Occurrence and mineral description

Occurrence and physical properties

Bianchiniite was found in only a few samples collected in the Sant'Olga level, Monte Arsiccio mine (43°58′N, 10°17′E), near Sant'Anna di Stazzema, Stazzema, Apuan Alps, Tuscany, Italy. The Monte Arsiccio mine exploited a pyrite ± baryte ± iron-oxide (magnetite, hematite and ‘limonite’) ore deposit located in the north-eastern sector of the Sant'Anna tectonic window, in the southern Apuan Alps. A review on its geological setting can be found in Biagioni et al. (Reference Biagioni, D'Orazio, Fulignati, George, Mauro and Zaccarini2020c) and references therein.

In type material, bianchiniite is associated with baryte, ‘hyalophane’ and ‘chlorite’, in a fracture of metadolostone (Fig. 1). After the first finding, which provided the material used for the type description, two additional specimens of bianchiniite were identified. The first one has the same associated minerals as observed in type material, plus galena in cuboctahedral crystals. Other phases identified in the same occurrence are sphalerite, derbylite, graeserite, siderite, dolomite, ankerite and quartz. Usually, crystals of bianchiniite are deeply altered to an X-ray amorphous earthy material (Fig. 1), in some cases admixed with microcrystalline rutile; in the second additional identified specimen, a relict of bianchiniite was observed at the core of a deeply altered crystal, confirming the identity of the original pseudomorphosed mineral. The crystallisation of bianchiniite is probably related to the circulation of As-rich hydrothermal fluids during the Alpine tectono-metamorphic event, under specific $f_{{\rm O}_ 2}$ conditions, within the metadolostone rock body embedded in the Monte Arsiccio ore deposit. The alteration affecting the large majority of the observed crystals of bianchiniite may suggest a successive alteration by low-T fluids, during the final stages of the evolution of the Monte Arsiccio mineral assemblages.

Fig. 1. (a) A loose aggregate of tabular crystals of bianchiniite, associated with ‘hyalophane’ and ‘chlorite’. Holotype material was sampled from this specimen. (b) Deeply altered earthy tabular crystals of bianchiniite associated with rhombohedral crystals of siderite and whitish microcrystalline fibrous aragonite. Both specimens from the Sant'Olga level, Monte Arsiccio mine, Stazzema, Apuan Alps, Tuscany, Italy (in a private collection).

Bianchiniite occurs as square tabular {001} crystals, up to 1 mm across. The colour is brown and the streak is brownish. The mineral is transparent, with a vitreous lustre. Owing to the small size of the grains available for the collection of mineralogical data, hardness was not measured. Bianchiniite is brittle, with a perfect {001} cleavage; fracture is irregular. Density was not measured; on the basis of the empirical formula and unit-cell volume refined from single-crystal X-ray diffraction data, the calculated density is 4.863 g/cm³.

The holotype material was embedded in epoxy and subsequently optical properties were determined on the polished crystal using reflected light microscopy. Reflectance was measured using an AVASPEC-UL2048×16 spectrometer attached to a Zeiss Axiotron UV-microscope (Swedish Museum of Natural History, Stockholm, Sweden), using a halogen lamp (100 W), with a measured field of 30 μm in diameter. In plane-polarised light, bianchiniite is grey in colour, non-pleochroic, with orange–yellow internal reflections. Between crossed polars, this mineral is weakly bireflectant, with a very weak anisotropy in shades of grey. Reflectance values (SiC as standard), measured in air, are given in Table 2. Mean refractive index, calculated according to the Gladstone–Dale relationship (Mandarino, Reference Mandarino1979, Reference Mandarino1981), is 2.044.

