I appreciate all of the hard work that the authors did in response to the last round of reviewer comments. One of the major changes is the new treatment of the Henry’s law partitioning of MVK (and MACR), which is much more realistic than what was assumed in the previous version. It also leads to a very different conclusion: these species contribute a negligible amount of aqueous SOA. Despite this, the results are significant and help to highlight the importance of the Henry’s law constant for aqueous SOA. While there are a few issues that need to be addressed (see below), I recommend that the paper be accepted after this.
1. The authors did a good job estimating the Henry’s law constants for MVK and MACR, but there are a number of minor problems:
(a) It’s easiest to understand the impact of salts by the ratio K(H)*/K(H), rather than the inverse, which the authors generally use (e.g., Figure 6). I recommend that the authors use the former. For example, Fig. 6 would be more intuitive if the data was plotted as K(H)*/K(H), which would show a decrease in effective Henry’s law constant with increasing K(S). Eq. 4 could be accordingly modified as log(K(H)*/K(H)) = -Ks[salt].
(b) The text sometimes incorrectly uses the ratio. For example, on Page 16, lines 9-12: “…we apply a maximum value enhancement of KH*/KH = 100 which seems applicable for a saturated ammonium sulfate solution and a Setchenov coefficient of KS ~ 0.5 kg mol-1...” The ratio should be 0.01 (1/100) and the word “enhancement” is misleading.
(c) Similarly, in Table 1 the formulas in the left column for K(H)* are backwards, i.e., the effective K(H)* should be lower than the physical K(H) due to salting out.
(d) Also in Table 1, the range of values for MVK is good. But the range of values for MACR is wrong and should be 0.065 - 6.5 M/atm (rather than 100 times higher than this). The model needs to be re-run if these incorrect values were used.
2. Section 3.3.3.
(a) The comparison of gasSOA with the the K(H)* = 0.01 x K(H) case is the most realistic case for MVK and MACR and shows that their aqueous reactions are a negligible source of SOA. But it is not clear if the next case, with an assumed 20 mM of aqSOA precursors from isoprene, is reasonable. (For comparison, what is the total aqueous concentration of MVK and MACR in the first scenario?) I understand how 20 mM was estimated, but what is the justification of this concentration? Is it reasonable for the isoprene scenario? For example, what is the equivalent mass of precursors and how much isoprene oxidation would be required to approximately give this mass?
(b) The 1 M aqueous organic precursor case seems very high for the model scenario; is there any actual data that suggests this is a reasonable amount? In this case the model results in Figure 8 show that the amount of SOA increases by a factor of approximately 2.5 (from 20 to (20+30) ng/m3, roughly) in the presence of modest amounts of particle water. Is there any evidence from isoprene chamber studies that this large enhancement occurs? Rather than assume an aqueous organic precursor concentration, the authors should use chamber data to constrain this concentration. If there are no chamber studies with particle water, this case should be highly qualified and its assumption more critically evaluated. I get the sense that the authors are using 1 M to account for all possible atmospheric organics that could form oligomers in the aqueous phase, but this is not reasonable since the gas-phase SOA bar in Figure 8 only results from isoprene reactions. It's not "fair" to compare gasSOA from isoprene with aqSOA from a wider set of precursors.
3. The authors revised their treatment of H2O2 photolysis, but there are a few issues with Eq. 2 that should be addressed:
(a) Eq. 2 only works if the absorbance measurement for MVK was done under the same conditions (pathlength, concentration) as the MVK setup in the illumination condition. If this is the case, it should be stated. If not, the authors need to account for differences in these conditions between the two set-ups.
(b) I'(0,lambda) in Eq. 2 is the photon flux transmitted through the solution (i.e., at a distance equal to the solution pathlength), which is an underestimate of the depth-averaged photon flux in the solution. This is probably why the Model fit doesn't match the experimental values in Figure 3. Instead of using I'(0,lambda) for the Model fit, the authors should instead use the depth-averaged photon flux at each wavelength (i.e., the average of the exponentially decaying I through the pathlength). This is a relatively easy fix. Also, a better variable name for I'(0,lambda) would be I'(lambda) (where "lambda" = the subscripted Greek letter). The zero subscript generally means the incident photon flux, which does not apply here.
(c) For future experimental work, I suggest the authors use a chemical actinometer in the same container as used for the experiments. This will account for the multiple passes of the photons through the container.
1. pg. 19, lines 27-29. To better calculate the SOA yield the amount of isoprene reacted should be simply calculated as k(OH+isoprene)[OH][isoprene]x(elapsed time) since the reactant amounts are constant.
2. (a) pg. 20, line 1. The upper bound of 20% is misleading since it is for an enormous (and largely hypothetical) aqueous organic concentration. Rather than give a range with no explanation, the values for the different cases should be given. (b) This unjustified “upper bound” of ~ 20% is too speculative to be included in the Conclusions (pg. 23, line 24).
3. pg. 22, lines 1-4. What concentration of oligomer precursors are you assuming for this ratio with O2? Is this based on the most realistic salting-out scenario for MVK and MACR, i.e., K(H)*/K(H) = 0.01? If not, why not?
pg. 17, line 20. Give a reference for the K(H) value for MACR here.
pg. 18, line 19. “low” should be “high”
pg. 21, line 28. To be consistent with my recommended K(H)*/K(H) ratio for MVK and MACR, the O2 ratio here should be inverted.
pg. 23, line 18. “are” should be “is”
pg. 23, line 22. “there” should be “their”
pg. 32, line 15. “photooxdiation” should be “photooxidation”
Table 3. The units of k are incorrect (given that these are all 2nd-order reactions).
Figure 1. The caption should indicate what the different colors represent.
Figure 2. (a) Extraneous numbers appear on the y-axes on the two left plots. (b) The caption should indicate that symbols are experimental results while lines are model results. (c) What do the different symbol colors represent? Replicate experiments?
Figure 4a. The line for the 20 mM MVK, low O2 case is not dashed in the figure (as it is in legend).
Figure 6. (a) Solution concentration labels (~1 mol kg-1, etc.) don't match up with the white boxes. (b) It is difficult to distinguish the blues and purples in the plot. What about using a linear color scale instead of a log scale?
Figure 7. The figure resolution is poor (at least on my screen).