Introduction
The minerals of the pearceite–polybasite group, general formula [M 6T 2X 7][M 10X 4] with M = Ag and Cu, T = As and Sb, and X = S, Se and Te, are fairly common in Nature. Eight members are known at present: argentopearceite (Sejkora et al., Reference Sejkora, Plášil, Makovicky, Škácha, Dolníček and Gramblička2020), argentopolybasite (Števko et al., Reference Števko, Mikuš, Sejkora, Plášil, Makovicky, Vlasáč and Kasatkin2023), benleonardite (Bindi et al., Reference Bindi, Stanley and Spry2015), cupropearceite (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a, Reference Bindi, Evain, Spry, Tait and Menchetti2007c), cupropolybasite (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a, Reference Bindi, Evain, Spry, Tait and Menchetti2007c), pearceite (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a), polybasite (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a), and selenopolybasite (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a, Reference Bindi, Evain and Menchetti2007b).
Their crystal structure can be described as a sequence, along the c axis, of two alternating layers: a [M 6T 2X 7]2– A layer and a [M 10X 4]2+ B layer (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a and references therein). As documented by Bindi et al. (Reference Bindi, Evain, Spry and Menchetti2007a), these minerals can exhibit three different polytypes: the high-temperature fast-ion conductivity form (Tac polytype), the partially ordered form (T2ac polytype), and the low-temperature fully ordered form (M2a2b2c polytype).
During an ongoing structural study of complex Ag(Cu)-bearing minerals in mineralogical collections from various museums, we found specimens containing Cu-free ‘polybasite’ (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a) from two different localities in North America. The same mineral has been recently approved by the International Mineralogical Association (IMA) as a new species with the name argentopolybasite (T2ac polytype) and the simplified formula Ag16Sb2S11 (corresponding to the structural formula [Ag6Sb2S7][Ag9AgS4]; Števko et al., Reference Števko, Mikuš, Sejkora, Plášil, Makovicky, Vlasáč and Kasatkin2023). Interestingly, the As-dominant silver end-member argentopearceite (Ag16As2S11; T2ac polytype), has also been recently discovered and described from the Lehnschafter mine, Czech Republic (Sejkora et al., Reference Sejkora, Plášil, Makovicky, Škácha, Dolníček and Gramblička2020). Those discoveries, in particular that related to argentopearceite, were surprising as it seemed that copper in the B layer was mandatory to stabilise these phases, especially when occurring in their disordered or partially ordered polytypes (Bindi et al., Reference Bindi, Evain and Menchetti2006a, Reference Bindi, Evain, Pradel, Albert, Ribes and Menchetti2006b; Reference Bindi, Schaper, Kurata and Menchetti2013).
Given the presence of ubiquitous twinning, mobile Ag+ and Cu+ cations, satellite reflections due to commensurately modulated structures (giving rise to different polytypes) or to composite modulated structures, partially occupied sites and strong pseudo-symmetries, it is mandatory to critically evaluate the structural models obtained. The existence of two Cu- and Se-free polybasite samples with excellent diffraction quality gives us the opportunity to critically elucidate this unusual crystal structure and the effect of Ag, Sb and Se on the B sites of the B layer.
Occurrence and chemical composition
The museum specimen (Fig. 1) used in the current work (Royal Ontario Museum, Canada; catalogue number M27183) is from Gowganda, Timiskaming District, Ontario, Canada, a well-established source of silver minerals. The geology of the deposit was studied and described by Andrews et al. (Reference Andrews, Owsiacki, Kerrich and Strong1986). Argentopolybasite occurs as black anhedral grains, up to 60 μm in length, with a black streak, associated closely with polybasite, calcite and chalcopyrite. An Ag-rich variety of polybasite from the same museum specimen was studied by Bindi and Menchetti (Reference Bindi and Menchetti2009), who did not realise the presence of argento-polybasite at that time. Argentopolybasite and polybasite show no macroscopic/microscopic differences.