Table 2. Reflectance data (%) for bianchiniite in air.*

* The reference wavelengths required by the Commission on Ore Mineralogy (COM) are given in bold

Chemical and micro-Raman data

Quantitative chemical data were collected using a Cameca SX50 electron microprobe at the Istituto di Geologia Ambientale e Geoingegneria, CNR, Rome, Italy. The following analytical conditions were used: wavelength dispersive mode, accelerating voltage = 15 kV, beam current = 15 nA and beam size = 1 μm. Standards (element, emission line) were: rutile (TiKα), vanadinite (VKα), magnetite (FeKα), GaAs (AsLα), InSb (SbLα), celestine (SrLα), baryte (BaLα), galena (PbMα) and fluorophlogopite (FKα). Chromium and Cl were sought but were below the detection limit. Chemical data are given in Table 3. Vanadium, Fe and As were considered as V³⁺, Fe³⁺ and As³⁺, respectively, on the basis of the crystal structure refinement (see below). On the basis of 12 anions per formula unit, the empirical formula of bianchiniite is (Ba_2.00Sr_0.04Pb_0.02)_Σ2.06(Ti⁴⁺_1.14V³⁺_0.44Fe³⁺_0.42)_Σ2.00[(As_3.96Sb_0.02)_Σ3.98O₁₀](O_1.18F_0.82), corresponding to the end-member formula Ba₂(Ti⁴⁺V³⁺)(As₂O₅)₂OF, in accordance with the CNMNC-IMA guidelines (Bosi et al., Reference Bosi, Hatert, Hålenius, Pasero, Miyawaki and Mills2019). This latter formula corresponds to (in wt.%) TiO₂ 9.20, V₂O₃ 8.63, As₂O₃ 45.57, BaO 35.32, F 2.19, sum 100.92, –O =  F –0.92, total 100.

Table 3. Electron microprobe data (in wt.% – average of 10 spot analyses) and atoms per formula unit for bianchiniite.

S.D. – standard deviation

Micro-Raman spectra of bianchiniite were collected in near back-scattered geometry with a Horiba Jobin-Yvon Xplora Plus apparatus, equipped with a motorised x–y stage and an Olympus BX41 microscope with a 100× objective (Dipartimento di Scienze della Terra, Università di Pisa). The 532 nm line of a solid-state laser was used. The minimum lateral and depth resolution was set to a few μm. The system was calibrated using the 520.6 cm^–1 Raman band of Si before each experimental session. Spectra were collected through multiple acquisitions (3) with single counting times of 60 s, with the laser power filtered at 50% (i.e. 12.5 mW). No thermal damage was observed. Back-scattered radiation was analysed with a 1200 gr/mm grating monochromator. The Raman spectrum of bianchiniite in the region between 100 and 1200 cm^–1 is shown in Fig. 2. The interpretation of the Raman bands is not straightforward, as very few Raman spectroscopic studies on diarsenites are available (e.g. Čejka et al., Reference Čejka, Bahfenne, Frost and Sejkora2008; Bahfenne et al., Reference Bahfenne, Rintoul, Langhof and Frost2012). Bands between 750 and 850 cm^–1 may be related to As–O_nbr stretching modes, where subscript ‘nbr’ indicates ‘non-bridging’ oxygen atoms in the (As₂O₅) groups, in agreement with Bahfenne et al. (Reference Bahfenne, Rintoul, Langhof and Frost2012) and the theoretical calculations of Tossell (Reference Tossell1997). In particular, the band at 766 cm^–1 seems to be the most probable candidate for this assignment, as it is very close to the value observed in fetiasite (770 cm^–1; Graeser et al., Reference Graeser, Schwander, Demartin, Gramaccioli, Pilati and Reusser1994) and interpreted by Tossell (Reference Tossell1997) as due to the As–O_nbr stretching vibrations. Other bands in this range, as well as the bands at 892 cm^–1 can be also related to other M–O modes, where M can be Ba or (Ti,V,Fe). Bands in the range between 480 and 660 cm^–1 are attributed to the As–O_br stretching mode, where subscript ‘br’ indicates ‘bridging’ oxygen atoms. More questionable is the correct attribution of the strong band at 658 cm^–1. Indeed, this band is similar to the one reported in the infrared spectrum of fetiasite (at 660 cm^–1; Graeser et al., Reference Graeser, Schwander, Demartin, Gramaccioli, Pilati and Reusser1994) and interpreted by Tossell (Reference Tossell1997) as due to the As–O_br stretching mode. However, these values do not match with the experimental data given for paulmooreite, with bands at 562 and 503 cm^–1, and with the theoretical ones reported by Tossell (Reference Tossell1997), who calculated stretching wavenumbers at 554 and 496 cm^–1. Consequently, the strong band at 658 cm^–1 could have another interpretation; for instance, it may be due to (Ti,V,Fe)–O vibrations. The band at 295 cm^–1 may be due to deformations of the As–O_nbr bonds (e.g. Bahfenne et al., Reference Bahfenne, Rintoul, Langhof and Frost2012). Other bands below 300 cm^–1 can be probably attributed to lattice or M–O bonds. In paulmooreite, bands at 186 and 138 cm^–1 were observed (Bahfenne et al., Reference Bahfenne, Rintoul, Langhof and Frost2012), comparable to those at 178 and 140 cm^–1 measured in bianchiniite. No spectral features were observed in the range 2800–4000 cm^–1, thus supporting the absence of (OH)^– and H₂O groups in bianchiniite.

