Kobellite homologues from the Boliden Au–Cu–(As) deposit, Sweden: jigsaw patterning via nanoscale intergrowths in chessboard structures

Abstract

Sulfosalt assemblages in a specimen from the Boliden Au–Cu–(As) deposit in northern Sweden, comprise micrometre to nanometre scale intergrowths of Se-rich izoklakeite and tintinaite with average formulae and calculated homologue number (N) given as: (Cu_1.88Fe_0.18)_2.06(Pb_22.92Ag_1.47Cd_0.01Zn_0.01)_24.41(Sb_13.12Bi_8.69)_21.8(S_50.19Se_6.43Te_0.12)_56.73, N = 3.83, and (Cu_1.31Fe_0.74)_2.05(Pb_10.58Ag_0.18Cd_0.05Zn_0.02)_10.83(Sb_10.2Bi_5.23)_15.43(S_32.22Se_2.46)_34.7, N = 2.05, respectively. Tintinaite coexists with (Bi, Se)-rich jamesonite. High-angle annular dark field scanning transmission electron microscopy (HAADF STEM) imaging reveals chessboard structures comprising PbS and SnS modules with the number of atoms in the octahedral (M) sites counted as: n1 = 18 and n2 = 8 for tintinaite and n1 = 30 and n2 = 16 for izoklakeite. The homologue number can be calculated using the formula: N = (n1/6)–1 and N = n2/4 for PbS and SnS modules giving N_Tti = 2 and N_Iz = 4. A new N = 3 homologue, defined by n = 12 and n = 24 SnS and PbS modules, respectively, is identified as single or double units within areas with intergrowths between kobellite and izoklakeite. HAADF STEM imaging also reveals features attributable to lone electron pair micelles within the Sb-rich kobellite homologues. Atomic-resolution EDS STEM chemical mapping of Pb–Bi–Sb-sulfosalts shows a correlation with crystal structural modularity. The maps also highlight sites in the SnS modules of tintinaite in which Sb > Bi. Coherent nanoscale intergrowths between tintinaite and izoklakeite define jigsaw patterns evolving from chessboard structures and are considered to have formed during co-crystallisation of the two phases. Displacement textures and crosscutting veinlets (a few nm in width) are interpreted as evidence for superimposed syn-metamorphic deformation and are associated with the redistribution of Bi and Se. Imaging and mapping using HAADF STEM techniques is well suited to characterisation of Pb–Sb–Bi-sulfosalt phases, offering largely untapped potential to unravel the evolution of chessboard structures with applications across mineralogy but also extending into allied fields.

Introduction

The kobellite homologous series is described by the formula (Cu,Fe)₂M_10N+6S_llN+13 for a 4 Å repeat where M = Pb, Bi and Sb; and N = homologue number (Makovicky and Mumme, 1986; Zakrzewski and Makovicky, 1986). Natural phases of the kobellite homologous series include kobellite and tintinaite with N = 2, and giessenite and izoklakeite with N = 4.

Tintinaite (Sb-rich) and kobellite (Bi-rich) are isostructural and form a continuous solid-solution pair with general formula: (Cu_2–xFe_x)[Pb_10–2z+x(Ag,Cu)_z(Sb_1–yBi_y)_16+z–x]₂₆S₃₅; 0 < x < 1; 0 < y < 0.67; 0 < z < 0.54 (Moëlo et al., 1995). Tintinaite from the type locality (TL) Tintina Mine, Yukon, Canada, was defined as the Sb-pure equivalent of kobellite from TL Vena, Sweden (Harris et al., 1968; Moëlo et al., 1984). The two species have space group Pnnm and cell parameters: a = 22.30 Å; b = 34.00 Å; and c = 4.04 Å for TL tintinaite (Harris et al., 1986), and a = 22.62 Å; b = 34.08 Å; and c = 4.04 Å for TL kobellite (Nuffield, 1948).

Izoklakeite, ideally [(Cu,Fe)₂Pb_26.5(Sb,Bi)_19.5S₅₇], is orthorhombic (space group Pnnm; a = 34.07 Å, b = 37.98 Å, and c = 4.072 Å, for specimen from Vena, Sweden; Zakrzewski and Makovicky, 1986) whereas its Bi-dominant analogue giessenite, ideally [(Cu,Fe)₂ Pb_26.4(Bi,Sb)_19.6S₅₇], is monoclinic (space group P2₁/n a = 34.34 Å; b = 38.05 Å; c = 4.06 Å and β = 90.33° for specimen from Bjorkåsen, Norway; Makovicky and Karup–Møller, 1986). Although no Sb-pure izoklakeite has been described, Bi-rich izoklakeite specimens with orthorhombic symmetry are documented from Lake Zervreila, Switzerland (Iz:Gie ratio 40:60; Armbruster and Hummel, 1987) and Gutaishan, China (Iz:Gie ratio 35:65; Li et al., 2019). The general formula for a quasi-solid solution between izoklakeite and giessenite was given as: (Cu_2–xFe_x)[Pb_26–2z+x(Ag,Cu)_z(Sb_1–yBi_y)_20+z–x]₄₆S₅₇; 0 < x < 1; 0.44 < y < 0.85; 0 < z < 2.2 (Moëlo et al., 1995), even though the exact Sb:Bi ratio at which the symmetry changes is unknown (Moëlo et al., 2008).

Distinguishing among these sulfosalt phases, and also compositionally similar but structurally unrelated phases such as jaskólskiite (Cu_xPb_2+x(Sb,Bi)_2–xS₅ (x ≈ 0.15; Zakrzewski, 1984; Makovicky and Nørrestam, 1985), or jamesonite (FePb₄Sb₆S₁₄) (Niizeki and Buerger, 1957) can be difficult based solely on compositional data (e.g. electron probe microanalysis; EPMA). Unequivocal identification from EPMA data is impossible if the phases are intergrown at the sub-micrometre scale.

A prior study (Li et al., 2019) demonstrated the utility of the high-angle annular dark field scanning transmission electron microscope (HAADF STEM) technique for visualising the crystal structures of coexisting kobellite and izoklakeite based on their characteristic modular structures. Here, we characterise intergrowths of Sb-rich kobellite homologues, modular minerals that extend down to the nanoscale and thus yield compositions at the scale of the microprobe beam that cannot be readily reconciled with one species or the other.

Boliden is a Palaeoproterozoic metamorphosed massive sulfide deposit in the Skellefte district of northern Sweden with characteristics interpreted as an ancient analogue to a modern high-sulfidation environment (Weihed et al., 1996). The deposit was known for its high gold grades (average 15 ppm), massive arsenopyrite orebodies, and highly complex ore mineral assemblages (Grip and Wirstam, 1970; Wagner and Jonsson, 2001). An important part of the deposit is represented by sulfosalt-rich veins that host, alongside kobellite, a variety of other ore minerals including several Se-bearing sulfosalts (e.g. tetrahedrite, bournonite, boulangerite and meneghinite), Bi-selenides and -tellurides (tellurobismuthite, tetradymite and laitakarite), Hg-rich gold–silver alloys, gudmundite and native bismuth (Ödman, 1941; Grip and Wirstam, 1970; Wagner and Jonsson, 2001; Wagner et al., 2007).

