Secondary uranyl arsenates–phosphates and Sb–Bi-rich minerals of the segnitite–philipsbornite series in the oxidation zone at the Prakovce-Zimná Voda REE–U–Au quartz-vein mineralisation, Western Carpathians, Slovakia

Abstract

This work is an investigation of the assemblages of supergene minerals occurring in hydrothermal REE–U–Au quartz-vein mineralisation at the Prakovce-Zimná Voda site, Slovakia. Heterogeneous uranyl arsenates and minor phosphates of the autunite group (nováčekite, kahlerite, threadgoldite, autunite, arsenuranospathite and chistyakovaite) together with scorodite and Sb–Bi-rich philipsbornite–segnitite-series minerals formed by oxidising fluids during decomposition and leaching of primary hypogene uraninite, brannerite and base-metal sulfides and sulfosalts. A progressive change of pH from acidic to near-neutral due to the gradual consumption of sulfides resulted in the formation of late phosphuranylite, pharmacosiderite and arseniosiderite. Goethite and other Fe oxides represent the latest hydrous ferric mineral phases and were formed after most of the As was already fixed in Fe arsenates. Antimony and Bi were taken up only into philipsbornite–segnitite and suggest unusual conditions during this process. X-ray absorption spectroscopy indicates that Sb in the philipsbornite–segnitite is fully oxidised (0.1–0.4 apfu Sb⁵⁺, octahedral coordination on the G site). Pentavalent Sb together with the presence of ferric oxides and arsenates and uranyl minerals suggest oxidative conditions during weathering. This study also indicates that hydrous ferric arsenates are dominant and stable secondary minerals in a supergene environment in a quartz vein rich in Fe and As accompanied by elevated concentrations of U, Pb, Sb, Bi, S, P, Ca and Ba under oxidising conditions.

Introduction

In oxidation zones of uranium deposits, uranyl minerals with autunite-type sheets (known as ‘uranium micas’) form during the oxidation–hydration weathering of uraninite from acidic solutions which have been generated by the simultaneous oxidative weathering of sulfides (Belova, 1975, 2000; Göb et al., 2013; Krivovichev and Plášil, 2013; Plášil, 2014). Once the sulfides are exhausted, the pH can increase, providing conditions that are conducive for the formation of uranyl minerals with phosphuranylite-type sheets. The mode and results of uraninite–sulfide weathering depend on local parameters, such as mineralogy and geochemistry (e.g. the mass ratio of uraninite to sulfides, or the presence or absence of carbonates) of the ores and the host rocks, their permeability (caused primarily by the rock porosity, or secondarily by the brittle tectonic structures), and the compositional evolution of the percolating ground water (Krivovichev and Plášil, 2013; Plášil, 2014).

Many oxidation zones of uranium and base-metal sulfide deposits, uranium mine tailings, and drainage systems also contain abundant Fe oxides and arsenates. The Fe oxides are typically represented by goethite and ferrihydrite, whereas common crystalline Fe arsenates include scorodite, kaňkite, symplesite, pharmacosiderite, arseniosiderite and yukonite (e.g. Drahota and Filippi, 2009; Lalinská-Voleková et al., 2022).

Goethite is stable over a large pH–Eh range and in oxidised conditions, and can precipitate at very low pH if pyrite is oxidised. It commonly contains appreciable amounts of As, which is the most abundant constituent other than Fe³⁺, followed by Al³⁺, Si⁴⁺ and Ca²⁺ (e.g. Göb et al., 2013; Paikaray, 2015; Herrmann et al., 2018). Ferric oxides can adsorb a remarkable amount of As(V) onto their surfaces under neutral and acidic conditions (e.g. Majzlan et al., 2007, 2011; Lalinská-Voleková et al., 2022); their affinity to As(III) is lower than that to As(V) (Cheng et al., 2009).

