The  paper report the effects of annealing and HF partial dissolution experiments on four different zircon samples of different age and radiation damage. With the help of various imaging and measuring techniques the changes due to annealing and partial dissolution are monitored and form the basis for a discussion of the factors that control the geometry of the partial dissolution processes and implications for the use of this mineral in geochronology. The paper provides a systematic insight into the behaviour of zircon during the widely applied chemical abrasion procedure, and will be a welcome addition to the literature on the subject.

Overall the paper is well prepared and structured. I find, however, that the present version could be improved by a good weeding of unnecessary words and sentences. Some expressions that the authors like to use frequently should be reconsidered. One example is ‘compositional zone’. I have made some suggestions in the file and discuss specific points below.

Although the results are very relevant for zircon U-Pb geochronology, the studied zircons have not themselves been dated, and so the specific data do not have a direct connection with the U-Pb behavior of the samples. Consequently, the discussion of implications for U-Pb dating presented in the paper is very generalized and in part trivial. I suggest to concentrate on the main substance of the study and avoid meandering off in inefficient discussions.

Some specific points:

Line 64;  ‘… poorly understood, and several outstanding questions remain. Do most zircon crystals predominantly dissolve from rim to core?’

Everyone who has done chemical abrasion will have noted very rapidly that zoned and metamict zircons do not dissolve that way. That can hardly be described as a poorly understood fundamental question.

Fig.2: The explanations are unclear. Are the grains shown in each of the four panels representing each of the sample? If yes, why does the second panel contain zircon from two different samples (blue and yellow circles). If not, what is the distinction? Overall I find this figure quite useless, not even as a decoration.

Fig.3:  Not sure about using the term ‘metamict’ once zircon has been annealed.  Can be confusing.

209:  ‘typical magmatic growth patterns’   It is true that many magmatic zircons have oscillatory zoning, but the same pattern can in part be seen in metamorphic zircon. Better to use descriptive terms.

287: ‘ This could imply that increasing the duration of the leaching step results in a more crystalline zircon residue due to the progressive dissolution of higher damage domains.’ Why the ‘could’? It seems to be the most logical explanation.

And then the following sentence; ‘We note, however, that only a small number of AS3 crystals survived 12 h of chemical abrasion, and only a small number of Raman analyses were made. We recommend further study to better evaluate this possibility.’ Sounds rather trivial.

316: ‘As evidenced by our SE images and discussed further below, μCT does not capture radiation damage zoning that does not result in a strong density contrast such as variations in radiation damage below the ~1×1018 α/g threshold.’ Suggest rephrasing to avoid the double negatives.

Fig. 9, caption: ‘interior compositional zone’. Compositional zone? As opposed to what?

367: ‘ All observed dumbbells are oriented parallel to the grain’s c-axis’. That is a surprising statement. Looking at the figures I would have assumed that they are all normal to the elongation of the crystal (= c-axis). Please elaborate to avoid confusion

Fig 13  caption: ‘The yellow arrow highlights the grain’s shell-like appearance because of significant dissolution in the grain’s interior.’ I see a highly resorbed grain, not much left of the shell.

‘images of dog-chewed zircon residues ‘  maybe a bit too colloquial?

419: ‘In a visual game of connect-the-dots …’     ??

422: ‘We see dumbbell-like fracture patterns again in sample Zr36 (Fig. 13b-III) where crosscutting fractures connect different oscillatory zones removed by dissolution to one another and to the grain surface’.  Rather convolute sentence, hard to understand. Please rephrase.

483: ‘… The long axes of deep, octahedral etch pits on (100) align with the crystal’s c-axis…’

I wonder why they are called octahedral. Those in Fig 16a look prismatic to me.

Fig. 19, caption: ‘ Projection on (100) looking down the a-axis. The c-axis is vertical to the page, and the a2-axis is horizontal’ Confusing: should it not be: ‘Projection on (100)

looking down the c-axis’ ? The same for (b): if the pane is 001 then the view must be parallel to a?  Or not?

640: ‘… suggest that a crystal’s bulk radiation damage also plays an important role.’  That seems a rather trivial discovery. What else could one expect?

641: ‘Crystal morphology plays a lesser role in that crystals with very high aspect ratios dissolve more slowly than more equant grains’   ??? That is not apparent in Fig. 18e, and would seem to contradict the higher solubility along the c-axis than along a.

Section ‘4.2.1 Zircon U-Pb ages’ lines 648 – 726: I find the discussion on this topic very generalized and superfluous. The experiments in this paper demonstrate the great variability between zircon of different geneses, compositions and ages. Yet many of the reflections made here focus on some idealized magmatic zircon in young systems. Because of such a simplification the discussion is trivialized and almost meaningless. Clearly, the lessons from the present experiments must be considered separately for each set of zircon used for geochronology. I recommend removing this section, it just detracts from the paper.

