Natromelansonite, Na₃Zr[Si₇AlO₁₉]⋅4–5H₂O, a new member of the rhodesite mero-plesiotype series from Mont Saint-Hilaire, Quebec, Canada

Abstract

Natromelansonite, Na₃Zr[Si₇AlO₁₉]⋅4–5H₂O, was found at the Poudrette (Demix) quarry, Mont Saint-Hilaire, Quebec, Canada in a highly altered pegmatite together with a clay mineral, steacyite, polylithionite and rhodochrosite. It occurs as an outer zone of tabular crystals to 0.1 × 0.3 × 1 mm in size flattened on (001). The inner zone is made of melansonite. The mineral is grey with white powder colour and vitreous lustre. The cleavage is parallel to {010}, perfect. The Mohs hardness is 3.5. The mineral has green fluorescence under short-wave ultraviolet light. The D_calc is 2.31 g/cm³. The infrared spectrum is reported. The composition (wt.%, average of 8 analyses) is Na₂O 10.08, K₂O 1.72, CaO 0.24, BaO 0.32, MnO 0.10, Al₂O₃ 6.86, Y₂O₃ 0.35, Yb₂O₃ 0.55, SiO₂ 56.23, ZrO₂ 14.78, H₂O 9.52, total 100.75. The empirical formula calculated on the basis of O = 23 apfu and H = 8 apfu is Na(□Na_0.38Ca_0.02Mn_0.01)₂(Na_0.70K_0.28Ba_0.02)_Σ1.00(Zr_0.91Y_0.02Yb_0.02)_Σ0.95(Si_7.08Al_1.02)_Σ8.10O₁₉⋅4H₂O. The mineral is monoclinic, P2₁/m, a = 6.5156(3) Å, b = 24.061(1) Å, c = 6.9759(6) Å, β = 90.453(5)° and V = 1093.61(9) ų and Z = 2. The strongest reflections of the powder X-ray diffraction pattern [d,Å(I)(hkl)] are: 12.02(100)(020), 6.97(89)(001), 6.51(39)(100), 3.416(37)(160), 3.062(42)($\bar{1}$61, 102, 161 and $\bar{1}$12), 3.018(38)(230 and 042), 2.864(40)(240 and 132). The crystal structure, solved and refined from single-crystal X-ray diffraction data (R₁ = 0.042), is based on a double sheet of tetrahedra (T) and a sheet of octahedra (O) that alternate along the [010] direction forming a TOT structure typical for members of the rhodesite mero-plesiotype series.

Introduction

Minerals of the rhodesite mero-plesiotype series (Ferraris and Gula, 2005; Cadoni and Ferraris, 2011), are often found in agpaitic and hyperagpaitic environments. One of the members of the series monteregianite-(Y), KNa₂Y[Si₈O₁₉]⋅5H₂O, has only been found at the famous alkaline igneous complex Mont Saint-Hilaire, Quebec, Canada (Chao, 1978; Ghose et al., 1987). It usually occurs in miarolitic cavities in igneous breccia and marble xenoliths (Horvath et al., 2019, and references therein) as tabular-bladed crystals. Therefore, when we came across a specimen from Mont Saint-Hilaire with what appeared to be typical monteregianite-(Y) crystals but from a highly altered pegmatite it immediately attracted our attention. Its examination showed that the crystals are zoned and composed of melansonite, KNa₂Zr[Si₇AlO₁₉]⋅4–5H₂O, which previously was found in only one specimen from a contact between a marble xenolith and the host nepheline syenite at Mont Saint-Hilaire (Gore and McDonald, 2023) and the new species natromelansonite, ideally Na₃Zr[Si₇AlO₁₉]⋅4–5H₂O. Natromelansonite is named as the Na-analogue of melansonite.

Both the new mineral and the name (symbol Nmso) have been approved by the Commission on New Minerals, Nomenclature and Classification of the International Mineralogical Association, proposal IMA 2023-076 (Lykova et al., 2024). The holotype specimen of natromelansonite (catalogue number CMNMC 90813) is from the collection of the Canadian Museum of Nature, Canada, and was originally labelled as “monteregianite-(Y)”. A part of the holotype used for structure determination was deposited at the Natural History Museum in Oslo under catalogue number KNR 44447.