At low pH and in highly concentrated solutions, scorodite is the most common ferric arsenate secondary mineral formed as a product of oxidation of arsenopyrite or arsenian pyrite. In addition to scorodite, an X-ray amorphous equivalent, termed pitticite, is commonly reported as an important As carrier (Drahota and Filippi, 2009; Das, 2019).

Under near-neutral conditions in the BaO–CaO–Fe₂O₃–As₂O₅–H₂O system, solutions with high As(V) and low Fe(III) concentrations precipitate arseniosiderite and yukonite intimately associated with pharmacosiderite (Paktunc et al., 2015). Several studies have reported the replacement of pharmacosiderite and scorodite by arseniosiderite and its coprecipitation with goethite (Paktunc et al., 2004; Sejkora et al., 2006; Filippi et al., 2007; Drahota et al., 2009). Detailed investigations of localities with secondary ferric arsenates (e.g. Sejkora et al., 2006) show that minerals of the pharmacosiderite supergroup occur at these sites as a minor, generally inconspicuous, phase and that at some localities, pharmacosiderite can be an important, or the dominant, carrier of As (Drahota et al., 2009, Majzlan et al., 2019). Such occurrences seem to be linked to slow weathering under near-neutral conditions. Hence, even at sites where substantial acidity is developed, pharmacosiderite could crystallise in microenvironments that are able to neutralise the acidity (Majzlan et al., 2019). In the presence of Ba, formation of bariopharmacosiderite is favoured and seems to persist even under long-lasting circumneutral conditions (Herrmann et al., 2018).

Supergene processes are also often responsible for the precipitation of alunite-supergroup minerals, for example in base-metal sulfide ore deposits, unconformity-type uranium deposits, pyritiferous rocks, mine drainage, acid soils and saprolite horizons (bauxite, laterite) and confined to the topmost part of the vadose or infiltration zone (Dutrizac and Jambor, 2000; Dill, 2001; Beaufort et al., 2005; Gaboreau et al., 2005, 2007; Welch et al., 2007, 2008, 2009; Adlakha and Hattori 2015). Among alunite-supergroup minerals, the Pb-rich members of the dussertite group, segnitite, ideally PbFe₃(AsO_3.5(OH)_0.5)₂(OH)₆ and philipsbornite, ideally PbAl₃(AsO_3.5(OH)_0.5)₂(OH)₆ form typically under oxidising and acidic weathering conditions (e.g. Birch et al., 1992; Rattray et al., 1996; Moura et al., 2007; Sejkora et al., 2009, 2011; Mills et al., 2014; Golebiowska et al., 2016; Pekov et al., 2016). The compositional variability of the philipsbornite–segnitite series and the coexistence of several solid solutions among arsenate, phosphate and sulfate alunite-supergroup end-members (notably plumbogummite, hidalgoite, kintoreite and beudantite) accompanied by minor Ba, Cu, Zn, Bi and Sb, has been described from different localities. Examples include Ba-rich philipsbornite from Cínovec, Czech Republic (David et al., 1990), Cu- and Zn-rich beudantite–segnitite-series minerals from Krupka and Jáchymov, Czech Republic (Sejkora et al., 2009, 2011), or Zn-rich segnitite from Broken Hill, Australia (Birch et al., 1992). Furthermore, to date, there have been only a few reported examples of either Bi or Sb, but not both, in alunite-supergroup mineral phases substituting within the octahedral and tetrahedral sites (e.g. Van Wambeke, 1975; Clark et al., 1986; Kolitsch et al., 1999; Mills et al., 2014; Golebiowska et al., 2016; Pekov et al., 2016; Števko et al., 2016; Sejkora et al., 2021).

In this work, we investigated a complex supergene uranyl autunite- and phosphuranylite-group mineral assemblage associated with a variety of supergene Fe arsenates and oxides from the Prakovce-Zimná Voda REE–U–Au quartz-vein mineralisation, Gemeric Unit, Western Carpathians, Slovakia. The aim of this work was a detailed mineralogical study focusing on secondary minerals. In addition, identification of unusual Bi- and Sb-rich minerals of the philipsbornite–segnitite series prompted a detailed crystal chemistry, micro-Raman and X-ray absorption spectroscopy (XAS) investigation to determine compositional variations, substitution mechanisms, valency, and the structural role of Sb and Bi within the investigated members of the dussertite group. Of significant note, the Bi content in the philipsbornite–segnitite series is the highest published Bi₂O₃ content in these minerals and natural samples reported to date.

