Introduction

Montanite is the earliest discovered tellurium (Te) oxysalt, first identified and described from the Highland Mining District, Montana, USA and David Beck’s mine, North Carolina (Genth 1868; cf. Table 1). These two localities are the modern-day equivalent of cotype localities, even though conventionally only the Highland Mining District has been reported as the type locality of montanite (as reported on databases such as Mindat.org). In particular, the North Carolina material was used to support the chemical analysis and determination of the water content. The ideal formula for montanite was given as Bi2TeO6·nH2O where n is 1 or 2 (Genth 1868), determined from wet-chemical analysis. It is worth noting that Genth (1868) defined montanite as containing one water molecule per formula unit, with two H2O (commonly presented as the formula of montanite for at least 100 years; Larsen 1921; Pasero 2021) also suggested, but in brackets. Standard for the time, chemical tests such as acid dissolution behaviour were also performed. Montanite was just the second mineral with a formula containing both Te and O (after tellurite, the orthorhombic form of TeO2 described in 1845 from Romania; Haidinger 1845). Frederick A. Genth described montanite and the mercury(I) tellurite magnolite (Genth 1877; Grice 1989; Roberts et al. 1989), a remarkable achievement considering that only five further Te–O minerals were discovered up until 1960. Montanite was defined as questionable (Q status) by the International Mineralogical Association (IMA) due to incomplete data and the lack of a designated type specimen (Pasero 2021).

Table 1 Summary of our three main synthetic samples and detail on reported montanite occurrences analysed and/or checked in this study
Full size table

While primary bismuth–tellurium minerals such as tetradymite (Bi2Te2S), joséite-A (Bi4TeS2) and joséite-B (Bi4Te2S) are relatively common (Ciobanu et al. 2009; Cook et al. 2021), the seven secondary Bi–Te–O minerals are generally rarer and/or poorly defined. The close association of montanite with tetradymite, as noted by Genth (1868), sometimes led to the conclusion that montanite is the only weathering product of tetradymite. In our opinion, it is therefore no surprise that most yellow–orange crusts on specimens containing Bi–Te(–S) minerals such as tetradymite, joséite-A and joséite-B were assigned as montanite specimens in the past. Indeed, montanite was the only described Bi–Te–O mineral until smirnite (Bi2Te4+O5) was discovered over a century later, though montanite does seem to be the most common Bi–Te–O mineral (Spiridonov and Demina 1984). In total, there are 32 Bi–Te primary minerals, which are of geochemical interest due to their association with gold (Ciobanu et al. 2005; Tooth et al. 2011; Brugger et al. 2016). These primary minerals include four Bi tellurides, six Bi telluride–sulfides, four Bi telluride–selenides and 18 additional Bi–Te minerals containing other metallic elements, which all may subsequently undergo weathering to produce secondary minerals. Along with the seven Bi–Te(–S)–O minerals that have been formally described (Table 2), Bi–Te–( +) minerals contribute substantially to the extraordinary diversity of Te minerals (Christy 2015; Missen et al. 2020).

Table 2 Details of the seven secondary Bi–Te–O minerals described to date
Full size table

In this paper, we report the results of our X-ray, electron microprobe, thermogravimetric and infrared spectroscopic studies on a combination of both natural (including type mineral) and synthetic samples (corresponding to the formula Bi2TeO6·nH2O) to determine whether montanite is a valid mineral. Additionally, for the first time, we determined the amount of water which can be structurally incorporated into montanite.

Specimens

Synthetic specimens

Both hydrothermal and solid-state syntheses were undertaken in our montanite synthesis experiments. Polycrystalline montanite was obtained from 0.2425 g Bi(NO3)3·5H2O and 0.0110 g KTeO(OH)5(H2O), i.e., a 1:1 Bi:Te mole ratio with Te6+ in excess (sample code SynPC1, i.e., Synthetic polycrystalline 1). In a different experiment, 0.2425 g Bi(NO3)3·5H2O with 0.0574 g Te(OH)6, i.e., a 2:1 Bi:Te mole ratio, was used in the synthesis (sample code SynPC2). For both experiments, the solids were added to a Teflon-lined steel vessel, using 1.18 g of water for SynPC1 and 2.30 g of water for SynPC2 to fill the vessels to an approximate \({\raise0.5ex\hbox{$\scriptstyle 2$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) loading. The vessels were then sealed in a steel autoclave and heated at autogenous pressure at 200 °C for 5 days, followed by furnace cooling. The contents of the Teflon vessels were filtered through a Büchner funnel, washing the solid products with mother liquor, then deionized water, ethanol and acetone, followed by drying in air.

