|In my view this is a much improved paper; the authors addressed my major concerns, and the new manuscript now discusses when the PMF approach is valid/necessary. I do think this discussion (of the PMF approach vs the tracer approach) could still be more thorough; further, the PMF results raise some questions about the heterogeneous chemistry that are currently not discussed. These concerns, which need to be addressed prior to publication, are described below.|
1) Discussion of PMF vs the marker-ion approach:
- Given that the authors are no longer arguing that previous AMS-based heterogeneous oxidation experiments have severely underestimated uptake coefficients, the text should clearly state so; plus lines 475-478 should be deleted. It may also help to cite my previous comment as evidence that the tracer and PMF approaches can sometimes agree (and Smith et al. 2007 as evidence that the tracer approach and molecular-ion approaches can sometimes agree as well).
- Similarly, I would recommend a change to the title, since "improvement" is too strong. The authors show that PMF analysis gives a higher value of gamma, but without independent confirmation that this higher value is correct (i.e., via molecular detection of the reactants) this new approach has not yet been shown to lead to an improvement in accuracy.
- Somewhere in the text (maybe in lines 299-315) it needs be highlighted that the drop in the PMF product factor is substantially greater than that of any of the individual ions. This is implied in the text, but never said explicitly - it’s important because it means that PMF (a highly complex mathematical procedure!) is required to analyze the oxidation kinetics, which the authors argue cannot be done by simpler means. Similarly, it should be stated that a large difference between the PMF-factor decay rate and the marker-ion decay rate is mathematically possible only when the two factors (reactants and products) are extremely similar.
- Page 24: I think this section needs to again highlight the usefulness of measurements (from CIMS, GC-MS, VUV-AMS, etc.) for the determination of heterogeneous loss rates. In truth such measurements probably are superior to both EI-AMS approaches discussed (tracer ions and PMF), and this should be discussed as the best way forward.
2) Discussion of the chemistry:
- There needs to be some discussion of the high value (2.74) of the uptake coefficient determined by the PMF approach. This value of gamma is much higher than those measured by other groups for OH-initiated reactions. Moreover, it means that every OH-particle collision leads to the reaction of almost 3 citric acid molecules, (Alternatively, as discussed by Lambe et al 2009, it could mean that some fraction of the parent molecule is actually in the gas phase, but this is highly unlikely for citric acid.) What is the likely mechanism for this? And why have such large values of gamma not been observed for other systems?
- There should also be some discussion of the role of multigenerational chemistry in the present experiments. The implication of this work is that only early-generation chemistry is being probed, leading to the small differences in the reactant mass spectrum and derived product mass spectrum. However, at the highest level of OH oxidation, >80% of the product factor has reacted away (Figs 5b and 6a). At this level of oxidation, later-generation products ought to be formed as well (see Smith et al. 2009; Wilson et al. 2012). That assumes that 1st generation products exhibit oxidation kinetics that are reasonably similar to that of citric acid; but they probably should given their chemical similarity. Yet no further evolution of the mass spectrum (which could show up in a 3rd PMF factor) is observed. Why is this?
- Lines 371-373 (and 402): There actually is some overlap between the OH exposures in the present study and those in Kessler et al. 2012, so the difference in exposures alone probably cannot explain the differences observed (i.e., differences in change to m/z 68). Some of the more subtle effects given in lines 404-422 may be at play here. The work of Che et al. (PCCP 11:7885), which finds uptake coefficients can vary with different timescales and concentrations, may also be helpful here.
Lines 124-131: The authors need to be specific about how AMS tracer ions are typically chosen – that they are usually the fastest-decaying ions in the spectrum (meaning the technique is equivalent to spectral subtraction).
Line 217: Ulbrich et al. 2009 should probably be cited here.
Lines 292-293: Here it is stated that m/z 68 consumption is observed; this is in direct contradiction with line 369, which states “no significant consumption of m/z 68 was observed”. Which is correct?
Line 349: This correction should be included in Eq. 7 (or at least referred to in the text just below the equation).
Line 349: the uncertainty value appears to be incorrect. (Possibly high by a factor of 10?)
Lines 398-403: I don’t follow this sentence, especially the part marked “2” – the numbers themselves (not the potential differences in product distributions) indicate that the rate constants are different.
Lines 432-433 (and Scheme 2): This is only one product; somewhere there should be text mentioning other possible reaction products. Might fragmentation products (which can look quite different) play a role? Also, the citation should be modified to reflect that this follows the general chemistry described by Atkinson et al. (since citric acid wasn’t studied directly).
Typos/language: Line 70 (“particles make”), 103 (lower case c), 112 (“has been widely”), 117 (“detection”).