Selenium-rich kobellite has been reported previously from Boliden (Moëlo et al., 1984 (reanalysis of a sample from Harris et al., 1968); Wagner and Jonsson, 2001; Mumme et al., 2013). Whereas the crystal structure of this Se-rich kobellite (Mumme et al., 2013) differs only in minor ways from that determined for kobellite by Miehe (1971), the Sb concentration can vary in Boliden kobellite. A Sb/(Bi+Sb) ratio of ~0.5 is reported by Moëlo et al. (1984) and Mumme et al. (2013) and ~0.6 by Wagner and Jonsson (2001). On this basis, the Boliden kobellite reported by Wagner and Jonsson (2001) will hereafter be referred to as tintinaite.

This contribution describes co-existing Se-rich kobellite homologues (izoklakeite and tintinaite) that form micro- to nanoscale intergrowths in a specimen from the sulfosalt-rich veins at Boliden. By comparison with documented occurrences worldwide, we show these are among the most Sb-rich varieties that also contain Bi yet reported. We discuss the nature of the nanoscale intergrowths and show the presence of both primary and secondary textures. The primary intergrowths lead to formation of a new kobellite homologue during sulfosalt co-crystallisation, whereas the secondary textures record nanoscale remobilisation of Bi and Se during vein shearing. We aim to show how HAADF STEM imaging of modular structures is an appropriate tool to validate their building blocks and provide atomic-scale insights into formation of new phases as well as the primary versus secondary/overprinting origin of observed intergrowths.

Analytical methodology

The sample investigated (specimen 1061) originates from the Boliden Mineral AB collection (owned by Boliden AB, Sweden), that formed the basis of the first description of the deposit by Ödman (1941). The sample was collected from the 250 m level E in the mine, in room 28 ~213–216 metres away from the western part. Ore samples were mounted, polished and examined under reflected light and in back-scatter electron mode using an FEI Quanta 450 Field Emission Gun scanning electron microscope equipped with a silicon-drift energy-dispersive X-ray spectrometer (Adelaide Microscopy, The University of Adelaide).

Quantitative compositional data (spot analyses) and element-distribution maps were obtained using a Cameca SX-Five electron probe microanalyser in the same laboratory. The instrument runs PeakSite v6.5 software for microscope operation and Probe for EPMA software for data acquisition and processing. The instrument is equipped with five tunable wavelength-dispersive spectrometers and was operated at 20 kV and 20 nA with a focussed beam. X-ray lines were as follows: S (Kα), Pb (Mα), Se (Lα), Te (Lα), Bi (Mα), As (Lα), Fe (Kα), Cu (Kα), Mn (Kα), Ag (Lα), Zn (Kα), Sb (Lα) and Cd(Lα). All standards used were from P&H and Associates: galena for Pb and S; chalcopyrite for Cu and Fe; synthetic guanajuatite (Bi₂Se₃) for Bi and Se; gallium arsenide for As; galena for Pb; rhodonite for Mn; silver telluride for Ag; HgTe for Te; sphalerite for Zn; stibnite for Sb; and greenockite for Cd. Count times were 20 s on peak for all elements except Pb and S (15 s). Due to the complexity of off-peak interferences, traditional two point background analysis was initially acquired (20 s high and 20 s low), then through a function of the Probe for EPMA software these were turned into multipoint backgrounds utilising an exponential curve for background fitting. On-peak interferences were corrected using a range of standards in Probe for EPMA. Average minimum detection limits (99% C.I. in wt.%) were 0.008 for S; 0.05 for Pb and Ag; 0.04 for Bi and Cd; 0.03 for Se, Te, As, Cu and Zn; and 0.02 for Te, Sb, Fe and Mn. Qualitative X-ray mapping utilising the same instrument were acquired using Probe Image at settings of 20 kV and 200 nA at a pixel resolution of ~1 μm. Pixel dwell time was set to 2000 ms for enhanced sensitivity. Intensity maps were then coloured using Surfer 10® software distributed by Golden Software.

Additional element-distribution mapping was undertaken by laser ablation inductively coupled plasma mass spectrometry (LA–ICP–MS) using a RESOlution–LR 193 nm excimer laser attached to an Agilent 7900x Quadrupole ICP–MS. The following suite of isotopes were measured: ³⁴S, ⁵⁷Fe, ⁵⁹Co, ⁶⁰Ni, ⁶⁵Cu, ⁷⁵As, ⁷⁷Se, ¹⁰⁷Ag, ¹²¹Sb, ¹²⁵Te, ¹⁹⁷Au, ²⁰⁶Pb and ²⁰⁹Bi. LA–ICP–MS trace-element distribution maps were generated by ablating sets of parallel line rasters in a grid across the sample surface. A square laser beam profile was used throughout with spot size varying between 11 and 16 μm to provide adequate spatial resolution for each map. Spacings between adjacent trenches and laser beam scan speed were modified to match spot size. Laser fluence was ~3.5 J/cm² using a repetition rate of 10 Hz. A 10 s background measurement was acquired before the ablation of each trench, followed by a 20 s delay after ablation. Two 74 μm-diameter spot analyses were made on STDGL3 and GSD-1G reference materials at 10 Hz before and after each map. All LA–ICP–MS trace-element distribution maps were processed with Iolite software.

Thinned <100 nm-thick foils for nanoscale investigation were prepared in situ from the surface of a polished mount attached to a copper grid using focused ion beam – scanning electron microscopy techniques. This was carried out on a FEI–Helios nanoLab dual focused ion beam and scanning electron microscope (FIB–SEM) at Adelaide Microscopy, The University of Adelaide. Extraction and thinning procedures followed Ciobanu et al. (2011).

HAADF STEM imaging and energy-dispersive spectroscopy (EDS) analysis reported here were performed using an ultra-high resolution, probe-corrected, FEI Titan Themis S/TEM in the same laboratory and previously used for imaging sulfosalts structures (e.g. Ciobanu et al. 2016; Li et al., 2019). This instrument is equipped with a X-FEG Schottky source and Super-X EDS geometry. The Super-X EDS detector provides geometrically symmetric EDS detection with an effective solid angle of 0.8 sr. Probe correction delivered sub-Ångstrom spatial resolution and an inner collection angle greater than 50 mrad was used for HAADF imaging with a Fischione detector. Image acquisition, including a drift-corrected frame integration (DCFI) package and processing, was performed using both DigitalMicrograph^TM and Thermo-Scientific Velox software. Various filters were used to eliminate noise. EDS mapping was performed using Thermo-Scientific Velox software, which utilises standard Cliff–Lorimer quantification and included absorption correction optimised for both the Super-X detector geometry and also the effect of sample holder shadowing for the double-tilt Super-X holder used.

Diffraction indexing was performed using Winwulff© (JCrystalSoft) software. Data are stored in the American Mineralogist Crystal Structure Database (http://rruff.geo.arizona.edu/AMS/amcsd.php). Crystal structure models were generated in CrystalMaker® and image simulations using STEM for xHREM ^TM (v4.1) software.