Crystals of synthetic montanite were obtained from a batch intended to prepare Bi6Te2O15 (Nénert et al. 2020). For that purpose, the binary oxides Bi2O3 and TeO2 were mixed in the stoichiometric ratio 3:1, thoroughly ground in an agate mortar and placed in a platinum crucible that was heated under atmospheric conditions within 4 h to 950 °C and cooled with a rate of 5 °C/h to room temperature. The light-yellow reaction product had a glassy appearance and was subsequently leached with water. Optical inspection under a polarising microscope revealed that the dried product contained a small fraction with tiny crystals not larger than 40 microns in size, from which a crystal was selected for single-crystal structure analysis. Given the high-temperature conditions used for the synthesis, it is believed that the water detected in the crystal-structure analysis was most likely incorporated into the structure during the leaching of the product. This sample is subsequently referred to as SynCeramic and was used for single-crystal analysis.

Natural specimens

Type specimens, i.e., the specimen(s) analysed by authors in a new mineral proposal for definition of the species, are an essential aspect of modern mineralogy (Nickel and Grice 1998). Strict guidelines exist for the designation of type specimens (Dunn and Mandarino 1987). Before the official designation of type specimens (first codified by Embrey and Hey 1970), the best candidates for re-analysis of old, poorly defined minerals are specimens which are attributable to the original describer (if known, in this case Frederick A. Genth). In the course of our investigation, samples in the Genth collection at Pennsylvania State University were identified, but were not available for study. Fortunately, samples from each original montanite locality were available at the NHM in London: BM 85116 from Highland Mining District, Montana and BM 1985,Nev336 from David Beck’s mine, North Carolina (Fig. 1). Both samples feature earthy montanite growing around relict tetradymite (Fig. 2) and have good historical provenance, BM 85116 coming from Genth himself (Rumsey et al. 2022, in preparation) and BM 1985,Nev336 traceable back to Genth. Following our investigation, these samples are now formally recognised jointly as neotype specimens for montanite by the IMA Commission on New Minerals, Nomenclature and Classification (IMA–CNMNC), Proposal 22-A (Miyawaki et al. 2022). Both electron microprobe analyses (EPMA) and powder X-ray diffraction (PXRD) scans were collected on the neotype montanite samples.

Fig. 1
figure 1

Montanite neotype specimens. a BM 85116 shown with its labels, b BM 85116 under an optical microscope and c optical image of BM 1985,Nev336. The silvery-grey mineral is tetradymite, the earthy yellow mineral is montanite, and the earthy reddish material is unidentified Fe (oxyhydr) oxides

Full size image
Fig. 2
figure 2

a and b Backscatter SEM images showing inhomogeneous natural montanite (shades of grey) surrounding striated grains of tetradymite (bright in back-scatter electron mode) on neotype specimens BM 85116 and BM 1985,Nev336. Orange text relates to location of EPMA points. c and d Backscatter SEM images of the two synthetic polycrystalline montanite samples, showing < 5 μm crystallites

Full size image

As part of our search for montanite, the Highland Mining District was visited by two authors (OPM and SJM). The area is now remediated. The remains of a small mill that processed ore from the Highland mine is the only remaining trace of the mine today. As such, we relied on old montanite specimens for our study, rather than collecting new material, highlighting the importance of museums in preserving historic specimens.

We compiled data on as many montanite specimens as we were able to easily locate, including collecting new data where possible (Table 1). Both PXRD and SEM work were performed on ‘montanite’ from Captains Flat, NSW, Australia (specimen numbers M2087 and M20910, Museums Victoria, Australia, collected in the 1880s). The online database Mindat.org lists 25 localities across five continents as localities for montanite (Mindat.org, 2021). However, some of these localities are yet to be analytically verified as montanite occurrences.

