Explicit simulation of chemical composition, size distribution and cloud condensation nuclei of secondary organic aerosol from α-pinene ozonolysis

Zhen Song, Chengqi Zhang, Hongru Shen, Hao Ma, Iida Pullinen, Defeng Zhao2025 · DOI

Abstract

Secondary organic aerosols (SOA) contribute significantly to cloud condensation nuclei (CCN), which depend on size distribution, chemical composition and hygroscopicity parameter ($\kappa$). However, how well current understanding of SOA formation can reproduce CCN concentrations and the influence of these factors on modelled CCN uncertainties are still unclear. In chemical transport models, it is difficult to address the issue due to model complexity and oversimplified representation of chemical mechanisms, particle size and $\kappa$. Here, we explicitly simulated CCN concentrations of SOA from $\alpha$-pinene ozonolysis, a benchmark system for SOA studies using a box model (PyCHAM). Using state-of-the-art treatment of chemical mechanisms, aerosol size and $\kappa$, we assessed how CCN as well as chemical composition, aerosol size and $\kappa$ can be modelled against measurement and evaluated the influence of these factors on CCN simulation. The model well simulated SOA mass concentration but overestimated O:C and H:C ratios, suggesting a lack of particle-phase chemistry. Highly oxygenated molecules contributed substantially to SOA mass and thus CCN. Modeled $\kappa$ closely aligned with measurements at moderate supersaturation (0.37 %) but overestimated $\kappa$ (by 19 %) at low supersaturation (~0.19 %) and underestimated $\kappa$ (by 21 %) at high supersaturation (0.73 %). The model well reproduced particle growth, but exhibited a wider and flatter size distribution compared with measurement. The simulated CCN concentrations agreed well with measurement at moderate to high SS (0.37–0.73 %) but had a significant bias at low SS. Sensitivity analysis highlights the importance of accurate representation of both size distribution and $\kappa$ for CCN prediction, especially at lower SS (<0.4 %).

Introduction

Secondary organic aerosols (SOA) are critical components of the atmosphere, acting as the "seeds" or Cloud Condensation Nuclei (CCN) around which water vapor condenses to form clouds. Understanding how volatile organic compounds (VOCs) like $\alpha$-pinene—the characteristic scent of pine forests—transform into these particles is essential for accurate climate modeling. However, current global chemical transport models (CTMs) often rely on simplified parameterizations of chemical mechanisms, particle size distributions, and hygroscopicity (the $\kappa$ parameter), which can lead to significant uncertainties in climate predictions.

This study utilizes an explicit box model, PyCHAM, to simulate the formation and CCN activity of SOA from $\alpha$-pinene ozonolysis. By avoiding the "shortcuts" typical of larger models and using high-fidelity chemical and microphysical descriptions, the researchers aim to identify specific gaps in our current theoretical understanding when compared against controlled chamber experiments.

Methods and Data

The research team employed the PyCHAM (CHemistry with Aerosol Microphysics in Python) box model. To ensure a high degree of chemical realism, they coupled the Master Chemical Mechanism (MCM) with the Peroxy Radical Autoxidation Mechanism (PRAM). This combination allows for the explicit tracking of thousands of species, including Highly Oxygenated Molecules (HOMs), which are vital precursors for aerosol growth.

Key microphysical processes such as gas-particle partitioning, wall losses, nucleation, and coagulation were accounted for using a "full-moving" sectional approach with 106 to 128 size bins. Furthermore, rather than assuming a fixed hygroscopicity, the UManSysProp tool was used to calculate $\kappa$ based on the evolving molecular composition of the simulated aerosol. The model was validated using data from a benchmark 8.5-hour dark $\alpha$-pinene ozonolysis experiment conducted in the SAPHIR chamber at Forschungszentrum Jülich, Germany.

