Launayite is a unique (Cu)-Pb-Sb(As) sulfosalt mineral, first described by Jambor (1967a, b) from Taylor Pit, Madoc, Ontario, Canada. Other very rare minerals from the same locality, identified and described simultaneously with launayite by Jambor, are playfairite, sorbyite, madocite, sterryite, twinnite, guettardite, dadsonite, and veenite. The crystal structure of dadsonite (Makovicky et al., 2006), sterryite (Moëlo et al., 2012), twinnite (Makovicky and Topa, 2012), guettardite (Makovicky et al., 2012), and veenite (Topa and Makovicky, 2017) were resolved in the meantime. The veenite crystal structure was obtained from Madoc material, and the rest of the crystal structures employed materials from other deposits. The crystal structure of launayite, playfairite, madocite, and sorbyite remained unknown up to now.

The first chemical formula published by Jambor for launayite was Pb22Sb26S61. X-ray powder diffraction indicated that the crystal is monoclinic, with unit cell parameters a=42.6(7), b=8.04(5), c=32.3(6) Å, β=102.02(25)°, and V=5370.8Å3. He noted that the b parameter was twice the length of a pseudocell axis (b=2b′=2×4.02 Å), based on the observation that diffraction rows corresponding to 2b′ were “extremely weak and barely discernible”. For the pseudocell, diffraction spots occurred only when h+k were even, leading him to suggest the space group C2, Cm, or C2/m (Jambor, 1967b).

In 1982, Jambor and co-authors reported new electron microprobe analyses of launayite from the same type material (Jambor et al., 1982), which revealed minor amounts of Cu, ranging from 1.3 wt % to 1.4 wt %. The authors stated that “its non-essential character is evident in that launayite has been synthesised in the pure PbS–Sb₂S₃ system” (Nekrasov and Bortnikov, 1977) and therefore concluded that the small amount of Cu was not significant and did not need to be considered in the ideal chemical formula.

The re-examination of the chemistry of tintinaite–kobellite series by microprobe analyses conducted and published by Moëlo et al. (1984) pointed to the conclusion that “minor amounts of Cu incorporated in a specific tetrahedral site is thought to stabilise some Pb sulfosalts at low temperatures”. Some examples were given, including launayite, with a new formula Pb10(Sb,As)13CuS30, based on results from Moëlo (1983).

The review of sulfosalt systematics published as a report of the sulfosalt sub-committee of the IMA Commission on Ore Mineralogy by Moëlo et al. (2008) placed launayite with formula CuPb10(Sb,As)13S30 and unknown crystal structure into an unclassified group of sulfosalt minerals.

In connection with our previous and ongoing analytical work on Madoc sulfosalt material, we performed additional microprobe analysis and a single-crystal study of this mineral. This study presents the crystallography and crystal chemistry of launayite from Madoc, Ontario, Canada. The present paper expands the knowledge and understanding of this very rare and complex mineral and its close structural relationship with rouxelite, while also contributing to the broader field of sulfosalt research.

Launayite is extremely rare, apparently unique up to now, and difficult to obtain due to the exhaustion of the Taylor Pit deposit. Upon request we were kindly provided with type material of launayite (sample MC 61062) by Inna Lykova, Laura Smyk, and Michael Bainbridge from the Canadian Museum of Nature for re-analysing and attempting to solve the crystal structure of launayite.

Figure 1 depicts the re-polished type specimen of launayite (MC 61062) which contains two black sulfosalt aggregates in embedding material. The backscattered electron image of aggregate no. 2 from Fig. 1 is presented in Fig. 2a. It shows one large launayite (Lau) grain (1) in contact with jamesonite (Ja) and a second small launayite (Lau) grain (2), which was mechanically extracted after EMPA analyses and after permission for crystal extraction was obtained. Additional launayite material, beside aggregate no. 1 from Fig. 1, was found in only two other samples, MC 12177 (grain #6) and MC 12178 (grain #2), in our larger ongoing study on Madoc.

