Quantitative chemical analyses were carried out using a Cameca SX100 electron microprobe at the National Museum of Prague (Czech Republic). Experimental conditions were as follows: WDS mode, accelerating voltage of 25 kV, beam current of 20 nA, and beam diameter of 0.7 µm. Standards (element, emission line) were as follows: Ag metal (AgLα), Bi₂Se₃ (BiMβ), CdTe (CdLα), chalcopyrite (CuKα, SKα), pyrite (FeKα), PbS (PbMα), NiAs (AsLβ), Sb₂S₃ (SbLα), Tl(Br,I) (TlLα), and ZnS (ZnKα). The peak counting times were 20 s for all elements, and 10 s for each background. Other elements (i.e., Au, Cl, Co, Ga, Ge, Hg, In, Mn, Ni, Sn, Se, and Te) were found to be below the detection limits (0.02 wt %–0.10 wt %). Matrix correction by PAP procedure (Pouchou and Pichoir, 1985) was applied to the data. As reported in other papers (e.g., Biagioni et al., 2020; Sejkora et al., 2021), there are several different approaches to calculate the chemical formula of the TGMs. Due to possible vacancies at the S(2) site in the Ag-rich members, the best approaches are as follows: (1) normalization on the basis of ΣMe = 16 apfu, assuming that no vacancies occur at M(2), M(1), and X(3), or (2) normalization on the basis of (As + Sb) = 4 apfu, taking into account that previous studies (e.g., Johnson et al., 1986) revealed that negligible variations with respect to the ideal number of X(3) atoms occur. The results of our study indicate that only very minor vacancies, if any, possibly occur at M(2), M(1), and X(3) sites, and, therefore, the first approach was selected for calculation. Chemical data for the grain of argentotennantite-(Fe) used for the single-crystal X-ray diffraction study are given in Table 2, whereas Table S1 in the Supplement reports the full dataset (386 spot analyses in total) of the chemical analyses of the described TGMs.

A single-crystal X-ray diffraction study of argentotennantite-(Fe) was performed on a small grain (0.035×0.030×0.025 mm) extracted from the rim of an aggregate of pyrargyrite and argentotetrahedrite-(Zn) (Fig. 3). Data collection was performed using a Bruker D8 Venture four-circle diffractometer equipped with an air-cooled Photon III detector and microfocus MoKα radiation (Centro per l'Integrazione della Strumentazione scientifica dell'Università di Pisa, University of Pisa, Italy). The detector-to-crystal distance was set to 38 mm. Data were collected using φ scan modes, in 0.5° slices, with an exposure time of 20 s per frame. A total of 216 frames were collected, and they were integrated with the Bruker SAINT software package using a narrow-frame algorithm. Data were corrected for Lorentz polarization, absorption, and background. Unit-cell parameters were refined on the basis of the XYZ centroids of 498 reflections above 20 σI with 5.52<2θ<56.30° as a=10.4365(5) Å, and V=1136.75(17) ų, with space group I4‾3m.

The crystal structure of argentotennantite-(Fe) was refined using Shelxl-2018 (Sheldrick, 2015) starting from the atomic coordinates of Johnson and Burnham (1985). The occurrence of the racemic twin was modeled, and the structure had to be inverted. The following neutral scattering curves, taken from the International Tables for Crystallography (Wilson, 1992), were used: Cu vs. Ag at the M(2) site, Cu vs. Fe at M(1), As vs. Sb at X(3), and S at the S(1) and S(2) sites. The scattering factors of Cu (Z = 29) and Fe (Z = 26) are similar, and, consequently, the site occupancy at M(1) was fixed on the basis of electron microprobe data, idealizing the occupancy at (Cu_0.67Fe_0.33). After several cycles of isotropic refinement, the R1 converged to 0.0842, confirming the correctness of the structural model. The anisotropic model for both cations and anions converged to R1=0.0384 for 246 reflections with F>4σF and 20 refined parameters. The details of data collection and refinement are given in Table 3. Fractional atom coordinates and equivalent isotropic displacement parameters are reported in Table 4. Table 5 reports selected bond distances, whereas weighted bond-valence sums, calculated according to Brese and O'Keeffe (1991), are shown in Table 6. The crystallographic information files (CIFs) are deposited in the Supplement of this paper.

