Crystal chemistry of zemannite-type structures: IV. Wortupaite, the first new tellurium oxysalt mineral described from an Australian locality Open Access

Tellurium (Te) is a very rare element (crustal abundance 1 ppb) that forms ~200 described minerals, making it the most anomalously diverse element when comparing crustal abundance to number of minerals (Christy, 2015). This high number of described minerals may be explained by the wide range of geochemical niches which Te fills (Brugger et al, 2012; Grundler et al., 2013; Krivovichev et al., 2020). While both a chemically and structurally diverse group of minerals, the majority (70 out of 100, 70%) of tellurium oxysalts have been described from a small geographical area of south-western North America. For instance, Moctezuma, Sonora, Mexico (21 new minerals) and Otto Mountain, California, USA (16 new minerals) have contributed 37 of the 100 known Te–O minerals between them. All other countries have contributed just 29 new Te–O minerals, of which Australia had previously contributed none. The only new Te minerals previously described in Australia were three rare tellurides found in Western Australia; namely honeaite (Au₃TlTe₂; Rice et al., 2016; Welch et al., 2017), kalgoorlieite (As₂Te₃; Rempel and Stanley, 2016) and saddlebackite (Pb₂Bi₂Te₂S₃; Clarke, 1997). Wortupaite thus represents the first tellurium oxysalt mineral to be described from Australia, and just the sixth new mineral in the Southern Hemisphere after brumadoite from Pedra Preta Pit, Bahia, Brazil (Atencio et al., 2008) and four minerals from the Tambo mine, Coquimbo, Chile: walfordite (Back et al., 1999), telluromandarinoite (Back et al., 2017), and tamboite and metatamboite (Cooper et al., 2019).

Wortupaite has a unique chemical composition amongst natural compounds, being the first mineral to contain only Mg, Ni and Te as the metallic cations. Keystoneite is the only other Ni-containing secondary Te mineral described to date, though keystoneite also contains Fe³⁺ (Missen et al., 2021). Wortupaite shares an isotypic framework with zemannite (Miletich, 1995a; Cametti et al., 2017, Missen et al., 2019a; Effenberger et al., 2023), kinichilite (Koyama and Nagashima, 1981; Miletich, 1995a), keystoneite (Missen et al., 2021) and ilirneyite (Pekov et al., 2018). All four of these minerals have a near 50% split of divalent (Zn²⁺, Mn²⁺ or Ni²⁺) and trivalent (Fe³⁺ and Mn³⁺) cations, making wortupaite the first natural zemannite-structured phase to have only divalent cations in the framework sites.

The name and formula (symbol Wor) of wortupaite (IMA2022–107) have been approved by the Commission on New Minerals Nomenclature and Classification (CNMNC) of the International Mineralogical Association (IMA) (Missen et al., 2023). Wortupaite is named for the type locality. The name ‘Wortupa’ means ‘shadow’ in the language of the local indigenous people, the Adnyamathanha. Type material is deposited in the collections of Museums Victoria, Melbourne, Australia with specimen number M2021.

Wortupaite occurs at the Wortupa gold mine (Fig. 1), 12 km north-north-east of Balcanoona Station, Flinders Ranges, South Australia, Australia (30.437966°S, 139.24237°E). Gold was discovered in 1899, however shallow workings provided poor returns and work ceased by 1905. The National Museum of Victoria (as Museums Victoria was then known) purchased the holotype sample on the 12^th of September, 1900 from James Mitchell. The mineralisation consists of irregular veins, lenses and pods of quartz and calcite, with scattered crystals of pyrite and disseminated malachite within scapolitised limestone (Hamisi et al., 2023) and calcareous siltstones of the Neoproterozoic Adelaide geosyncline (Coats and Blissett, 1971). Wortupaite is the first new mineral described from this locality.

Wortupaite is intimately associated with nickel telluride mineral melonite (NiTe₂), upon which it is found growing. Calcite is the gangue mineral on the holotype specimen. Other minerals previously identified from the Wortupa mine include quartz, native gold, pyrite, chalcocite, malachite, azurite and goethite, as well as scapolite (Coats and Blissett, 1971), however none of these were observed on the holotype specimen.

The first evidence of Te secondary minerals at the Wortupa mine was reported by Higgin (1899) when he noted a green coating associated with melonite and suspected that it was “probably a tellurite of nickel”. Wortupaite formed after oxidation of melonite, from which its major component elements (Ni and Te) were derived, probably under near-surface conditions. The Mg²⁺ in wortupaite was most likely to have been leached from minerals such as magnesite or Mg-silicates in associated sedimentary rocks, which then mixed with the Ni- and Te-rich fluids produced on the surface of weathering melonite to form wortupaite.