Fig. 2. Micro-Raman spectrum of bianchiniite in the range between 100 and 1200 cm^–1.

Crystallography

Powder X-ray diffraction data of bianchiniite were collected using a 114.6 mm Gandolfi camera and Ni-filtered CuKα radiation (Dipartimento di Scienze della Terra, Università di Pisa). Table 4 gives the observed powder X-ray diffraction lines, along with the calculated pattern based on the structural model discussed below. Unit-cell parameters were refined in a tetragonal setting on the basis of 16 unequivocally indexed reflections using the software UnitCell (Holland and Redfern, Reference Holland and Redfern1997), giving the following results: a = 8.7894(10), c = 15.637(2) Å and V = 1208.0(2) Å³.

Table 4. X-ray powder diffraction data (d in Å) for bianchiniite.*

* Notes: intensity (I) and d _hkl were calculated using the software PowderCell 2.3 (Kraus and Nolze, Reference Kraus and Nolze1996) on the basis of the structural model given in Table 6. Only the reflections with I _calc > 2 are given, if not observed. Intensities were estimated visually: vs = very strong; w = weak; vw = very weak. The eight strongest reflections are given in bold.

Single-crystal X-ray diffraction data were collected using a Bruker Smart Breeze diffractometer equipped with an air-cooled Photon II CCD detector, with graphite-monochromatised MoKα radiation (Dipartimento di Scienze della Terra, Università di Pisa). The detector-to-crystal distance was set at 50 mm. A total of 653 frames were collected in ω and φ scan modes in 0.5° slices, with an exposure time of 25 s per frame. Intensity data were corrected for Lorentz-polarisation factors and absorption using Apex3 (Bruker AXS Inc., 2016). Bianchiniite is tetragonal, with unit-cell parameters a = 8.7266(4), c = 15.6777(7) Å and V = 1193.91(12) Å³. The c:a ratio is 1.7965. Reconstructed precession images showed very weak reflections, occurring at low 2θ values and apparently doubling the a parameter (Fig. 3), supporting a superstructure with a’ = 2a and c’ = c. Owing to their weakness, they were ignored and the crystal structure was solved and refined in the tetragonal cell reported above. However, their possible meaning is discussed below.

Fig. 3. Reconstructed precession image of the hk0 reciprocal lattice plane. Weak not-indexed reflections at lower angles can be observed (some of them are indicated by thin dotted arrows). Only positive h and k indices are shown.

