Zagamiite, CaAl₂Si_3.5O₁₁, the Hexagonal High-Pressure CAS Phase with Dominant Si, as a Mineral from Mars

Round 1

Reviewer 1 Report

This is an excellent and long-awaited full description of zagamiite, a new high-pressure mineral. The paper is written well, but, in my opinion, a bit loosely (some misprints and recommendations are given below). The ms can be accepted after minor revisions.

L56

remove semicolon after the formula of liebermannite

L122

by not By (no capital letter)

L124

zagamiite, not Zagamiite (no capital letter)

L158

insert point at the end of the sentence

L182-183

“The structure represents a potentially extremely dense packing”.

In fact, “It is based upon close packing of Ca and O atoms that follow the hexagonal ABABAB = [AB]³  sequence (six layers per unit cell), but with different occupancies of the close-packed layers by Ca, which generates a threefold superstructure along the c axis” (see: DOI: 10.31857/S0869605521050038; pages 47-49). 

L262

insert comma after ‘however’

L264

‘zagamiite defines’ better: ‘zagamiite is defined’

L270

small ‘a’ after ‘:’

Author Response

Reviewer 1

This is an excellent and long-awaited full description of zagamiite, a new high-pressure mineral. The paper is written well, but, in my opinion, a bit loosely (some misprints and recommendations are given below). The ms can be accepted after minor revisions.

-> Thank you for the thorough reading and the positive comments!

L56

remove semicolon after the formula of liebermannite

-> Done

L122

by not By (no capital letter)

-> Done

L124

zagamiite, not Zagamiite (no capital letter)

-> Done

L158

insert point at the end of the sentence

-> Done

L182-183

“The structure represents a potentially extremely dense packing”.

In fact, “It is based upon close packing of Ca and O atoms that follow the hexagonal ABABAB = [AB]³  sequence (six layers per unit cell), but with different occupancies of the close-packed layers by Ca, which generates a threefold superstructure along the c axis” (see: DOI: 10.31857/S0869605521050038; pages 47-49). 

-> Yes, wording changed to ‘close packing’ -  thank you!

L262

insert comma after ‘however’

-> Done

L264

‘zagamiite defines’ better: ‘zagamiite is defined’

-> Done

L270

small ‘a’ after ‘:’

-> Done

Reviewer 2 Report

The authors conducted x-ray diffraction experiment on zagamite dominant meteorite sample. They successfully refined the structure and found it is deficient in Na. The manuscript is well-written. Discussions on crystal chemistry and formation mechanism are sufficient. By considering the following minor comments, I am happy to see it published in the near future.

The shock pressure was estimated as approximately 29 GPa. I am wondering if the authors have observed metamorphic seifertite around stishovite? Why the former was not shown in the meteorite?

The powder pattern looks poor in signal/noise ratio but the refinement is remarkably good. Figure 3 has shown ticks for zagamite and stishovite. But I could not find peaks for maskelynite, which took up 15-20% of the whole volume. Please explain the absence of maskelynite.

I have another concern about the refinement. Peaks like 104, 112, 201, etc. deviate greatly from predicted results. I guess zagamite samples have textures. Please also provide the original 2D XRD pattern in Figure 3. If they do have a lot of textures, did the authors correct them during refinement?

From the aspect of crystal chemistry, how were CA-O, Al-O and Si-O polyhedron distorted from the refined structure? This may add a short paragraph in the discussion.

Author Response

Reviewer 2

The authors conducted x-ray diffraction experiment on zagamite dominant meteorite sample. They successfully refined the structure and found it is deficient in Na. The manuscript is well-written. Discussions on crystal chemistry and formation mechanism are sufficient. By considering the following minor comments, I am happy to see it published in the near future.

Response: Thank you for the thorough reading and the very helpful comments!

The shock pressure was estimated as approximately 29 GPa. I am wondering if the authors have observed metamorphic seifertite around stishovite? Why the former was not shown in the meteorite?

