Introduction
Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) describe the new mineral natromelansonite, Na3Zr(Si7AlO19)⋅4–5H2O, from the Poudrette (Demix) Quarry, Mont Saint-Hilaire, Québec, Canada, and compare it to the related species melansonite, (Na,□)□2KZrSi8O19⋅5H2O, from the same locality. It is the intent of the present comment to address concerns raised regarding the behaviour of Al and Na in both minerals.
The rhodesite group
Both melansonite and natromelansonite belong to the rhodesite mero-plesiotype series, which includes rhodesite, delhayelite, hydrodelhayelite, macdonaldite, monteregianite-(Y), seidite-(Ce), fivegite, melansonite and natromelansonite. In addition to the naturally occurring members, several synthetic compounds also fit this classification (Cadoni and Ferraris, Reference Cadoni and Ferraris2009, Reference Cadoni and Ferraris2010). These minerals are all mesoporous silicates, in which the structures are all based on framework arrangements. Common crystal-structure features amongst members of this series include:
(1) An infinite double-sheet silicate layer composed of eight- and four-membered rings, reminiscent of those seen in members of the apophyllite group.
(2) Continuous channels occupied by low-charge cations coordinated in part by OH groups, H2O groups, and halogens (with coordinations > 6).
(3) Alternating octahedral–tetrahedral layers producing a TOT topology reminiscent of that occurring in phyllosilicates.
The octahedral layer is of particular interest as it is extremely versatile and so is able to accommodate a wide variety of cations of various charges (Na, Ca, Sr, Y, REE, Zr and Ti). This, and to a lesser extent variability amongst the channel occupants, results in the chemical diversity exhibited amongst members of this series.
The role of Al
The aforementioned double-sheet silicate layer forms the backbone of the structure and is comprised exclusively of corner-sharing tetrahedra. Though Si dominates the tetrahedra in this layer, many species in this series, namely delhayelite, hydrodelhayelite, fivegite, melansonite and natromelansonite, have been reported with non-trivial concentrations of Al which has been invariably partitioned into branching tetrahedra that serve to connect the two layers of tetrahedra that constitute the double sheet. The occurrence of Al in these branching tetrahedra is also observed in the crystal structures for melansonite and natromelansonite.
Both melansonite [avg (range): 6.32 (5.36–6.79) wt.% Al2O3, n = 15] and natromelansonite [6.89 (6.35–7.28) wt.% Al2O3, n = 8] contain approximately equal concentrations of Al, these representing ~1 atom per formula unit (apfu) when calculated on the basis of 19 anions (water excluded). However, the significance of the Al was considered differently by the studies describing the two individual minerals: while Gore and McDonald (Reference Gore and McDonald2023) consider Al to be non-essential in melansonite, Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) consider it to be essential in the related species natromelansonite, a fact reflected in the proposed ideal chemical formula for the mineral. There is no dispute that Al is present in both minerals and that it is partitioned into specific tetrahedral sites, specifically those related to branching tetrahedra. This approach was based on the greater average bond lengths in comparison to the other Si–O bonds in the structure and their corresponding bond valence sums of 3.63 and 3.65 valence units (vu) in melansonite and natromelansonite, respectively. Though consideration was given to attributing a mixed occupancy (Al0.5Si0.5) to the branching Si(3) site in the description of melansonite, ultimately, the decision was made that there were insufficient data to definitively prove this, primarily owing to the similarity in the X-ray scattering factors for Al and Si. As such, the site was treated as being Si dominant, i.e. Al was not considered as being essential to the mineral and herein lies the point of contention. Though the question of whether Al is essential or not may appear trivial to some, it does precisely underpin the question of what the mineral is, which in turn directly relates to the conditions necessary for its formation. For example, one could ask the question that if an attempt were made to synthesise this mineral in an environment devoid of Al, could melansonite still be formed? It is our contention that if Al is not considered essential, then under the correct conditions the answer to this question would be yes. By extension, if one equally considers Al to be non-essential to natromelansonite, then it too should be possible to synthesise the mineral under Al-free conditions. Also relevant to this communication is the fact that the data provided by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) for U-bearing melansonite indicate only 0.62 apfu Al, well below the 1 apfu suggested by their revisions to the ideal formula and lower than all other published data for Al in these minerals: 1.02 in natromelansonite and 0.93 in melansonite, both from Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), and 0.99 in melansonite, Gore and McDonald (Reference Gore and McDonald2023). These data clearly indicate variability in the concentration of Al in melansonite, supporting our interpretation that Al should be considered as a non-essential element. It should also be noted that in table 1 of Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), the mean value for Al2O3 is given as 6.29 wt.%, despite the range being listed as 5.53–5.89 wt.%. This discrepancy requires clarification to properly ascertain the concentration of Al in the melansonite observed in this study.
