This study underscores the critical role of accurately representing semi-volatile organics in global scale models for simulating aerosol-climate interactions. The volatility distribution of organic compounds is pivotal as it governs their phase partitioning. The authors explored the sensitivity of CCN to volatility assumptions of organic compounds. Findings indicate that while the process model shows high CCN sensitivity to volatility uncertainties, the global model, particularly with a detailed 9-bin VBS, reveals that particle number concentration and CDNC are largely unaffected by major shifts in volatility. Most of the mass in lower volatility bins remained in the particle phase. The SOA mass burden, N, and radiative forcing were found to be more responsive to uncertainties in semi-volatile bins than in low-volatile bins. Furthermore, the study identified that IRFari is sensitive to volatility uncertainties in a detailed 9-bin VBS setup, while the ERF becomes sensitive in a simplified 3-bin setup. Comparing the 9-bin and 3-bin VBS setups, the study found notable differences in N100, CDNC, and SOA mass concentrations. The 3-bin setup showed higher N100 and CDNC but lower SOA mass concentrations compared to the 9-bin setup. The choice of volatility calculation method also impacted the accuracy of the model, affecting either particle number concentration or SOA mass concentration.

Overall, the results from this manuscript are interesting; my main concerns are the description of the model and methods as well as the explanation of some of the results, which were hard to follow, as mentioned below. Those parts would benefit from reorganizing and clarifications.

60: “GISS II’ GCM (Farina et al., 2010), ECHAM-SALSA (Mielonen et al., 2018), CESM2 (Tilmes et al., 2019), GEOS-CHEM (Fritz et al., 2022), GFDL AM (Zheng et al., 2023)” please spell out all the models.

60: Citing Farina et al (2010)as GISS II GCM needs double checking, according to Farina, “The primary tool in this work is the so-called “unified” chemistry-climate-aerosol model [Liao and Seinfeld, 2005], which was developed on the basis of the Goddard Institute for Space Studies General Circulation Model II′ (GISS II′ GCM) [Rind and Lerner, 1996] and includes online tropospheric chemistry [Mickley et al., 1999; Wild et al., 2000] and bulk aerosol thermodynamics modules [Adams et al., 1999; Nenes et al., 1999]. ” it was based on GISS II GCM with other configurations to make it, as Farina et al called it, “Global Chemical Transport Modeling” , please differentiate with GISS ModelE, which is currently more well acknowledged in the modeling community.

64-66: please cite Donahue et al. (2006) as well since it was the first time the original concept was presented

66-69: this is a very long complex sentence, please consider re-writing to shorter ones for easier understanding and readability

83: does SALSA have a full name? please spell it out

98-107: Unfortunately, I had trouble following the description of the particle growth model even after reading it 4 times. Please, please re-write or include an illustration. Currently, the major confusions for me are :

• particle and nanoparticles need to be specific, in the first sentence they’re both used. What size are we talking about?
• the modes aren’t specified either, please differentiate nucleation mode, Aitken mode, and accumulation mode in the text.
• “Particle phase is assumed to form an ideal solution and no particle phase processes are included.” this is also confusing considering you mentioned particle properties, and particle phase repeatedly, if particle phase processes aren’t considered, how are properties changed and what’s going on with coagulation, etc. in the next sentence?
• “Aitken and accumulation mode particle diameters and number concentrations are set constant. Self-coagulation decreases nucleation mode number concentration but its impact to particle size is ignored.” But why? You’re talking about a model that simulates “particle growth and survival to CCN size” How do you set diameters to constant and ignore self-coagulation’s impact on size?

Perhaps re-writing this description more clearly would help clear the confusion.

109-110: “stoichiometric coefficient of the respective bin.” Can you list them? If they’re listed later, please refer to them here too

117-134: let’s walk through the logic here:  model+microphysics module SALSA is mentioned, model description, then aerosol module HAM description, then HAM’s two options M7 & SALSA, then SALSA in more detail. It does not flow well. Please consider reorganizing the order, for example, from big to small: main model->aerosol module->aerosol microphysics module options.

135-149: I see there are more descriptions for SALSA, which is good. I recommend going through model descriptions (2.1-2.2) as a whole section and seeing the logic and flow to make it clearer, and better connection with the whole model. It’s very unclear at the moment how they’re connected. They’re just paragraphs of information with little connections. For instance, reading up to this point, it’s unclear how MCOLNAG plays a role in the global model and its relations to SALSA.

205: “This is why the VBS setup of several global models have simpler VBS representations.” Please specify which ones

301-306: Fig 5 shows that there’s a shift in mass, but barely changed for N and CDNC. Could you please explain why this is? This paragraph only described the results but there’s no discussion of the reason why.

310: This is interesting, but which factors? Could you please elaborate?

319-323: “The difference in mass concentration primarily comes from the third bin (bin3) of the 3-bin VBS setup, which contributes to the lower SOA mass concentration in the 3-bin VBS setup, while the mass concentration in the other two lower volatile bins is the same between the two VBS setups. Differences between the two setups are sensitive to how the volatilities for each bin have been chosen in the 3-bin VBS setup.”

• It’s hard to follow what this is trying to say. Which is the third bin, it is not clear in Fig 6.
• Which are the two lower volatility bins and which are you comparing against, the a vs b and c?

Similar problem for Figure 7. The referred bins are hard to follow. I recommend that you add a summary table of all the experiments for better reference in the text.

Bonus question: do you think your results would hold true in other models, why and why not (and what factors would make the difference)?

This manuscript explores the sensitivity of aerosol growth rates (in a box model) as well as SOA mass, CCN, and radiative forcing (in a global model) to uncertainties in SOA volatility distributions. The analysis generally seemed sound, though the overall science contribution is generally minor (e.g., tying the box and global model results together to learn more about *why* SOA mass and CCN change in opposite directions could have made the paper a larger contribution). I’m fine with the manuscript being published once the authors have responded to my comments.

