Jing Li and co-authors have performed computational simulations on HIO3-HNO3-NH3 clustering relevant to UTLS (upper troposphere - lower stratosphere) new-particle formation, including also some comparisons to H2SO4-HNO3-NH3 clusters. This is a highly relevant and timely topic for atmospheric chemists and physicists. For example, there is an ongoing discussion, even debate, on the relative roles of iodine oxyacids (HIO3, HIO2) versus iodine oxides (e.g. I2O5) in promoting particle formation. Engsvand and Elm recently showed, using similar methods as the authors, that in combination with bases such as amines, the latter are more efficient (https://chemrxiv.org/engage/chemrxiv/article-details/67b4894bfa469535b9db7fa5). However, the present manuscript suggests that this comparison should probably be broadened to include also species like HNO3, which might possibly interact more favourably with the oxyacids (or not, this remains to be seen). Overall, the article is generally well written and easy to follow, the simulation methods are broadly appropriate (I especially commend the authors for both thinking carefully about appropriate temperature-dependent boundary settings, and also reporting these so clearly), and I’m happy to recommend publication of the study in ACP. I only have some minor questions and comments that I’d like the authors to briefly address. 

Scientific (minor) issues

1)Concerning the discussion of temperature on page 2, any gas-to-particle nucleation mechanism will be more efficient at low temperatures if the concentrations of participating vapors are kept constant. This follows straight from the lesser role of entropy at lower temperatures. The catch is that most potentially nucleating vapor concentrations (in the real world, if not always in simulations or laboratory experiments) also tend to decrease (often very strongly) with temperature. Is the IA-driven mechanism somehow especially efficient in this regard? For example, is it known that IA concentrations in the air decrease less with temperature compared to other potentially nucleating vapours? Or does the IA-related nucleation rate at constant concentrations increase much more steeply with decreasing temperature than most competing nucleation rates? Just saying that it is shows “remarkable efficiency” at low temperatures doesn’t really say that much, the same is arguably true for almost any mechanism. Please elaborate on this.

2)The authors use a aug-cc-pVTZ basis set (and the associated pseudopotential) for I atoms, and a 6-311++G(3df,3pd) basis set for other atoms (C, O, N, H). Previously, the use of imbalanced basis sets (specifically, large basis sets on I atoms and small basis sets on other atoms) has been shown to lead to catastrophically large biases in favour of forming bonds with iodine, see e.g. Finkenzeller et al, https://www.nature.com/articles/s41557-022-01067-z, for a discussion on this. Now, the difference in size between 6-311++G(3df,3pd) and aug-cc-pVTZ is not that dramatic, for example for C atoms its 39 vs 46 basis functions. So I don’t expect the present results to be qualitatively incorrect because of this issue - especially as the final energies are then corrected by coupled-cluster calculations, which do consistently use the same basis set for all atoms. However, a few test calculations comparing e.g. the structures and binding energies (both pure DFT energies and coupled-cluster corrected energies on top of structures optimised with different basis sets) of the smallest HIO3 - containing clusters obtained with the authors’ approach, and with aug-cc-pVTZ for all atoms (with the PP for iodine of course) also at the DFT stage, might be warranted, to check whether the bias in the present results is negligible or not. 

3)Just to confirm: when the collision rates are multiplied by 2.3 to account for long-range attractions, also the evaporation rates go up by the same fact, right? (They should, by detailed balance, equation S2 in the authors own supplement. I.e. I just want the authors to confirm that the multiplication by 2.3 is applied to both the collision and the evaporation rates.)

4)Concerning the discussion in section 3.2, of course the nucleation rate goes up with the concentration of participating species. This is inevitable and obvious. Thus it is not an actual discussion-worthy result that J goes up (“exhibits a positive correlation”) with [IA], or that the J rate with NA present is higher than the rate of the otherwise identical system with NA absent. Now, the numerical values themselves are of course interesting, e.g. the fact that even 1E9 per cm3 of NA substantially increases the rate is a valid results. But please reformulate this so that mathematically inevitable consequences of how ACDC works are not reported as novel or “notable” results. 