Wortupaite forms pale yellowish green needles and prisms, often in clusters and blocky masses (Fig. 2). The needles may reach 25 μm in length, but are generally shorter with a lower length to diameter ratio. Wortupaite has prismatic {10|$\bar{1}$|0} forms, with prisms terminated by {10|$\bar{1}$|1} (Fig. 3). The c:a ratio is 0.8149 (single-crystal X-ray diffraction data) and no evidence for twinning was observed. Wortupaite has a pale-green streak, and individual crystals are vitreous, whereas masses have a more earthy lustre. Wortupaite does not fluoresce. Mohs hardness was not determinable but is likely to be < 3, analogous to zemannite. The fracture is uneven, tenacity is brittle and neither cleavage nor parting was observed. The density was calculated to be 4.42 g/cm³ for the empirical formula and unit cell volume refined from single-crystal X-ray diffraction data. The indices of refraction of wortupaite are higher than the liquids available for their measurement. The average refractive index (n_ave) was calculated from the Gladstone‒Dale compatibility index as 1.907, using the unit cell used for the crystal-structure refinement. ω and ɛ were calculated from n_ave after measuring the ω–ɛ birefringence [0.075(10), using a Bertrand lens and a 530 nm gypsum plate]. Wortupaite is uniaxial (+), with ω(calc) = 1.882(10) and ɛ(calc) = 1.957(10). Neither dispersion nor orientation was observed. Faint pleochroism was observed, with O = green–blue, E = yellow with green tinge and O > E.

X-ray photoelectron spectroscopy (XPS) was conducted using a Nexsa Surface Analysis System at the Monash X-ray Platform (MXP) of Monash University, Clayton, Australia, and data were processed using Avantage software (both by Thermo Fisher Scientific). X-rays were generated with a Mg target producing Kα radiation with a beam diameter of 100 μm. The binding energy scale was referenced to amorphous C1s at 284.5 eV. A full-range spectrum was collected prior to collecting detailed scans in the Ni2p (845–885 eV) and Te3d regions (560–600 eV).

The XPS spectrum in the Te3d region shows two peaks at 575.3 and 585.9 eV corresponding to the 3d_5/2 and 3d_3/2 peaks of Te⁴⁺, based on comparison with the reference compound TeO₂ that displays Te⁴⁺ peaks at 576.3 and 586.7 eV (Thermo Fisher, 2013–2021).

The XPS spectrum in the Ni2p region shows a peak at 855.2 eV and a satellite at 861.4 eV, then a peak at 872.6 eV and a satellite at 878.7 eV which correspond to the 2p_3/2 and 2p_1/2 peaks of Ni²⁺, respectively, based on comparison with the reference compound NiO that shows a combined peak at 855.1 eV, satellite at 860.5 eV and then a second peak at 872.4 eV with satellite at 879.5 eV (Thermo Fisher, 2013–2021). No evidence of any Ni³⁺ was observed in the XPS spectra.

Chemical analytical data were collected using electron probe microanalysis (EPMA) on a JEOL JXA-8530F Field Emission Electron Microprobe (wavelength dispersive spectroscopy mode, 15 kV, 5 nA, 20 μm beam diameter), at the School of Earth Sciences, University of Melbourne, Australia (see Table 1). All other elements were below detection limits, including Zn. The ratio of atoms per formula unit (apfu) calculated on the basis of 9 O apfu for the anhydrous part is Mg_0.57Mn_0.04Ni_2.26Fe_0.13Te_3.00O₉⋅3H₂O and the empirical formula is (Mg_0.57Ni_0.39Mn_0.04)_Σ1.00(Ni²⁺_1.87Fe³⁺_0.13)_Σ2.00(Te⁴⁺O₃)₃⋅3H₂O. The ideal formula is MgNi²⁺₂(Te⁴⁺O₃)₃⋅3H₂O, which requires MgO 5.58, NiO 20.68, TeO₂ 66.27, H₂O 7.48, total 100 wt.%. H₂O content was calculated from the crystal structure as 3 H₂O groups pfu, although up to 6 H₂O pfu is the maximum possible hydration (see Crystal structure discussion below). It is worth noting that with 3 H₂O groups pfu, the total of the EPMA data is 99.99 wt.%.