The role of Na
The analytical data given for Na in melansonite in our original study was low compared to that calculated from refinement of the crystal structure. This was attributed to a combination of migration of Na under the electron beam during analysis and to the extremely thin nature of the crystals (<2 μm). Both of these points were noted by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024); importantly demonstrating the loss of intensity in NaKα as a function of time in natromelansonite. The latter suggests that the migration of Na in melansonite, as suggested by Gore and McDonald, is thus likely to be an issue. Subsequent refinement of the crystal structure also indicated the Na(2) site to be only partially occupied. This, coupled with the need to achieve electroneutrality in the mineral, accomplished through the generation of vacancies via the coupled substitution Y3+ + Na+ ↔ Zr4+ + □ [when compared to monteregianite-(Y)] was the reason that the occupancy factor of this site was fixed at 28%. The *.cif for melansonite is available from the Depository of Unpublished Data on the MAC website (https://www.mineralogicalassociation.ca/depository/), as was noted in our original publication.
The new sample of melansonite studied by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) provided an opportunity to collect additional chemical information on a larger grain of higher quality, mitigating some of the challenges faced in the original study of the mineral. The resultant data show a mean value of 8.20 wt.% Na2O compared to our observed 2.82 wt.% Na2O (with the ideal formula originally proposed for melansonite requiring 4.01 wt.% Na2O). Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) also give a mean value for K2O of 2.48 wt.%, which compares well with our reported value of 2.64 wt.% (the ideal formula proposed for melansonite requiring 6.10 wt.%). These lower-than-ideal K values are unsurprising owing to the Na-rich environment in which natromelansonite crystallised. It also indicates that a portion of this Na is probably substituting for K in melansonite observed in both the holotype material and the sample presented in Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), preserving the occurrence of vacancies in the Na(2) site. The vacancy-dominant nature of the Na(2) site in melansonite is the mechanism that was invoked for charge-balance requirement of the mineral, relative to that in the related species monteregianite-(Y) (the latter requiring the coupled substitution Y3+ + Na+ ↔ Zr4+ + □). While we recognise that Si ↔ Al substitution occurs in melansonite, we continue to support our contention that the dominant mechanism for charge-balance control in melansonite is the generation of vacancies, primarily with respect to the Na content in the Na(2) site, and to a lesser extent the K content. It is also noteworthy that the U-bearing melansonite examined by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) has a lower Na content than melansonite observed in the same study, this being a mean value of 6.46 vs. 8.20 wt.%, respectively. This is coupled with a lower Al content in the sample (as mentioned above) which further corroborates the idea that the generation of vacancies at the expense of Na is the dominant mechanism through which melansonite maintains electroneutrality.
Closing remarks
The research conducted by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), and their discovery of natromelansonite has provided valuable insight into the crystal chemistry of rhodesite mero-plesiotype series minerals in agpaitic systems. It has also shown that our understanding of the crystal chemistry of these minerals is incomplete and in some instances contradictory, particularly with respect to the roles that Al and Na play in their structure and stability. It has also continued to exemplify that this series is extremely versatile (growing from five members in 2003 to nine today) and that there are probably many additional members that will be discovered in time, as was initially proposed by Hesse et al. (Reference Hesse, Liebau and Merlino1992). A comprehensive crystal-chemical study of the series as a whole would serve to elucidate the less-well understood aspects of all members and would provide more satisfactory answers to the questions raised in the most recent publications.