Chemical analyses of launayite and associated minerals were carried out using a JEOL “Hyperprobe” JXA 8530F field emission gun electron microprobe (FE-EPMA) at the Central Research Laboratories of the Natural History Museum, Vienna. The analyses employed JEOL and Probe for EPMA software (WDS mode, 25 kV, 20 nA, 1.5 µm beam diameter, count time 15 s on peak and 5 s on background positions). The following emission lines and standards were used: AsLα and TlLα (lorándite, TlAsS₂), PbMα (galena), AgLα (Ag metal), SbLα and SKα (stibnite), HgLα (cinnabar), and CuKα (chalcopyrite). Other elements such as Bi, Fe, and Cl were sought but not detected. Proper empirical correction was made for the interference of the third order of the SbLα line with the analytical AsLα line. Under the described analytical conditions, the detection limits for the measured elements in the launayite matrix were as follows (expressed in wt %): Cl 0.03; S, Cu, and Fe ∼0.04; As, Ag, and Sb ∼0.06; Tl 0.08; and Hg, Tl, Pb, and Bi ∼ 0.1.

Based on all 54 analytical measured points of launayite, we defined four main groups of analyses, represented by their mean values and presented in Table 1: (1) all 54-point analyses combined, (2) a group with low Cu and As contents, (3) a group with high Cu and As contents, and (4) a group for the grain used in the single-crystal X-ray diffraction (SCXRD) study. All four groups show good charge balance and are characterised by low standard deviations for the measured elements. The detected amounts of Ag and Tl are very low but clearly over the detection limits. The high Cu and As values and the low Cu and As values are not correlated. Variations of As are correlated with variations of Sb and indicate an As3+↔Sb3+ substitution.

The empirical formulas for all groups were calculated based on 108 apfu, Z=4 (from the SCXRD study). The empirical formula for the grain used in the SCXRD study is Cu2.078Ag0.059Tl0.057Pb20.404Sb22.830As2.772S59.80.

A plane-polarised optical image (Fig. 2b) of aggregate no. 2 reveals twin lamellae of launayite in the large (Lau1) grain but none for the small (Lau2) launayite grain. The latter grain was extracted mechanically from the polished sample and used for the SCXRD study.

Intensity data of a small needle-shaped fragment were collected at room temperature using a STOE StadiVari diffractometer system equipped with an EIGER2 1M CdTe detector. Finely collimated MoKα radiation was employed, utilising finely sliced ω scans and a detector distance of 70 mm. Data were processed using STOE X-Area software, and intensities were scaled with STOE LANA (Koziskova et al., 2016).

The structure of launayite was solved by the dual-space method implemented in SHELXT (Sheldrick, 2015a), which revealed most of the atom positions, and was subsequently refined using SHELXL (Sheldrick, 2015b). In subsequent cycles of the refinement, remaining atom positions were deduced from difference Fourier syntheses by selecting from among the strongest maxima at appropriate distances.

Details on data collection and results of the structure refinement of launayite are given in Table 2, and full details are given in the CIF file deposited as a Supplement (S1_launayite.CIF file). The fractional atomic coordinate, occupancies, anisotropic atomic displacement parameters, charge distribution (CD) values, bond valence sum (BVS), and coordination values for all sites are compiled in Table 3. Table 4 gives selected Me–S bond lengths.

The asymmetric unit of launayite (Fig. 3) contains 48 cation and 60 anion sites. The metal sites consist of 21 Pb, 25 Sb, and 2 Cu positions. In total, 7 Pb sites are mixed with As (Me15 to Me21): 1 Sb site is mixed with Pb[(Sb,Pb)9], 4 Sb sites are mixed with As [(Sb,As)14, (Sb,As)16, (Sb,As)17, and (Sb,As)18], and 10 Sb sites are split (Sb7, Me8, Me12, M15, and Sb20–Sb25). Cu sites, Cu1 and Cu2, are both split as well (Table 3), and their CD and BVS suggest that Ag is not present at these sites. There are no As-dominant sites present in the structure. Although the (Sb,As)17 site contains a significant amount of As (Sb_0.57As_0.43), arsenic is not featured in the ideal formula of launayite.

Final refinement led to R1=0.0946 for 11 471 unique reflections. The resulting structural formula obtained from SCXRD study for the asymmetric unit, Cu2Pb20.330Sb23.024As2.689S60, Z=4, is close to the EMPA empirical formula, Cu2.078Ag0.059Tl0.057Pb20.404Sb22.830As2.772S59.80. The validation of refinement process is given in the Supplement as S2_checkcif_launayite.pdf file.

