Secondary organic aerosol (SOA) may contribute significantly to cloud condensation nuclei (CCN), yet relevant research of explicit simulation remains relatively limited. While existing SOA modeling studies predominantly concentrate on mass concentration, this work specifically investigates the CCN activity of SOA, thereby advancing our understanding of SOA's role in CCN formation. My specific comments are as follows:

(1) Although the paper is titled " Explicit simulation of chemical composition, size distribution and cloud condensation nuclei of secondary organic aerosol from α-pinene ozonolysis", it only provides detailed descriptions for the size distribution and CCN simulations, with inadequate description on the chemical composition of SOA. For instance, the number of species involved in gas-particle partitioning in the model remains unspecified. Furthermore, no information is provided regarding whether the gaseous concentrations of these species were characterized with experimental observations or have undergone laboratory validation. The authors should provide a list of substances involved in gas-particle transformation in supplement file.

(2) Accurate simulation of CCN critically depends on both number concentration and particle size distribution. Notably, the authors employed two distinct methods for number concentration: when modeling CCN, they utilized observation-derived fitting results, whereas for SOA mass simulation, they adopted a nucleation scheme based on C20H30O17 molecule. Why were these two methods applied separately? Are the simulation results from these two approaches comparable?

(3) Line 116: Please specify the exact model of the DMA in the SMPS. Also, provide the specific model of the AMS, and similarly, specify the models of other equipment used.

(4) Line 119: The authors state that "The SS calibration and κ parameter calculations followed Zhang et al. (2023)," but later in the results section, it is mentioned that κ was measured. The authors should explain how κ was measured in the experimental section.

(5) Line 140: Please provide the specific formula used to calculate k_i,j, as well as the range and basis for the values of γ and α in this study.

(6) Line 165: The authors mention that the aerosol particle size was divided into 128 bins, but later state that it was divided into 106 bins. This inconsistency should be clarified, and the aerosol bin division should be explained in detail in the methods section.

(7) Lines 196-198: The authors compare measured and simulated values of α-pinene to indicate the capability of PyCHAM with the MCM + PRAM mechanism to describe the gas-phase chemistry of α-pinene ozonolysis. To validate the model's performance in simulating the MCM gas-phase reactions after incorporating the HOMs module, comparing only the reactants is insufficient. It is recommended to also compare the temporal evolution of other major product concentrations, particularly the simulation performance for HOMs.

(8) The authors attribute the overestimation of simulated O/C and H/C ratios to the lack of consideration of particle-phase reactions in the model. However, in Figure S6, the simulated HOMs are generally higher than the measured values, especially for ions with m/z above 400. Yet, the total SOA mass concentration is simulated well, implying that the simulation underestimates other components while overestimating HOMs. Clearly, the overestimation of HOMs would lead to higher O/C ratios. Additionally, the authors should analyze the reasons for the overestimation of HOMs in the simulation compared to observations (Figure S6).

(9) It is difficult to observe the differences between the simulated and observed particle size distributions in Figure 5. It is recommended to supplement the figure with a two-dimensional curve showing the particle number concentration as a function of particle size at a specific time.

(10) How were the κ values in Figure 6 measured? This is not explained in the text. Furthermore, why does the measured κ value show a sudden decrease at the second hour, while the simulated value does not exhibit such a change? As shown in the figure, κ values differ under different SS conditions, so what SS was used to determine the simulated κ?

(11) When SS = 0.19%, the simulated CCN concentration is much higher than the measured value. The authors attribute this overestimation to the wider and flatter particle size distribution in the simulation. Why does this overly broad particle size distribution not cause significant deviations under other high SS conditions?

Song et al. 2025 provide an investigation into cloud condensation nuclei sensitivities that is novel in taking a near-explicit approach to gas- and particle-phase composition, allowing this composition to affect CCN. It is a meaningful investigation for the atmospheric science community, as it demonstrates the abilities and limitations of this bottom-up approach. An accurate version of such an approach would be valuable to the community to provide a benchmark against which parameterised approaches could be compared or even trained, therefore the work contributes toward realising such a standard. Given this novelty and significance, and that the method appears sound, I recommend publication in Atmospheric Chemistry and Physics after attention is paid to the suggested revisions below.