Technical or language issues:

-Figure 5b. How can the two pie charts corresponding to NA=1E9 and IA=1E6 (2nd pie chart from the left in both rows) be different? NH3, T, CS are the same in these runs, as are NA and IA - why are the branching ratios different? Is this a typo, or a bug, or what?

-The last sentence on page 1 (ending with “undisclosed”) seems to be missing some words, should this be “has led to… REMAINING undisclosed”? Also, the word choice is odd: “undisclosed” implies purposeful keeping of secrets (by a sentient actor, typically a human), while what the authors presumably mean is that this facet of the natural world has simply not yet been discovered or understood. 

-Page 6, it’s trivially true that SA-NA-NH3 clusters do not form halogen bonds - they do not contain halogen atoms! The first sentence on page 6 thus sounds a bit odd, and could use some rephrasing. (The general point that XBs make the IA-containing clusters substantially stronger is of course valid and worth making, I’m just commenting on the formulation here.)

Jing Li and coworkers have investigated the nucleation behavior of the combined iodic acid (IA), nitric acid (NA), and ammonia (AM) systems up to hexamers. They have done this under upper tropospheric and lower stratospheric conditions. They compare this with the already studied, equivalent systems with iodic acid replaced by sulfuric acid (SA). These types of combined systems are relevant to study due to the complexity of the real world atmosphere, where different chemical species can either work in synergy enhancing their nucleation beyond what two separate nucleation pathways can provide, or they can hinder further nucleation reducing the total nucleation.

This paper suggests that nitric acid can play a vital role in enhancing the nucleation exhibited by iodic acid, and that when studying iodine-driven nucleation outside of the lower troposphere / boundary layer one should keep the impact of this in mind.

Overall, I will recommend publication, with the caveat that the following concerns regarding the discussion of the results are addressed, I have seen the first review, and have tried to avoid reiterating any concerns discussed there unless I had something extra to add:

1) Regarding the concentrations used:

The authors have found different studies from litterature detailing concentration measurements from different parts of the atmosphere. I would recommend that the authors have a more detailed discussion of this in the methods section, where they go through exactly what has been measured in different areas of the troposphere and stratosphere. This is because the authors cite works with for example measurements that just reach the lower troposphere of IA, but uses measurements from the upper troposphere / lower stratosphere for NA. Please correct if I am wrong, but the beginning of the free troposphere is a few kilometers above ground (depending on the local conditions), while the upper troposphere is quite vaguely defined, but could potentially be several kilometers higher, or is it just the free troposphere up to the tropopause?

Thus, I would like a more specific and detailed discussion of where you use concentrations directly, and where exactly they have been measured, and where you have to assume that for example the concentration in the upper troposphere is equivalent to the concentration at the start of the free troposphere. That is, what kind of concentration ranges are measured where in the atmosphere.

2) Regarding the choice of method:

I am missing some discussion on the expected accuracy of the quantum chemical calculations carried out in the study. For example, DLPNO-CCSD(T) with a triple zeta basis set is often used in lieu of the "gold-standard" CCSD(T), but how accurate is it for heavy atoms such as iodine where relativistic effect can start to become relevant. Is it expected to overestimate, or underestimate, and if so, do we have any knowledge of how much? You also use a pseudo-potential for iodine, how accurate is this choice of simplification for clusters, is it known? If not, please explicitly say that you assume transferability of benchmark results from other chemical species if that is the case.

As the other reviewer has pointed out, you also mix basis sets within the same calculation (6-311++G(3df,3pd) and aug-cc-pVTZ-PP), which one should be careful of. aug-cc-pVTZ is defined for the other atoms too, why not use that one?

Thus in general, I would like to see the "Quantum Chemistry Calculation" sections expanded, with a discussion of how the different assumptions used in your study affects your final binding energies, because we have to make some assumption / simplifications to make the calculations feasible. However, it is important to keep in mind the accuracy of the QC, because even small errors can significantly affect the ACDC nucleation rate.