A comparison between the experimental X-ray powder diffraction data (Jambor, 1967b) and the calculated pattern using VESTA 3 (Momma and Izumi, 2011), based on the launayite crystal structure in Debye–Scherrer geometry with Cu Kα radiation (λ=1.540598 Å), is presented in Table 5 and shows a good agreement.

The launayite crystal under investigation is built from distinct modulated domains which can be derived from a common simplified parent structure with C2/m symmetry and b∼4 Å (4 Å being the approximate edge length of an MS₆ octahedron in sulfosalt structures). The model of this simplified structure was generated by halving the b axis, increasing the symmetry, merging duplicate atoms, removing minor positions, and moving the atoms onto the reflection planes at y=0 and y=1/2, respectively. Henceforth basis vectors and hkl indices will be given with respect to this parent structure.

The parent structure of launayite is closely related to the b∼4 Å parent structure of rouxelite (Orlandi et al., 2005; Topa et al., 2025), with isomorphic C2/m symmetry. The cell parameters of the parent structures are compiled in Table 6.

A boxwork description of rouxelite was first given by Orlandi et al. (2005), then by Makovicky and Topa (2009), and was later improved by Topa et al. (2025). In such a description (Fig. 4), both structures are defined by wall-like modules (the C layers, green in Fig. 4) parallel to (001) and limiting portions of alternating rod-like modules: PbS-based A modules (grey in Fig. 4) and SnS-based B modules (red in Fig. 4). For an introduction to PbS-like and SnS-like modules, see e.g. Makovicky (1993). From a structural point of view, extending in the [010] direction, the A and B rods in the parent structures of launayite and rouxelite can be regarded as isotypic. They possess pb12/m1 and pb121/m1 rod group symmetry (the subscript “b” in the Bravais symbol indicates the direction with translational symmetry), respectively, and together form A/B layers extending in the (001) plane with c12/m1 layer group symmetry. The C-centring of the C2/m space group corresponds to a translation by 1/2a+1/2b. For the A rod, which is located on a 2₁-screw rotation axis, translation by 1/2b is equivalent to a 2-fold rotation about [010]. In contrast, the B rods are symmetric by 2-fold pure rotation, which means that their origin may be located on either y=0 or y=0.5. The C-centring exchanges the origin as indicated by different shades of red in Fig. 4.

Even though the C walls likewise feature c12/m1 layer group symmetry in launayite and rouxelite, they are structurally fundamentally different. Notably, in rouxelite, they are formed by three cation–anion planes, whereas in launayite they are only two planes wide. Thus, launayite may be regarded as an inferior (N=2) homologue of rouxelite (N=3). This is reflected by the C-layer-to-C-layer distance, which is derived from csin⁡(β), being distinctly smaller in launayite (ca. 31 Å) than in rouxelite (ca. 34 Å). Accordingly, the cell volume is reduced by ca. 9 % (see Table 6). Moreover, the rouxelite C walls possess octahedrally coordinated Hg atoms (with minor Ag substitution) at the 2/m positions and characteristic rods of pb121/m1 symmetry, where the lone pairs of one-sidedly uncoordinated S atoms point towards the 2₁ axis. The precise nature of these rods is not yet established, as they might contain minor amounts of O and/or feature S vacancies (for details see Topa et al., 2025). Both features are missing in launayite. In both cases, out-of-phase protuberances add short fourth and third cation–anion layers, respectively.

For deeper insights into the difference between the C walls, it is useful to adopt a rod-based perspective. In rouxelite, the C walls have been decomposed into PbS-based C rods by Topa et al. (2025), as shown in Fig. 4 (right). To avoid confusion between the C wall and C rods, the meaning will henceforth always be specified. Moreover, we will indicate the archetype in the subscript and designate them as C_PbS rods. The C_PbS rods are located on general positions of the parent structure and can be seen as two double layers related by a 2₁-screw rotation resulting in pb121/m1 pseudo-symmetry. Adjacent C_PbS rods in rouxelite are connected by the Hg (+ minor Ag) and S rods with an additional Pb position (see uncoloured background in Fig. 4, right) described above. C_PbS rods connected by Hg are at the same position with respect to [010], and those connected by the S rods are translated by 1/2b.

