify from the limited set of variables certain links to nitrate, such as RH, T, etc, but the explanation provided in this manuscript for the cause for the link are not novel and not justified in some cases. Finally, the conclusion that various parameters, such as RH trends, phase state transitions, and thermodynamic constraints, need to be considered is nothing new. Most researchers addressing the question of what explains PM1 nitrate would run a full thermodynamic analysis that incorporates these various aspects parameters, that means having measurements of key gas phase species, such as ammonia and nitric acie. Finally, some of the discussion on water uptake (efflorescence/deliquescence) and interpretation of AMS data to infer wet aerosol morphology is also questionable. I do not recommend publishing this paper.
General Comment
A fundamental problem with the interpretation of this data is attempting to link nitrogen emissions (eg, NOx) to PM1 nitrate. The authors tend to directly link NOx and other processes to particulate nitrate. But the process is more complicated. NOx can react to produce many species, such as nitric acid (HNO3), PAN, etc) so although a correlation between NOx and PM1 nitrate may exist, it would be very difficult to quantify the linkage through a mass balance. Nitric acid itself can partition between the gas and particle phases, and in this case only total nitrate (TNO3=pNO3 + HNO3) is conserved. NO2 conversion to PAN and back again to NO2 depends on T; this complexity also cannot be considered in the analysis. Thus the authors are simply left with a correlation analysis, ie a certain percent of the PM1 nitrate variability is explained by NOx variability.
Considering just pNO3 and HNO3, since this is the focus of the paper, these species can change concentrations back and forth (ie, partition) due to changes in ambient conditions, such as T, RH, other aerosol species concentrations, all of which affect ALWC and pH. ALWC can be estimated from measured aerosol composition, which is what has been done in this paper. But factors that affect pH are not considered, and pH strongly affects HNO3/pNO3 partitioning (this is discussed at times in the paper). A thermodynamic equilibrium model can predict this partitioning and pH and is really required for a more in-depth analysis. Because partitioning back and forth between the gas and particle phase can be fast (10s of minutes), just discussing changes in observed pNO3 severely limits the interpretation since there is a missing component, gas phase HNO3.
Here is a specific example for this. Line 211-213 states: “The positive correlation between relative humidity (RH) and particulate nitrate is well established, primarily due to the enhancement of heterogeneous and aqueous-phase pathways such as N2O5 hydrolysis under high-RH conditions (Liu et al., 2020)”. (Also, see discussion that go to the end of the paragraph, line 229.). This statement is highly qualified and not informative. For example, instead of concluding that this observation is linked to N2O5 hydrolysis, much of the RH affect could be due to a change in HNO3/pNO3 partitioning, which is a complex function of RH, T, and particle composition, ie pH, none of which are discussed. There is no way to know if N2O5 hydrolysis had any effect at all, so the statement is speculation.
Other comments.
Define NOR, NOx oxidation ratio.
Line 186 to 187 states; “NO2, the primary precursor of nitrate, was identified as the most significant contributor, accounting for 31.6% nitrate formation during winter (Fig. 2a).” What was the other source of the pNO3?
Line 238-240 states: “Under strict equilibrium thermodynamics, the concurrent drop in daytime RH would induce immediate particle efflorescence (crystallization), expelling liquid water and shutting down the aqueous volume required for this freshly formed HNO3 to partition into the aerosol phase.” What is the expected RH for efflorescence and does the ambient measured RH every reach these values? There are publications on water uptake of ambient aerosols, which may provide some insights. How does particle mixing state effect this. In summary, is the actual ambient particle efflorescence RH known – or is this all speculation.
Lines 260 to 268. Typically, the AMS is operated with a dryer at the AMS inlet – so it measures properties of dried particles. Was that the case for this study? If so, how can AMS data then be used to infer the behavior (eg, shape) of a wet ambient particle?
Line 267 and on. Sulfate’s behavior is in part different from pNO3 because it is nonvolatile, it does not shift between particle and gas phases. There is confusion on particle water uptake and loss (hysteresis behavior), which is demonstrate by the following line (line 265-267) that states: The average RH during the campaign was 47.7 ± 16.3%, and 48.9 ± 14.5% during 7–10 AM—sufficient for nitrate deliquescence but typically below the threshold for sulfate.” First this implies sulfate and nitrate are externally mixed, is this reasonable? What is the effect of internal mixing of the salts and also OA aerosol on deliquescence? Second, avg RH is not what matters, peak RH matters, since once above the deliquescence point the particle will remain wet as long as the RH does not drop below the efflorescence point. Since RH time series are not given, this is hard to interpret. (There is average data in Fig S4, which seems to show the RH does not go below the efflorescence RH of ammonium sulfate or ammonium nitrate). Furthermore, internally mixed particles do not strictly follow the sharp changes in water uptake/loss of pure salts.The following line is not supported by the data and is unclear: Line 289-290 states: “Together, these results provide, for the first time, field-based evidence that RH impacts on nitrate formation are dynamic and modulated by aerosol phase transitions, not merely RH magnitude.”
The authors could use thermodynamic parameters, like Henry’s law constants, or sensitivity analysis with a thermodynamic model, to support the following statement (which is likely not true): Lines 309-311: “At subfreezing temperatures, thermodynamic equilibrium shifts so heavily toward the particle phase that nearly all available gas-phase precursors (HNO3 and NH3) have likely already partitioned; once the gas-phase reservoir is depleted, further temperature decreases cannot yield more particulate nitrate.”
Line 312 to 314 states, .. “at subfreezing temperatures, aerosols can transition into highly viscous semi-solid or glassy states, which restrict bulk diffusion and suppress heterogeneous reactivity (Meng et al., 2024; Shiraiwa et al., 2011).” At what T does this happen, and is it likely for the T’s and RH of this study?
The following line appears to be speculation, (what data supports it): Line 318… “Therefore, while low temperatures generally enhance nitrate formation through thermodynamic favorability, extremely low temperatures introduce phase-related limitations that partially counteract this effect. These findings clarify why a positive temperature–nitrate relationship is observed overall, despite the apparent suppression of reactivity under very cold conditions.” I would suggest the situation is much more complicated than that. Example, have you considered temperature effects on activity coefficients – just to name one.