Powder X-ray diffraction data could not be collected due to the paucity of available material. Consequently, powder X-ray diffraction data, given in Table 7, were calculated through the software PowderCell 2.3 (Kraus and Nolze, 1996) using the structural model of argentotennantite-(Fe) discussed below.

Chemical data for the holotype grain of argentotennantite-(Fe) used for the single-crystal X-ray diffraction study are given in Table 2 and correspond to the following formula: Ag_3.36(9)Cu_6.75(9)Fe_1.88(2)Zn_0.07(1)As_2.29(23)Sb_1.66(22)S_12.82(5). Taking into account the results of the crystal structure refinement (see below), the crystal chemical formula can be written as ^M(2)(Ag_3.36Cu2.70)Σ6.06 ^M(1)(Cu_4.05Fe_1.88Zn0.07)Σ6.00 ^X(3)(As_2.29Sb1.66)Σ3.95S(1)+S(2)S_12.82. The end-member formula of argentotennantite-(Fe) is Ag₆(Cu₄Fe₂)As₄S₁₃ (Z=2), corresponding to (in wt %) Cu 14.70, Ag 37.42, Fe 6.46, As 17.33, S 24.10, and total 100.00.

The chemical compositions for other argentotennantite-(Fe) grains and for associated Ag-rich tennantite-(Fe) and argentotetrahedrite-(Zn) are given in Table S1. Late-stage rims of TGM crystals (Figs. 3 and 4) are represented by zoned argentotennantite-(Fe) to Ag-rich tennantite-(Fe), with a negative correlation between Ag and As (Fig. 6b) and a less distinct negative correlation of Ag and S contents (Fig. 6c). All compositions belong to Fe- and As-dominant members with S contents >12.5 apfu (Fig. 6a–c). The silver contents of argentotennantite-(Fe) are found to lie in the range of 3.00–3.68 apfu; the mineral is Fe-dominant (1.64–1.93 apfu), with only minor Zn (up to 0.25 apfu) and traces of Cu²⁺, Pb, and Cd of up to 0.08, 0.04, and 0.01 apfu, respectively. Arsenic (1.98–3.25 apfu) is partly substituted by Sb (0.69–1.97 apfu), and S contents were found to be in the range 12.54–13.07 apfu. For Ag-rich tennantite-(Fe), Ag contents are in the range of 2.23–2.99 apfu; it is Fe-dominant (1.30–1.95 apfu), with minor Zn (up to 0.60 apfu) and, locally, traces of Pb and Cd of up to 0.04 and 0.01 apfu, respectively. The calculated Cu²⁺ contents occasionally reach up to 0.27 apfu but are usually only around 0.05 apfu. The dominant As (2.66–3.82 apfu) is accompanied by Sb (0.11–1.22 apfu), and S contents were determined to be in the range 12.61–13.18 apfu.

The relatively uniform chemical composition of early-stage TGM crystals (Figs. 3 and 4) is significantly different (Fig. 6) and corresponds to argentotetrahedrite-(Zn) (Sejkora et al., 2022), with 3.21–3.29 apfu Ag as the A-chemical constituent. The dominant Me²⁺ element is Zn (1.46–1.92 apfu), accompanied by minor Fe (0.02–0.43 apfu) and Cd (0.02–0.05 apfu); calculated Cu²⁺ contents are low (up to 0.09 apfu). The extent of AsSb₋₁ substitution is limited to 0.60 apfu As, and contents of S (12.76–13.16 apfu) are close to the ideal value of 13 apfu. The average formula (28 spot analyses) can be expressed as (Ag_3.25Cu2.73)Σ5.98[Cu_4.00(Zn_1.73Fe_0.19Cu_0.04Cd0.04)Σ2.00]_Σ6.00(Sb_3.76As0.26)Σ4.02S_13.02.