Powder X-ray diffraction (XRD) data were collected using an Oxford Xcalibur 2 system with Sapphire2 CCD and MoKα radiation (50 kV and 1 mA) at 293(1) K. A Gandolfi-type randomised crystal movement was achieved by rotations on the φ and ω axes. Data are presented in Table 2. The unit-cell parameters refined from the powder XRD pattern with whole pattern fitting using ChekCell software (Laugier and Bochu, 2004) are a = 9.327(9) Å and c = 7.5946(3) Å.

Single-crystal XRD data were collected on the micro-focus macromolecular MX2 beamline at the Australian Synchrotron, part of ANSTO. A 5 × 5 × 15 μm prism of wortupaite was selected from specimen M2021 by crushing an aggregate of crystals in oil then selecting a single-crystal fragment. We tested a number of crystal fragments from the type specimen, with the chosen crystal being the best diffractor despite providing only an average of 5 distinct spots per frame. Data were collected at 100 K by a Dectris EigerX 16M detector using monochromatic radiation with a wavelength of 0.71073 Å, collecting a full 360° of data with 1° step size and 1 second per step. The data were processed using XDS (Kabsch, 2010), XPREP (Bruker, 2001) and SADABS (Bruker, 2001), finding 1830 reflections with R_int of 0.1031. Despite this low number of reflections, the crystal structure of the framework of wortupaite was readily refined, and a model for the arrangement of the channel species is presented below. Structure solution of the average structure of wortupaite was carried out by direct methods using SHELXT (Sheldrick, 2015a) followed by structure refinement in SHELXL (Sheldrick, 2015b). All atom positions and anisotropic displacement parameters for Ni and Te only were refined to final R₁ and wR₂ values (all data) of 0.0558 and 0.1393, respectively. Further details of data collection and structure refinement are provided in Table 3. Atomic coordinates and displacement parameters are shown in Table 4, selected bond-lengths in Table 5 and a bond-valence analysis in Table 6, using parameters of Gagné and Hawthorne (2015) for all atoms except Te (Mills and Christy, 2013). The crystallographic information files have been deposited with the Principal Editor of Mineralogical Magazine and are available as Supplementary material (see below).

The space-group in which zemannite-like minerals and compounds crystallise in has generally focused on two hexagonal space-groups: centrosymmetric P6₃/m and non-centrosymmetric P6₃. Recent work has proposed a trigonal P3 symmetry for zemannite due to a different ordering scheme of the M²⁺ and M³⁺ cations within the framework of the zeolitic structure (Effenberger et al., 2023). The influence of framework cation ordering and also the H-bonding network leads to a non-centrosymmetric refinement in zemannite (Cametti et al., 2017, Missen et al., 2019a). However, synthetic zemannite-like compounds Na₂[Co₂(TeO₃)₃]⋅3H₂O and Na₂[Zn₂(TeO₃)₃]⋅3H₂O (Miletich, 1995b), which have only one framework transition metal cation crystallise in centrosymmetric P6₃/m. Refinements in both space groups were explored for wortupaite, with the centrosymmetric refinement preferred for the final refinement as the non-centrosymmetric setting did not give a stable structural refinement. The issue with the non-centrosymmetric refinement included a Flack parameter close to ½, and the presence of non-positive definite atoms such as the channel Mg²⁺ and framework O sites). Additionally, lower symmetry space groups are also possible for zemannite, with recent examples of lower symmetry found in monoclinic twin domains in zemannite-structured K[(Cu²⁺,Mn²⁺,Mn³⁺)₂(TeO₃)₃]⋅2H₂O (Eder et al., 2023b). Refinements in P2₁/m and P|$\bar{6}$| were also undertaken but did not result in improved statistics for wortupaite.

The crystal structure of wortupaite (Fig. 4) is similar to that of zemannite, with a characteristic host-guest structure (Cametti et al., 2017; Missen et al., 2019a; Effenberger et al., 2023). Like zemannite, wortupaite is a zeolitic tellurite mineral containing isolated neso Te⁴⁺O₃^2– pyramids (Christy et al., 2016), linked into a three-dimensional microporous framework by face-sharing Ni₂O₉ octahedra. The Te⁴⁺O₃^2– pyramids feature a typically bimodal bond-length distribution, with three short strong bonds forming the trigonal pyramid with an average Te–O distance of 1.85 Å, and four (2×2) long, weak secondary Te–O bonds with an average length of 3.03 Å (Christy and Mills, 2013) resulting in a total bond-valence sum of 4.26 valence units (vu). Nickel (with minor substitution by Fe³⁺, represented in the crystal structure as M1) is coordinated by six O anions with a bimodal distribution: three shorter Ni1–O2 bonds with length 2.02(2) Å and three longer Ni1–O1 bonds with length 2.10(2) Å.

