The is an interesting study of the effect of the anneal-leach (chemical abrasion) procedure on the structure and U-Pb concordance of zircon using micro-Raman spectroscopy, SIMS and APT. The main revelation of the Raman data is the association of zones of high radiation damage with spots that gave discordant data for both annealed (but unleached) and unannealed samples, which is what might be expected. The APT data show no evidence of Pb mobility or zoning at the sub-micron scale although as the authors point out there could be sampling bias in that highly damaged material may have failed to produce suitable APT samples. Given the universal adoption if the chemical abrasion method in U-Pb geochronology for mitigating Pb loss in ancient zircon and the generally poorly understood mechanisms of Pb loss, I think that the work is worth publishing after a few improvements to the figures and perhaps a more balanced discussion of Pb loss as suggested below. The manuscript does not strike me as too long for a technical note but if it needs to be shortened there could be less discussion of the APT results given that these show a generally uniform distribution of U and Pb atoms.

Even after several generations of U-Pb geochronology there still seems to be uncertainty in the community about the mechanism(s) of Pb loss in zircon, with some associating Pb loss directly with metamictization. The following is my personal assessment of the problem, based partly on Das and Davis (2010, doi:10.1016/j.gca.2010.06.029) which I hope the authors and other readers will find useful. After a lifetime of experience, I have never found a case of low-temperature volume diffusion of Pb in zircon that was not associated with chemical alteration. I have analyzed many high-U zircon grains with metamict structure that gave concordant results from internally homogenous domains. Zircon alteration was documented by Tom Krogh (1975, Carnegie Institution of Washington Yearbook, 74, 619-623) who showed that it required a highly damaged crystal structure and an external source of fluid, which alters zircon from the surface inwards or along cracks generated in zoned zircon by differential expansion of the higher U zones.

Alteration can only occur after the degree of damage becomes high enough and its age should be approximated by the lower intercept of the Pb loss line. The alteration is usually associated with an increase in elements not normally found at high abundance in zircon such as Ca.  I routinely measure 88Sr as a proxy for alteration when doing LA-ICPMS analyses, although normal zircon has small amounts of Sr and a peak could also correspond to an inclusion. Alteration of metamict domains is usually incomplete, suggesting that the source of the fluid is limited, perhaps from residual magmatic water. Alteration is readily recognizable, usually as an amoeboid front penetrating into the metamict phase, which is dark under both BSE and CL. Since, as shown by Nasdala et al (2006, doi: 10.2138/am.2006.2241), highly damaged zircon is bright under BSE but dark under CL, alteration is much easier to recognize under BSE. This is problematic for some communities of geochronologists who exclusively use CL. Provided one can recognize and avoid alteration, analyzing only unaltered zircon no matter how damaged should give a concordant result. This was the idea behind the air abrasion method where exposed surfaces of crack-free zircon grains are mechanically removed hopefully leaving unaltered material. Since many ancient zircon populations are extensively cracked, air abrasion usually results on a highly selected sample that may not be representative of the whole population. Mattinson (2005) solved this problem by first annealing a picked fraction to partially remove the radiation damage and then leaching in HF, using the same procedure as for dissolution but for a shorter period of time to differentially dissolve alteration. This has the advantage that alteration along internal cracks can be removed, not just on the grain surface. One problem is that the annealing process does not completely restore the crystalline structure so the previously metamict but unaltered phase will also dissolve although at a slower rate than altered zircon. Because the Raman emission from a single undamaged crystal is polarized but is unpolarized from a damaged but annealed crystal, I assume that damaged domains are reconstituted after annealing as a polycrystalline assemblage that remains relatively soluble. This can result in a great deal of mischief when applied to zoned Archean zircon. Most zircon from felsic rocks shows micron-scale oscillatory zonation presumably because of the high magma viscosity and differential diffusion rates of U vs Zr. High-U zones will become metamict but crucially, significant daughter Pb will be displaced by alpha recoil into adjacent low U, relatively undamaged  zones. Applying chemical abrasion to such a sample results on a comb structure where the high-U zones have been dissolved. The result of analysis is a reversely discordant datum because of excess Pb recoiled from the high-U zones. Because the 238U decay chin emits 8 alpha particles while the 235U decay chain emits 7, the average recoil distance of a 206Pb atom will be larger than for 207Pb, which means that the measured 207Pb/206Pb age of the reversely discordant point can be slightly younger than the true age. Cases like this make me very careful about applying the chemical abrasion to highly damaged or very ancient zircon. Chemical abrasion was originally developed to remove discordance from Phanerozoic zircon where it succeeded brilliantly. Over most of this age range one relies completely on the 206Pb/238U system and the degree of radiation damage is usually relatively limited. The leach procedure is more effective than air abrasion but should not result in extensive dissolution of the sample grain.

Although I realize that SHRIMP is used here as an analytical tool, CA treated zircon is meant to be analyzed using ID-TIMS. As noted above, discordant and concordant domains should typically coexist in the same grain and the in situ nature of SHRIMP, especially its 1 micron penetration depth, should allow alteration-free domains to be targeted based in imaging without pre-treatment of the sample.

The sampling depth of Raman appears similar to that for SHRIMP (Fig 1). Was there any reason not to perform micro-Raman spectroscopy on the spots before SHRIMP analysis? Presumably the temperature of the pit during sputtering is low enough that there is no annealing.

Line 35: ‘vanishingly’, non-scientific term.