Uranium occurrences in the Gemeric Unit, Western Carpathians

The Gemeric Unit is the major thick-skinned crustal-scale unit in the Central Western Carpathians, composed of Early Palaeozoic to early Late Carboniferous, mostly low-grade metasediments and metavolcanics intruded by small bodies of Permian granites (Plašienka et al., 1997). Radioactive anomalies and related uranium occurrences in the Gemeric Unit have been a subject of exploration campaigns, prospecting reports, and mineralogical–geochemical research (Varček, 1977; Novotný and Čížek, 1979; Rojkovič and Novotný, 1993; Rojkovič, 1997; Novotný et al., 1999; Donát et al., 2000; Ferenc et al., 2018; Ondrejka et al., 2023a, 2023b). Two principal genetic groups of U mineralisation can be recognised in the Gemeric Unit: (1) stratiform U–Mo deposits and occurrences developed usually in Permian volcano-sedimentary complexes, e.g. Košice (Kurišková-Jahodná), Krompachy (Petrova Hora), Novoveská Huta and Stratená (Rojkovič and Novotný, 1993; Rojkovič, 1997; Kohút et al., 2013) and Cu±U infiltration mineralisation in the copper-bearing Permian sandstones at Novoveská Huta, Šafárka occurrence with uraninite, coffinite, U–Ti oxides, molybdenite, pyrite and chalcopyrite (Ferenc et al., 2022) as primary dominant minerals; and (2) quartz±apatite veins with U–REE±Au and Mo mineralisation, developed in Early Palaeozoic rocks of the Gemeric basement and generally situated in the proximity of the Permian granites, e.g. Betliar, Čučma, Gemerská Poloma (Krátka Dolina), Henclová, Hnilec (Peklisko), and most notably Prakovce-Zimná Voda (Varček, 1977; Rojkovič et al., 1997; Števko et al., 2014; Ferenc et al., 2018; Ondrejka et al., 2023a, 2023b).

Geological setting

The Prakovce-Zimná Voda occurrence is located in the Lower Palaeozoic metamorphic rocks of the Bystrý Potok Formation, a part of the Gelnica Group in the Gemeric Unit (Fig. 1). The metamorphic rocks were intruded by Hummel granites that outcrop 600 m to the south-west of the investigated occurrence. These igneous rocks are leucocratic peraluminous granites with S-type affinity and a relatively high degree of fractionation, evidenced by higher P concentrations in K-feldspar and rare lithophile elements (Li, Rb, Cs, B, Sn, W, Nb and Ta). They originated and were emplaced during the post-Variscan, Permian (260–270 Ma) extension (e.g. Villaseñor et al., 2021, and references therein). Moreover, numerous hydrothermal quartz veins with Sb–Au mineralisation occur in the exocontact of the granites (e.g. Zlatá Idka and Poproč deposits near the Prakovce-Zimná Voda occurrence; Kobulský et al., 2011).

Figure 1. (a) Simplified geological-tectonic sketch-map of the Western Carpathians. (b) A geological map of the area investigated (modified after Bajaník et al., 1984). (c) A close-up view on the veins, alteration and host rocks (modified after Donát et al., 2000; from Ondrejka et al., 2023b, reproduced with permission).