Methods

EPMA analysis

Small fragments were removed from both neotype samples and mounted in a probe block. Quantitative chemical spot analyses were performed on a Cameca SX100 electron microprobe (WDS mode, 20 kV, 20 nA, 1 μm beam diameter and PAP matrix correction) at the Imaging and Analysis Centre, Core Research Laboratories, NHM. EPMA results are summarised in Table 3, shown with comparison to results from previous studies. Consistent with a hydrated, earthy and inhomogeneous material (likely including adsorbed as well as structural water) totals are < 100 wt%. Montana montanite (BM 85116) contains systematically higher Te and lower Bi amounts compared to North Carolina montanite (BM 1985, Nev336), consistent with the observations of Genth based on wet-chemical analyses (1868).

Table 3 Chemical analysis data for montanite (averages)
Full size table

Based on 6 O anions per formula unit coupled with the maximum of 0.67 H2O per formula unit in natural montanite, the empirical formula for BM 85116 montanite is Bi1.89Ca0.02Te1.04S0.01O6.67H1.33, which may be simplified to (Bi1.89Ca0.02)Σ1.91(Te,S)Σ1.05O6·0.67H2O. Using the same approach for BM 1985, Nev336, the empirical formula is Bi2.03Ca0.01Te0.96Cu0.06O6.67H1.33 (P forms < 0.01 atoms per formula unit and is not included in the formula), which simplifies to (Bi2.03Ca0.01)Σ2.04(Te,Cu)Σ1.02O6·0.67H2O The ideal formula is Bi2TeO6∙0.67H2O, requiring 71.29 wt% Bi2O3, 26.87 wt% TeO3 and 1.84 wt% H2O, total 100 wt%.

IR spectroscopic analysis

Attenuated total reflection (ATR) traces were collected on a Pike Technologies GladiATR with a Bruker Tensor 27 system on both the SynPC1 and SynPC2 samples. The samples show similar features, with SynPC2 shown in Fig. 3. The background and IR traces were determined from an average of eight scans. Scans were taken from 400 to 4000 cm−1, with the transmittance recorded every 2 cm−1. These spectra clearly showed the broad main tellurate vibrational band centred at 634 cm−1, the dominant feature of the pattern (consistent with the location of tellurate IR bands in minerals such as mcalpineite-2O (also 634 cm−1); Missen et al. 2022). Tellurate bending modes were not visible as they occur at energies below 400 cm−1. Weak, broad vibrational bands at 1629 cm−1 and 3249 cm−1 may be attributed to the H–O–H bend and the O–H vibration stretch, respectively. The IR spectrum of montanite is somewhat similar to that of hydrated tellurate-sulfate bairdite, which has tellurate bands between 666 and 716 cm−1 and bands at 3117 and 1613 cm−1 related to O–H bonds within the structure (Kampf et al. 2013). The double band with peaks at 1388 and 1320 cm−1 is attributable to the nitrate ion, present as a minor impurity in the products, perhaps from a remnant Bi-nitrate which was not fully removed by washing (consistent with a 1345 cm−1 NO3− stretch in YCu(TeO3)2(NO3)(H2O)3; Mills et al. 2016). However, no evidence for nitrates was observed in the PXRD pattern.

Fig. 3
figure 3

IR trace of synthetic montanite (SynPC2)

Full size image

Thermogravimetric analysis (TGA)

Thermogravimetric analysis (TGA) was undertaken on sample SynPC1 on a Perkin Elmer TGA 8000. The sample was first dried in a drying oven at 50 °C for 3 h, before a 10.274 mg subsample was loaded into the TGA alumina pan under N2 atmosphere. The sample was analysed across the 50–800 °C temperature range, beginning by holding the sample at 50 °C for 10 min, then heating from 50 to 400 °C at a rate of 5 °C/min, and finally heating from 400 to 800 °C at a rate of 10 °C/min. The TGA curve of SynPC1 is shown in Fig. 4.

Fig. 4
figure 4

TGA trace of montanite synthetic sample SynPC1, showing loss of first H2O and then TeO3

Full size image

Crystallography

Powder XRD

Powder diffraction scans were performed on the polycrystalline synthetic montanite samples SynPC1 and SynPC2. Representative samples of the bulk material were ground in a mortar and pestle, fixed with small amounts of petroleum jelly on zero-background silicon wafers, and measured with CuKα1,2 radiation in Bragg–Brentano geometry on a Panalytical X’PertPro system.