Results

Specimen petrography – micrometre-scale characterisation

Both polished blocks analysed (1061 and 1061a) feature lamellar intergrowths between Pb–Sb–Bi-sulfosalts, identified as tintinaite and izoklakeite, and Bi–Pb chalcogenides which are readily distinguishable from the sulfosalts by their colour and brightness in reflected light (Fig. 1a–c). The bismuth minerals are associated with minor chalcopyrite, either interstitial within, or at the margins of the sulfosalts, and with pyrite, a major component of polished block 1061 (Supplementary Fig. S1). The three areas selected for nanoscale study (all in polished block 1061a) targeted contacts between acicular tintinaite lamellae and the Bi–Pb chalcogenide and the mutual contacts between the two sulfosalt species (Fig. 1d–f). Bi–Pb chalcogenide species are dominated by a new member of the aleksite series with the formula PbBi₆(Te,Se)₆(S,Se)₄ calculated from EPMA. The composition and crystal structure of this phase will be reported in detail elsewhere.

Figure 1. Reflected light microphotographs (a–c) and back-scattered electron (BSE) microscopy images (d–f) showing micrometre-scale intergrowths between tintinaite (Tti), izoklakeite (Iz) and Pb–Bi-chalcogenides (Pb-Bi-chalc) in sample/polished block 1061a. Note minor chalcopyrite (Ccp) is also present. Areas studied at the nanoscale are marked on the overall map of the sample (a). (d–f) Detailed images of areas 1–3 showing the stamps of FIB cuts from which the analysed S/TEM foils were prepared as well as the profiles for EPMA spot analyses (red lines). Tti‒tintinaite.

Trace-element distribution maps from LA–ICP–MS (polished block 1061) show that the sulfosalts concentrate minor Ag and Cd, whereas Se is a major component of the Bi-chalcogenide (Fig. 2a). A pyrite grain with a corroded margin at the mutual boundary with sulfosalts displays a Co-enriched rim but patchy distribution of As and Ni. Specks of relatively high Au concentration occur across the boundary between pyrite and the sulfosalt. Such textures are indicative of small-scale trace-element remobilisation during an overprinting event. This is also illustrated by the presence of laitakarite, [(Bi₄(Se,S)₃], among other Bi-chalcogenide species from the tetradymite series, which occurs along trails crosscutting pre-existing phase boundaries (e.g. between sulfosalt and aleksite series species, as seen on BSE images and EPMA maps; Fig. 2b, c). Fine-scale intergrowths between izoklakeite and tintinaite are readily seen on EPMA element-distribution maps for Pb (Fig. 2d).

Figure 2. (a) LA–ICPMS element-distribution maps, (b) BSE image, and (c, d) EPMA element-distribution maps showing micrometre-scale distribution of main and minor/trace elements within intergrowths of tintinaite (Tti), izoklakeite (Iz) and Pb–Bi-chalcogenides (sample 1061). Note that Se is richest in the Pb–Bi-chalcogenide in (a, c, d). Silver is higher in the sulfosalt intergrowths whereas Au occurs as minute inclusions (high intensity spots) more abundant across a transect from pyrite into the Bi-mineral intergrowths in (a). (b) Inclusion of laitakarite (Ltk; Bi₄Se₃) at the junction between lamellae of Pb–Bi-chalcogenide and tintinaite. (c) Contact between tintinaite and Pb–Bi-chalcogenide is marked by fractures crosscutting the sulfosalt. These are shown by relative depletion in Pb, Bi and Sb and enrichment in Se. Such areas of element redistribution are accompanied by formation of laitakarite and clausthalite (Cth) marked by a red outline. (d) Tintinaite with interstitial Pb–Bi-chalcogenide. Note small area of izoklakeite (Iz) with the highest Pb concentration.

Chemical composition

Compositional data for the two sulfosalts was obtained from both polished blocks. Mean compositions of izoklakeite and tintinaite are given in Tables 1, and 2–3, respectively. The complete EPMA datasets for kobellite homologues and associated (Bi, Se)-rich jamesonite are compiled as Supplementary Tables S1, S2 and S3. A selection of published data for kobellite homologues used for comparison are given in Supplementary Table S4.

Table 1. Electron probe compositional data for izoklakeite.

* Weighted mean. Mdl –method detection limit; S.D. – standard deviation

Table 2. Electron probe compositional data for tintinaite.

Mdl –method detection limit; S.D. – standard deviation

Table 3. Electron probe compositional data for tintinaite in three areas of sample 1061a, highlighting local variation.

Mdl – method detection limit; S.D. – standard deviation

To achieve an optimal correlation between micro- and nano-scale characterisation, EPMA of sample 1061a was carried out in areas immediately adjacent to the site of FIB cutting (Table 3). Analysis spots along short profiles are marked on Fig. 1d–f.

The ideal end-members of the two homologues as defined by Zakrzewski and Makovicky (1986) give a range of M ²⁺(Pb,Cd,Zn) relative to M ³⁺(Sb,Bi) ratios (cations with octahedral and prismatic coordination) correlated to the presence of atoms in tetrahedral coordination (T = Cu⁺ and Fe²⁺) by the formula:

$$N_{homologue} = {\rm x}\displaystyle{{6M^{2 + } + 3M^{3 + }} \over {4M^{3 + }-M^{2 + }}} + ( {1 - x} ) \displaystyle{{5M^{2 + } + 2M^{3 + }} \over {4M^{3 + }-M^{2 + }}}, \;{\rm where}\,{\rm x} = \displaystyle{{T^ + } \over {( {T^ + { + } T^{2 + }} ) }}\;\,{\rm and\ }( {1 - {\rm x}} ) = \displaystyle{{T^{2 + }} \over {( {T^ + { + } T^{2 + }} ) }}.$$

For the sake of simplicity, Ag⁺ is included in the M ²⁺ site and only Cu⁺ is considered.

The M ²⁺:M ³⁺ ratio is higher for x = 0 than for x = 1 (0.86 and 0.63 for kobellite–tintinaite, 1.56 and 1.3 for izoklakeite). The lowest ratio is obtained for T = 0, where ${{M^{2 + }} \over {M^{3 + }}}$ = ${{8N-8} \over {2N + 14}}$, i.e., 0.44 for N = 2 (kobellite or tintinaite) and 1.09 for N = 4 (izoklakeite). For N calculation we used the substitution: Ag⁺ + (Sb, Bi)³⁺ → 2 Pb²⁺, as considered for the formula calculation (Moëlo et al. 1995).

Izoklakeite has a mean composition Iz₆₀Gie₄₀ (Table 1) with the empirical formula: (Cu_1.88Fe_0.18)_Σ2.06(Pb_22.92Ag_1.47Cd_0.01Zn_0.01)_Σ24.41(Sb_13.12Bi_8.69)_Σ21.8(S_50.19Se_6.43Te_0.12)_Σ56.73. This corresponds to a Pb(Ag,Cd,Zn)-depleted, (Sb,Bi)-enriched composition relative to the ideal N = 4 homologue, (Cu,Fe)₂Pb₂₆(Sb,Bi)₂₀S₅₇, with substitution factor x = 1 (Zakrzewski and Makovicky, 1986). The mean homologue number N is 3.83 using calculations based on 105 atoms per formula unit (apfu) (Table 4).

Table 4. Calculated N values¹ for sulfosalts in this study compared with literature data.

