Thomas et al. present a seasonal study on the composition of ultrafine particles in the Eastern Amazon. Strictly speaking, on the composition of the organic fraction in ultrafine particles, as no information on inorganic ions (sulfate, nitrate, ammonia) or metals is provided. In this respect, the title should contain that information, especially because in the light of publications of the Smith group, one might expect the measurements with the TD-CIMS including inorganic ions. In this study, they sampled particles with a combination of a nano-DMA and a sequential spot sampler for offline analysis. Samples from three different months were merged and analysed by liquid chromatography / high-resolution mass spectrometry with negative-mode electrospray ionization. The samples from June represent the late wet season, September the dry season and December the late dry season with strongest biomass-burning activities. They found in all three samples strong contributions from isoprene-derived organosulfates. For the late wet season, they also detected compounds that are likely emerging from fragments of biological spores. The December sample showed strongest aromaticity (degree of unsaturation) and highest average carbon oxidation state. Overall, the manuscript presents substantial novel data, the results are discussed in a balanced and appropriate way with plenty of relevant references. The presentation quality is good to excellent – I would only suggest to adapt to the colour logic of the AMS community (Figure 1 and 7 with CHO as green; CHON and CHOS are already good with blue and red). Apart from this minor thing, I have one important comment on the volatility classification, as described in the next paragraph. I think this can be resolved, so overall, there is one major issue to resolve and a few minor clarification or changes necessary.

Volatility: the authors use the volatility classification of Li et al. (2016) – the molecular corridor approach. This is justifiable, because this approach is the only volatility classification that allows to calculate C* for compounds that include all three relevant heteroatoms: oxygen, nitrogen, and sulphur. However, the Li et al. parametrisation tends to overestimate C* up to a few orders of magnitude, as pointed out by Isaacman-VanWertz and Aumont (2021) and Thoma et al. (2022). In line 363, the authors classify the tricarboxylic acid MBTCA as semi-volatile based on Li et al. (2016). Thoma et al. (2022) argue instead (on the same molecule) that with using the SIMPOL.1 model, which specifically considers the carboxylic acid functionality, MBTCA is a low-volatile compound. Almost four orders of magnitude lower C* between the two different approaches – with this bias one might come easily to wrong conclusions about new particle formation and growth. In my opinion, the SIMPOL.1 model is more trustworthy as it specifically accounts for functional groups. If the sulphur-containing species that were used in the Li et al. parametrisation are no organosulfates, this might also cause systematic uncertainties. However, due to the polarity of organosulfates, the conclusion that these compounds are low-volatile and contribute to particle mass appears plausible. I think this issue can be resolved with an appropriate discussion of the uncertainty regarding volatility prediction.

Further comments:

1. line 46-48: what happens to organic aerosol partitioning during convective downdrafts? BVOCs that were nucleating/condensing in the upper troposphere at -50°C might re-evaporate when reaching the surface layer?
2. l. 61-62: how does MBTCA and pinic acid relate quantitatively to isoprene SOA markers?
3. l. 95: 30 meters is a very long inlet. This is problematic especially for ultrafine particles. Were size-dependent inlet line losses quantified? If not, they should be estimated and it should be critically mentioned that the overall 5-70 nm range is systematically biased towards larger particles. On the same issue: what about multiple-charged particles and their contribution? Overall, if a differential mobility analyser was used for size selection, the associated uncertainties should be discussed in the methods section.
4. l. 99: shouldn´t the blank also be sampled for 24 hours? I guess the two hours were chosen to get daily blanks; the shorter sampling duration might cause blanks appearing too low? Also in this section on the sampling, I miss the information on the sampling times. Midnight to midnight?
5. l. 116: m/z in italics (check the whole manuscript).
6. l. 127: This sounds as if the raw data were investigated manually for peaks? I strongly suggest free non-target analysis software such as mzmine. With this software the risk of missing peaks is much reduced and files can be processed in batch mode. Also it allows to use data-bases for compound identification, and possibly identify oxidation products of VOCs mentioned in line 225-228.
7. Generally, chapter 3.3 should describe the scan modus (e.g.  full-scan with data-dependent MS2 in discovery mode).
8. l. 165: a resolution of >100k should enable to measure four digits behind the comma with enough precision.
9. l. 168-170: organosulfates generally ionize very efficiently in negative ESI. Hence, the peak height can be misleading with comparison to other non-sulfate-containing organic compounds.
10. l. 182: It should be possible to evaluate in the dataset whether isomers are present.
11. l. 269: methacryloyl peroxynitrate (two words)?
12. l. 286: The H/C of this CHOS specie does not fit to PAHs. I did not find this formula in Jiang et al. where PAH-OS have H/C of around 1.
13. l. 289-292: these two sentences are contradictory.
14. l. 355-360: I find it interesting that the source of sulfate is not clear. Can this somehow be figured out based on SO2 measurements or the variability between SO2 and other marine markers -> answering whether the sulfate is coming from the ocean or from anthropogenic sources on land?
15. l. 369: I doubt that the particles nucleated over the ocean where the biogenic VOCs are missing.

References:

Isaacman-VanWertz, G. and Aumont, B.: Impact of organic molecular structure on the estimation of atmospherically relevant physicochemical parameters, Atmos. Chem. Phys., 21, 6541–6563, https://doi.org/10.5194/acp-21-6541-2021, 2021.

Thoma, M., Bachmeier, F., Gottwald, F. L., Simon, M., and Vogel, A. L.: Mass spectrometry-based Aerosolomics: a new approach to resolve sources, composition, and partitioning of secondary organic aerosol, Atmos. Meas. Tech., 15, 7137–7154, https://doi.org/10.5194/amt-15-7137-2022, 2022.