Two quartz veins (termed Western and Eastern) with REE–U–Au mineralisation are hosted by fine-grained micaceous phyllites, with interbeds of fine-grained metaquartzites. The Western vein has an E–W strike, total length of ~90 m with an average dip of 65° to the S and conforms to the metamorphic schistosity of the host rocks. The thickness of the vein ranges from 3–30 cm. The vein is slightly corrugated and segmented by transverse faults into 3–55 m long segments, and the termination of the vein on both tails is also tectonic. Drilling surveys showed a tectonic zone cutting the vein at a depth of 7–10 m. The main vein body is accompanied by short veinlets with U–Au mineralisation and quartz veins without ore mineralisation. Along the contact in a zone of 2–8 m thickness, the host rocks are intensively argillitised, locally silicified, and impregnated by pyrite. Supergene alteration of pyritised rocks has caused their limonitisation.

In addition to the dominant quartz, the following hydrothermal minerals have been identified at the Prakovce-Zimná Voda occurrence (Rojkovič et al., 1997; Ondrejka et al., 2023b): uraninite, brannerite, arsenopyrite, pyrite, gold, rutile, bismuth, bismuthinite–stibnite, tintinaite–kobellite, gersdorffite, cobaltite, glaucodot, molybdenite, galena, tetrahedrite-(Fe), fluorapatite, monazite-(Ce), monazite-(Nd), monazite-(Sm), monazite-(Gd), xenotime-(Y), ‘xenotime-(Gd)’, hingganite-(Y), muscovite, chlorite and members of the tourmaline group. The maximum U content detected in the ore is 11,850 ppm (3080 ppm on average) and the maximum Au content is 164 ppm, with 25 ppm on average (Rojkovič et al., 1997). The REE orthophosphates are exceptionally rich in middle rare earth elements (especially Gd) and precipitated in response to the alteration of the primary uraninite, brannerite and fluorapatite by low-temperature hydrothermal fluids. Further details and description of the uraninite and REE minerals can be found in Ondrejka et al. (2023b).

Analytical methods

Electron probe micro-analyses (EPMA) in wavelength-dispersive spectrometry (WDS) mode were performed with a JEOL JXA – 8530FE electron microprobe (Earth Science Institute of the Slovak Academy of Sciences, Banská Bystrica, Slovakia) using 15 kV accelerating voltage and 15–20 nA beam current. The beam diameter varied, usually from 3–5 μm. A more focused <1–3 μm beam was used for small heterogeneous domains, however, even this beam size was not adequate for analysis of small domains. The microprobe was calibrated with natural and synthetic standards (Supplementary table S1), and the raw counts (20 s for all elements and 10 s for each background) were converted to wt.% of oxides using the ZAF matrix correction. Elemental contents in the mineral formulae are expressed in atoms per formula unit (apfu). The autunite-type minerals (U–As–P) were calculated based on 12 oxygen atoms. The pharmacosiderite (Pmsd), bariopharmacosiderite (Bpsd) and arseniosiderite (Assd) empirical formulae were normalised on the basis of (P + As + Si + S) = 3 apfu and zeolitic H₂O content was calculated from the nominal formula with H₂O = 6.5 (Pmsd), H₂O = 5 (Bpsd), H₂O = 3 (Assd) and OH, O on ideal stoichiometry. Scorodite (Scd) was calculated on the basis of (P + As + Si + S) = 1 and H₂O content was taken from the nominal formula as H₂O = 2. The alunite-supergroup minerals nomenclature used is after Bayliss et al. (2010) and the formulae of philipsbornite (Pbn) and segnitite (Sgt) were normalised on the basis of Σ(T + G) = 5 cations and the total H₂O content calculated assuming OH + F = 7 apfu. All mineral compositional plots were achieved using the CorelKit plug-in for CorelDRAW (Zhang et al., 2023).

The X-ray absorption near-edge structure (XANES) spectroscopy data were collected at the beamline of the Synchrotron Radiation Laboratory for Environmental Studies (SUL-X, Angströmquelle Karlsruhe, Germany) in the synchrotron radiation source ANKA. A silicon (111) crystal pair with a fixed beam exit was used as a monochromator. The X-ray beam was aligned to an intermediate focus, and then collimated by slits located at the distance of the intermediate focus to ~100 × 100 μm and subsequently focused with a Kirkpatrick-Baez mirror pair to ~50 × 50 μm at the sample position.