¹Calculations used the formula for the tintinaite–kobellite and izoklakeite–giessenite series given by Moëlo et al. (1995): tintinaite–kobellite: (Cu_2–xFe_x)[Pb_10–2z+x(Ag,Cu)_z(Sb_1–yBi_y)_16+z–x]_Σ26S₃₅, with 0 ≤ x ≤ 1; 0 ≤ y ≤ 0.67; 0 ≤ z ≤ 0.54, and izoklakeite–giessenite: (Cu_2–xFe_x)[Pb_26–2z+x(Ag,Cu)_z(Sb_1–yBi_y)_20+z–x]_Σ46S₅₇, with 0 ≤ x ≤ 1; 0.44 ≤ y ≤ 0.85; 0 ≤ z ≤ 2.2, with addition of the following as indicated: * excess Cu on Pb; ** excess Cu on (Sb+Bi); ***Ag incorporated on (Sb,Bi) sites.

The mean tintinaite composition is Tti₆₆Kbl₃₄ with the empirical formula: (Cu_1.31Fe_0.74)_Σ2.05(Pb_10.58Ag_0.18Cd_0.05Zn_0.02)_Σ10.83(Sb_10.2Bi_5.23)_Σ15.43(S_32.22Se_2.46)_Σ34.7 and calculated N = 2.05 (Tables 2 and 4). Tintinaite is richer in Pb(Ag,Cd,Zn) and slightly poorer in (Sb,Bi) than the ideal N = 2 homologue, (Cu,Fe)₂Pb₁₀(Sb,Bi)₁₆S₃₅, with substitution factor x = 1 (Zakrzewski and Makovicky, 1986), thus opposite to the trend shown by izoklakeite.

The Sb-rich sulfosalt occurring as the darkest lamellae on BSE images from FB area 3 (Fig. 1f) contains Fe and was identified as jamesonite (Fe_0.99Cu_0.01)_1.00(Pb_3.89Cd_0.03)_3.91(Sb_5.14Bi_1.03)_6.17 (S_13.14Se_0.78)_13.92 (Supplementary Table S4). Notably, Se is present in this third sulfosalt, albeit at a lower concentration (2.81 wt.%) than in izoklakeite (4.82 wt.%) or tintinaite (3.3 wt.%).

There are, however, subtle compositional variations within each sulfosalt from the two polished blocks and, for tintinaite, also between the three analysed areas (Fig. 3; Tables 1–3; Supplementary Tables S1 and S2). Both izoklakeite and tintinaite show a slight relative depletion in Pb (Ag,Cd,Zn) in the pyrite-rich sample (1061) compared to the sulfide-poor sample (1061a). Tintinaite coexisting with izoklakeite (area 2, 1061a) is lowest in Pb(Ag,Cd,Zn), whereas tintinaite coexisting with (Bi, Se)-rich jamesonite is lowest in Sb+Bi but highest in Pb(Ag,Cd,Zn) (Fig. 3). Izoklakeite displays a more marked Pb(Ag,Cd,Zn) variation between the two samples than tintinaite.

Figure 3. Binary plots of Bi+Sb–Ag vs. Pb+2Ag+Cd+Zn (apfu) showing subtle differences in chemistry of izoklakeite and tintinaite between the two analysed samples (as labelled). Data from Tables 1–3 and Supplementary Tables S1–3. For comparison, data for kobellite–tintinaite from Moëlo et al. (1984), Mumme et al. (2013) and Wagner and Jonsson (2001) are plotted and labelled as Bol1, Bol2 and Bol3, respectively.

Nanoscale characterisation

Three areas of the Boliden specimen were investigated in four S/TEM foils obtained from polished block 1061a as marked on Fig. 1a: one (foil #1) from area 1, two (foils #2 and #3) from area 2, and one (foil #4) from area 3 (Fig. 4a–d). Contacts between tintinaite and the Pb–Bi-chalcogenide are sharp (Fig. 4a,b), with a thin sliver of Sb-rich Bi sulfosalt, a few μm wide observed in foil #1 which is obtained from the area where no μm-scale lamellae of tintinaite are identified. Foils obtained across the contacts between tintinaite and izoklakeite (Fig. 4c,d) display intergrowths between the two phases, and feature μm- to sub-μm-scale patches of the Pb–Bi-chalcogenide. EDS STEM maps across the three phases in foil #1 show a relative enrichment of Ag in tintinaite and the Pb–Bi-chalcogenide, whereas the Sb-rich Bi-sulfosalt is also enriched in Fe (Fig. 4e), consistent with jamesonite. Selenium is concentrated within the Pb–Bi-chalcogenide.

Figure 4. (a–d) Images showing the four foils as labelled (from polished block 1061a, details in text) which have targeted contacts between the three components in the lamellar intergrowths, tintinaite (Tti), izoklakeite (Iz) and Pb–Bi-chalcogenide. Foils #1 and #2 are shown both before (left) and after (right) thinning. (e) HAADF STEM image and EDS STEM element-distribution maps of the grain boundary between tintinaite and Pb–Bi-chalcogenide (marked by rectangle ‘map 1’ in (a)) showing the presence of an Sb-rich bismuth sulfosalt lamella between the two phases. Note the relative depletion in Bi, Ag and Se, and the marked Fe enrichment of this lamella relative to tintinaite. The contacts between tintinaite and izoklakeite in foils #3 and #4 are marked by the presence of small Pb–Bi-chalcogenide inclusions.

Nanoscale intergrowths between tintinaite and izoklakeite are widespread along the direct contacts between the two phases (Fig. 5). Typically, the intergrowths consist of lamellae of one species within the other forming rhythmic banding within fine-grained aggregates (Fig. 5a–c). The banding is however irregular, with rhythms that change across the same grain, and with lamellae thickness that varies from a few nm to several tens of nm in width. The Pb–Bi-chlacogenide is embayed into the sulfosalt aggregates (Fig. 5b). Sharp changes in banding patterns are observed across grain contacts (Fig. 5c). EDS STEM maps of sulfosalt intergrowths show significant compositional changes in some elements, particularly Sb > Pb or S > Se in tintinaite versus izoklakeite, whereas the variation in Bi is more subtle (Fig. 5d,e). Izoklakeite is richer in minor Ag but poorer in Fe relative to tintinaite. Clusters of Bi–Se nanoparticles (NPs) decorate the margins of tintinaite lamellae which are either truncated or displaced by fractures (Fig. 5e).

Figure 5. Intergrowths between tintinaite (Tti), izoklakeite (Iz) and Pb–Bi-chalcogenide as HAADF STEM images (a–c) and EDS STEM maps 2 and 3 (d, e) marked on Fig. 4c and d. The mapped areas are marked by the rectangles as labelled. Note that contacts display smaller grains with fine, nanometre–scale lamellar intergrowths between the two sulfosalts. (d) (e) Maps showing the relative increase in Pb, Fe and S and decrease in Se between tintinaite and izoklakeite. Remobilisation of Se (circled area) along the contacts between the finest intergrowths.

Chemical structural modularity: the main building modules in tintinaite and izoklakeite

PbS and SnS modules are readily identified on the [001] zone axis of tintinaite and izoklakeite (Fig. 6a–d). These are recognised on HAADF STEM images as blocks with the number of heavy atoms M (bright dots on images) counted as: n1 = 18 and n2 = 8 for tintinaite, and n1 = 30 and n2 = 16 for izoklakeite (overlays on Fig. 6a,c). Using the formulae M _6(N+1)S_7N+9 and M _4NS_4(N+1) (M = Bi, Sb and Pb) (Makovicky and Mumme, 1986; Zakrzewski and Makovicky, 1986), we can calculate the homologue number N from the M sites as: N = n1/6–1 and N = n2/4 for PbS and SnS modules giving N _Kbl = 2 and N _Iz = 4. The atomic arrangements for the larger SnS modules are shown as given by Li et al. (2019) and Makovicky and Mumme (1986).