A second specimen containing argentopolybasite (Fig. 2) was collected independently at the IXL mine, Silver Mountain mining district, Alpine County, California (Clark and Evans, Reference Clark and Evans1977), by Kyle Beucke and given for study to one of the authors (FNK). Associated minerals in the specimen include pyrite, argento-pearceite and acanthite. Both argentopolybasite and argentopearceite occur as black anhedral grains in a quartz gangue.
The chemical composition of argentopolybasite from both occurrences was determined using wavelength-dispersive analysis (WDS) by means of a JEOL JXA-8200 (University of Florence, Italy) and a JEOL JXA-8530F (Natural History Museum of Vienna, Austria) electron probe micro-analyser. Major and minor elements were determined at a 25 kV accelerating voltage and a 20 nA beam current, with a spot size of 2 μm (no surface damage was observed when using these conditions). For the WDS analyses of the argentopolybasite from Gowganda the following lines were used: SKα, FeKα, CuKα, ZnKα, AsLα, SeLα, AgLα, SbLβ, TeLα, AuMα and PbMα. For the WDS analyses of the argentopolybasite from the IXL mine the following lines were used: SKα, CuKα, AsLα, SeKα, AgLα, SbLα and PbMα. The standards employed for Gowganda were: native elements for Cu, Ag, Au and Te; galena for Pb; pyrite for Fe and S; synthetic Sb2S3 for Sb; synthetic As2S3 for As; synthetic ZnS for Zn; and synthetic PtSe2 for Se. The standards employed for the IXL mine were: lorándite for As; galena for Pb; chalcopyrite for Cu and S; native silver for Ag; stibnite for Sb; and synthetic Cu2Se for Se. The detection limit for minor elements was 0.01 wt.%.
The argentopolybasite fragments from both specimens were found to be homogeneous within analytical error. The average chemical compositions (N = 5 and 4 for argentopolybasite from Gowganda and IXL mine, respectively) are reported in Table 1. On the basis of 29 atoms, the formula can be written as (Ag16.00Cu0.02)Σ16.02(Sb1.95As0.04)Σ1.99S10.99 and (Ag16.04Cu0.02Pb0.01)Σ16.07(Sb1.37As0.56)Σ1.93(S10.46Se0.54)Σ11.00, for argentopolybasite from Gowganda and IXL mine, respectively.
Note: n.a. = not analysed; b.d.l. = below detection limit
X-ray crystallography
Unit-cell parameters for argentopolybasite from Gowganda are: a = 26.384(2), b = 15.232(1), c = 24.148(2) Å, β = 90.03(1)° and V = 9704.6(1) Å3, indicating the M2a2b2c polytype (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a). On the contrary, hexagonal unit-cell parameters for argentopolybasite from IXL mine are: a = 15.061(2), c = 12.308(2) Å and V = 2417.8(2) Å3, indicating the T2ac polytype (Bindi et al., Reference Bindi, Evain, Spry and Menchetti2007a). Given the much better diffraction quality and the fact that the trigonal polytype was already reported by Števko et al. (Reference Števko, Mikuš, Sejkora, Plášil, Makovicky, Vlasáč and Kasatkin2023), we present here only the structural data for argentopolybasite-M2a2b2c from Gowganda.