The μ-XAS spectra at the SbL ₁ and L ₃ edges were measured in transmission (standard) and fluorescence (samples) mode in energy steps of 5 eV in the region from –150 to –50 eV relative to the absorption edge, of 2 eV in the region from –50 eV to –20 eV, of 0.5 eV from –20 eV to +20 eV, and with a k step of 0.05 from +20 eV to +400 eV (~k = 10). The intensity of the primary beam was measured by an ionisation chamber. Fluorescence intensities were collected with a seven element Si(Li) solid-state detector with the energy window set to the Sb lines. Data were dead-time corrected, summed for all seven channels and divided by the input intensity, which was measured in an ionisation chamber prior to the sample analysis. The collected data were processed by ATHENA (Ravel and Newville, 2005).

The reference compounds used for the determination of the oxidation state and coordination geometry were natural stibnite (Sb₂³⁺S₃), synthetic schafarzikite (FeSb₂³⁺O₄), tripuhyite (FeSb⁵⁺O₄), Sb(V)-doped fresh ferrihydrite, and Sb⁵⁺-doped ferrihydrite aged to goethite.

The Raman spectra were obtained using a Thermo Scientific DXR3xi Raman Imaging microscope at the Slovak National Museum, The Natural History Museum in Bratislava, Slovakia. Several lasers were used during investigation, mainly 785 nm and 633 nm with 100× objective, 25 μm confocal pinhole and an EMCCD detector. The spectra were acquired from approximately the same spots as the EPMA at a laser power of 0.2−10 mW between 0.05 and 2 s (80 scans for a cycle) in ranges of 50−1800 cm^–1. The processing of spectra was carried out using the Thermo Fisher Scientific OMNIC v. 9.11 software package. Spectra of segnitite and philipsbornite were also measured on a LabRAM HR800 spectrometer (Horiba Jobin-Yvon), through an Olympus BX41 microscope confocally coupled to a Czerny-Turner type monochromator (focal length 800 mm). Laser wavelengths of 532 and 633 nm were used for excitation, the Raman-scattered light was collected in 180° geometry through a long-working distance 100×0.8 objective and dispersed by diffraction grating with 600 gr/mm onto a cooled CCD chip. The system resolution was <6 cm^–1. The Rayleigh line of excitation and a Teflon standard were used for calibration. The measurement conditions were adjusted to avoid thermal damage of delicate mineral aggregates. Spectra were pre-processed in Labspec5 software (Horiba Jobin-Yvon) and deconvoluted to single peaks by Gaussian–Lorentzian band shapes in PeakFit software version 4.12. The Raman activity of vibrational modes was calculated on the Bilbao crystallographic server (Kroumova et al., 2003) using the crystallographic data for philipsbornite (Cooper and Hawthorne, 2012).

Powder X-ray diffraction (XRD) analysis was performed using a Bruker D8 Advance diffractometer using CuKα radiation generated at 40 kV and 40 mA and an energy-dispersive Sol-XE detector. The beam was collimated with a slit assembly of 0.3°-6 mm–0.3°-0.2 mm, with primary and secondary Soller slits at 2.5°. The sample was scanned as random powder using a Si holder from 2 to 65°2θ with step size 0.02°2θ and 3 s counting time. Data were processed with the software DIFFRAC.EVA and PDF2/2010 database.

Results

Sample description and primary ore minerals

The minerals investigated were identified in 16 thin sections and mounts sampled from the REE–U–Au site (Western vein). Primary hypogene minerals, sometimes visible by the unaided eye, occur in fine-grained quartz with abundant stains and veinlets of rusty Fe oxides and arsenates (Fig. 2). Uranium minerals, base-metal sulfides and sulfosalts form clusters irregularly distributed within the vein. Here, macroscopic uraninite is the dominant primary mineral together with abundant brannerite and arsenopyrite accompanied by pyrite, gold, Ni–As–Co–Bi–Pb–Cu sulfides and sulfosalts, especially gersdorffite, galena, minerals of the stibnite–bismuthinite series and tintinaite–kobellite. Hydrothermal phosphates include fluorapatite, monazite- and xenotime-group minerals (Ondrejka et al., 2023a, 2023b).