Figure 6. High-resolution HAADF STEM images and corresponding fast Fourier transform (FFT) patterns for tintinaite (Tti) and izoklakeite (Iz) on [001] zone axis (labelled in square brackets). The SnS (n = 8 and 16 for Tti and Iz) and PbS (n = 18 and 30 for Tti and Iz) building modules are shown as overlays (green and red, respectively) highlighting the number of bright dots corresponding to heavier atoms; less brighter atoms at the corners of the modules are shown in yellow. The modules are depicted in (a) and (c) use the approach of Li et al. (2019) on the left, and Makovicky and Mumme (1986) on the right.

The crystal structure models show a very good correlation with the STEM simulations and atomic-scale images for the two sulfosalts (Fig. 7). The structures (Miehe, 1971; Makovicky and Mumme, 1986) consist of PbS and SnS modules (referred to as rods in earlier descriptions of Makovicky and Mumme, 1986; Zakrzewski and Makovicky, 1986) that alternate on both a and b directions (highlighted by overlays on crystal models, STEM simulations and images in Fig. 7). Tetrahedral-coordinated cations (T = Cu and Fe) are nested at the module interfaces. The HAADF STEM images show patterns of dark lines, particularly visible through the centre of the SnS modules in izoklakeite. These correspond to weaker/longer Sb(Bi)–S bonds between adjacent polyhedra.

Figure 7. Crystal models, STEM simulations of kobellite and izoklakeite on [001] zone axis as labelled. Images for tintinaite and izoklakeite shown at the bottom of the figure. Simulations were obtained using the structures of Miehe (1971) and Makovicky and Mumme (1986) for kobellite and izoklakeite, respectively. The SnS and PbS building modules are highlighted by overlays. Weaker/longer Sb(Bi)–S bonds between adjacent polyhedra interpreted as lone electron pair micelles (e.g. Makovicky and Mumme, 1986) are marked by grey lines on atom-fill models. These correspond to dark lines on HAADF STEM images for tintinaite and izoklakeite and are highlighted by yellow lines across the overlays. Note that such features are visible through the middle of the SnS modules in izoklakeite, whereas in tintinaite they also occur through the PbS modules.

High-resolution EDS STEM element distribution maps and their overlays show a good match with atom-fill models for the two sulfosalts (Figs 8, 9, and Supplementary Figs S2 and S3). Overlays of Sb–Pb–Bi–Fe distribution maps and a crystal model for tintinaite show an excellent correlation with the two building modules (Fig. 8a). The distribution of Fe in the tetrahedral sites flanking the corners of the SnS module and in the kinks of the PbS module are very well depicted for tintinaite. Overlays of Sb–Pb–Bi distribution maps with the HAADF STEM image also reproduce the two modules, albeit with less accuracy than for tintinaite (Fig. 8b). Nonetheless, the distribution of SnS modules is well resolved for izoklakeite.

Figure 8. Atomic resolution map overlays and models for tintinaite_[001] and izoklakeite_[001] as labelled. Extended element-distribution maps and additional overlays for tintinaite and izoklakeite are shown in Supplementary Figs S2 and S3. Models were obtained using the structures of Miehe (1971) and Makovicky and Mumme (1986) for kobellite and izoklakeite, respectively. Note the match between the atomic arrangement and the building modules SnS and PbS shown as overlays. Circles show the atoms at the corners of these modules (Fe, Cu) and (Cu, Fe, Ag) for tintinaite and izoklakeite, respectively. Such atoms are exceptionally well depicted by the Fe maps for tintinaite.

Figure 9. (a) Element-distribution maps for Pb, Sb, Bi and Fe in tintinaite (crops from maps in Supplementary Fig. S2) showing the distribution of these elements throughout the structure whereby four Bi/Sb sites lie at the centre of the SnS modules; these are surrounded by Pb atoms. (b) Crystal model for the asymmetric unit cell of kobellite from Miehe (1971), there are seven Bi/Sb sites (labelled by numbers) of which two have 0.5/0.5 ratios, whereas the others, except site #1 where Sb is twice as much as Bi (0.33/0.67), have occupancies in which Bi > Sb. The map for Sb suggests, however, that occupancy is Sb > Bi for sites #3, 4 and 5. This is in agreement with the higher Sb/(Sb+Bi) ratio in Boliden tintinaite compared with the kobellite specimen studied by Miehe (1971); 66 mol.% and 44 mol.% tintinaite, respectively.

In closer detail, the EDS STEM element-distribution maps for Pb, Sb, Bi and Fe in tintinaite show the distribution of these elements throughout the structure whereby four Bi/Sb sites are at centre of the SnS modules; these are surrounded by Pb atoms (Fig. 9a). Taking the crystal model for the asymmetric unit cell from Miehe (1971), there are seven Bi/Sb sites of which two have 0.5:0.5 ratios, whereas the others, except site #1 where Sb is twice as much as Bi (0.33:0.67), have occupancies in which Bi > Sb (Fig. 9b). The map for Sb suggests, however, that occupancy is Sb > Bi for sites #3, 4 and 5 (Fig. 9a). This agrees with the higher tintinaite component in the Boliden N = 2 kobellite homologue compared with the specimen studied by Miehe (1971) – 66 mol.% and 44 mol.% tintinaite, respectively.

Nanoscale intergrowths and defects in tintinaite–izoklakeite

Considering that b _Kbl~Tti and a _Iz are equal (~34 Å), coherent syntactic intergrowths along these directions are commonly formed (Fig. 10a–c). In some cases, however, the boundaries between such intergrowths show stepwise displacements with small-scale re-shaping and are accompanied by a protrusion of Bi-rich stringers (Fig. 10a). Likewise, the boundaries between grains featuring rhythmic intergrowths of variable width are curvilinear, with small adjustments along the (010_Tti and (100)_Iz planes (Fig. 10b,c). At higher magnification, the adjustments are illustrated by misorientations in the building modules (Fig. 10d) and swells along boundaries (Fig. 10e), as well as loops and kinks of the smallest-scale arrays of one species within another (Fig. 10f). Defects along (100) planes in izoklakeite show a doubling of the building modules and their displacements at the edges of the defects or along mutual contacts between the two sulfosalts (Fig. 10g,h).

Figure 10. (a–f) Images showing various types of intergrowths between tintinaite (Tti) and izoklakeite (Iz). (a) Tintinaite lamella with straight and irregular contacts (arrowed). (b, c) Curvilinear interfaces between grains with lamellar intergrowths of different thickness. Direction of planes for boundaries and intergrowths given in brackets. (d, e) At higher magnification, the grain boundary adjustments (dashed line) are illustrated by misorientations in the building modules (in d) and swells along boundaries (in e). (f) Loops and kinks are common along smallest arrays of one sulfosalt into another, in this case izoklakeite within tintinaite. (g) Defect along (100) plane showing doubling of PbS modules in izoklakeite (image at the top). Underneath, left side, a detail showing the building modules (red outline for PbS, also numbered, and green for SnS) are shown to be displaced from one another along a strike-slip plane. At the edge (right side) of the defect, the displacement is by ½ b along the (100) plane. (h) Module displacement within izoklakeite affecting the contact with tintinaite.