Response: Seifertite was not observed in the melt pockets. In shergottites seifertite often formed via a solid-state transformation from precursor silica, not crystallized from a melt.

The powder pattern looks poor in signal/noise ratio but the refinement is remarkably good. Figure 3 has shown ticks for zagamite and stishovite. But I could not find peaks for maskelynite, which took up 15-20% of the whole volume. Please explain the absence of maskelynite.

Response: Maskelynite is shock-amorphized feldspar, a vitreous material but it is not background in the same sense as the slope of the Rayleigh-broadened primary beam and cannot well be treated as background. Eight years ago, the best options appeared to us to model maskelynite by using crystalline feldspar as starting model but allowed peak profiles to become broad enough to fit this vitreous signal, then we switched the weighted refinement of maskelynite (and only that ‘phase’!) to Pawley refinement, thus, we took it out of the Rietveld fit. The estimate of 15-20vol% comes from the initial Rietveld fit. Obviously, a structure analysis of maskelynite cannot be conducted this way but a volume estimate based on an equivalent feldspar cell is expected to be a reasonable approximation because the composition is the same and the density at least similar.

However, presently background frame subtraction with Dioptas allows us for reducing the signal from maskelynite, resin, and the glass-slide by using a diffraction image of maskelynite and subtract it pixel-per-pixel from the diffraction frame of zagamiite. Thus, diffraction profiles are not skewed (as would happen if a fitted background is removed). We repeated the Rietveld refinement and changed the text to:

‘The frame shown in Figure 3a was used for Rietveld refinement where a 2-dimensional diffraction pattern that was generated only by amorphized feldspar (‘maskelynite’), resin, and the glass substrate of the thin secion was subtracted as background frame. This pixel-by-pixel subtraction reduces the structured background and vitreous signal without distorting the Bragg peak profiles or removing crystalline diffraction signal. After integration with Dioptas [22] the remaining background was fitted with a 5^th order polynomial. Stishovite (not visible at the surface of the section at that location but present underneath zagamiite) exhibits superposition of coarser crystallites (Fig 3a). Therefore, for structure analysis and refinement of zagamiite, stishovite was first refined with the Rietveld approach, then with the Pawley approach, and finally zagamiite was refined with the Rietveld approach. Thereby, an upper limit of any potential non-powder statistical contribution from stishovite is fitted before the structure of zagamiite is analysed. The Rietveld refinement of zagamiite converged to a wRp of 6.4% (including background) and is shown in Figure 3b’

Our focus is in the assessment of the structure of zagamiite not the phase proportions, which would be better done in reflection geometry with lower diffraction volume). Thus, the relevant measure of goodness of fit is R_F of zagamiite rather than wRp. We removed the statement about the phase proportions as not relevant here: The diffraction volume sampled in transmission regions of zagamiite, zagamiite + stishovite, and maskelyite (see Fig 2).

The diffraction frame is shown in Figure 3a and the integrated and the refined modeled pattern is shown in Figure 3b.

Comment: I have another concern about the refinement. Peaks like 104, 112, 201, etc. deviate greatly from predicted results. I guess zagamite samples have textures. Please also provide the original 2D XRD pattern in Figure 3. If they do have a lot of textures, did the authors correct them during refinement?

Response: Calculated and observed |F(104)|, |F(112)|, and |F(2010| deviate by 13%, 12 %, and 6%, respecitvely. Overall R_F was 14.2%. For information and clarification we add the table of calculated and observed |F(hkl)| to the cif. Deviations from expected F’s are owed to  a) insufficient background fit (i.p. around 8 to 10 deg tth where 104 and 112 are diffracting), b) overestimation of the signal of coarser stishovite grains (see above). Texture of zagamiite was tested during Rietveld refinement but found insignificant: e.g. 006 is not well matched but 004 is well matched, 202 and 203 are well matched but 206 less so. In sum, there is no indication for systematic deviations between calculated and observed |F(hkl)|’s for particular zones. For clarification, the list of calculated and observed Fhkl is added to the cif file.