The statistical tests on the distribution of |E| values (|E ² – 1| = 0.877) suggested the occurrence of an inversion center. The examination of systematic absences indicated the space group symmetry I4/mcm. The crystal structure of bianchiniite was solved using Shelxs-97 (Sheldrick, Reference Sheldrick1997) and refined using Shelxl-2018 (Sheldrick, Reference Sheldrick2015). Scattering curves for neutral atoms were taken from the International Tables for Crystallography (Wilson, Reference Wilson1992). During the first step of the crystal structure solution, the positions of Ba, Ti, As, and one anion position were found. Difference-Fourier synthesis maps located three additional anion positions. The isotropic structural model refined to R ₁ = 0.038, thus suggesting the correctness of the structural model. Taking into account the electron microprobe data, the site occupancy factors (s.o.f.) at the Ba(1) and As(3) sites were refined using the Ba and As scattering curves only, respectively. The M(2) site is occupied by Ti, V and Fe. Owing to the similar atomic number of Ti (Z = 22) and V (Z = 23) and the slightly higher atomic number of Fe (Z = 26), the s.o.f. was refined using the Ti vs. Fe scattering curves. In addition, taking into account the different crystal-chemical behaviour of Ti and (Fe,V) (with the former giving rise to the off-centre displacement phenomenon; e.g. Megaw, Reference Megaw1968), their positions were refined independently, resulting in the splitting of the M(2) site into M(2a) and M(2b). Titanium was located at M(2a); indeed this position shows the largest polyhedron distortion value, estimated as Δd = (d _max – d _min), where d _max and d _min represent the maximum and minimum bond distance, respectively, whereas M(2b) has a more regular coordination environment (see below). A bond-valence calculation performed on the isotropic model indicated the likely mixed (O,F) nature of the O(1) and O(4) sites. Owing to the very similar scattering factors of O (Z = 8) and F (Z = 9), the s.o.f. was fixed on the basis of chemical data and bond-valence requirements. The refinement statistical factor was slightly improved to R ₁ = 0.026. Successively, an anisotropic model for cations lowered it to R ₁ = 0.016. After several cycles of anisotropic refinement for all atoms, the final statistical factor R ₁ converged to 0.0134 for 555 unique reflections with F _o > 4σ(F _o) and 34 refined parameters. Details of the intensity data collection and crystal structure refinement are given in Table 5. Atom coordinates, s.o.f. and displacement parameters are reported in Table 6, whereas selected bond distances are listed in Table 7. Weighted bond-valence sums, calculated using the bond parameters of Gagné and Hawthorne (Reference Gagné and Hawthorne2015) for cation–oxygen pairs and of Brese and O'Keeffe (Reference Brese and O'Keeffe1991) for cation–fluorine pairs, are shown in Table 8. The crystallographic information file, including reflection data, has been deposited with the Principal Editor of Mineralogical Magazine and is available as Supplementary material (see below).

Table 5. Crystal data and details of data collection and crystal structure refinement for bianchiniite.

Table 6. Sites, Wyckoff positions, site occupation factors (s.o.f.), fractional atomic coordinates and equivalent isotropic displacement parameters (Ų) for bianchiniite.*

* Note: the refined site scattering at M(2a) + M(2b) matches the site population (Ti_0.57V_0.22Fe_0.21) obtained from electron microprobe analysis.

Table 7. Selected bond distances (in Å) for bianchiniite.

Table 8. Weighted bond-valence balance (in vu) for bianchiniite.*

* Notes: left and right superscripts indicate the number of equivalent bonds involving anions and cations, respectively. For sites with mixed site occupancy, the bond valence sums have been weighted.

Crystal structure description

The crystal structure of bianchiniite is formed by columns of corner-sharing M(2) octahedra running along c, connected, in the {001} plane, by (As₂O₅) dimers. Trigonal pyramidally coordinated As³⁺ ions form {001} layers, with their apex pointing alternatively up and down. Lone-electron-pair micelles of As³⁺ atoms alternate, along c, with layers of intercalated Ba atoms (Fig. 4). Consequently, the crystal structure may be described as layered, with heteropolyhedral layers hosting Ba, (Ti,V,Fe) and As atoms, separated by layers occupied by lone-pair electrons and the anion hosted at O(4) connecting successive M(2) octahedra along c.

Fig. 4. Crystal structure of bianchiniite as seen down [110] (a) and [001] (b). Symbols: large violet spheres = Ba(1) site; green spheres = F(1) site; red spheres = O(2)–O(4) sites. Light blue octahedra = M(2) site [only the M(2a) position is shown]. For the sake of clarity, As(3) sites belonging to different heteropolyhedral layers are shown in dark magenta and pink, respectively. Blue ellipses indicate the location of As lone-pair electrons. Black dashed lines indicate the unit cell. Bracket indicates one heteropolyhedral layer.

Barium is twelve-fold coordinated, with ten bond distances in the range between 2.79 and 2.98 Å. Two additional bonds with the mixed (F,O) site F(1) at 3.22 Å complete the coordination environment, that can be described as a bicapped pentagonal prism. The average bond distance is 2.942 Å, to be compared with a calculated value of 2.985 Å based on ionic radii for ^[XII]Ba²⁺, ^[IV]O^2–, ^[II]O^2– and ^[II]F^– given by Shannon (Reference Shannon1976). The bond-valence sum at the Ba atom is 2.16 vu, in keeping with the expected value.