The match between the values of b _Kbl and a _Iz at ~34 Å allows formation of intergrowths between tintinaite and izoklakeite (Fig. 11). The alternating PbS and SnS modules that form a zig-zag line along (100_Tti//(010)_Iz planes are shown on the model in Fig. 11a. The change between tintinaite and izoklakeite along a _Tti and b _Iz is achieved by increasing the width of these modules whereas their heights remain constant (~17 Å). The HAADF STEM image shows a perfect match with the model (compare Fig. 11a and b). Oblique contacts between [001]_Tti and [001]_Iz are often observed (Fig. 11c). In this case the alignment between (010)_Tti and (100)_Iz planes features small-scale areas of mismatch, recognisable by misfit of the two types of modules and their respective morphologies (Fig. 11d,e). Formation of defects and nano- to micro-fractures are both associated with element remobilisation and accompany this type of low-angle misfit between the two sulfosalts (see below).

Figure 11. Schematic model (a) and image (b) showing coherent contacts between izoklakeite (Iz) and tintinaite (Tti). PbS and SnS building modules are shown as red and green overlays. The match between b _Tti and a _Iz features alternating PbS and SnS modules that form a zig-zag line. (c) An oblique contact between [001]_Tti and [001]_Iz. Parallel arrays of tintinaite within izoklakeite are arrowed. The misalignment a*_Tti and b*_Iz is also shown in the FFT (inset). (d, e) Alignment between (010)_Tti and (100)_Iz planes features small-scale areas of mismatch, recognisable by misfit of the two types of modules and their respective morphologies.

The second most common type of modular intergrowths within either tintinaite or izoklakeite is observed along planes from the <110> family (Fig. 12). Single arrays of smaller PbS modules along (1$\bar{1}$0) planes in izoklakeite (Fig. 12a) can be interpreted as modules corresponding to an N = 3 homologue (N3), i.e. from n1 = 24 and n2 = 12 defined by PbS and SnS modules, albeit with corner-sharing of atoms between PbS modules along (100)_Iz (Fig. 12b). Arrays of thicker SnS modules along (1$\bar{1}$0) planes in tintinaite (Fig. 12c) can be interpreted as two PbS modules of the N3 homologue alternating with SnS modules of tintinaite (Fig. 12d). Likewise, module disorder and formation of the N3 homologue are observed when the intergrowths comprise changes in thickness or branching between arrays of different orientations (Fig. 12e–h).

Figure 12. (a, d) Single and double arrays of modular intergrowths in izoklakeite (Iz) and tintinaite (Tti) along planes from the <110> family showing PbS (n = 24; blue) and SnS (n = 12; yellow) modules corresponding to the N = 3 homologue. Circles represent dots/atoms unaccounted for. (a, b) A single array (blue line in (a)) of the N = 3 homologue shown as PbS_N3 (blue)–SnS_N3 (yellow) modules within izoklakeite (PbS_Iz and SnS_Iz modules outlined in red and green). (c, d) A double array (blue line in (c)) of PbS_N3 modules intergrown with SnS_Tti modules in tintinaite. (e–h) Module disorder involving slabs of N = 3 homologue in intergrowths comprising changes in thickness or branching between arrays of different orientations as labelled in izoklakeite and tintinaite.

A range of textures crosscutting the main crystal structural motifs of the two sulfosalts and leading to redistribution of Bi and Se are also depicted (Fig. 13). Modifications of the PbS module in both izoklakeite or tintinaite, e.g. giving them an S-shaped or irregular appearance, are present as single defects (Fig. 13a–c), or occur along nanofractures displaying Bi enrichment (Fig. 13d). Some lamellae (~26 Å-wide) feature building modules which, although syntactic with host tintinaite, are different to kobellite homologues (Fig. 13e). These are associated with localised enrichment in Bi and Se, and depletion in Sb (Fig. 13f).

Figure 13. Defects and textures crosscutting the main crystal structural motifs of the two sulfosalts and leading to redistribution of Bi and Se in izoklakeite (Iz) and tintinaite (Tti). (a) S-shaped defect derived from modification of PbS modules (red outline) in izoklakeite attributable to shear–displacement along (100)_Iz planes. SnS_Iz module outlined in green. Note a slab of n = 24 dots/atoms of equal intensity with the PbS_Iz along the displacement plane. (b, c) Defects showing irregular aggregation of PbS modules that are interpreted as dilational jogs in izoklakeite and tintinaite. (d) Nanoscale fracture in izoklakeite displaying Bi remobilisation. (e) Image showing a ~26 Å-wide lamella featuring building modules, which, although syntactic with host tintinaite, are different to kobellite homologues. A possible interpretation of the modules along the lamellae are given as overlays underneath the image. The SnS and PbS modules in tintinaite are also shown as green and red overlays. (f) EDS STEM maps of image in (e) showing localised enrichment in Bi and Se, and depletion in Sb.

Discussion

Compositional variation in kobellite homologues: comparison with other occurrences

The compositional variation observed in species from the tintinaite–kobellite and izoklakeite–giessenite series from previously published studies spanning a range of magmatic–hydrothermal and metamorphosed deposits (Fig. 14; Supplementary Table S3) reflects the ability of modular structures to adapt to changes in physical and chemical parameters of fluids during ore formation.

Figure 14. Diagram showing compositional variation in kobellite homologues from Boliden and comparison with other occurrences. Data from Tables 1–3 and Supplementary Table S3, References: [1] Harris et al. (1986); [2] Moëlo et al. (1995); [3] Zakrzewski and Makovicky (1986); [4] Armbruster and Hummel (1987); [5] Li et al. (2019); [6] Graeser and Harris (1986); [7] Makovicky and Karup–Møller (1986); [8] Moëlo et al. (1984); [9] Mumme et al. (2013); [10] Wagner and Jonson (2001); [11] Harris et al. (1968); [12] Miehe (1971); [13] Pršek and Peterek (2008). See text for further explanation and Table 4 for locality details.

The Boliden izoklakeite, Iz_~60Gie_~40, is richer in Sb (highest Iz component) than all previously published data (Table 1, 4 and Supplementary Table S3), including the type specimen from Izok Lake, Canada (Harris et al., 1986) with composition Iz₅₃Gie₄₇.

Likewise, the average tintinaite composition, Tti ₆₆Kbl_34, shows this is also rich in Sb, comparable to Pedra Luz, Tt_i65Kbl₃₅ (Moëlo et al., 1995), and higher than the Tti₅₁Kbl₃₉ composition from Boliden given by Wagner and Jonsson (2001) (Tables 2, 4 and Supplementary Table S3). The Tti:Kbl ratio in tintinaite analysed here shows a Tti:Kbl range between 64:36 and 69:31, clearly representing the highest Sb-rich composition reported if we exclude the TL tintinaite which lacks any Bi (Harris et al., 1986). Previously reported data for the N = 2 homologue considered kobellite from Boliden have Tti:Kbl ratios midway through the series, i.e. 50:50 (Moëlo et al., 1984) and 52:48 (Mumme et al. (2013), highlighting the constraints of local conditions on compositional variation in a single deposit. All studies of the homologue N = 2 from Boliden, including the present data, affirm these are Se-rich specimens (~3–4 wt.%; Tables 2 and S3). The izoklakeite reported here from Boliden is also Se-rich (~5 wt.%; Table 1). The Se-rich environment from which these sulfosalts have formed is further emphasised by the presence of 2.81 wt.% Se in Bi-rich jamesonite (Table S4). Selenium-bearing assemblages of jamesonite and tintinaite are also reported from Úhorná, Spišsko Gemerské Rudohorie Mts., Slovakia (Pršek and Peterek, 2008). The Korunka Mine specimen used for refinement of the kobellite crystal structure (Miehe, 1971) is from the same sulfosalt rich, hydrothermal vein system (~1200 veins) in Eastern Slovakia as the Úhorná specimen. As at Boliden, we note that the Tti:Kbl ratio changes from 60:40 to 44:56, emphasising the presence of both tintinaite and kobellite in the same orefield.