PXRD scans on natural montanite samples were performed at the Natural History Museum using an X’Pert Pro MPD system (Panalytical) with CoKα1,2 radiation. Tiny montanite fragments were removed from each neotype and carefully separated from tetradymite and ground in an agate mortar. The PXRD patterns were collected in transmission mode on a Kapton foil. The small sample amounts meant that broad low-intensity reflections of the Kapton foil were observed at low angles (< 20° 2θ); thus, additional measurements on an empty Kapton foil were also undertaken to ensure that no (supercell) reflections for montanite were hidden in this low-angle region. The software Highscore Plus (Panalytical) was used for data evaluation and Rietveld refinement.

A comparison of PXRD data is listed in Table 4 showing the match between the pattern calculated from SynCeramic montanite with the neotype specimens, SynPC1 montanite and other examples, while the match between SynPC1 and the PXRD patterns of the neotype specimens is also shown in Fig. 5. In the case of samples which were not analysed by Rietveld refinement, unit-cell parameters were refined from the PXRD patterns using the Chekcell software (Laugier and Bochu 2004).

Table 4 Powder X-ray diffraction data for natural and synthetic montanite samples, including Rietveld data where applicable
Full size table
Fig. 5
figure 5

PXRD patterns of natural montanite neotypes: a BM 85116 and b BM 1985,Nev336 shown with c synthetic montanite (SynPC1). hkl indices of the five most intense peaks are labelled

Full size image

Single-crystal XRD (SCXRD)

A 10 × 40 × 40 μm single crystal was studied using a Bruker APEX-II CCD single-crystal X-ray diffractometer (Bruker-AXS, Madison, WI, USA) operating with MoKα radiation at 295(2) K across an angle range of 2.58–27.53° θ to determine the crystal structure of montanite. Data collection strategies were optimised with APEX-3 (Bruker 2016), and the collected data (omega-scans with 30 s exposure time for 0.5° frames) were subsequently reduced with SAINT (Bruker 2016). Absorption correction was performed using the semi-empirical multi-scan method with SADABS (Bruker 2016). The crystal structures were solved using SHELXT (Sheldrick 2015a) identifying the heavy-atom sites. Subsequent refinement against F2 was performed using SHELXL (Sheldrick 2015b). Despite picking the best crystal available from the synthesis, the Rint of 0.128 is relatively high. Twinning was identified and a twin matrix of [0 1 0/1 0 0/0 0 -1] was used in the final refinement. All heavy atoms were refined anisotropically for the final refinement. Te–O bond lengths were refined using a soft restraint due to unusual bond lengths resulting from unconstrained refinements, perhaps related to the elongate displacement parameters of the Te atoms. The occupancies of Oxygen Water (OW) sites were refined consecutively and then fixed at 0.40 and 0.25 in the final refinement, and the displacement parameters of the OW sites were fixed in the same manner. The final refinement converged to R1 and wR2 values (all data) of 0.0841 and 0.1040, respectively.

Full refinement details are shown in Table 5, atomic coordinates and displacement parameters in Table 6, Bi–O and Te–O bond lengths in Table 7, and a bond valence analysis in Table 8.

Table 5 SXRD refinement details of synthetic montanite (SynCeramic)
Full size table
Table 6 Atomic coordinates and displacement parameters for synthetic montanite (SynCeramic)
Full size table
Table 7 Bond-length table (Å) for synthetic montanite (SynCeramic)
Full size table
Table 8 Bond-valence table for synthetic montanite (SynCeramic)
Full size table

Evidence for a supercell and Rietveld refinement

Using a significantly increased exposure time (2 min per frame), the synthetic sample of montanite used for structural determination (SynCeramic) shows evidence of a second crystal domain with a minor contribution to the diffraction data (highlighted by yellow circles in Fig. 6). More importantly, a supercell (a = 15.7366(6) Å and c = 5.6011(5) Å), in the form of reflections with faint intensities and a ratio of \(\sim \sqrt{3}\) between strong reflections in the (ab)* planes, is evident (highlighted by red circles in Fig. 6). However, all attempts to undertake refinements in the threefold supercell were unstable. Natural samples of montanite, including the neotype samples, are typically polycrystalline with minimal evidence of the supercell based on analysis of PXRD scans, except for barely visible superstructure reflections in the region where 2θ < 20°. PXRD scans of polycrystalline synthetic montanite (SynPC1 and SynPC2) show the same features as natural montanite. All the reflections in scans of natural montanite may be indexed by the P\(\overline{6 }\) subcell with a ≈ 9.10 Å and c ≈ 5.57 Å.