 778: ‘…chemically abraded residues are more crystalline than their annealed counter parts…’: Isn’t that the logical relationships since CA removed the less crystalline domains? And: ‘… radiation damage is annealed hydrothermally during HF leaching…’. Speculative?

867: ‘Increasing the leaching temperature from 180 °C to 210 °C or increasing the leaching duration leads to the development of more extensive dissolution networks in higher damage grains ‘  This is a rather trivial conclusion. Something CA-users did not observe before?

870: ‘More crystalline zircon samples lack fracturing related to radiation damage zoning.’ Another trivial statement. The conclusions should focus on the important aspects of the research, not on trivialities.

Aug. 13, 2022   F. Corfu

Comments on the manuscript “Chemical abrasion: The mechanics of zircon dissolution” by Alyssa J. McKanna and co-authors, submitted for publication in Geochronology (#gchron–2022–19)

Lutz Nsdala

Generalities:

This is an interesting contribution whose results deserve to be brought to the attention of the international community. During the last 15 years, several teams have undertaken extensive dry-annealing experiments in order to understand why applying Mattinson´s CA–TIMS technique is advantageous in many cases; that is, why dry annealing prior to HF etching improves the selective removal of more radiation-damaged zircon during HF etching? The present study focuses more on the second step: What exactly happens during the HF attack? Observations such as preferred dissolution and preferred volume shrinking along a certain crystallographic direction, and the obvious contribution of fluid-driven recovery processes, are novel and valuable. The results are of interest and for the most part sound, and I am happy to suggest acceptance for publication after minor (mostly technical) revision.

Main comments:

The main shortcoming of the present research is perhaps that there is no chemical information on the samples studied. The present paper presents results of rather sophisticated studies including micro-CT and precise dimension estimates, but the most basic information remains hidden. There are quite a few journals that would refuse to publish a study on samples whose chemical compositions are unknown; for good reasons, as I think.

In particular, there are no U and Th concentrations available that, along with U–Pb ages, could have been used to calculate realistic time-integrated alpha doses. Instead, authors fiddle with equation #2 of Váczi and Nasdala (2017) in attempting to estimate self-irradiation doses from Raman FWHMs. Which is based on a serious fallacy of thinking: The equation of Váczi and Nasdala (2017) – that by the way has never been proposed as a “calibration”, as present authors claim in line 223 – refers solely to Sri Lankan gem zircon with its particular damage-annealing history, whereas other zircon populations may have appreciably different FWHM-dose relationships.

In contrast, comparing alpha doses that were calculated from U, Th and age, with the observed FWHM values, would have opened up valuable opportunities for characterising the initial samples and their annealing histories. This chance was missed. Also, because of the unavailability of chemical compositions, it cannot be evaluated whether HF etching has removed completely some interiors of the grains but has left the rest fully unchanged, or did un-dissolved remnants experience some chemical leaching?

Minor comments:

- Present authors rely on the results of Palenik et al. (2003) and Váczi and Nasdala (2017) who both have claimed that the FWHM of the n₃(SiO₄) Raman band of Sri Lankan zircon has a maximum (“saturation”) value of about 35 cm^–1. On the other hand, sample AS3 yielded FWHMs of about 48 cm^–1 (Fig. 5a, Table S1). This apparent contradiction should be discussed, and actually such extremely broadened Raman spectra should be shown (at least in a supplementary figure).

It has been suspected that Sri Lankan zircon of elevated degrees of radiation damage have experienced preferred annealing, which might explain the “saturation” as a particular feature of Sri Lankan zircon. In conclusion, it is not surprising that other zircon, not affected by the particular “Sri Lankan” annealing history, may indeed show further band broadening. This has been rarely observed thus far but is valuable; so it should been shown and discussed.

- Line 183, it is not really of relevance for the reader to learn about the equipment of the Princeton Raman system; instead, it should merely be stated which particular laser was actually used in the present study. Presumably red, as the green laser tends to induce Er³⁺-related photoluminescence obscuring the Raman spectrum? The laser power at the sample surface (so not laser output but power measured behind the objective) needs to be reported.

- Line 185: I wished Horiba would stop implementing the main silicon band in their auto-calibration procedure. However, if authors believe the friendly Horiba service person who did so was right, they should provide a published reference for the 520.7 cm­^–1 value they used for calibration. Other authors have used 520.5 or 521.0 cm^–1, and none of these values has ever been appropriately supported. Note that the oriented Si wafer is provided for alignment purposes but pretty unsuitable for wavenumber calibration: Visible light penetrates only a few µm into Si, which is why increasing the laser power, or changing the laser colour, changes absorption-induced heating and causes a shift of the Raman band. Why not using a truly known signal instead? The Rayleigh line has a Raman shift of 0 cm^–1; a perfect calibration means.