Key Results

The explicit model demonstrated a high degree of correlation with experimental measurements for total SOA mass ($R^2 = 0.97$) and particle number concentration ($R^2 = 0.99$). A major finding was the significant contribution of HOMs, which accounted for approximately 43% of the simulated SOA mass. Without the PRAM mechanism to account for these molecules, the model failed to replicate the observed aerosol growth rates.

However, discrepancies emerged in the chemical and physical details:

  • Chemical Composition: The model consistently overestimated O:C and H:C ratios. This suggests that the real-world system involves "particle-phase" chemistry, such as oligomerization (where small molecules link into chains), which is not yet captured in the gas-phase-centric explicit model.
  • Size Distribution: While the model captured the general growth trends, the resulting size distribution was "wider and flatter" than the sharp peaks observed in the chamber.
  • Hygroscopicity ($\kappa$): The simulated $\kappa$ (~0.17) matched observations at moderate supersaturation (0.37%) but showed significant biases at the extremes—overestimating $\kappa$ by 19% at low supersaturation (0.19%) and underestimating it by 21% at high supersaturation (0.73%).

CCN Prediction and Sensitivity

At moderate-to-high supersaturations (0.37%–0.73%), the simulated CCN concentrations agreed well with measurements. However, the analysis revealed this was partly due to a compensatory effect where errors in size distribution and $\kappa$ balanced each other out.

The most critical failure occurred at low supersaturation (<0.4%), where the model overpredicted CCN concentrations by a factor of four. In this regime, CCN activation is highly sensitive to the largest particles in the distribution. The "wider" simulated size distribution meant the model predicted more large particles than were actually present, leading to the observed bias.

Conclusion

This work demonstrates that while explicit models like PyCHAM can accurately track bulk aerosol properties, they still struggle with the fine details of size and chemistry that govern cloud formation at low supersaturations. To improve climate models, future research must focus on integrating explicit particle-phase chemistry and refining the representation of the nucleation-to-growth transition. Addressing the sensitivities at low supersaturation is paramount, as this is the regime where atmospheric cloud formation is most frequent and where our current theories diverge most significantly from reality.

Summary updated

Abstract

Secondary organic aerosols (SOA) contribute significantly to cloud condensation nuclei (CCN), which depend on size distribution, chemical composition and hygroscopicity parameter (κ). However, how well current understanding of SOA formation can reproduce CCN concentrations and the influence these factors on modelled CCN uncertainties are still unclear. In chemical transport models, it is difficult to address the issue due to model complexity and oversimplified representation of chemical mechanisms, particle size and κ. Here, we explicitly simulated CCN concentrations of SOA from α-pinene ozonolysis, a bench-mark system for SOA studies using a box model (PyCHAM). Using state-of-the-art treatment of chemical mechanisms, aerosol size and κ, we assessed how CCN as well as chemical composition, aerosol size and κ can be modelled against measurement and evaluated the influence of these factors on CCN simulation. The model well simulated SOA mass concentration but overestimated O:C and H:C ratios, suggesting lack of particle-phase chemistry. Highly oxygenated molecules contributed substantially to SOA mass and thus CCN. Modeled κ closely aligned with measurements at moderate supersaturation (0.37 %) but overestimate κ (by 19 %) at low supersaturation (~0.19 %) and underestimate κ (by 21 %) at high supersaturation (0.73 %). The model well reproduced particle growth, but exhibited wider and flatter size distribution compared with measurement. The simulated CCN concentrations agreed well with measurement at moderate to high SS (0.37–0.73 %) but had a significant bias at low SS. Sensitivity analysis highlights the importance of accurate representation of both size distribution and κ for CCN prediction especially at lower SS (&lt;0.4 %).
BibTeX
@misc{song2025explicit,
  author = {Zhen Song and Chengqi Zhang and Hongru Shen and Hao Ma and Iida Pullinen and Defeng Zhao},
  title = {{Explicit simulation of chemical composition, size distribution and cloud condensation nuclei of secondary organic aerosol from α-pinene ozonolysis}},
  year = {2025},
  doi = {10.5194/egusphere-2025-4393},
}
Figure