Fig. 6
figure 6

Reconstructed hk\(\overline{1 }\) plane from SCXD of the SynCeramic sample, showing faint reflections found at a positioning ratio of ~ √3 between strong reflections in the (ab)* planes. The grid refers to the subcell, and reflections marked in red define a threefold supercell; reflections marked in yellow originate from a second crystal domain with minor contribution

Full size image

To show the lack of necessity of a supercell to adequately describe the montanite crystal structure, Rietveld refinements were undertaken on both natural (both neotypes) and synthetic (SynPC1) samples. These Rietveld refinements show a satisfactory match (d-spacings identical, peak heights vary) for the PXRD scan from the model developed from the structure of SynCeramic montanite. The unit-cell parameters obtained from the Rietveld refinements of these samples are a = 9.0313(8) Å, c = 5.6114(7) Å and Rwp = 0.048 (BM 85116, includes ~ 5% quartz), a = 9.037(1) Å, c = 5.606(3) Å and Rwp = 0.029 for BM 1985,Nev336 (Fig. 7a) and a = 9.1024(12) Å and c = 5.5693(11) Å with Rwp = 0.083 for SynPC1 (Fig. 7b).

Fig. 7
figure 7

Rietveld refinement showing a satisfactory match (peak heights vary) between the powder X-ray diffraction pattern of a natural sample BM 1985,Nev336 and b synthetic sample SynPC1, using the crystal structure data of synthetic montanite (SynCeramic) as the model. Peaks in shaded area (amorphous) not used in the final refinement

Full size image

Results

Synthesis results

Hydrothermal experiments under mild conditions produced polycrystalline montanite, sometimes admixed with amorphous Bi–Te oxides (based on broad, low-intensity peaks sometimes observed in the PXRD scans). The Bi(NO3)3·5H2O–Te(OH)6 synthesis produced a PXRD pattern with a lower signal-to-noise ratio, indicating poorer crystallinity. Large single crystals of montanite do not grow readily (as evidenced by the abundance of polycrystalline powders observed in natural montanite samples) and the only method to date by which montanite crystals have been produced is the high-temperature synthesis described here.

Structure description and comparison

The crystal structure of montanite is described with the subcell model. It consists of a three-dimensional network of BiOn (n = 6 or 7) and TeO6 polyhedra, with half of the Bi3+ and all Te6+ cations in trigonal-prismatic coordination (Fig. 8a). The Bi2O6 polyhedra are much more asymmetrical than those of TeO6, probably due to the steric effects of the stereoactive 6s2 lone pair, and Bi1O7 polyhedra are pseudo-trigonal-prismatic with an additional bond to reach 7-coordination. There are two Bi and three Te sites located at Wyckoff positions 3 k and 3j (both with m.. symmetry) for Bi1 and Bi2, and located at 1c (\(\overline{6 }\)), 1f (\(\overline{6 }\)) and 1b (\(\overline{6 }\)) for Te1, Te2 and Te3, respectively. Bi–O distances are found in three pairs to form the triangular prisms, with four Bi–O bonds approximately planar and the other pair found on the opposite side of the Bi3+ cation to the lone pair. Triangular prism Bi1–O distances have an average of 2.33 Å, while the average Bi2–O distance is longer at 2.42 Å, leading to a decreased bond valence of 2.52 valence units (vu). Bi1 cations additionally form one bond to the OW1 site with a length of 2.643(6) Å, leading to a final Bi1 average bond-length of 2.37 Å and bond valence of 3.25 vu. Despite this, there are no further O atoms within the Bi2-bonding sphere found at typical distances for Bi–O secondary bonding.

Fig. 8
figure 8

a Crystal structure of montanite (SynCeramic sample in the subcell), in a projection along the c-axis. Bi1O6 polyhedra are turquoise, Bi2O6 polyhedra are light-blue and TeO6 trigonal prisms are red, each with O atoms shown as white spheres. OW1 atoms and its bonds are yellow, and OW2 atoms are orange. b TeO6 trigonal prisms in SynCeramic montanite (subcell) shown with displacement ellipsoids at the 50% probability level

Full size image

Tellurium cations are all coordinated by six crystallographically related O atoms in a trigonal-prismatic configuration (Fig. 8b). Te1–O1 bonds are 1.92(3) Å in length (leading to a bond valence sum of 6.03 vu), Te2–O2 bonds are 1.93(3) Å (5.86 vu) and Te3–O3 bonds are longest at 1.95(3) Å (lowest valence sum of 5.69 vu), respectively. H2O groups are found in small cavities within the structure (OW1⋅⋅⋅O1 = 2.43 Å; OW2⋅⋅⋅O2 = 2.27 Å), with additional void space occupied by the stereoactive 6s2 lone pair of the Bi3+ cation. Unlike the OW1 sites, which form a bond to the Bi1 cation, the OW2 sites are isolated from the rest of the structure, with their nearest neighbour being Bi1 sites at a distant 4.329(9) Å.