The structures of zemannite and wortupaite differ with regards to the valence of the framework cations, geometry of M(H₂O)₆ clusters in the channels and also by the presence of two (rather than one) extra-framework cation site within the channels for charge balance. In zemannite (and all other zemannite-framework minerals described to date, see below), half of the framework cations are divalent and the other half are trivalent, leading to an ordered arrangement (Missen et al., 2019a; Effenberger et al., 2023). In wortupaite, all the framework cations are divalent and Ni-dominant, with no ordering by definition. The wortupaite framework has a higher negative charge (–2 in ideal wortupaite) than zemannite (–1), leading to a different arrangement of channel species than the half-occupied octahedral Mg(H₂O)₆ complexes typical of zemannite, kinichilite, ilirneyite and keystoneite.

Due to the low levels of diffraction from wortupaite crystals and correspondingly low number of reflections collected in the refinement, the refinement of channel species is not unambiguous. The data are subject to Fourier truncation effects which means that the highest peaks in the difference-Fourier map (i.e. those in the channel) may originate from not having sufficient Fourier coefficients. In this section, we present the best model we refined: the model with EPMA fitted across three metal sites (one framework site, two channel sites). The crystallographic information file of the unrefined model (M1 as Ni, M2 as Mg and M3 as Ni) is also provided in Supplementary material. Once the framework species are refined, the largest remaining peak in the difference-Fourier map is at (0 0 ¾) and with a site-scattering value of ~8.6 e^–, indicating less than full occupation by the M2²⁺ cation, modelled initially as a partially occupied Mg²⁺ site (again noting that the high value of z i.e. ¾ is potentially due to Fourier truncation effects). The largest peak following the refining of the first peak is at (0 0 0.86), with site scattering of ~5 e^–, modelled initially as a partially occupied Ni²⁺ site (M3²⁺ cation). Both of these sites were refined with fixed U_iso values of 0.05. Refinement allowing the site occupancies to vary results in an initial tunnel composition of M2_0.60M3_0.43, close to the ideal cation requirement of 1.0 cations pfu. Thus, the required M²⁺ content of the channels, (Mg_0.57Ni_0.39Mn_0.04)_Σ1.00 pfu, as indicated by the EPMA data, can be reasonably attributed to two sites, i.e. M2(2a) (Mg_0.42Ni_0.08)_Σ0.5 pfu at (0 0 ¾) and M3(4e) (Mg_0.15Ni_0.31Mn_0.04)_Σ0.5 pfu at (0 0 0.846(6)) based on the site-scattering factors. Fitting occupancies with the EPMA data changes the R indices by less than 0.05%.

The highest remaining difference-Fourier peak (site-scattering factor ~2 e^–) is at an appropriate distance from the channel cation sites and from framework O1 and O2 sites to be an OW site, defined as 50% occupied to avoid unreasonably short OW–OW distances and again positionally refined using a fixed U_iso value of 0.05. The clear preference for the OW sites in wortupaite is for an arrangement around M2 showing significant distortion away from octahedral coordination (Fig. 5). The addition of the half-occupied OW site to the model lowers the R₁(all) to 0.054 from 0.076. Given the limited nature of the dataset, it is possible that this site is related to disorder or Fourier truncation effects, but we have chosen to present the final model including the O sites for completeness of the M2 coordination sphere (M2–OW: 2.13(5) Å × 6) (Fig. 5). It is worth noting that the quarter-occupied M3 (Ni-dominant) channel site is not expected to be coordinated by the half-occupied OW site in its refined position (which with M3 would display a one-sided coordination featuring three M3–OW bonds at 1.77(5) Å and three at 2.36(6) Å). Both the one-sided coordination and bonds to OW with distance <1.80 Å are essentially impossible for O ligand geometry around the M3 transition metal-dominant cation. Instead, M3 may be expected to be coordinated by two additional 12i sites (‘OW2’ and ‘OW3’) for 6-fold coordination, but such low-scattering sites were not locatable in the Fourier map of the available refinement data. This possibility also suggests that wortupaite may have a hydration > 3H₂O groups per formula unit (up to 6H₂O groups pfu possible as a maximum). We also explored the possibility for the OW sites (isotropic displacement parameter of the single position, if refined, is > 0.05 Ų) to be disordered over multiple sites – which would allow for a more symmetric stereochemical arrangement of OW sites around M3, as well as a reasonable geometry and bond-valence sum, however this attempted refinement was unstable. In the synthetic zemannite-like compound of Eder et al. (2023a) with the formula K₂[Co₂(TeO₃)₃]⋅2.5H₂O, channel cation positions were found in positions similar to those of the OW sites in wortupaite. We tried to refine a model with the channel cation sites away from the centre of the channel, but the refinement was also unstable.