- Line 189–190, quoting ~2 µm spatial resolution upon using a “400 to 100 µm confocal pin hole” is too general and imprecise. With the confocal hole set to 400 µm, the spatial resolution will be much poorer than 2 µm lateral, and with a 100 µm confocal hole, the LabRAM Evolution should – provided the beam alignment is well done – reach its maximum performance close to the Rayleigh or Sparrow criterions, well below 1 µm (see detailed study by Kim et al., 2020, Current Applied Physics 20:71–77). The above resolution estimate does of course only apply if the Olympus 100× / n.a. 0.90 objective was used (this essential information is missing as well).

- Page 8, caption of Fig. 3a: The wording “SEM images…” is too imprecise. State explicitly which images are CL, BSE, SE.

- Lines 218–219, the expression “Si–O asymmetric stretching band” is a contradiction in itself. To vibrate asymmetrically, it takes at least two bonds, whereas the stretching vibration of a single Si–O bond does not have any symmetry. Should be reworded to “asymmetric SiO­₄ stretching”.

- Fig. 18, in subfigures a–d the main x-axis label is missing.

- Lines 768–769, “…suggest that annealing or radiation damage is not simply the inverse of damage accumulation”; and again lines 891–892, “Raman analyses of annealed grains suggest that dry annealing is not the inverse of radiation damage accumulation”. These statements are presented as if they were a result of the present study, which is decidedly not the case. In contrast, it has been emphasised may times already during the last 20 years that annealing does not inverse the damage accumulation process (mismatch of Raman FWHMs and Raman shifts, mismatch of unit-cell parameters a and c, etc.). Something like “our observations confirm again that…” would be more appropriate and more honest.

- Same in lines 250–252, “We note that none of the samples have achieved complete structural recovery after annealing at 900 °C for 48 h, since all measured peak widths are broader than that of synthetic zircon”. This is not a new finding but merely another well-known fact, here merely reconfirmed again. People who wish to achieve near-complete annealing of zircon do this at 1400°C for four days or so.

By the way, Raman analyses do not suggest anything; only the results of Raman analyses suggest something (wording in line 891 should be improved).

-Line 800, a typo? Instead of “verses” there should be “versus”.

- Line 983, the second author is Dr Chutimun (first name) Chanmuang N. (family name), so her name should be abbreviated “Chanmuang N., C.”; same in line 1012. Similarly, in line 1050 the second author should be quoted as “Van den haute, P.”

- Table S1, the “empirical correction” of Raman shift values for band downshifts that are claimed to be caused by laser-induced heating remains fully vague (unclear). Something like this has never been observed and reported before; so at least some comprehensible documentation is needed. How about simply decreasing the laser-power density, to reduce laser-induced heating to a negligible extent??

- Table S1, in view of the fact that many authors present and discuss uncorrected FWHMs, applying FWHM correction in the present article is meritorious. However, a correction assuming a FWHM of the IPF (instrumental profile function) of 1.5 cm^–1 (I assume this is the meaning of the unitless “1.5” in footnote c?) most likely has led to an overcorrection. The FWHM of the IPF of the LabRAM Evolution with its 800 mm focal length, in case the 1800 grooves per millimetre diffraction grating is used, is about 0.8 cm^–1 in the red and about 1 cm^–1 in the green range of the electromagnetic spectrum. Authors should test this – and hence the experimental band broadening of their system – by measuring the width of the Rayleigh line, Kr lamp or similar emission lines, or even better narrow Raman bands of known widths (unstressed diamond, Ba nitride).

Unfortunately I was unable to find Raman shifts and corrected FWHMs of synthetic zircon as measured by present authors. However, data points in Figs. 5 and 6 imply that synthetic zircon would have a FWHM of the n₃(SiO₄) band of about 1 cm^–1, which is well below the real value of about 1.7–1.8 cm^–1. This seems to support my above suspicion of FWHM overcorrection. Let me assume authors have fitted an FWHM of 2.1 cm^–1. This value, corrected for a (too large) 1.5 cm^–1 IPF, yields an overcorrected FWHM of 1.0 cm^–1. In contrast, a fitted FWHM of 2.1 cm^–1, corrected for a 0.8 cm^–1 IPF, yields a quite realistic FWHM of 1.8 cm^–1. Anyway, data for synthetic zircon should be added to Table S1.

- Table S1: Reporting FWHM values with two decimal digits fakes an unrealistic accuracy. As the uncertainty is 10% for narrow FWHMs, reporting 4.4 instead of 4.38 cm^–1 will do the job.