Occurrence and general appearance

Natromelansonite was found at the Poudrette quarry (Demix quarry), Mont Saint-Hilaire, Quebec, Canada in a highly altered pegmatite in nepheline syenite (Horvath et al., 2019, and references therein). It occurs on rhodochrosite together with polylithionite and beige aggregates of poorly diffracting clay mineral and steacyite microcrystals. The specimen was collected by Gilles Haineault in 2004.

Natromelansonite occurs as an outer zone of tabular crystals up to 0.1 × 0.3 × 1 mm in size (Fig. 1) flattened on (001) where the inner zone is melansonite. The boundary between the two species is sharp and marked by a thin zone of uranium-bearing melansonite (Fig. 2). The observed crystal forms are dominant pinacoid {001} with two other pinacoids {001} and {100}.

Figure 1. Grey tabular prismatic crystals composed of natromelansonite and melansonite on a beige aggregate of a poorly diffracting clay mineral and steacyite microcrystals. Specimen CMNMC 90813. Photo: François Génier.

Figure 2. Back-scattered electron image of a crystal composed of melansonite and natromelansonite zones. Bright zone – U-bearing melansonite. Polished section, specimen CMNMC 90813.

Physical and optical properties

Natromelansonite is grey with a white powder colour and vitreous lustre (Fig. 1). The cleavage is parallel to {010}, perfect. The fracture is stepped. The Mohs hardness is 3.5. The mineral has green fluorescence under short-wave ultraviolet light. The density calculated using the empirical formula and unit-cell volume refined from the single-crystal X-ray diffraction (XRD) data is 2.31 g/cm³.

Natromelansonite is optically biaxial (+), α = 1.510(2), β = 1.516(2), γ = 1.524(3), 2V(meas.) = 86(3)° (from a spindle-stage extinction curve) and 2V(calc.) = 82°.

Experimental methods

Chemical data were obtained using a JEOL 8230 SuperProbe electron microscope equipped with five WDS spectrometers (University of Ottawa – Canadian Museum of Nature MicroAnalysis Laboratory, Canada) with an accelerating voltage of 15 kV and a beam current of 10 nA. Both natromelansonite and melansonite are unstable under an electron beam, so a larger beam diameter of 20 μm was used to minimise the damage. The matrix correction method was ZAF or Phi-Rho-Z calculations (Armstrong, 1988). The following reference materials were used: albite (NaKα), sanidine (KKα, AlKα and SiKα), diopside (CaKα), tephroite (MnKα), sanbornite (BaLα), YIG (YLα), YbPO₄ (YbLα), zircon (ZrLα) and UO₂ (UMα). Both the peak and the background count times were 20 s for all elements. The intensity data were corrected for Time Dependent Intensity loss (or gain) using a self-calibrated correction for KKα, SiKα and NaKα. The time/counts analysis showed significant Na migration in natromelansonite (Fig. 3).

Figure 3. Time dependent intensity loss for NaKα in natromelansonite (8 analyses).

The Fourier-transform infrared (FTIR) spectrum of natromelansonite was obtained at the Canadian Conservation Institute, Canada using a Bruker Hyperion 2000 microscope interfaced to a Tensor 27 spectrometer with a wide-band mercury cadmium telluride (MCT) detector. A small grain of natromelansonite was mounted and compressed on a low-pressure diamond anvil microsample cell and analysed in transmission mode. The spectrum was collected between 400–4000 cm^–1 by averaging 256 scans at a 4 cm^–1 operating resolution.

Powder X-ray diffraction (PXRD) data were collected at the Canadian Museum of Nature, Canada using a Bruker D8 Discover microdiffractometer equipped with a DECTRIS EIGER2 R 500K detector and IμS microfocus X-ray source (CuKα₁, λ = 1.54060 Å) with the Kα₂ contribution removed using the ‘Strip Kα2’ tool in Bruker Diffrac.EVA V4.3. The instrument was calibrated using a statistical calibration method (Rowe, 2009). A powder ball ~200 μm in diameter, mounted on a fibre pin mount, was analysed with continuous Phi rotation and 10° rocking motion along the Psi axis of the Centric Eulerian Cradle stage.

Single-crystal XRD studies were carried out at the Natural History Museum, University of Oslo, Norway on a Rigaku XtaLAB Synergy-S diffractometer equipped with a HyPix 6000HE detector (MoKα, λ = 0.71073 Å) operating at 50 kV and 1 mA at room temperature. The data were collected and processed using CrysAlis Pro software (Rigaku Oxford Diffraction, UK).