The crystal structure of argentotennantite-(Fe) agrees with the general features of the members of the tetrahedrite isotypic group, i.e., a collapsed sodalite-like framework formed by corner-sharing M(1)S(1)₄ tetrahedra with cages hosting S(2)-centered M(2)-octahedra, surrounded by four X(3)S(1)₃ trigonal pyramids (e.g., Johnson et al., 1988).

The M(2) site shows a planar triangular coordination, with an average 〈M(2)–S〉 distance of 2.342 Å. The refinement of its site occupancy points to Ag_0.53(4)Cu_0.47(4), corresponding to a mean atomic number (MAN) of 38.54 electrons, to be compared with the value obtained during electron microprobe analyses, i.e., ∼39 electrons (38.99 normalizing the sum of Ag and Cu to one atom at the M(2) site). In other isotypic Ag members of the tetrahedrite group, e.g., argentotetrahedrite-(Zn) (Sejkora et al., 2022) and zvĕstovite-(Zn) (Sejkora et al., 2021), the M(2) site was found to be split. It is very probable that, in argentotennantite-(Fe), such a splitting also occurs, as testified to by the high Ueq value observed at this site, i.e., 0.076(2) Ų, as well as the high residuals around the M(2) site and the M(2)–S distances, physically too short when the site is occupied by Ag. A split model suggests the preferential occupancy of Cu at M(2a) and Ag at M(2b), as supported by observed average bond distances. However, some correlations higher than 0.8 were found during the refinement of the split model, and consequently an unsplit model is presented. The bond-valence sum is 1.36 valence units (v.u.).

The tetrahedrally coordinated M(1) site has an average 〈M(1)–S〉 bond distance of 2.329 Å. Electron microprobe analysis agrees with an ideal (Cu_0.67Fe_0.33) site occupancy, with a minor replacement of these elements by Zn. On the basis of this proposed site occupancy, the bond-valence sum at the M(1) site is 1.60 v.u., larger than the theoretical value of 1.33 v.u.

The X(3) site is a mixed (As, Sb) position, with an average 〈X(3)–S〉 distance of 2.340 Å. Considering the ideal As–S and Sb–S distances of 2.26 and 2.45 Å, respectively, calculated on the basis of the bond parameters of Brese and O'Keeffe (1991), the observed average distance should correspond to the site occupancy (As_0.58Sb_0.42), to be compared with the results of crystal structure refinement (As_0.53Sb_0.47) and electron microprobe analysis (As_0.58Sb_0.42). The bond-valence sum, calculated on the basis of the refined site occupancy, is 3.18 v.u.

The S(1) and S(2) sites are fully occupied by S. The S(2)-centered octahedron has a volume of ∼15 ų, in accord with an S(2)M(2)₆-occupied octahedron (e.g., Welch et al., 2018; Sejkora et al., 2022; Biagioni et al., 2026). The bond-valence sums of S atoms at S(1) and S(2) are 2.25 and 3.48 v.u., respectively. Both S atoms are overbonded, mainly due to the too-short Ag–S distances. This is particularly true for the S atom hosted at the S(2) position, resulting in an exceptionally high overbonding. This feature has been observed in all of the other known Ag members of the tetrahedrite group, e.g., in argentotetrahedrite-(Zn) (bond-valence sums ranging between 3.00 and 3.30 v.u. for the three refined samples), zvĕstovite-(Zn) (3.24 v.u.), and zvĕstovite-(Fe) (2.76 v.u.) (Sejkora et al., 2021, 2022; Biagioni et al., 2024). This is due to the short M(2)–S(2) bond distance, definitely shorter than the ideal 〈[III]Ag–S〉 distance, i.e., 2.56 Å. The physically unrealistic distances may be due to some disorder affecting both the S(2) position (e.g., Sejkora et al., 2022) and the M(2) site. The former shows a high Ueq value (0.12 Å). A high value of Ueq could be related to a positional disorder or to an incorrect refinement of the site occupancy at S(2). As no significant vacancies are suggested by the chemical analyses, by the refinement of its site occupancy (pointing to a full occupancy), or by the volume of the S(2)-centered octahedron, the former hypothesis is more probable. This could be in accord with Peterson and Miller (1986), who proposed that, in order to achieve longer Ag–S distances, S(2) should be displaced from the 2a position. Indeed, as in argentotetrahedrite-(Fe), in argentotennantite-(Fe), a residual at the 8c (x, x, x) position, with x=0.0448, was also observed in the difference-Fourier map. Moreover, as discussed above, the Ueq value at the M(2) site is high and also suggests a splitting of this position into at least two sub-sites able to achieve physically sound bond distances (e.g., Biagioni et al., 2024).