Launayite features rods which in projection along [010] look practically identical to the C_PbS rods in rouxelite (Fig. 4, left). However, here the double layers are related by 2-fold pure rotation, making these rods of the SnS kind (pb12/m1 symmetry, as opposed to in pb121/m1 in rouxelite). Hence, these rods will be designated as C_SnS rods. While in the parent structure of rouxelite, the C_PbS rods are located on a general position, the C_SnS rods in the parent structure of launayite are located on 2/m. Moreover, launayite features D rods with pb121/m1 symmetry, which alternate with the C rods (blue background in Fig. 4, left). Since the D rods are only two atom sheets wide, they cannot be said to be of the PbS or SnS archetype (Makovicky and Topa, 2009). Observe that the combination of the symmetries of the C_SnS rods and D rods in launayite (and of the C_PbS rods, Hg (+ minor Ag) rods, and S rods in rouxelite) both result in c12/m1 symmetry of the C walls. Their structures are, however, fundamentally different. These comparisons show that the rod-based approach is superior for the structural comparisons in this family of minerals.

A consequence of the different location of the symmetry elements of the C walls in launayite (in C_SnS and D rods) and rouxelite (between C_PbS rods) is that if one A/B layer is fixed, in the adjacent A/B layer the positions of the A and B rods are exchanged; i.e. the adjacent A/B layer in rouxelite is shifted by ca. 1/4a in launayite vs. rouxelite, which results in a distinctly more acute β angle in launayite (see Table 5). A further consequence of the shifted A/B layers is that the protuberances of the C walls are located next to the 2 axes in launayite but halfway between the 2 and 2₁ axes in rouxelite (see Fig. 4).

At the interface between A rods, B rods, and C walls, Cu atoms are located in tetrahedral coordination in both cases.

In summary, the simplified parent structures of launayite and rouxelite are built of isotypic A/B layers and fundamentally different C walls. Thus, they can be considered members of a family of merotypes (Ferraris et al., 2008), a generalisation of polytypes, where members of a family share only a part of the layers and where other layers may differ.

The description above was given with respect to modules of PbS- and SnS-like archetype structures. From a polytype/merotype approach, a different picture is obtained. Thereto, we consider the features/atomic positions that both structures have in common: foremost, the A/B layers, but also the protuberances of the C walls and the tetrahedrally coordinated Cu atoms. These are combined to layers X with c12/m1 symmetry, as shown with a red background in Fig. 5. The remaining atoms are attributed to layers Y, again with c12/m1 symmetry (blue background in Fig. 5), which correspond to the C walls without the protuberances. Here, from a symmetry point of view, both approaches are equivalent, and henceforth we will stick to the crystal-chemical interpretation. In other cases, though, both may lead to fundamentally different descriptions, as we have shown for owyheeite (Stöger et al., 2023).

Launayite and rouxelite not only share structural features, but their diffraction patterns likewise show similarities. While in both cases the reflections corresponding to the C2/m parent structure are strong and well defined (Fig. 6a, blue squares), they possess additional superstructure reflections in the k=n+1/2 planes, which indicate a doubling of the b axis (Fig. 6b) and which can be assigned to multiple domains in both cases.

The major domain of the launayite crystal under investigation is described by a modulation wave vector q1=1/2b*, indicated by red circles in Fig. 6b, which is compatible with monoclinic symmetry. Observe that there are twice as many reflections of this domain in the k=n+1/2 planes as in the k=n planes because the former contains first-order satellites from both adjacent planes. The second-order satellites, which would appear at places h+k odd in the k=n planes, are not observed (see red circles in Fig. 6a), which explains an excessive number of unobserved reflections in the structure refinements. Thus, in contrast to what one might infer from the 1/2b*-component of the q1-vector, this structure is a 4-fold superstructure because the second-order satellites are located at the systematic absences of the C2/m space group.

Two minor twin domains can be described as using the pair of modulation wave vectors q2=1/2a*+1/2b* and q2′=-1/2a*+1/2b*. These two domains are triclinic 2-fold superstructures. Examples of such reflections are indicated in Fig. 6b by black arrows.

The launayite structural data presented here are based on the q1 monoclinic structure. The same superstructure was observed for rouxelite, though as a minor domain (Topa et al., 2025). The dominant domains of rouxelite were a pair of triclinic twin domains with modulation wave vectors q3=1/2a*+1/2b*+1/2c* and q3′=-1/2a*+1/2b*-1/2c*.