General comments

1) The box model and global model results are not tied well together. They feel like completely orthogonal analyses in the manuscript. What can we learn from using both tools? What box-model simulations can help elucidate some of the findings of the global model?

2) If I were to choose the C* (Csat) of the 3 volatility bins to best capture both particle growth (CCN) and SOA mass, I’d choose: one low volatility bin that condenses ~irreversibly and contributes to growth of small particles (Csat of 1E-2 ug m-3 and lower grouped together) and two bins that span the range of atmospheric OA concentrations (maybe Csat of 0.5 and 5 ug m-3). Unless the model includes aging through the VBS bins, the 5.32E2 ug m-3 bin is going to be irrelevant for SOA formation during warmer months when monoterpenes are being emitted. I was surprised to not see any reasoning through what an ideal 3-bin scheme might be to capture both growth and mass variability. The readers are not left with a way forward for SOA modeling.

Specific comments

L26-28: These values are very large. When I read the abstract, I thought these were global forcings, but I think they are just over the boreal forest regions, so this needs to be clear here.



L26-28: What do the uncertainty ranges here represent? Why are the uncertainty ranges for the ERFs so large (relative to the central value and the IRF uncertainty ranges)?

L32-33: The total atmospheric aerosol loading by mass is dominated by coarse aerosol: dust and sea spray, not organics. The papers cited for the range are for submicron aerosols, mostly at continental sites, so “total atmospheric aerosol loading” is not precise. 

L43: I don’t think the consensus is currently that aromatics are the main anthropogenic SOA precursors (S/IVOCs are in the mix now), though I guess you say “VOC” in the sentence.

L67: Partitioning is a phase transition, not a reaction.

L72: Dominate what?

L108-110: So the vapor concentrations are held fixed? There are no feedbacks as condensation sinks evolve? No volatility-dependent growth-rate differences between smaller and larger particles? Is this realistic enough to gain useful insight?

L130: I’m confused because one sentence says that it’s a grid model, the next says that it’s a spectral model.

Section 2.2: Semi-volatile condensation/evaporation across size bins is a very stiff numerical system. Small particles will adjust very quickly to be in equillibrium with gas-phase semi-volatile species while accumulation/coarse-mode particles take orders-of-magnitude longer. How does the model handle this? Is there some sort of hybrid solver scheme or do you take very short timesteps to explicitly resolve the cond/evap kinetics of the smallest particles? The text says the APC scheme, but can you state a bit more? 

L165: Csat is stated as a mass concentrations elsewhere in the paper. Is “x” in eqn 2 actually a mole fraction or is it a mass fraction. The Donahue VBS usually uses mass fraction with mass-based concentrations, while Raoult’s law would use mole fraction with partial pressures (proportional to molar concentrations rather than mass concentrations). Using mole fractions with mass concentrations would be an unusual hybrid. 

L188: This is a very low Hvap that came out of not-fully-constrained fits of smog-chamber experiments. It’s inconsistent with Hvap values for SOA C* values of interest, e.g., https://doi.org/10.1021/es902497z (Epstein et al., EST, 2010). 

Section 3.1. Did you explore the interplay between volatility and the relative amount of mass split between the growing nucleation mode and the pre-existing larger models, e.g., https://doi.org/10.5194/acp-11-9019-2011 (Pierce et al., ACP, 2011)?

Figures 2 and 4: While it does say at line 231 in the methods that the results will focus on the boreal regions, most readers skim articles and don’t read methods in detail. Please add a note to the caption says that the figure is limited to boreal regions to highlight the sensitivity to monoterpene SOA (or similar).

Figures 3, 5, 6, 7, 8, and 9: Are these limited to just the regions above boreal forests (e.g. the regions in the maps of figures 2 and 4). Please state this explicitly in the captions. This information is critical to understanding these figures. Note that Figures 8 and 9 says “boreal region”, but I think it’s just the boreal forest region (if it is a more general “boreal” northern region beyond just the forests, then the region needs to be defined). Also, what does “close to ground level” mean specifically?

Figure 3: I’m a bit confused about how there is non-trivial mass in the particle phase in the 1E3 ug m-3 bin in the VBSx10 simulation (which would make this bin 1E4 ug m-3, right?). It would need to be very cold or very high concentrations in order to get much aerosol mass in that bin. But I wouldn’t expect much SOA precursor emissions when it’s very cold, so in a weighted average, I’d expect very little mass in the particle phase in this bin. Can you please check the numbers in this figure?

Figures 5, 6, and 7: Are the concentrations on the x axis for ambient conditions or normalized to STP (e.g. ug sm-3)? This needs to be specified because it makes about a factor-of-5 difference in the concentrations at 200 hPa.

Figure 7: Too many colors are similar here, especially the reds, and I can’t tell what is what. 

Figures 8 and 9 and the associated discussion. What are the total IRF and ERF of SOA in the base case (i.e., base relative to a case with SOA shut off)? This will give more context to the sensitivity of the forcings to SOA volatility here. 

Figures 8 and 9: Error bar depicts 1 standard deviation of what? The different grid cells over boreal forests? Is it still temporally averaged or is temporal variability also part of the standard deviations?

Section 3.2.4: To increase the precision of writing, please try to always state what the IRFs and ERFs are relative to (in most cases in the discussion, it’s the base simulation).

L405-406: I disagree that the ERF is only sensitive when the 3-bin version is used. Looking at Figure 8b, the ERF for VBSx10 and VBSx0.1 are at least as large as all of the IRF values in Figure 8a.