The structural formula of argentotennantite-(Fe), based on electron microprobe data and crystal structure refinement, could be written as ^M(2)(Ag_3.18Cu2.82)Σ6.00M(1)(Cu_4.00Fe2.00)Σ6.00X(3)(As_2.12Sb1.88)Σ4.00S(1)+S(2)S₁₃ (Z=2).

Johnson et al. (1987) proposed the following relation between the unit-cell parameter a and the chemistry of the studied sample: a (Å) = 10.379 + 0.082(Ag) - 0.01(Ag²) - 0.009(Cu*) + 0.066(Hg) - 0.038(As) + 0.144(Bi). On the basis of the formula derived from the structural analysis, the calculated unit-cell parameter is a=10.46 Å, which aligns with the observed value of 10.44 Å.

The name “arsenofreibergite” was introduced by Biagioni et al. (2020) as a series name to indicate the As analogues of the freibergite-series minerals. At that time, only argentotennantite-(Zn) from the Kvartsitoviye Gorki deposit, Kazahstan, was formally approved (Spiridonov et al., 1986), whereas Števko et al. (2018) had reported chemical data corresponding to argentotennantite-(Fe). Both minerals are characterized by the ideally fully occupied S(2) site, but no crystallographic data were available. Only recently were the first crystal chemical data on members of the arsenofreibergite series made available due to the description of kenoargentotennantite-(Fe) (Biagioni et al., 2026) and due to the present work, where argentotennantite-(Fe) is described. These data allow some comparison among these species (Table 8).

According to Spiridonov et al. (1986), the type material of argentotennantite-(Zn) has the chemical formula (Ag_5.67Cu0.33)Σ6.00(Cu_4.15Zn_1.52Fe_0.37Pb_0.01Cd0.01)Σ6.06(As_2.14Sb1.89)Σ4.03S_12.90, to be compared with its Fe analogue, whose chemical formula, as reported above, is (Ag_3.36Cu2.70)Σ6.06(Cu_4.05Fe_1.88Zn0.07)Σ6.00(As_2.29Sb1.66)Σ3.95S_12.82. The former has a higher Ag content, reflected in a larger unit-cell parameter, i.e., a=10.58 Å, expanded with respect to that observed in argentotennantite-(Fe), a=10.44 Å. In both species, a significant amount of Sb occurs, with As / (As + Sb) atomic ratios of 0.531 and 0.579 for the Zn and Fe isotypes, respectively. A non-negligible content of Sb also occurs in the keno-counterpart of argentotennantite-(Fe), i.e., kenoargentotennantite-(Fe), whose empirical formula has been reported by Biagioni et al. (2026) as (Ag_4.41Cu1.59)Σ6.00(Cu_4.02Fe_1.22Zn_0.39Hg0.37)Σ6.00(As_2.80Sb1.19)Σ3.99S12.12□0.88, with an As / (As + Sb) atomic ratio of 0.702. The presence of a significant amount of Sb in kenoargentotennantite-(Fe) has been interpreted considering the fact that, following Barton and Skinner (1978), the stability field of As₂S₃ is at a higher f(S₂) than the Ag₂S–Ag equilibrium, and, consequently, As³⁺ could not coexist with Ag metal; on the contrary, below 300 °C, the equilibrium Sb₂S₃–Sb is below the Ag₂S–Ag equilibrium, suggesting that minor Sb seems necessary to stabilize kenoargentotennantite-(Fe). This hypothesis cannot be applied to argentotennantite-(Fe), where no hints of Ag–Ag bonding were observed. It is indeed possible that the occurrence of Sb is also necessary for steric reasons.