The symmetry relationships of the launayite and rouxelite domains with their parent C2/m structures are shown in Fig. 7. To show the common and distinct superstructure types, the structures of launayite and rouxelite are combined in the same family tree, even though the parent structures are technically not isotypic (but rather merotypes). The symmetry of the q1 superstructure is a klassengleiche (same point group, different lattice) subgroup of index 4 and type P2/a (left branch in Fig. 7). The symmetry descent can be decomposed into two klassengleiche steps of index 2: loss of the C-centring from C2/m to P2/a, followed by a doubling of the b axis. The latter can be more precisely designated as an isomorphic group/subgroup relationship, as not only the point group but also the affine space group type is retained (hence the code i2 in Fig. 7).

The symmetry of the q2 superstructure is a general (loss of point and translation symmetry) subgroup of index 4. It can be decomposed into a translationengleiche (different point group, same lattice) descent to the space group C1‾, followed by an isomorphic descent to a structure with lattice basis (a-b,2b,c) (middle branch in Fig. 7). Since we did not determine the structure of the q2 domain, we cannot rule out a loss of inversion symmetry to C1; however, there is no reason to assume that it exists. The q3 superstructure (major domain of rouxelite) is derived analogously, but there the lattice basis is (a-b,2b,c-b) (right branch in Fig. 7).

To understand the formation of the different superstructures, it is crucial to note that by doubling the b axis, translation symmetry is lost and the A and B rods now possess two different potential origins which are schematised in Fig. 8 by hatching. In the q1 superstructure, the b components of the origins of all rods are equal. In the q2 and q3 superstructures (where the q vector possesses an a* component of 1/2), the origins are shifted by b (of the parent structure) for each translation along the a axis of the parent structure. In the q3 superstructure (with an additional c* component of 1/2 in the q vector), there is an additional shift along b when crossing the C wall.

Figure 8 highlights a fundamental issue with comparing modulations/superstructures of non-isotypic basic structures. An alternative setting of the launayite parent structure that still maintains comparability with rouxelite is (a,b,-a+c). This setting is just as valid, though we decided against it, because the corresponding monoclinic angle β∼131.9° is larger than 120°. However, in that setting (see blue arrow in Fig. 8), the minor domain features an origin shift of b when translating along c, as in rouxelite, which means that in this setting the modulation wave vector is identical to the major domain of rouxelite. Thus, it makes little sense to compare modulation wave vectors in distinct structures.

In summary, the superstructures are caused by a doubling of the translation period of the A and B rods in the [010] direction, and the cell and symmetry are due to the distribution of the origins of the A and B rods in the structure.

The major q1 superstructure of launayite is shown in Fig. 9. The reflection symmetry is lost, and the layer group symmetries of the C walls and the A/B layers are reduced to p12/a1. The rod groups of the A modules are reduced to pb-1 (loss of the 2₁-screw rotation), and those of the B modules are reduced to pb121 (loss of inversion). The latter therefore appear in two orientations with respect to inversion, indicated by different shading in Fig. 9. In contrast, in the q2 and q3 superstructures, all rotational symmetry is lost and only centres of inversions remain. For details, see our recent publication on rouxelite (Topa et al., 2025).

Observe that the major domain features the same 2/m point group as the parent structure; therefore modulation cannot be responsible for the twinning observed in the (Lau1) grain (see above). Since we only collected diffraction data of an untwinned crystal (Lau2), the twin law cannot be stated unambiguously. However, in rouxelite we observed twinning by reflection at (100). We suspect that the same is true for launayite. This twinning suggests a reversal of the orientation of the A/B layers with respect to [100], which can be rationalised by partial [100] reflection symmetry of the halved C walls: given a C wall, the adjacent A/B layer can be reflected at [100] without major distortions.

Given the intricate nature of launayite's crystal structure, the proposed cation distributions were evaluated using element-specific bond-length trends, following the approach of Trömel (1980) and Berlepsch et al. (2001a, b). This technique has been applied recently to the crystal structure of rouxelite from Monte Arsiccio Mine (Topa et al., 2025), which serves as a point of comparison with launayite. The procedure involves comparing pairs of contrasting bond lengths – those lying roughly within a plane and those extending above and below it – found in coordination octahedra and in mono-capped trigonal prisms (or split octahedra). The “in-plane” bonds correspond to equatorial positions (such as the shared base of two opposing pyramids or the base of a capping pyramid), while the “out-of-plane” bonds point toward axial ligands, more or less perpendicular to these “planes” which run parallel to the [010] zone axis and share the same orientation as the well-developed lone electron micelles.