Te6+ cations coordinated by O anions in a trigonal-prismatic configuration have never been previously observed (Christy et al. 2016), meaning that montanite has unique Te6+–O coordination polyhedra (Fig. 8b). A trigonal-prismatic environment is considerably rarer than octahedral (Housecroft and Sharpe 2004), but is observed for elements such as for Sb in pyrochlore minerals like roméite (Grey 2020) when distortion is imposed on 6-coordinate sites in the structures. Outside of near-regular Te6+O6 octahedra, the only other Te6+ coordination environments known are extremely rare examples of Te6+O4 tetrahedra and Te6+O5 trigonal bipyramids observed in synthetic compounds Cs2[TeO4] (Weller et al. 1999), Cs2K2[TeO5] (Untenecker and Hoppe 1986) and Rb6[TeO5][TeO4] (Wisser and Hoppe 1990).

The presence of water in synthetic montanite (SynCeramic) is somewhat surprising as the montanite crystals were the product of a ceramic reaction performed at high temperatures. We hypothesise that the water is either incorporated into the structure upon cooling and exposure to moisture in the atmosphere or, most likely, during the leaching process. The occupancy of both OW sites is low, leading to the final formula of Bi2TeO6·0.22H2O. If both H2O sites are unoccupied, the formula is anhydrous Bi2TeO6, while if both OW sites are occupied, the endmember formula is Bi2TeO6·\({\raise0.5ex\hbox{$\scriptstyle 2$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\)H2O. The observed weight loss of ~ 1.08 wt% (0.39 H2O) determined by TGA on SynPC1 is consistent with this range of hydration. Thus, the final formula of montanite is best described as Bi2TeO6·nH2O where 0 ≤ n ≤ \({\raise0.5ex\hbox{$\scriptstyle 2$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\). Sejkora et al. (2004) analysed a sample of montanite (verified by PXRD) by thermogravimetric analysis and determined that montanite contains two H2O groups per formula unit (pfu) through a two-step dehydration process. Structural considerations in this study show that stoichiometrically, only \({\raise0.5ex\hbox{$\scriptstyle 2$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) H2O groups are possible pfu. We hypothesise that this additional water results from strongly adsorbed water contained in micro-cavities within the polycrystalline montanite, a phenomenon also observed in some clay minerals (e.g., Feng et al. 2018). The incorporation of other Bi–Te–O minerals into the analysis is unlikely as none of smirnite, chekhovichite (Bi2Te4+4O11) or pingguite (Bi6Te6+2O15) contain H2O groups that might have contributed to a higher total of emitted water in TGA analysis.