The image is added to Figure 3 (Good point – thank you!)

From the aspect of crystal chemistry, how were CA-O, Al-O and Si-O polyhedron distorted from the refined structure? This may add a short paragraph in the discussion.

Response: Thank you! This is a good idea. We added following statement to the end of ‘Results’:

‘The higher Si content shortens the T-O distances to 3x1.655 Å + 1x 1.844 Å relative to those in synthetic CAS [10] of 1.699 and 1.877 Å, respectively . Thus, we propose that in zagamiite site T is dominated by Si (not excluding minor Al but this cannot be as-sessed from the structure). The Si-O distances for site 6g (M1 in [10]) are equal to those found by Gautron et al. [10] and are unuusally long. We propose the same reason as Gautron et al. [10]: bond distances appear extended in average because of the high amount of vacancies on that site. (Al,Fe)-O distances for M2 are larger in zagamiite than in the synthetic CAS-phase by 17% and 30 %, probably because of the presence of Fe2+ in the zagamiite type material. In zagamiite the (Ca,Na)-O distances are shorter by 5% and 19% than in the CAS-phase. With ~10% Na on that site an expansion is expected. However, this site is quite anisotropic and the difference in contraction along the shorter and longer bond distance is consistent with this point. It is proposed that the actual bond coordination of Na is lower than that of Ca for this site..’

Reviewer 3 Report

The manuscript in review by Ma et al. reports on the high pressure phase of zigamiite mineral, CaAl2Si3.5O11, by combining several techniques such as FE-SEM, EBSD, EPMA and synchrotron XRD. It is of much interest to report this new mineral for the mineralogy community. The manuscript is generally well written. I have, however, several objections on the structural analysis.

I recommend the manuscript for publication after minor revision where these points should be treated appropriately:

Ø  First of all, in section 2, the information about how the XRD data was collected is scarce. When reviewing the manuscript there is no mention about the transmission geometry and if the data was collected using thin sections deposited on glass substrate, etc. Only in the section 3 the authors give information about the size of the crystals and further on that “ synchrotron diffraction patterns of zagamiite were powder-like…”.

Ø  A few more details about the XRD data analysis are also required since the XRD pattern showed in Figure 3 rises some concerns. Here, it would be recommended to show raw 2D image together with integrated pattern in order to better asses the quality of the diffraction pattern. It can be included in the supplementary materials.

Ø  The authors mention that “Maskelynite was modeled based on a compressed albite-structure with broad profiles. In the final refinement of the structure of zagamiite the F(hkl) of maskelynite were not weighted. This approach proved more robust than attempt to include maskelynite into the background”.

In which way the background was fitted in the Rietveld refinements and in this regard, one has to give the R-factors corresponding to the background and the profile function used in the refinement. Is the size broadening considered during the refinement?

Ø  There is no mention about the removed part in the XRD pattern corresponding to the 2θ ~14. It should be discussed at least in the legend of the Figure 3.  

Author Response

Reviewer 3

The manuscript in review by Ma et al. reports on the high pressure phase of zigamiite mineral, CaAl2Si3.5O11, by combining several techniques such as FE-SEM, EBSD, EPMA and synchrotron XRD. It is of much interest to report this new mineral for the mineralogy community. The manuscript is generally well written. I have, however, several objections on the structural analysis.

I recommend the manuscript for publication after minor revision where these points should be treated appropriately:

Response: Thank you for the thorough reading and the very helpful comments!

Ø  First of all, in section 2, the information about how the XRD data was collected is scarce. When reviewing the manuscript there is no mention about the transmission geometry and if the data was collected using thin sections deposited on glass substrate, etc. Only in the section 3 the authors give information about the size of the crystals and further on that “ synchrotron diffraction patterns of zagamiite were powder-like…”.