The resulting x–y scatter plot (Fig. 10) reveals that the Pb-bearing capped prisms generally plot near the diagram's diagonal, indicating only minor disparities between paired bond lengths in the plane. In contrast, greater deviations from the ideal Pb curve are observed for the axial bonds, likely reflecting local distortions caused by adjacent semimetal polyhedra or positional splitting of the central cation.

Similar to the findings in rouxelite (Topa et al., 2025), the bond pairs within Sb-centred octahedra do not conform to the expected hyperbolic trend. Instead, they follow a nearly linear pattern intersecting the Sb curve, implying that the sums of the opposing in-plane bond lengths are consistently maintained constant (roughly between 5.6 and 5.7 Å). This supports a model of highly symmetrical in-plane ligand arrangements which are aligned with the crystallographic b axis.

The Armbruster–Hummel plot (Armbruster and Hummel, 1987) shown in Fig. 11 (used to distinguish between Sb- and Pb-dominated sites by plotting the mean of the three shortest bond lengths against the next two) displays a clear separation between the two types of coordination environments.

The resulting structural formula obtained from the SCXRD study for the asymmetric unit, Cu₂Pb_20.31Sb_23.028As_2.662S₆₀, Z=4, is close to EMPA empirical formula, Cu_2.078Ag_0.059Tl_0.057Pb_20.404Sb_22.830As_2.772S_59.80. The resulting ideal formula Cu₂Pb₂₀(Sb,As)₂₆S₆₀ (Z=4) confirms the stoichiometric formula given by Moëlo (1983). Launayite undergoes limited homovalent and heterovalent substitutions, such as Sb3+→ As³⁺ and 2Pb2+→ Me3++Me¹⁺, where Me³⁺ is Sb and As; where Me¹⁺ is Ag (0.059 apfu), Tl (0.057 apfu), and Cu^extra (0.078 apfu); and where Cu^extra is (Cu - 2).

The ternary plot (in atomic percent) in the (Cu + Ag + Tl) - (Pb + Hg) - (Sb + As) system (Fig. 12) based on all measured points from Buca della Vena (#20) and Monte Arsiccio (#171) from Topa et al. (2025), as well as launayite from this work (Table 3), illustrates a clear separation between the different groups in terms of Me¹⁺, Me²⁺, and Me³⁺ content. Note that the “inclined narrow line” shape of the launayite data points to a narrow Me¹⁺ variation together with a larger Me³⁺ variation and smaller Me²⁺ variation. This indicates that Sb → As substitution is stronger than 2Pb2+→ (Ag,Tl,Cuextra)1++ (Sb,As)³⁺ substitution, which is not important in launayite. Chemically and structurally, Hg is present only in rouxelite, but Cu is essential for both minerals. Chemically, Ag, Tl, and As are present in launayite and rouxelite but bear no relevance for the ideal formula. Whether As stabilises launayite remains to be seen. No distinct position of Ag and Tl could be located by x-ray diffraction in launayite.

Several binary plots (in atomic percent) using the same set of data points as described above are presented in Fig. 13a–d. These plots reveal clearly distinct chemical compositions for the two minerals, along with the internal compositional variations within each. Figures 12 and 13c show clear trends that indicate a substitution of Pb with (Sb,As) without a change in Me¹⁺. This trend is very small, spanning ±0.43 Pb atomic % and ±0.75 (Sb,As) atomic %, and we have not been able to identify a known substitution mechanism or propose how the charge balance for this substitution occurs. Table 7 shows mixed Pb/Sb positions which could be interpreted as possible candidates for such a substitution mechanism; however, we have no evidence for the mechanism of this very limited substitution or for how charge balance occurs.

Empirical, structural, and ideal formulae of launayite and rouxelite, along with their unit cell parameters, are presented in Table 6. The unit cell volume of rouxelite is larger than that of launayite, and substitutional mechanisms are more pronounced in rouxelite than those in launayite, owing to its wider C walls as described above.