Discussion

As shown in Fig. 6, superstructure reflections with faint intensities are observed in the X-ray diffraction pattern of synthetic montanite (SynCeramic), defining a threefold supercell with asupercell and bsupercell axes related by \(\sqrt{3}\) to the asubcell and bsubcell axes. As previously noted, on the basis of the current SCXRD dataset, it was not possible to obtain a satisfactory structure model under consideration of the supercell reflections. Higher quality data in terms of intensity statistics and resolution would be required to obtain a full understanding of the superstructure, which may be related to a number of reasons, such as the behaviour of the stereoactive lone pair of Bi3+ cations and the concomitant distortions from an ‘ideal’ symmetry. Behaviour of the H2O groups in void space may also relate to the superstructure. Whereas the true symmetry of the superstructure could not be derived from the current data, the subcell shows hexagonal symmetry and relates to an averaged structure, as indicated by the anisotropic displacement ellipsoids of heavy atoms (cf. Fig. 8b), the very short OW⋅⋅⋅O distances in the cavities and the deviations from the expected bond balance sums for the cations. Nevertheless, there is a more than satisfactory match between H2O content from TGA and within the structure, and between the PXRD patterns of both natural (including the neotypes; see Table 4 and Fig. 5) and synthetic montanite samples to the calculated pattern from the structure data of the SynCeramic sample using the subcell model. As a result of these observations, PXRD may be reliably used to determine if a (historical) specimen denoted as montanite does, in fact, contain montanite. Two samples listed in the International Centre for Diffraction Data (ICDD) database for montanite (ICDD 00-057-0626 from Župkov, Slovakia, Sejkora et al. 2004; and ICDD 00-038-0417 from the Novo-Boevskoe wolframite deposit, Ural Mountains, Russia, Pokrovskii and Yunikov 1967), along with a Richard Gaines specimen reported to bear a Genth label, presumably from Highland, Montana (Williams and Cesbron 1985) all match perfectly with the XRD pattern calculated on the basis of the SynCeramic SXRD data. Of these PXRD patterns, Sejkora et al. (2004) present the most detailed pattern; however, many of the reflections are not readily indexed by either the montanite subcell or supercell, suggesting that the montanite sample was intermixed with minority phase(s). Refining the unit-cell parameters on basis of the recorded PXRD patterns results in minor variation away from a = 9.1195(14) Å and c = 5.5694(8) Å, with the largest difference recorded for the Genth label-bearing specimen of Williams and Cesbron (1985), namely a = 9.02(1) and c = 5.598(1) Å. In general, the natural montanite samples have a smaller unit cell parameter a (average 9.042 Å) and larger parameter c (5.573 Å) compared to the synthetic samples (averages 9.101 and 5.553 Å, respectively). This variation in unit-cell parameters may be related to the ability of natural montanite to contain other cations such as Pb2+, Ca2+, or Cu2+ or to a varying structural water content.

Tetradymite is one of the most common primary minerals in both Bi and Te ores. It is even the major ore mineral in rare localities such as Dashuigou, Sichuan, China (Mao et al. 2002). The weathering products of tetradymite and other Bi–Te(–S) phases are generally yellow or orange powdery ochres, commonly observed coating rocks around Bi–Te(–S) ore. As a result of labelling any such earthy minerals as montanite, some historic ‘montanite’ specimens were incorrectly identified and contain a different Bi–Te oxysalt mineral. For instance, ‘montanite’ specimens from near Captains Flat, NSW, Australia (collected in the 1880s; Museums Victoria specimen numbers M2087 and M20910) turned out to be bodieite (Bi2(Te4+O3)2(SO4); only described in 2018; Kampf et al. 2018) following inspection by energy-dispersive spectroscopy (EDS) and PXRD. It is likely that the earthy porosity of montanite leads to its ability to adsorb non-structural water, potentially explaining why montanite is historically reported as Bi2TeO6·2H2O. Sejkora et al. (2004) attributed the first mass loss in a TGA analysis of montanite to 0.91 H2O molecules per formula unit, slightly greater than the structural maximum of 0.67 H2O pfu, perhaps due to adsorbed rather than structural water. Their subsequent finding of an additional 1.03 H2O molecules lost (total 1.94 H2O molecules) between 240 and 620 °C could be due to the beginning of a loss of volatile Te oxides (observed for SynPC1 montanite between 455 and 498 °C).

Understanding the crystal structures and compositions of the various Bi–Te(–S)–O minerals is important for determining the mobility of Bi and Te in the environment, along with the potential for these minerals to act as sinks for Bi and Te in the weathering zone (Golebiowska et al. 2011; Hou et al. 2005; Keim et al. 2018; Leverett et al. 2003). While the structures of five out of seven Bi–Te–O minerals are now known (except chiluite (Bi3Te6+Mo6+O10.5) and yecoraite (Fe3+3Bi5(Te6+O4)2(Te4+O3)O9·9H2O), both of which are reported to contain additional cations), future work may focus on determining thermodynamic properties of the Bi–Te secondary minerals. Some data exist for pingguite (Nénert et al. 2020), but other Bi–Te–O minerals have no reported thermodynamic data to our knowledge. Understanding the stability fields of these minerals would allow for better predictions of the behaviour of Bi and Te in the weathering zone. Overall, Bi–Te oxysalts are a fertile field of study both in mineralogy for describing new species and in environmental geology for characterising weathering Bi–Te(–S) ores, with further work required to fully characterise this group of secondary minerals.