Furthermore, if a choice of method is simply based on precedence / to be comparable to other results, this should be explicitly laid out.

3) Further on the choice of method:

You say that you do spin-orbit coupling calculations using a specific DFT functional, however, you do not directly say how this is done, just that is done in Gaussian. Please comment further on what methods Gaussian is using to do this, because there is a veritable ocean of different ways to calculate SOC even if you restrict it to using DFT.

4) On the size of simulation system:

You calculate clusters with up to 6 monomers, however the nucleation rate derived from ACDC can be highly dependent on this choice. Thus an evaluation / discussion of the impact of this choice would be prudent, see for example: https://doi.org/10.1021/acs.jpca.3c00068. I don't expect you to do any calculations, but I recommend discussing what possible changes that increasing the simulation system further could bring. I.e. would we expect it to have converged to the "true" nucleation rate?

5) On the sulfuric acid comparison system:

Maybe I just did not notice, but you only cite/refer to the SA-NA-AM system in your introduction. I think you are comparing to the experimentally derived nucleation rates (and later experimentally derived concentrations), but to be honest this was / is quite unclear to me. So please clarify that you are doing this, either in your methods or directly in the discussion.

6) On the discussion of the cluster formation rate:

In general for the enhancement strength, you define the enhancement as the ratio of the nucleation rate with NA and the one without. I disagree on this choice of definition, because unless NA is actively hindering (and is doing it enough to counteract potential NA-AM nucleation) the nucleation you will always have an enhancement of 1 or above with this definition. Likewise, if you add 10^11 more molecules that can collide, then it is probable that even a small amount will stick even if it is for a brief period of time, if this happens on the boundary clusters, you would find that the clusters cross the boundary and contribute to nucleation. Thus it should always be larger than 1. It is just measuring the logical consequence of adding more molecules to the system, with the potential to capture if it hinders nucleation significantly.

I would suggest that this type of discussion of enhancement would be more suitable if you compared: IA-NH3 and NA-NH3 nucleation co-occurring (i.e. non-interacting nucleation) compared with the simulation of IA-NA-AM (i.e. interacting nucleation). Thus the ratio: J(IA-NA-AM)/(J(IA-AM)+J(NA-AM)) would measure what I would refer to as enhancement. With the caveat that running two entirely separate nucleation simulations does not give the same results as a simulation with two separate nucleation channels. However, this is not possible in ACDC as far as I remember.

As the other reviewer suggested, a reformulation of this section is in order.

7) Further on the discussion of the cluster formation rate:

Many of your differences are well within one order of magnitude in e.g fig 3a. How accurate would you expect the ACDC calculations to be, given your choice of QC method?

Likewise for fig. 3b, how significant is this difference, compared to the uncertainties in the input QC?

8) Line 213: 

"This finding indicates that the IA–NA–NH3 ternary pathway dominates in regions where IA level is limited, while NA is abundant, aligning well with the conditions in the focused UTLS. More broadly, such scenario characterized by scarce IA and rich NA also exists in higher atmosphere, such as ~20 km (i.e., the bottom of near space). This NA-enhanced mechanism is likely a vital source of fresh particles in this region."

I guess, but how much ammonia is going to be present there?

Extra comments:

Line 52: In your citation for the gas-phase mixing ratios of NA you cite Popp et al, who do measurements in the lower stratosphere. However you also cite Wang et al 2023, which is the paper: "Mechanistic understanding of rapid H2SO4-HNO3-NH3 nucleation in the upper troposphere". I assume this is a mistake? The Wang et al. study does cite papers that have conducted measurements / modeling that could support the argument, however, then you should cite those papers directly.

Line 58: You choice of systems is exclusively 1:1 or acid-dominated. While there has been indications elsewhere that this is the case, I would prefer it if you comment on this choice of systems. Not necessarily much, but it would be prudent to comment on the fact that you don't allow any base dominated clusters (at least this is how I read it)