Mirroring the Boliden kobellite homologues, distinct ranges of composition for the same sulfosalt within a given district or province are also reported (Moëlo et al., 1995). At Boliden, co-crystallisation of the two sulfosalts is reflected by local variations in the Sb:Bi ratios in tintinaite throughout the three analysed areas in sample 1061a, i.e. the lowest ratio of 66:36 in area 2 when co-existing with izoklakeite, and highest (69:31) when co-existing with jamesonite in area 3 (Table 2).

The compositional data for the Boliden specimens (Tables 1–3) are compared in Fig. 14 with published data from other localities (list in Table 4) and ideal end-members for the two series (Supplementary Table S3, and references therein). The relative proportions of Pb, Bi and Sb, the three main elements in the M sites defining the structural modules for the two homologues, are depicted on the plot of Pb*/(Pb*+Bi+Sb–Ag) versus Bi/(Sb+Bi) where Pb* = (Pb+Ag+Cd+Zn) in Fig. 14. Locality details are given in Table 4.

Izoklakeite data separates into two groups: (i) ‘Bi-rich izoklakeite’ encompassing the specimen from Zervreila Lake, Switzerland (Armbruster and Hummel, 1987) and material from Gutaishan, China (Li et al., 2019), and (ii) izoklakeite from the type locality (Harris et al., 1986), Vena, Sweden (Zakrzewski and Makovicky, 1986) and the specimens from Boliden reported here. In addition to these, specimens from France and Portugal (Moëlo et al., 1995) are defined from chemistry alone. Among these, Bazoges (France) includes both types of izoklakeite. Notably, Boliden izoklakeite plots at the lower end of the Bi/(Sb+Bi) range and within the central part (x = 1) of the N = 4 homologue band.

Specimens with greater Bi of homologue N = 2, kobellite sensu stricto, form a tight cluster with 0.55 > Bi/(Sb+Bi) > 0.7 including data from Korunka Mine (Miehe, 1971); Gutaishan (Li et al., 2019); Vena, Sweden and Raleigh, USA (Moëlo et al., 1984); Les Chalanches, France; and Pedra Luz, Portugal (Moëlo et al., 1995). Two of the Boliden specimens at Bi/(Sb+Bi) ≈ 0.5 (Moëlo et al., 1984; Mumme et al., 2013) lie at the boundary between kobellite and tintinaite. The other tintinaite data include specimens from Rossland, Canada (Moëlo et al., 1984), Pedra Luz (Moëlo et al., 1995), Úhorná, Slovakia (Pršek and Peterek, 2008), data from Boliden (Wagner and Jonsson, 2001, and the present work, and the Sb-pure tintinaite from Tintina (Harris et al., 1986; Moëlo et al., 1984). The tintinaite–kobellite data plots along the middle to upper part of N = 2 band, with Korunka Mine kobellite at x = 0.

Primary nanoscale intergrowths: jigsaw patterns evolving from chessboard structures

Widespread nanoscale intergrowths between tintinaite and izoklakeite (Figs 3, 10–12) account for compositional fields in terms of M ²⁺(+Ag¹⁺) and M ³⁺ cations when analysed at the micrometre-scale (Fig. 3). It is however interesting to note that the range of N values, 2.01–2.08 for tintinaite (Tables 2, 3) are slightly in excess of the ideal N = 2, whereas the range of N values for izoklakeite, 3.76–3.94 (Table 1) are in deficit relative to the ideal N = 4. This mirrors results for N values of kobellite homologues elsewhere (Table 4). Challenges in calculation of N values from EPMA data are introduced by excess of Cu over the 2 apfu in the tetrahedral site. In addition to Pb and (Sb,Bi) correction due to Ag incorporation (see above), attributing excess Cu to either Pb or (Sb,Bi) sites drives significant changes of N (Table 4; Fig. 14), as observed for kobellite from Raleigh (Moëlo et al., 1984) and kobellite and izoklakeite from Les Chalanches (Moëlo et al., 1984; 1995). The Bi-rich izoklakeite from Zervreila Lake (Armbruster and Hummel, 1987) gives N = 4.08 and 4.54, if Ag is incorporated within the (Sb,Bi) or Pb sites, respectively. The N excess of tintinaite, and N deficit of izoklakeite, may indicate that, at the scale of the electron probe spot, the analysed volume of each sulfosalt integrates a minor fraction of the other sulfosalt as a syntactic intergrowth.

Difficulties in attributing specific schemes of (Sb,Bi,Pb) ordering in izoklakeite that markedly differ in chemistry were discussed in detail by Armbruster and Hummel (1987) using the examples of Bi- and Sb-rich varieties from Lake Zervreila and Vena (Makovicky and Mumme, 1986). Although this does not affect the identification of building modules by HAADF STEM imaging of either Bi- or Sb-rich izoklakeite and kobellite and tintinaite, as shown for specimens from Gutaishan (Iz₃₅ and Kbl₆₁; Li et al., 2019) and Boliden (Iz₆₀ and Tti₆₀; Figs 6, 7), it is nevertheless reflected in the atom-resolution mapping of izoklakeite and tintinaite (Fig. 8).

A novel feature of the sulfosalt intergrowths from Boliden are the SnS and PbS modules with n = 12 and 24, respectively, imaged as arrays of combined SnS and PbS modules in either izoklakeite or tintinaite (Fig. 12). These correspond to modules suggested for a hypothetical N = 3 member of the kobellite series (Makovicky and Mumme, 1986) with an estimated structural formula (without Ag) that can be given as: (Cu_2–xFe_x)Pb_18+x(Sb,Bi)_18–xS₄₆. The structure of the N = 3 homologue is illustrated as Fig. 15.

Figure 15. Proposed structure for the N = 3 homologue. The structural formula (without Ag) can be considered as (Cu_2–xFe_x)Pb_18+x(Sb,Bi)_18–xS₄₆. Estimated unit cell parameters are a = 34.1 Å, b = 41.5 Å and c = 4.05 Å.

Structural versatility of these sulfosalts is moreover reflected in the ability to combine modules corresponding to different homologues such as SnS_Tti=Kbl and PbS_N=3. Other structures can be derived forming a broader family of Pb–Bi–Sb sulfosalts considered as kobellite homeotypes (Makovicky and Balić-Žunić, 2023, and references therein).