A small argentopolybasite fragment (0.055 × 0.040 × 0.032 mm in size) was extracted from the M27183 sample and mounted on a 5 μm diameter carbon fibre, which was, in turn, attached to a glass rod. As pearceite–polybasite minerals are usually twinned (Bindi et al., Reference Bindi, Nespolo, Krivovichev, Chapuis and Biagioni2020 and references therein), a full diffraction sphere was collected at ambient temperature using an Oxford Diffraction Xcalibur 3 single-crystal diffractometer. Refined unit-cell parameters are: a = 26.384(2), b = 15.232(1), c = 24.148(2) Å, β = 90.03(1)° and V = 9704.6(1) Å3 (Z = 16). Intensity integration and standard Lorentz-polarisation corrections were done with the CrysAlis RED (Oxford Diffraction, 2006) software package. The program ABSPACK of the CrysAlis RED package was used for the multi-scan absorption correction. Subsequent calculations were conducted with the JANA2006 program suite (Petříček et al., Reference Petříček, Dušek and Palatinus2006). The refinement of the structure was carried out in the space group C2/c starting from the atomic coordinates given by Bindi and Menchetti (Reference Bindi and Menchetti2009) for the crystal structure of Ag-rich polybasite-M2a2b2c. Site-scattering values were refined using scattering curves (Wilson, Reference Wilson1992) for neutral species for the Sb sites (Sb vs As) and for the Cu sites (Ag vs Cu) of the B layer (hereafter labelled B sites). The Sb sites and the B sites were found to be fully occupied by antimony and silver, respectively, and their occupancies were fixed during subsequent refinement cycles. With the introduction of twinning by metric merohedry (see Evain et al., Reference Evain, Bindi and Menchetti2006) the refinement smoothly converged to R = 0.103 for observed reflections [2σ(I) level], including all the collected reflections in the refinement. Indeed, the peculiar geometry of the pseudo-orthorhombic unit cell (with a ≈ 3½⋅b) makes a {110} twinning very probable. Based on this refinement, the analyses of the difference-Fourier synthesis maps suggested an additional twin law with a twofold axis, perpendicular to the previous threefold axis as a generator twin element, thus leading to a second-degree twin. The introduction of only three new parameters (the new twin volume ratios) dramatically lowered the R value to 0.055, although the new domains were rather small [volume percentages: 4.12(4)%, 3.05(2)% and 2.54(3)%].
A non-harmonic approach with a Gram–Charlier development of the Debye–Waller factors up to the third order (Johnson and Levy, Reference Johnson, Levy, Ibers and Hamilton1974; Kuhs, Reference Kuhs1984; Bindi and Evain, Reference Bindi and Evain2007) was then used to describe the electron density in the vicinity of four Ag atoms (i.e. Ag15, Ag22, Ag24 and Ag29) for which residues were found in the difference-Fourier maps. Using this approach, with anisotropic atomic displacement parameters for all the atoms and no constraints, the residual value settled at R = 0.0334 (Rw = 0.0915) for 11,867 independent observed reflections [2σ(I) level] and 443 parameters, and at R = 0.0380 (Rw = 0.0945) for all 18,461 independent reflections. Experimental details are given in Table 2. Atomic coordinates and bond distances are reported in Table 3 and 4, respectively. The crystallographic information file has been deposited with the Principal Editor of Mineralogical Magazine and is available as Supplementary material (see below).
Results and discussion
As illustrated in previous structural studies (Bindi et al., Reference Bindi, Nespolo, Krivovichev, Chapuis and Biagioni2020 and references therein), the crystal structure of the monoclinic polytype (M2a2b2c) of argentopolybasite (Fig. 3) can be described as the sequence, along the c axis, of two alternating layers: a [M 6T 2X 7]2– A (or A') layer and a [M 10X 4]2+ B (or B') layer (A,B and A',B' being related by a c glide reflection).
In the [M 6T 2X 7]2– A (or A') layer, each silver cation is three-fold coordinated by sulfur. The Sb atoms are the top of a trigonal pyramid with the three S atoms forming a base. In the [M 10X 4]2+ B (or B') layer, the 18 independent Ag atoms show different environments from quasi-linear to quasi-tetrahedral (Table 4). Argentopolybasite shows the presence of Ag at all the B structural positions of the B layer. The B positions [B1, B2 (4e) and B3 (8f)] are generally occupied by copper in pearceite–polybasite minerals in a nearly perfect linear coordination (see Bindi et al., Reference Bindi, Nespolo, Krivovichev, Chapuis and Biagioni2020 and references therein). In argentopolybasite, the mean Ag–S bond distances at the B sites, i.e. 2.399, 2.398 and 2.397 Å, for B1, B2 and B3, respectively, are very close to those observed for the linearly-coordinated Ag atoms of the B layer (Table 4) and exhibit bond-valence sums close to 1.00 (0.94–0.95 valence units, Brese and O'Keeffe, Reference Brese and O'Keeffe1991). In the trigonal T2ac polytype of argentopolybasite (Števko et al., Reference Števko, Mikuš, Sejkora, Plášil, Makovicky, Vlasáč and Kasatkin2023) the two B sites (labelled Ag11 and Cu12) show mean bond distances of 2.244 and 2.230 Å in the refined structure model of these authors (R = 0.0741; four-fold twinning was modelled). The two sites also show partial occupancy: Ag11 = 0.687(10)Ag and Cu12 = 0.82(3)Cu. This is bizarre because the B sites always act as a fully-occupied, ordered part surrounded by mobile cations in the B layer (Bindi et al., Reference Bindi, Evain and Menchetti2006a, Reference Bindi, Nespolo, Krivovichev, Chapuis and Biagioni2020). Furthermore, the occupancy of the B sites is strictly correlated with the way the mobile part (liquid-like structure) has been modelled and the approach considered to deal with the pervasive twinning.