Results

Chemical data

Chemical data on natromelansonite, melansonite and uranium-bearing melansonite are given in Table 1. All the formulae are calculated on the basis of O = 23 per formula unit (pfu) and H = 8 pfu. The H content is estimated to be 8 pfu for simplicity. The empirical formulae written taking into account the structural data are as follows: natromelansonite Na(□_0.58Na_0.38Ca_0.03Mn_0.01)₂(Na_0.70K_0.28Ba_0.02)_Σ1.00(Zr_0.91Y_0.02Yb_0.02)_Σ0.95(Si_7.08Al_1.02)_Σ8.10O₁₉⋅4H₂O; melansonite Na(□_0.495Na_0.36Ca_0.06Mn_0.05Fe²⁺_0.035)₂(K_0.40Na_0.27Ba_0.03)_Σ0.70(Zr_0.90U_0.01Y_0.01Yb_0.01)_Σ0.93(Si_7.12Al_0.93)_Σ8.05O₁₉⋅4H₂O; and uranium-bearing melansonite Na(□_0.65Na_0.285Ca_0.05Mn_0.015)₂(K_0.52Ba_0.02)_Σ0.54(Zr_0.65Y_0.12U_0.09Yb_0.05)_Σ0.91(Si_7.57Al_0.62)_Σ8.19O₁₉⋅4H₂O.

Table 1. Chemical data for natromelansonite and melansonite.

*Calculated from the stoichiometry; dash means the content of a constituent is below the detection limit.

The ideal end-member formula for natromelansonite is Na₃Zr[Si₇AlO₁₉]⋅4H₂O, which requires Na₂O 12.24, Al₂O₃ 6.71, SiO₂ 55.35, ZrO₂ 16.22, H₂O 9.48, total 100 wt.%.

Natromelansonite does not react with an aqueous HCl solution at room temperature.

Infrared spectroscopy

The infrared (IR) spectrum of natromelansonite (Fig. 4) shows IR bands of O–H stretching (in the range from 3275 to 3490 cm^–1) and O–H–O bending (at 1638 cm^–1) vibrations characteristic of H₂O molecules. The bands at 1023 and 1160 cm^–1 correspond to symmetric and asymmetric Si–O stretching vibrations. The band at 622 cm^–1 can be assigned to Zr–O stretching vibrations. The band assignment was made in accordance with Chukanov and Chervonnyi (2016). The obtained spectrum is very similar to that of monteregianite-(Y) reported by Chukanov (2014).

Figure 4. Infrared spectrum of natromelansonite.

X-ray diffraction data

The indexed PXRD data are given in Table 2. Parameters of the monoclinic unit cell refined from the powder data are: a = 6.5156(3) Å, b = 24.061(1) Å, c = 6.9759(6) Å, β = 90.453(5)° and V = 1093.61(9) ų.

Table 2. Powder X-ray diffraction data for natromelansonite.*

*The strongest lines are given in bold

¹ For the calculated pattern, only reflections with intensities >2 are given.

² Calculated from PXRD Rietveld unit-cell refinement.

The single-crystal XRD data were indexed in the P2₁/m space group with the following unit-cell parameters: a = 6.51296(10) Å, b = 24.0944(4) Å, c = 6.97551(11) Å, β = 90.7466(13)° and V = 1094.55(3) ų. The structure was refined to R ₁ = 0.042 on the basis of 4430 independent reflections with I > 2σ(I) using the SHELXL-2018/3 program package (Sheldrick, 2015). Crystal data, data collection and structure refinement details are given in Table 3, atom coordinates, equivalent displacement parameters, site occupancy factors and bond-valence sums (BVS) in Table 4, selected interatomic distances in Table 5, and anisotropic displacement paraments in Supplementary Table S1. The crystallographic information file has been deposited with the Principal Editor of Mineralogical Magazine and is available as Supplementary material (see below). It was also deposited in the Inorganic Crystal Structure Database (ICSD; #CSD 2307107).

Table 3. Crystal data, data collection information and structure refinement details for natromelansonite.

*Located 0.70 Å away from the Zr1 site; the second highest peak is 1.74 e ^–/ų.

Table 4. Coordinates and equivalent displacement parameters (U _eq, in Ų) of atoms, site occupancy and bond-valence sums for natromelansonite.

^a BVS have been formally calculated taking into account the refined site occupancy, using bond-valence parameters of Gagné and Hawthorne (2015).

^b The anisotropic displacement parameters for Na3, Na4 and K5 atoms were refined constraining site movements and occupancies to each other (the EADP command).