Line 139-140: Regarding the common Pb correction in SHRIMP, there should be no significant primary common Pb in zircon, especially not for near-concordant early Archean aged grains. As I understand it, the small 204Pb signal measured by SHRIMP is derived from Pb in the Au coating, which should have a fixed isotopic composition.

Line 238-239: ‘This reconfirms that radiation damage lowers the Pb retention performance of zircon.’ This statement seems to imply that radiation damage leads to discordance, for which I see no direct evidence. It is a pre-condition necessary for low temperature Pb loss following alteration. It might have been worthwhile to search for any evidence of alteration using high resolution BSE images of the SHRIMP target areas associated with discordant data.

Line 276: ‘We know from the U-Pb data that the discordant spots in this study have lost Pb.’ This is a tautology.

Line 277: ‘…these data’.

Line 286-287: ‘Our results are hence consistent with radiation-damage-enhanced volume diffusion of Pb out of the zircon.’ Same comment as for Line 238-239 above. If radiation damage did lead to diffusive loss of Pb then it is hard for me to see how annealing damaged zones followed by leaching would consistently remove discordant domains while partially preserving those with damage. Alteration is very soluble in HF and should be selectively removed.

Fig. 1 This figure is confusing and could be composed with more care. It is hard to associate subfigures with their labels. What I presume is the scale bars look like a sub-figure.

Fig. 2 Again, more care should be put into this figure. The colours for same of the groupings make the ellipses virtually invisible. The figure contains a great deal of information that might be easier to absorb using a legend rather than having to read the caption to interpret which points are associated with which process.

Fig 3: Along with the textural maps in Fig 3, it would be useful to show high resolution CL and BSE images of both the spots and their whole grains before SHRIMP analysis to see the context of the spots.

Don Davis

I enjoyed reviewing this manuscript.

This study uses a combination of SHRIMP, Raman and APT measurements to evaluate the mechanisms of Pb loss in Zircons. It evaluates the effects of chemical abrasion including separately evaluating the impacts of annealing and partial dissolution. It is shown that all fully processed samples contain only concordant zircon while unprocessed and partially processed contain both concordant and partially discordant. The paper then evaluates potential mechanisms of homogeneous and heterogeneous Pb loss and implications for the usefulness of APT analysis of Pb distributions for evaluating the closure of the U-Pb system.

The results of this manuscript are of interest to the geochronology community and should be published after a few minor comments and suggestions as well as making the paper shorter to fit within the format of this kind of publication. In addition, the conclusion that for Zircons that have not experienced >1400 C Granulite facies metamorphism does not show Pb/Y/Al microstructures in APT data regardless of whether they are concordant or discordant and that homogeneity in Pb distribution is not equal to no Pb loss/closure is slightly underplayed by the authors and the relevance of this result could be made more strongly.

I have paid particular attention to the APT results which are of substantial interest for the use of APT to evaluate Pb disturbance in zircons.

Typically studies of Zircon also use a combination of CL and EBSD to evaluate domains within the Zircon to identify areas of alteration, zonation, strain or specific microstructures. It would be interesting to see the CL and EBSD data above the areas of discordant zircon where the APT samples failed. However, the authors already apply an impressive array of techniques to the samples studied and I am aware not all laboratories have access to all techniques, so I am not necessarily suggesting that they do additional analyses but perhaps a brief justification as to why these established techniques were not applied is appropriate or if CL maps are readily available/acquirable then adding those in would be welcome.

APT has also been used to provide U-Pb model ages of Zircon that are broadly consistent with SHRIMP/SIMS measurements using U and Pb isotope ratios, and the samples here were chosen as they should be old enough to have detectable variation in U/Pb to determine their age. However, no model ages using APT data are presented. Could the authors either add in APT model ages to show agreement with measured APT Pb isotope ratios with their SHRIMP data or state why these calculations were not possible to produce. It would also be interesting to show this data to support the author’s inference that up to 10% discordant and concordant zircon cannot be distinguished by APT.

Some sentences throughout the manuscript particularly in the introduction and discussion are missing citations where they are referencing the literature.

Line 71-73. It was not clear to me how the sentence regarding baddeleyite flows on from the preceding sentence and seems out of place. This paragraph should be refined to make the relevance of the ionization efficiency and orientation dependance in baddeleyite to the present study clear.

Line 200. It was unclear to me what constituted a successful atom probe analysis as no information was given as to dataset size. I’m not sure if supplementary materials are permitted but, if possible, it would be worth adding a table in line with Blum et al’s 2018 recommendations on the best practices for reporting atom probe analyses of geological materials.

Line 214-216. Pb isotope peaks were detectable. However, were corresponding U peaks also observable in the APT datasets? It is hinted that this is the case later in the manuscript but if so these should also be presented.

Line 216. It is very interesting that all trace elements are homogenous except for Li but might be worth explicitly mentioning Y and Al which formed clusters in previous studies. I note that the absence of Al and evaluation of Y is noted later in the discussion but may worth briefly noting in the results as well.

Line 329. ‘APT not be used for determining closure’ despite my earlier comment regarding making this point more strongly I think here it needs the caveat of APT not be used in isolation or as the sole method used for determining closure. I think it would also be worth adding a recommendation of combining APT with CL/Raman/EBSD to determine potential alteration/discordance.

Figure 3. There are clear zones within the Raman data within the SHRIMP pits of the discordant zircons, if possible, would the authors be able to show where each APT dataset was taken from within the pit and which ones failed? Were they at the interface or within one or other of the domains.

Figure 4. Add the U peaks to the figure.

Table 4. If possible, add U/Pb isotopes ratios to the table as well as model ages.

I hope these comments are of benefit to the authors

Luke Daly