You have
Access
Introduction
Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) describe the new mineral natromelansonite, Na3Zr(Si7AlO19)⋅4–5H2O, from the Poudrette (Demix) Quarry, Mont Saint-Hilaire, Québec, Canada, and compare it to the related species melansonite, (Na,□)□2KZrSi8O19⋅5H2O, from the same locality. It is the intent of the present comment to address concerns raised regarding the behaviour of Al and Na in both minerals.
The rhodesite group
Both melansonite and natromelansonite belong to the rhodesite mero-plesiotype series, which includes rhodesite, delhayelite, hydrodelhayelite, macdonaldite, monteregianite-(Y), seidite-(Ce), fivegite, melansonite and natromelansonite. In addition to the naturally occurring members, several synthetic compounds also fit this classification (Cadoni and Ferraris, Reference Cadoni and Ferraris2009, Reference Cadoni and Ferraris2010). These minerals are all mesoporous silicates, in which the structures are all based on framework arrangements. Common crystal-structure features amongst members of this series include:
(1) An infinite double-sheet silicate layer composed of eight- and four-membered rings, reminiscent of those seen in members of the apophyllite group.
(2) Continuous channels occupied by low-charge cations coordinated in part by OH groups, H2O groups, and halogens (with coordinations > 6).
(3) Alternating octahedral–tetrahedral layers producing a TOT topology reminiscent of that occurring in phyllosilicates.
The octahedral layer is of particular interest as it is extremely versatile and so is able to accommodate a wide variety of cations of various charges (Na, Ca, Sr, Y, REE, Zr and Ti). This, and to a lesser extent variability amongst the channel occupants, results in the chemical diversity exhibited amongst members of this series.
The role of Al
The aforementioned double-sheet silicate layer forms the backbone of the structure and is comprised exclusively of corner-sharing tetrahedra. Though Si dominates the tetrahedra in this layer, many species in this series, namely delhayelite, hydrodelhayelite, fivegite, melansonite and natromelansonite, have been reported with non-trivial concentrations of Al which has been invariably partitioned into branching tetrahedra that serve to connect the two layers of tetrahedra that constitute the double sheet. The occurrence of Al in these branching tetrahedra is also observed in the crystal structures for melansonite and natromelansonite.
Both melansonite [avg (range): 6.32 (5.36–6.79) wt.% Al2O3, n = 15] and natromelansonite [6.89 (6.35–7.28) wt.% Al2O3, n = 8] contain approximately equal concentrations of Al, these representing ~1 atom per formula unit (apfu) when calculated on the basis of 19 anions (water excluded). However, the significance of the Al was considered differently by the studies describing the two individual minerals: while Gore and McDonald (Reference Gore and McDonald2023) consider Al to be non-essential in melansonite, Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) consider it to be essential in the related species natromelansonite, a fact reflected in the proposed ideal chemical formula for the mineral. There is no dispute that Al is present in both minerals and that it is partitioned into specific tetrahedral sites, specifically those related to branching tetrahedra. This approach was based on the greater average bond lengths in comparison to the other Si–O bonds in the structure and their corresponding bond valence sums of 3.63 and 3.65 valence units (vu) in melansonite and natromelansonite, respectively. Though consideration was given to attributing a mixed occupancy (Al0.5Si0.5) to the branching Si(3) site in the description of melansonite, ultimately, the decision was made that there were insufficient data to definitively prove this, primarily owing to the similarity in the X-ray scattering factors for Al and Si. As such, the site was treated as being Si dominant, i.e. Al was not considered as being essential to the mineral and herein lies the point of contention. Though the question of whether Al is essential or not may appear trivial to some, it does precisely underpin the question of what the mineral is, which in turn directly relates to the conditions necessary for its formation. For example, one could ask the question that if an attempt were made to synthesise this mineral in an environment devoid of Al, could melansonite still be formed? It is our contention that if Al is not considered essential, then under the correct conditions the answer to this question would be yes. By extension, if one equally considers Al to be non-essential to natromelansonite, then it too should be possible to synthesise the mineral under Al-free conditions. Also relevant to this communication is the fact that the data provided by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) for U-bearing melansonite indicate only 0.62 apfu Al, well below the 1 apfu suggested by their revisions to the ideal formula and lower than all other published data for Al in these minerals: 1.02 in natromelansonite and 0.93 in melansonite, both from Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), and 0.99 in melansonite, Gore and McDonald (Reference Gore and McDonald2023). These data clearly indicate variability in the concentration of Al in melansonite, supporting our interpretation that Al should be considered as a non-essential element. It should also be noted that in table 1 of Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), the mean value for Al2O3 is given as 6.29 wt.%, despite the range being listed as 5.53–5.89 wt.%. This discrepancy requires clarification to properly ascertain the concentration of Al in the melansonite observed in this study.