^c The occupancy was fixed in the final refinement cycles.

Table 5. Selected interatomic distances (Å) in the structure of natromelansonite.

Description and discussion of the crystal structure

Natromelansonite is pseudo-orthorhombic. Its monoclinic unit cell (P2₁/m) is very close to the orthorhombic cell (Pmma) of melansonite (Gore and McDonald, 2023) and is not doubled as was described in monteregianite-(Y) (Ghose et al., 1987; Table 6). Group P2₁/m is a translationengleiche subgroup of Pmma and such pseudosymmetry in monteregianite-related minerals was predicted by Ghose et al. (1987).

Table 6. Comparative data for natromelansonite and related minerals.

The natromelansonite crystal structure consists of two different types of sheets normal to the b axis: (a) a double sheet of tetrahedra (T) (Fig. 5) and (b) a sheet of octahedra (O) (Fig. 6). The T and O sheets alternate along the [010] direction forming a TOT topology (Fig. 7) typical for members of the rhodesite mero-plesiotype series (Ferraris and Gula, 2005; Cadoni and Ferraris, 2011), including melansonite, KNa₂Zr[Si₇AlO₁₉]⋅4–5H₂O (Gore and McDonald, 2023), monteregianite-(Y), KNa₂Y[Si₈O₁₉]⋅5H₂O (Chao, 1978; Ghose et al., 1987), delhayelite, K₄Na₂Ca₂[AlSi₇O₁₉]₂F₂Cl (Sahama and Hytönen, 1959; Pekov et al., 2009), hydrodelhayelite KCa₂[AlSi₇O₁₇(OH)₂]⋅3H₂O (Dorfman and Chigarov, 1979; Pekov et al., 2009), and fivegite, K₄Ca₂[AlSi₇O₁₇(O_2–xOH_x)][(H₂O)_2–xOH_x]Cl (Pekov et al., 2011). Members of the series are characterised by nearly identical T sheets which may show a different ratio between upwards- and downwards-pointing tetrahedra which is related to the number, charge and coordination number of the cations in the O sheet and to the Si/Al ratio within the silicate sheet. Most of the compounds have a T:O ratio of 8:19 (T = Si and Al). Monoclinic and orthorhombic members are known.

Figure 5. (a) Double sheet of tetrahedra (T) in the structure of natromelansonite; (b) the single sheet consisting of four- and eight-membered rings. The structure was visualised using CrystalMaker® software.

Figure 6. Sheet of octahedra (O) in the structure of natromelansonite.

Figure 7. General view of the crystal structure of natromelansonite. Projection on (100). The unit cell is outlined. T and O sheets are indicated. Yellow and purple spheres are extra-framework Na and K atoms, respectively.

The O sheet in natromelansonite (Fig. 6) consists of alternating edge-sharing fully occupied Zr(1)O₆ and Na(1)O₄(H₂O)₂ octahedra and a highly distorted partially occupied (40%) Na(2)O₄(H₂O)₂ octahedron.

The T sheet is composed of two single silicate sheets of the apophyllite type with four- and eight-membered rings (Fig. 5a). They are formed by Si1-, Si2- and Si3-centred tetrahedra with the <Si–O> distance in the range of 1.608–1.610 Å, and a larger Si4-centred tetrahedron with the <Si4–O> distance of 1.664 Å, with a mixed Si/Al occupancy Si_0.5Al_0.5. The BVS at the Si4 site is 3.65. valence units (vu), which confirms mixed occupancy at the site. The T sheet can be also described using the structure hierarchy concept (Hawthorne et al., 2019) as based on the 3-connected plane net 4.82 plus an oikodoméic (up/down) operation (u³d)₁(u⁴du²d)₁ with all d tetrahedra occupied by Al atoms.

The two single sheets are linked via two vertex-sharing Si4-centred tetrahedra (Si4–O10–Si4, Fig. 5b), which resulted in the creation of large channels. Interstitial Na and K atoms occur in the channels in a split site with occupancies of 0.54, 0.21 and 0.25 for the Na3, Na4 and K5 sites, respectively. The Na3–Na4, Na3–K5 and Na4–K5 distances are 2.28 Å, 1.60 Å and 0.84 Å, respectively. Sodium is the dominant extra-framework cation.