Figure 2. Spatial and cross-cutting relationship of a solitary aggregate of ferric arsenates (FeAs), U minerals and a younger ferric oxide vein (Feox).

Supergene minerals

Uranyl arsenates and phosphates

Uranyl arsenates and phosphates represent common secondary products of uraninite and brannerite weathering in the oxidised parts of the REE–U–Au vein mineralisation investigated. They occur as large (<4 mm, mostly 50–150 μm in size) lath-shaped aggregates in close vicinity to altered botryoidal uraninite and arsenopyrite (Fig. 3a–c) or as anhedral clusters in a matrix of ferric oxides, arsenates and quartz (Fig. 3d–e). They also form aggregates (3–4 mm in size) of tabular crystals of yellow–green colour in fissures and cavities of the quartz veins (Fig. 4). These uranyl phases are heterogeneous, with complex substitution mechanisms within the autunite-type sheet minerals with a general crystal-chemical formula A^{n +}[(UO₂)(TO₄)](H₂O)_mX _k. The identified end-members, based on the dominant occupancy of A and T sites, are predominantly arsenates: (meta)kahlerite Fe²⁺(UO₂)₂(AsO₄)₂(H₂O)_8–12; (meta)nováčekite I–II Mg(UO₂)₂(AsO₄)₂(H₂O)_8–12; and the arsenuranospathite Al_1–x□_x[(UO₂)(AsO₄)]₂(H₂O)_20+3xF_1–3x to chistyakovaite Al[(UO₂)(AsO₄)]₂(F,OH)(H₂O)_6.5 series. Less abundant minerals include the uranyl phosphates threadgoldite Al[(UO₂)(PO₄)]₂(OH)(H₂O)₈ and autunite Ca[(UO₂)(PO₄)]₂(H₂O)₁₁. In addition, minerals with the phosphuranylite-type sheet are represented here by the rare phosphuranylite CaK(H₃O)₃(UO₂)[(UO₂)₃(PO₄)₂O₂]₂(H₂O)₈. Phosphuranylite forms pseudomorphs after uraninite in the vein portions where macroscopic sulfides are rare, or completely leached out. It was only identified based on powder XRD data, with the unit-cell parameters: a = 15.833 Å, b = 13.745 Å, c = 17.304 Å and V = 3766 ų.

Figure 3. Back-scattered electron images of primary and secondary mineral assemblages from REE–U–Au vein mineralisation at Prakovce-Zimná Voda, Gemeric unit, Western Carpathians, Slovakia. (a) Globular aggregate of uraninite I (Urn I) and its secondary generation consisting of altered rim and inner altered domains (Urn II), associated with secondary uranyl arsenates-phosphates (U–As–P) and quartz (Qz). (b) The ore minerals show signs of decomposition. Arsenopyrite (Apy) and uraninite (Urn) are being replaced by weathering rims of uranyl arsenates-phosphates (U–As–P) and scorodite (Scd). Another mineral is xenotime-(Y) (Xtm). (c) Pharmacosiderite (Pmsd) replacing arsenopyrite (Apy) in quartz (Qz) and muscovite (Ms). (d) Scorodite (Scd) and uranyl arsenates-phosphate (U–As–P) form complete replacement after arsenopyrite(?) in quartz (Qz). (e) Large uranyl arsenates-phosphates cluster (U–As–P) in a heterogeneous aggregate of hydrous ferric oxides/ arsenates (Feox/FeAs) and pharmacosiderite (Pmsd) in quartz (Qz). (f) Colloform aggregate of hydrous ferric oxides (Feox) precipitated in the cavity in quartz (Qz). (g) Large aggregate of alunite-supergroup minerals, uraninite (Urn) and uranyl arsenates-phosphates (U–As–P) scattered in fine-grained hydrous ferric oxides (Feox) and hydrous ferric arsenates (FeAs), muscovite (Ms) and quartz (Qz). (h) Alunite-supergroup mineral crystals with colloform banding with concentric zonation rich in philipsbornite (Pbn) in their central parts and segnitite (Sgt) in the outer parts. Other minerals are uraninite (Urn) and scorodite (Scd). (i) Replacement texture of large aggregate of segnitite (Sgt) and tiny relics of philipsbornite (Pbn) rimmed by uranyl arsenates/ phosphates (U–As–P) in iron oxyhydroxide (Gth), scorodite (Scd) and quartz (Qz) groundmass. Another mineral is uraninite (Urn). (j) Replacement texture; the dark relics are philipsbornite (Pbn) and the light colloform and fibrous crystals are segnitite (Sgt) associated with uranyl arsenates/ phosphates (U–As–P) in iron oxyhydroxide (Gth), scorodite (Scd) groundmass. IMA–CNMNC approved mineral symbols according to Warr (2021).