This is supported by some other natural occurrences of (Ag, As)-rich TGMs. For instance, Ixer and Stanley (1983) reported the chemistry of a grain of TGMs corresponding, on the basis of ΣMe = 16 apfu, to the composition (with rounding errors) Ag_6.02(Ag_1.91Cu_1.85Zn_1.44Fe_0.48Cd_0.16Pb0.16)Σ6.00(As_2.15Sb1.87)Σ4.02S_12.12 [As / (As + Sb) = 0.535], whereas Zakrzewski (1989) reported two chemical analyses corresponding to Fe- and Zn-rich members of the arsenofreibergite series with As / (As + Sb) atomic ratios of 0.500 and 0.502, respectively. In the zvĕstovite series, the As / (As + Sb) ratio is 0.604 in zvĕstovite-(Zn) (Sejkora et al., 2021) and 0.896 in zvĕstovite-(Fe) (Biagioni et al., 2024).

As discussed by Johnson et al. (1988), the chemical composition of TGMs affects the framework rotation and the percent of distortion from ideality. In particular, the two most important parameters for the stability of TGMs are the size of the “spinner blades” (in the sense of Wuensch, 1964 – see Fig. 4 in his paper) and that of the X(3) cations. The replacement of Cu by Ag at the M(2) site increases both the rotation angle of the framework and its distortion. Figures 8 and 9 in Johnson et al. (1988) show that TGMs with M(2) occupied by Ag and X(3) hosting As show high values of both of these parameters, suggesting their instability owing to high structural strain. However, one can see that substituting As with Sb helps in reducing such a strain, thus allowing the existence of (Ag, As)-bearing TGMs. This is in line with Fig. 6b, which shows a negative correlation between the Ag content and the As / (As + Sb) ratio.

For this reason, even if the end-member formula of argentotennantite-(Fe) is defined to be Ag₆(Cu₄Fe₂)As₄S₁₃, its actual formula would be more correctly represented by the formula Ag₆(Cu₄Fe₂)(As,Sb)₄S₁₃.

Tetrahedrite-group minerals with chemical compositions corresponding to that of argentotennantite-(Fe) were previously reported. To the best of our knowledge, the first chemical data corresponding to argentotennantite-(Fe) were reported by Števko et al. (2018) for part of the zoned TGM rims of argentotetrahedrite-(Zn) grains (Sejkora et al., 2022) from the Kremnica Au–Ag epithermal deposit, Slovakia. After recalculation of published chemical analyses (Števko et al., 2018) on the basis of ΣMe = 16 apfu, 11 spot analyses corresponded to argentotennantite-(Fe) with 3.02–3.77 apfu Ag, 1.31–1.98 apfu Fe, and S contents in the range of 12.54–12.99 apfu. Other parts of these rims are formed by kenoargentotennantite-(Fe), with 3.24–4.50 apfu Ag and 11.70–12.49 apfu S (13 spot analyses), or Ag-rich tennantite-(Fe), with 0.67–2.99 apfu Ag (28 spot analyses). Later, Kolitsch et al. (2019) gave, for the Ag- and As-richest portions of zoned grains (10 µm in size) hosted in fluorite from the Grube Clara, Germany, the empirical formula (Ag_3.36Cu_2.64)Cu₄(Fe_1.61Zn_0.39)(As,Sb)₄S₁₃. Ondruš et al. (2003) published, for “argentotennantite” from Jáchymov, Czech Republic (grains up to 25 µm), contents of S in the range of 12.46–12.69 apfu, but after recalculation on the basis of ΣMe = 16 apfu, S contents were only 12.06–12.44 apfu, and so this phase may correspond to kenoargentotennantite-(Fe), with 3.17–4.29 apfu Ag and 1.52–1.84 apfu Fe.