The presence of H₂O⁰ molecules is confirmed by the BVS at the O11, O12, O13 and O14 sites (0.10, 0.04, 0.32 and 0.01 vu, respectively; Table 4) and by the presence of O–H stretching bands and H–O–H bending bands in the IR spectrum of natromelansonite (Fig. 4). Some of these sites (O12, O14) are partially occupied, as it was observed in the crystal structures of monteregianite-(Y) (Ghose et al., 1987) and melansonite (Gore and McDonald, 2023). In general, water content varies significantly in members of the rhodesite mero-plesiotype series (Ferraris and Gula, 2005; Cadoni and Ferraris, 2011). In addition, the variable partial occupancy of multiple sites means that the water content varies from grain to grain in one mineral. Furthermore, the anisotropic thermal vibrations of ligand water molecules are very pronounced which complicates proper refinement of these sites in the structure. Thus, the H content is estimated to be 8 pfu in the ideal end-member formula for simplicity, and that number is used in the formula calculations. But the actual water content seems to be variable in the range 4–5 H₂O pfu in both melansonite and natromelansonite.

The resulting structural formula of natromelansonite is ^Na1Na^Na2(□_0.60Na_0.40)₂ ^Na3+Na4+K5(Na_0.75K_0.25)Zr(Si₇Al)O_18.8OH_0.2⋅4.46H₂O.

The idealised formulae of melansonite and natromelansonite

The idealised formula of melansonite was given as (Na,□)□₂KZrSi₈O₁₉⋅5H₂O in the original publication (Gore and McDonald, 2023). However, detailed examination of the published data combined with our data shows that the formula should be reconsidered by taking into account the behaviour of Al and Na in melansonite. One Si-centred tetrahedron (Si(2)O₄) in the structure of melansonite is significantly larger than the others with the <Si–O> distance of 1.66 Å, indicating that Al atoms are ordered at this site as they are in natromelansonite, and the idealised occupancy at the site should be Si_0.5Al_0.5. The published chemical data showing ~1.0 atoms per formula unit (apfu) of Al (6.32 wt.% Al₂O₃) corroborate this. The Si2 site in melansonite is analogous to the Si4 site in the structure of natromelansonite where Si(2)O₄ tetrahedra link single silicate sheets together. Similar ordering was observed in delhayelite, hydrodelhayelite (Pekov et al., 2009) and fivegite (Pekov et al., 2011) – three other members of the rhodesite mero-plesiotype series (Cadoni and Ferraris, 2011).

The Na content (0.71 apfu, 2.82 wt.% Na₂O) in melansonite from the original find from a marble xenolith at Mont Saint-Hilaire seems to be unexpectedly low which could be explained by migration of Na under the electron bean as was suggested by Gore and McDonald (2023) and the very high count time of 100 seconds. As we have shown, Na migration in melansonite and natromelansonite is significant even with a 20 μm beam (Fig. 3). Additionally, the crystals are very thin, only ~2 μm, which further complicated obtaining realisable data on Na content in the mineral. Melansonite from our find is characterised by a significantly higher Na content (8.20 wt.% Na₂O, Table 1) both due to (1) a larger crystal size which allowed us to reduce Na migration by using a larger beam diameter (20 μm) compared to a 10 μm beam used by Gore and McDonald (2023) and (2) lower contents of admixed cations Ca (0.92 wt.% CaO) and Ba (0.59 wt.% BaO) in contrast to 1.75 wt.% CaO and 1.35 wt.% BaO in melansonite from the original find (Gore and McDonald, 2023).

Furthermore, the occupancy at the Na site in the O sheet of melansonite was fixed at 28% and the refined site-scattering value at the site is unknown (Gore and McDonald, 2023). The full dataset including the cif was not provided by the authors of the description so the reasoning behind this value is unclear, but it is probable that it could be somewhat higher as it was observed in both natromelansonite and monteregianite-(Y) (Ghose et al., 1987).

Considering all the above, (□Na) instead of □₂ in the formula would much better represent the actual distribution of the cations at the Na2 site. Furthermore, the water content in both minerals is variable in the range 4–5 H₂O pfu. Thus, the melansonite formula should be written as Na(□Na)KZr(Si₇Al)O₁₉⋅4–5H₂O or, in a simplified form, KNa₂Zr(Si₇Al)O₁₉⋅4–5H₂O and the formula of its Na-analogue natromelansonite as Na(□Na)NaZr(Si₇Al)O₁₉⋅4–5H₂O or Na₃Zr(Si₇Al)O₁₉⋅4–5H₂O.