Response: Yes, this is a quite valid point. These data were collected and analysed 8 years ago. Meanwhile, we have retrieved all relevant information: The diffraction frame is added to Figure 3. We add the following information about the data collection process.

‘Synchrotron diffraction data were collected at the undulator beamline 13-IDD (GSECARS, APS, Argonne National Laboratory) using a microfocused beam (3×4 µm²) of wavelength 0.4133 Å and a MAR165 CCD area detector. Sample detector distance and geometric correction factors were determined using GSE-ADA [21]. The calibra-tion was used for integrating the diffraction data with Dioptas [22].

Data were collected in transmission on the type material thin section. A diffraction frame was taken for maskelynite through the resin-covered glass slide and used as background frame. Synchrotron diffraction patterns of zagamiite were powder-like for the given diffraction volume as shown in Figure 3a.’

Ø  A few more details about the XRD data analysis are also required since the XRD pattern showed in Figure 3 rises some concerns. Here, it would be recommended to show raw 2D image together with integrated pattern in order to better asses the quality of the diffraction pattern. It can be included in the supplementary materials.

 Response: Right. We add the diffraction frame to Figure 3. We also added to the description of data collection and analysis (see below).

Ø  The authors mention that “Maskelynite was modeled based on a compressed albite-structure with broad profiles. In the final refinement of the structure of zagamiite the F(hkl) of maskelynite were not weighted. This approach proved more robust than attempt to include maskelynite into the background”.

In which way the background was fitted in the Rietveld refinements and in this regard, one has to give the R-factors corresponding to the background and the profile function used in the refinement. Is the size broadening considered during the refinement?

Response: The original data analyses was conducted eight years ago and the fitting of maskelynite was the only way of handling these data without resorting to fitting a background with very high number of fix points or high order (we used eight fix points that time). With better tools of background frame subtraction in Dioptas we can now remove glass + resin + vitreous sample (maskelynite) better and we present this here as a more straightforward analysis.

We changed figure 3 accordingly and we added to the Description the following statement:

‘The frame shown in Figure 3a was used for Rietveld refinement where a 2-dimensional diffraction pattern that was generated only by amorphized feldspar (‘maskelynite’), resin, and the glass substrate of the thin secion was subtracted as background frame. This pixel-by-pixel subtraction reduces the structured background and vitreous signal without distorting the Bragg peak profiles or removing crystalline diffraction signal. After integration with Dioptas [22] the remaining background was fitted with a 5^th order polynomial. Stishovite exhibits superposition of coarser crystallites (Fig 3a). Therefore, for structure analysis and refinement of zagamiite, stishovite was first refined with the Rietveld approach, then with the Pawley approach, and finally zagamiite was refined with the Rietveld approach. Thereby, an upper limit of any potential non-powder statistical contribution from stishovite is fitted before the structure of zagamiite is analysed. The Rietveld refinement of zagamiite converged to a wRp of 6.4% (including background) and is shown in Figure 3b’

Background is included in the weighted refinement factor wRp. Background was fitted with a 5^th order polynomial. We changed the text accordingly.

For data with high or complicated background R_F is more informative than wRp. We provide both parameters. We added the list of extracted observed |F(hkl)| and calculated ones in the cif. We also added the profile terms to the cif in a comment line. With the given grainsize (Figure 2b), X-ray energy, and diffraction volume no size-related broadening is observed. This is indirectly seen by the rather small values of the Cagliotti terms. Note that the point spread of the MAR CCD causes comparably larger instrumental U,V,W, than modern hybrid pixel array detectors for the same high X-ray energy of 30 keV, so  the observed profiles are largely instrumental for the present set of data.

Ø  There is no mention about the removed part in the XRD pattern corresponding to the 2θ ~14. It should be discussed at least in the legend of the Figure 3.  

Response: There is nothing removed or excluded: See the diffraction image in Figure 3a. This Q-range does not exhibit marked diffraction of zagamiite or stishovite.