Wortupaite is structurally most closely related to the two synthetic zemannite-like compounds, Na₂[Co₂(TeO₃)₃]⋅3H₂O and Na₂[Zn₂(TeO₃)₃]⋅3H₂O (Miletich, 1995b), which also have frameworks with a formal –2 charge. Other related compounds include a selenite analogue of the zemannite structure type, K₂[Ni₂(SeO₃)₃]⋅2H₂O (Wildner, 1993), and an unpublished Sr-dominant Co zemannite, Sr_x[Co₂(SeO₃)₃]⋅yH₂O (Wildner, 1993), the latter featuring a fully occupied divalent channel cation (Sr²⁺). Ni–O bond lengths in the Ni²⁺O₆ octahedra of K₂[Ni₂(SeO₃)₃]⋅2H₂O are two triplets of 2.057(3) Å and 2.137(4)Å (average 2.097 Å). The Na⁺-containing compounds such as Na₂[Co₂(TeO₃)₃]⋅3H₂O usually have Na⁺ cations situated near the edges of the microporous channels, rather than in the centre as is typical for zemannite. A series of mainly cryptocrystalline zemannite-like compounds with Na⁺ in the centre of the microporous channels (compositions ranging from Na_1.25Fe³⁺_0.75Zn_1.25(TeO₃)₃⋅3H₂O to Na_0.55Fe³⁺_1.44Zn_0.56(TeO₃)₃⋅3H₂O) have also been synthesised (Missen et al., 2019b). In the studied sample of wortupaite, two partially-occupied channel cation sites were located [at different positions with coordinates in the form (0 0 z)]. The only previously described naturally occurring Ni-bearing Te oxysalt keystoneite, ideally Mg_0.5Ni²⁺Fe³⁺(TeO₃)₃⋅4H₂O, has Ni–O bonds with an average length of 2.085 Å. In addition, several new zemannite-like compounds have recently been reported for the first time, including Ni-bearing Na₂[Ni₂(TeO₃)₃]⋅2.5H₂O and K₂[Ni₂(TeO₃)₃]⋅H₂O (Eder et al., 2023a) and the first Jahn-Teller distorted framework in a zemannite-like structure observed in K[(Cu²⁺,Mn²⁺,Mn³⁺)₂(TeO₃)₃]⋅2H₂O (Eder et al., 2023b). The Ni-bearing compounds have Ni–O distances of 2.064 and 2.069 Å, respectively – close to the average 2.06 Å bond length in the wortupaite framework. There are likely to be many more zemannite-structured compounds yet to be discovered as minerals or to be synthesised in the laboratory.

We thank Professor Irina Galuskina for handling our manuscript and Structures Editor Professor Pete Leverett and two anonymous reviewers for their helpful and insightful comments which significantly improved the manuscript. This study has been partly funded by The Ian Potter Foundation grant “tracking tellurium” to S.J.M. Support funding was provided to OPM by an Australian Government Research Training Program (RTP) Scholarship, a Monash Graduate Excellence Scholarship (MGES) and a Robert Blackwood Monash-Museums Victoria scholarship. Allan Pring (Flinders University) is thanked for providing access to other potential wortupaite-bearing specimens. This research was undertaken in part using the MX2 beamline at the Australian Synchrotron, part of ANSTO, and made use of the Australian Cancer Research Foundation (ACRF) detector. The authors acknowledge use of the facilities and the assistance of James Griffith at the Monash X-ray Platform and of Graham Hutchinson (University of Melbourne) for the EPMA data. We also thank the South Australian Department for Environment and Water for providing guidance of the meaning of the name ‘Wortupa’. Field work at Wortupa, located in the Vulkathunha-Gammon Ranges National Park, was conducted via a permit from the Department for Environment and Heritage.

The supplementary material for this article can be found at https://doi.org/10.1180/mgm.2023.64.

The authors declare none.