The role of Na
The analytical data given for Na in melansonite in our original study was low compared to that calculated from refinement of the crystal structure. This was attributed to a combination of migration of Na under the electron beam during analysis and to the extremely thin nature of the crystals (<2 μm). Both of these points were noted by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024); importantly demonstrating the loss of intensity in NaKα as a function of time in natromelansonite. The latter suggests that the migration of Na in melansonite, as suggested by Gore and McDonald, is thus likely to be an issue. Subsequent refinement of the crystal structure also indicated the Na(2) site to be only partially occupied. This, coupled with the need to achieve electroneutrality in the mineral, accomplished through the generation of vacancies via the coupled substitution Y3+ + Na+ ↔ Zr4+ + □ [when compared to monteregianite-(Y)] was the reason that the occupancy factor of this site was fixed at 28%. The *.cif for melansonite is available from the Depository of Unpublished Data on the MAC website (https://www.mineralogicalassociation.ca/depository/), as was noted in our original publication.
The new sample of melansonite studied by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) provided an opportunity to collect additional chemical information on a larger grain of higher quality, mitigating some of the challenges faced in the original study of the mineral. The resultant data show a mean value of 8.20 wt.% Na2O compared to our observed 2.82 wt.% Na2O (with the ideal formula originally proposed for melansonite requiring 4.01 wt.% Na2O). Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) also give a mean value for K2O of 2.48 wt.%, which compares well with our reported value of 2.64 wt.% (the ideal formula proposed for melansonite requiring 6.10 wt.%). These lower-than-ideal K values are unsurprising owing to the Na-rich environment in which natromelansonite crystallised. It also indicates that a portion of this Na is probably substituting for K in melansonite observed in both the holotype material and the sample presented in Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), preserving the occurrence of vacancies in the Na(2) site. The vacancy-dominant nature of the Na(2) site in melansonite is the mechanism that was invoked for charge-balance requirement of the mineral, relative to that in the related species monteregianite-(Y) (the latter requiring the coupled substitution Y3+ + Na+ ↔ Zr4+ + □). While we recognise that Si ↔ Al substitution occurs in melansonite, we continue to support our contention that the dominant mechanism for charge-balance control in melansonite is the generation of vacancies, primarily with respect to the Na content in the Na(2) site, and to a lesser extent the K content. It is also noteworthy that the U-bearing melansonite examined by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024) has a lower Na content than melansonite observed in the same study, this being a mean value of 6.46 vs. 8.20 wt.%, respectively. This is coupled with a lower Al content in the sample (as mentioned above) which further corroborates the idea that the generation of vacancies at the expense of Na is the dominant mechanism through which melansonite maintains electroneutrality.
Closing remarks
The research conducted by Lykova et al. (Reference Lykova, Rowe, Poirier, Friis and Barnes2024), and their discovery of natromelansonite has provided valuable insight into the crystal chemistry of rhodesite mero-plesiotype series minerals in agpaitic systems. It has also shown that our understanding of the crystal chemistry of these minerals is incomplete and in some instances contradictory, particularly with respect to the roles that Al and Na play in their structure and stability. It has also continued to exemplify that this series is extremely versatile (growing from five members in 2003 to nine today) and that there are probably many additional members that will be discovered in time, as was initially proposed by Hesse et al. (Reference Hesse, Liebau and Merlino1992). A comprehensive crystal-chemical study of the series as a whole would serve to elucidate the less-well understood aspects of all members and would provide more satisfactory answers to the questions raised in the most recent publications.
Competing interests
The authors declare none.