Figure 4. Microphotographs of uranyl arsenates–phosphates. (a) Yellow acicular crystals of arsenuranospathite–chistyakovaite. (b) Well-formed tabular crystal of (meta)nováčekite in a fine-grained rusty mass. Photo Š. Ferenc.

The composition and site occupancies of the principal cations in autunite-type minerals are as follows: T site = 2.0–24.4 wt.% As₂O₅, 0.16–1.85 apfu As; 0.4–14.0 wt. % P₂O₅, 0.07–1.78 apfu P, which shows the complete substitution of PAs_–1 (Fig. 5a); and Fe, Mg, or Al as additional cations at the A site (≤5.6 wt.% FeO, ≤0.77 apfu Fe; ≤3.9 wt.% MgO, 0.83 apfu Mg; and ≤3.9 wt.% Al₂O₃, 0.6 apfu Al; Fig. 5b). Some measurements also detected Ca as an additional cation (3.5 wt.% CaO, 0.56 apfu Ca; Fig. 5c). The uranyl minerals studied lack monovalent cations, which are common in autunite-group minerals (Fig. 5d). The concentrations of Na₂O + K₂O are below 0.4 wt.%, that of BaO + PbO below 1.5 wt.%, and ZnO + CoO is always less than 0.8 wt.% (Table 1, Supplementary table S2).

Figure 5. Compositions of uranyl arsenates–phosphates expressed in ternary diagrams (in apfu). Abbreviations are as follows: nováčekite – Nvč, kahlerite – Kah, threadgoldite-autunite – Tdg-Aut, arsenuranospathite-chistyakovaite – Aush-Cak (Warr, 2021).

Table 1. Representative compositions (wt.%) from EPMA of uranyl arsenates and phosphates.*

* Abbreviations: kahlerite – Kah; nováčekite – Nvč; arsenuranospathite–chistyakovaite – Aush–Cak; threadgoldite – Tdg; autunite – Aut; segnitite (Sgt); philipsbornite (Pbn). Ferric arsenates: scorodite – Scd; pharmacosiderite – Pmsd; bariopharmacosiderite – Bpsd; arseniosiderite – Assd. Unspecified arsenates – FeAs; Ferric oxide – Feox.

n.a. – not analysed; ‘–’ below detection.