Structural arrangements in kobellite homologues have been described as ‘cyclically twinned’ (Makovicky, 1985), ‘boxwork’ (Makovicky, 1997), ‘chessboard’ (Li et al., 2019) or quasi-chessboard (Makovicky and Balić-Žunić, 2023). Chessboard structures are highly adaptable to accommodating nanoscale intergrowths between modules of variable sizes and combinations of thereof (Figs 11, 12), leading to smallest scale 2D jigsaw-like patterns.

Such physical–chemical behaviour can be considered as a type of microstructural evolution in complex nonlinear systems, a topic of interest from both theoretical and engineering viewpoints. For instance, quasiperiodic two-phase nanowire chessboard structures are common in a variety of materials from decomposing alloys to oxide ceramics and have been modelled as tweed structures with an underlying chessboard feature evolving during pseudospinodal decomposition (Ni and Khachaturyan, 2009). Chessboard-like supramolecular structures on Au₍₁₁₁₎ surfaces designed for chemical functionalisation of bio-sensors and molecular electronics (Dou et al., 2015), or broadband chessboard surfaces with a double-layer for an electromagnetic gap band structure (Mighani and Dadashzadeh, 2016), are among materials with block modularity akin to those discussed here.

Lone electron pair micelles – depicting Sb-rich kobellite homologues

The wide variety of distorted polyhedra, ranging from six to nine-fold coordination environments, is the basis for crystal structural modularity shared by Pb–Bi–Sb-sulfosalts (Makovicky, 1997, and references therein). This is due to variable stereochemical activity of the lone electron pair of Pb²⁺, Bi³⁺ and Sb³⁺ which, statistically, has a decreasing influence on distortion of the coordination polyhedra with increasing atom size Sb < Bi < Pb (Armbruster and Hummel, 1987). Makovicky and Mumme (1986) discussed the formation of lone electron pair micelles as a means to structurally accommodate stereochemical active pairs along interfaces in sulfosalts with reference to kobellite homologues. These concepts were subsequently taken to defining modularity in Pb–Sb-sulfosalts based on the arrangement of micelles into ‘fish-bone’ zigzag layers (e.g. Moëlo et al., 2000; Biagioni et al., 2016).

An outstanding result of the present study is depiction of the dark line patterning through the middle of SnS modules on the HAADF STEM images of Sb-rich homologues (Figs 6, 7). These are attributable directly to the lone electron pair micelles in the structures, as outlined for example in izoklakeite (Makovicky and Mumme, 1986). Notably, these features were not identified on the HAADF STEM images of Bi-rich kobellite homologues from Gutaishan (Li et al., 2019), and thus confirm the stronger influence of Sb relative to Bi atoms in terms of stereochemical activity of the lone electron pair (Armbruster and Hummel, 1987).

Ultralow thermal conductivity is a material property linked to the effect of lone electron pair micelles in strengthening bond anharmonicity as assessed in a synthetic MnPb₁₆Sb₁₄S₃₈ sulfosalt (Dawahre et al., 2020). This study acknowledged the lack of thermal transport behaviour studies of Pb–Sb-sulfosalts despite their significant potential in terms of technological applications, ranging from thermoelectric energy conversion to thermal barrier coatings (e.g. Dittrich et al., 2009).

Secondary textures; a record of metamorphic deformation

The range of shear, twist, jogs and displacement textures occurring as defects, crosscutting veinlets no more than a few nm wide, or along crystallographic planes in kobellite homologues (Figs 10g,h and 13) are secondary features attributable to superimposed syn-metamorphic deformation. These are associated with the redistribution of Bi and Se (Fig. 13f) and led to late formation of the Bi chalcogenides identified in areas of the samples bearing evidence of trace-element remobilisation at the micrometre scale (Fig. 2).

Such interpretations are concordant with assertions that sulfosalt-rich vein ores (containing kobellite homologues) experienced extensive ductile deformation post-dated by decomposition textures leading to formation of Se-rich assemblages (including laitakarite) at Boliden (Wagner and Jonsson, 2001). The twists, shears and displacements of building modules or the low-angle imbrication boundaries in kobellite homologues resemble chain-width disorder in non-conventional pyriboles produced during syn-tectonic events (e.g. Konishi et al., 2010; Campo-Rodriguez et al., 2024). Twists and shears recorded by pyrite during brittle to ductile transition are among nanoscale textures associated with Au and accompanying trace element (re)mobilisation in brecciated ores (Ehrig et al., 2023).

Summary and implications

The findings of the present study are summarised below.

1. (1) Sulfosalt assemblages from Boliden (Sweden) comprise micro- to nano-scale intergrowths of Se-rich izoklakeite (Iz₆₀Gie₄₀) and tintinaite (Tti₆₆ Kbl₃₄) with average formulae and calculated homologue number (N) given as: (Cu_1.88Fe_0.18)_Σ2.06(Pb_22.92Ag_1.47Cd_0.01Zn_0.01)_Σ24.41(Sb_13.12Bi_8.69)_Σ21.8(S_50.19Se_6.43Te_0.12)_Σ56.73, N = 3.83, and (Cu_1.31Fe_0.74)_Σ2.05(Pb_10.58Ag_0.18Cd_0.05Zn_0.02)_Σ10.83(Sb_10.2Bi_5.23)_Σ15.43(S_32.22Se_2.46)_Σ34.7 and calculated N = 2.05, respectively. The two sulfosalts are also Se-rich, 4.83 wt.% and 3.3 wt.%, for izoklakeite and tintinaite.

2. (2) HAADF STEM imaging reveals chessboard structures comprising PbS and SnS building modules with the number of atoms (bright dots on images) in the octahedral (M) sites counted as: n1 = 18 and n2 = 8 for tintinaite, and n1 = 30 and n2 = 16 for izoklakeite. The homologue number can be calculated using the formula: N = (n1/6)–1 and N = n2/4 for PbS and SnS modules giving N _Tti = 2 and N _Iz = 4.

3. (3) A new homologue, defined by n = 12 and 24 for SnS and PbS modules, corresponding to the N = 3 member of the kobellite series is identified as single or double units within areas with intergrowths between tintinaite and izoklakeite.

4. (4) HAADF STEM imaging of Sb-rich kobellite homologues reveals features attributable to lone electron pair micelles.

5. (5) The first atomic-resolution chemical mapping of Pb–Bi–Sb-sulfosalts shows a correlation with crystal structural modularity. The EDS STEM maps also highlight sites in the SnS modules of tintinaite in which Sb > Bi. Although this is only a qualitative assessment, the data is concordant with the overall Sb-rich composition of the Boliden specimen.

6. (6) Widespread coherent nanoscale intergrowths between kobellite homologues define jigsaw patterns evolving from chessboard structures and are considered to have formed during co-crystallisation of the two phases.

7. (7) A range of displacement textures and crosscutting veinlets (a few nm in width) are interpreted as evidence for superimposed syn-metamorphic deformation and are associated with the redistribution of Bi and Se.

Whereas crystal–structural modularity in Pb–Bi-sulfosalts is readily predictable from principles established some decades ago (Makovicky, 1997, and references therein), imaging and mapping using HAADF STEM techniques is a novel and ideally suited way to characterise such phases and their relationships down to the nanoscale (Ciobanu et al., 2016). These tools hold untapped potential to unravel intricate pattern evolution of chessboard structures with applications across mineralogy but extending into allied fields of materials science and nanotechnology.