• I recommend ‘Explicit’ in the title be changed to ‘Near-explicit’ if the gas-phase chemistry is being referred to. Potentially ‘Process-level’ would be more accurate since several processes are being considered.
• Line 29 and elsewhere, please check grammar in phrases like ‘exhibited wider and flatter size distribution’ so that they are grammatically correct, e.g.: ‘exhibited a wider and flatter size distribution’ or ‘exhibited wider and flatter size distributions’
• Line 55 and surrounding text. Whilst the authors explain the limitations of 3D models to identify CCN sensitivities (perhaps too much given that it is well known that such models depend on higher physicochemical resolution experiments and models to develop their parameterisations), the workflow between the presented work and the 3D models is not well explained, leaving the reader with some uncertainty about the significance of the current work. I think I understand this workflow, and therefore why the work is important, and have included my thoughts in my opening paragraph of this review. I recommend that the authors be less critical of 3D models and more constructive in describing why work like theirs can, eventually (once a sufficiently accurate model is verified), support the 3D model development. I think the revised messaging in this section of the introduction also needs to be folded into the summary of the study in the final paragraph of the introduction, where the aims of the study are stated. The current aims of the study are not very significant at all, as past studies have identified key components in the alpha-pinene system (such as Roldin et al. 2019), and others have identified key sensitivities in CCN evolution (such as McFiggans et al. 2006 (doi.org/5194/acp-6-2593-2006), if this last article is not yet cited in the introduction I think it, or a more recent article that updates its findings, should be cited)
• Line 60 and elsewhere, when referring to the chemistry, please use ‘near-explicit’ rather than ‘explicit’ to help clarify the type of mechanism applied (MCM papers often use the ‘near-explicit’ description)
• Lines 75-76 the dynamic transfer mentioned is based on thermodynamic absorption partitioning theory, so I’m not sure there is a distinction between the approaches as currently named (to justify the ‘or’ in this sentence). If the authors want to distinguish between equilibrium partitioning and dynamic partitioning, please make this distinction clearer
• Line 76, I think a more balanced introduction to the importance of particle-phase reactions can be provided, and would be helpful for the reader, for example Lopez et al. (2025) doi.org/1039/d5ea00062a suggest that for the alpha-pinene ozonolysis system at relatively low temperatures, particle-phase reactions play a minor role in SOA evolution.
• Line 86, when discussing the simulation of SOA composition I think the articles of Roldin et al. 2019 (PRAM paper) and Pichelstorfer et al. (2024) (autoAPRAM-fw) paper deserve to be mentioned as they reflect advances in HOMs simulations with comparisons against CIMS observations.
• Aims in final paragraph of introduction, please see my point three above.

• Section 2.2: the version of PyCHAM used should be stated (provided in the setup file of PyCHAM), and the relevant input files used in the study should be provided via a URL in the data availability section, with a reference to this section provided in the main text. Figure S2 indicates that best agreement with [SOA] is gained when the walls do not absorb gases. The main text should justify why a Cw>0 is therefore used and whether the chosen Cw>0 is physically realistic.

• Around line 170 and its paragraphs, I agree with the other reviewer that more detailed is needed about how the two different approaches to applying PyCHAM influence our interpretation of the results: what conclusions can we still make with this approach and what conclusions are no longer available (but would have been if the same approach was used throughout)

• Line 190, given that particle sizes vary by a factor of 10 in this study, and that particle losses to wall are sensitive to size (e.g. data contained for chambers here https://www.eurochamp.org/simulation-chambers), there either needs to better justification for using a size independent Beta_flec (e.g. a supplementary material showing size independence across particle sizes), or a size-dependent particle loss rate to walls needs to be used.

• Line 245, I don’t think the evidence presented up to this point justifies the comment about likely attribution, and possible attribution seems more fitting at this point. Also around this line, the authors mention the activity coefficient. Could there be some more discussion, with reference to the literature, about whether a higher solubility (represented by the activity coefficient) of the partitioning components could bridge the gap to observed SOA. In Section 2.2 please state the values used for activity coefficients and accommodation coefficients in simulations.

• Line 267, again I think ‘likely attributed’ is not yet supported by this stage of the evidence. ‘Possibly attributed’ at this stage would be more suitable.

• Line 281, rather than ‘alter’, can you tell the reader which direction the particle-phase processes mentioned in the referenced works take the H:C and O:C values.

• Around line 240, could the authors provide some more quantification of the monomer:dimer ratio (observed and simulated). Comparing this ratio between observation and simulation, could the authors please, discuss, e.g. around where O:C and H:C are discussed, whether inaccuracies in the gas-phase chemistry could explain the O:C and H:C discrepancy?

• Results and Conclusion section. Given that studies like this are necessary to assess how helpful process-level modelling currently is to informing 3D model parameterisations, there should at least be a discussion around what the results mean to this workflow. E.g., is all, or some parts, of the simulation sufficiently accurate as to be used to inform 3D models? Where should future work be focused to close any remaining gaps before we can reliably apply these process models to gain 3D model parameterisations?

• Figure S6, in the y-axis title is the word normalized. The method for normalising both observation and simulation results needs to be stated.

Simon P. O’Meara