Arsenic displays the trigonal pyramidal coordination typical of As³⁺. The <As–O> bond distance, 1.772 Å, agrees with the average distance reported by Majzlan et al. (Reference Majzlan, Drahota, Filippi, Bowell, Alpers, Jamieson, Nordstrom and Majzlan2014), i.e. 1.782 Å. It is worth noting that in the (As₂O₅) dimers, As–O_br is definitely longer than the two other As–O_nbr bonds, i.e. 1.838 and 1.740 Å, respectively. This feature was observed by Cooper and Hawthorne (Reference Cooper and Hawthorne2016) in schneiderhöhnite, where a bridging O atom has a comparatively large valence contribution from non-As atoms. These authors examined the observed contribution in other polymerised As³⁺ oxysalts (see Table 5 of Cooper and Hawthorne, Reference Cooper and Hawthorne2016), reporting that schneiderhöhnite showed the largest bond-valence contribution from non-As atoms in known minerals (i.e. 0.28 vu). Bianchiniite shows a still larger contribution, represented by the incident bond valence from Ba²⁺ (0.38 vu), partially compensated through the elongation of the As–O_br distance. The bond-valence sum at the As(3) site is 3.04 vu.

Titanium, along with minor V and Fe (both assumed in the +3 oxidation state – see below), are hosted at the M(2) site. As revealed by the crystal structure refinement, this site is actually split into two sub-positions, namely M(2a) and M(2b). The former is more distorted (Δd = 0.327 Å) than the latter (Δd = 0.140 Å) and this geometrical feature suggests that, as discussed above, Ti⁴⁺ may prefer the M(2a) position. The refined site scattering value at M(2a) + M(2b) is 23.04 electrons, to be compared with 23.06 electrons calculated from chemical data. Bond-valence sums at the two split sub-sites was calculated assuming Ti⁴⁺ ordered at M(2a) along with minor V³⁺ and Fe³⁺, and a mixed (V,Fe) occupancy for M(2b). As it is very difficult to establish a partitioning of V³⁺ and Fe³⁺ over these two sub-positions (both elements have a very similar ionic radius, i.e. 0.640 and 0.645 Å, respectively; Shannon, Reference Shannon1976), an equal distribution of both cations at the two split sites was assumed. As a result, the following site populations are proposed (with rounding errors), which match the observed site scatterings: ^{M (2a)}(Ti_0.57V³⁺_0.08Fe³⁺_0.07) and ^{M (2b)}(V³⁺_0.14Fe³⁺_0.14). On this basis, taking into account the ionic radii given by Shannon (Reference Shannon1976), one may expect that M(2a) should have the smaller average bond distance. On the contrary, M(2b) is slightly smaller that M(2a), i.e. 1.969 vs. 1.989 Å, respectively. This discrepancy may be related to the larger degree of polyhedral distortion of M(2a) with respect to M(2b) (the mean bond distance increases with increasing polyhedral distortion; e.g. Bosi, Reference Bosi2014) and the average nature of the ligands coordinating these two split positions. The bond-valence sums at these two split positions [i.e. 2.63 and 0.91 vu, respectively, for M(2a) and M(2b)], agrees with a mixed [Ti⁴⁺,(Fe,V)³⁺] occupancy.

The four independent anion sites have bond-valence sums ranging between 1.20 and 2.10 vu. The O(2) and O(3) sites, with bond-valence sums of 2.06 and 2.10 vu, host O^2– anions, whereas the other two sites are underbonded. This underbonding is slight at the O(4) site (1.68 vu), suggesting a partial replacement of O^2– by a monovalent anion, whereas at the fourth anion site the underbonding is severe, with 1.20 vu. This supports the dominance of a monovalent anion at this position. These results are in keeping with the occurrence of F, as detected during electron microprobe analysis; accordingly, this anion position was labelled as F(1). The site population at O(4) was fixed at (O_0.80F_0.20), whereas F(1) was modelled as (F_0.70O_0.30). According to the s.o.f., it appears that the occupancy of M(2a) is probably related with the occurrence of F^– at F(1) and O^2– at O(4), whereas when M(2b) is occupied it is associated with O^2– at F(1) and F^– at O(4).