To try to elucidate this discrepancy, we carried out a multi-regression analysis on 67 samples of the pearceite–polybasite group which were studied by both electron probe micro-analyser and single-crystal X-ray diffraction (Bindi et al., Reference Bindi, Nespolo, Krivovichev, Chapuis and Biagioni2020, references therein and unpublished data). The analysis allows modelling of the B–S distance of the linearly-coordinated Ag atom in the B layer on the basis of the Ag, Sb and Se contents, i.e. B–S (Å) = 2.101(8) + 0.213(5)Ag atoms per formula unit (apfu) + 0.037(5)Sb apfu + 0.012(2)Se apfu. The observed (derived from structure refinements) versus calculated (with the equation above) bond distances are shown in Fig. 4. Although there is not a large variation of the distribution of the B–S distances mainly due to the rarity of Cu-free members, the two values obtained by Števko et al. (Reference Števko, Mikuš, Sejkora, Plášil, Makovicky, Vlasáč and Kasatkin2023) for holotype argentopolybasite-T2ac are clearly off the main trend. This could imply either that the crystal they used for the structural study is not that analysed by electron microprobe or that the site occupancy and/or twinning was not modelled properly. To corroborate the hypothesis of a likely problem with the structure refinement there is the fact that the unit-cell and the overall geometry of the mineral studied by Števko et al. (Reference Števko, Mikuš, Sejkora, Plášil, Makovicky, Vlasáč and Kasatkin2023) perfectly fit the trend expected for Cu-free polybasite and argentopolybasite sensu stricto. In the structure of these minerals, disregarding the polytype, the B−S bond distances are mainly lined along the c axis (Fig. 3). This means that when Ag substitutes for Cu in these structural positions an increase in the c parameter and a consequent increase in the volume of the unit cell are expected. This is easily observable by plotting the unit-cell volume as a function of the copper content obtained by electron microprobe. To compare all the members belonging to the pearceite–polybasite group, the variation of the hexagonal subcell volume (i.e. a ≈ 7.5 and c ≈ 12 Å) was considered. Argentopolybasite of the present investigation and that studied by Števko et al. (Reference Števko, Mikuš, Sejkora, Plášil, Makovicky, Vlasáč and Kasatkin2023) (green and red circles in Fig. 5) are an excellent fit for the general trend observed for the minerals of the pearceite–polybasite group.
The discovery of argentopolybasite and argentopearceite demonstrates that there could be other surprises hidden in this complex group of minerals. It is critical to further study them for the advancement of the knowledge about processes relevant for Earth and to share that knowledge with others such as in environmental and material sciences.
Acknowledgements
We thank Federica Zaccarini, one anonymous reviewer, Pete Leverett, and the Associate Editor Frantisek Laufek for their constructive comments. LB wishes to thank MIUR-PRIN2017, project “TEOREM deciphering geological processes using Terrestrial and Extraterrestrial ORE Minerals”, protocol 2017AK8C32. FNK thanks the Harvard University Center for Nanoscale Systems (CNS), which is supported by the National Science Foundation under NSF award no. ECCS-2025158.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1180/mgm.2023.30.
Competing interests
The authors declare none.