Ferric arsenates and oxides

The most abundant ferric arsenates and oxides are scorodite, goethite and pharmacosiderite. The presence of X-ray amorphous iron oxides is very probable; however this was not verified conclusively. Rarer phases include bariopharmacosiderite and arseniosiderite. All minerals were analysed by EPM and the identification of scorodite and goethite was also confirmed by micro-Raman spectroscopy (see Raman spectroscopy section). The ferric arsenates and oxides usually form a fine-grained groundmass and earthy aggregates surrounding the primary ore minerals in quartz–muscovite gangue (Fig. 3a). They also form coatings, pseudomorphs, fissures and cavity fillings, weathering selvages and irregular domains of variable composition around the hypogene ore minerals (Fig. 3b–f), and they are generally rich in Fe, As, Sb, Bi, (Ca, Ba and K) (Fig. 6, Supplementary table S3). Arsenopyrite is partially-or-completely replaced by scorodite, pharmacosiderite + autunite-type arsenates (Fig. 3b–d). The average As content in the ferric oxides is 3.3 wt.%, and the highest content encountered is 10.4 wt.% (Table 1). The Sb content is lower, up to 1.4 wt.% with an average of 0.5 wt.%. Among arsenates, the compositional heterogeneity is controlled by variations in Ca (≤13.7 wt.% CaO, 1.9 apfu Ca in arseniosiderite), Ba, K, Al and P (≤9.9 wt.% BaO, 0.6 apfu Ba; ≤8.2 wt.% K₂O, 1.4 apfu K; ≤7.4 wt.% Al₂O₃, 1.1 apfu Al; and ≤3.4 wt.% P₂O₅, 0.4 apfu P in pharmacosiderite–bariopharmacosiderite), Al and P (≤5.3 wt.% and ≤3.3 wt.% in unspecified arsenates). Homogeneous scorodite shows slightly increased concentrations of P (≤4.6 wt.% P₂O₅, 0.1 apfu P), and Al (≤5.4 wt.% Al₂O₃, 0.2 apfu Al) and thus represents a limited solid solution towards the variscite end-member (Fig. 7, Table 1). The higher totals including calculated H₂O content (>105 wt.%) might be related to the beam-induced damage during accumulation of the EPMA data rather than the possible occurrence of kaňkite FeAsO₄⋅3.5H₂O (cf. Majzlan et al., 2012).

Figure 6. Compositions of the supergene mineral assemblage in: (a) Fe–As–Sb ternary diagram; and (b) Fe–Bi–Sb ternary diagram (in atomic wt.%). Abbreviations are: arseniosiderite – Assd, pharmacosiderite – Pmsd, scorodite – Scd, segnitite-philipsbornite – Sgt-Pbn (Warr, 2021).

Figure 7. Composition of supergene minerals in an Al/(Al + Fe) vs P/(P + As) compositional diagram. The selection of alunite-supergroup minerals end-members containing P + As in the T site is restricted to Pb-dominant end-members of the dussertite and plumbogummite groups, the variscite group excluding yanomamite, Fe–Al members of pharmacosiderite group and mitridatite group containing Ca–Fe-dominant end-members (atomic proportions).

Composition of alunite-supergroup minerals and speciation of Sb by X-ray absorption spectroscopy

The fine-grained rusty masses consist not only of goethite, scorodite and pharmacosiderite, but also alunite-supergroup minerals. Compositionally, they can be assigned to a mixture of two Pb end-members of the dussertite group: segnitite [PbFe₃(AsO_3.5(OH)_0.5)₂(OH)₆] and philipsbornite [PbAl₃(AsO_3.5(OH)_0.5)₂(OH)₆]. They occur as a part of a secondary assemblage in quartz–muscovite gangue or in close vicinity to the secondary uranyl arsenates–phosphates (Fig. 3g). Two distinct intergrowth textures were observed. Most of the alunite-supergroup mineral aggregates show colloform banding with concentric compositional zonation preferentially rich in philipsbornite in their central parts and segnitite in the outer parts (Fig. 3h). In some instances, the replacement texture of subhedral–anhedral microcrystalline aggregates is recognised (Fig. 3i–j).

The alunite-supergroup minerals are compositionally heterogeneous (Table 1, Supplementary table S4), mainly due to the extensive Fe–Al variations and the presence of Sb- and Bi-rich domains. The Fe concentrations on the G site range from 9.0 to 27.8 wt.% Fe₂O₃ (0.8–2.3 apfu Fe) and Al ranges from 3.1 to 15.3 wt.% Al₂O₃ (0.5–2.0 apfu Al). There is a strictly core–rim controlled distribution of Al³⁺ by Fe³⁺, where Al dominates in the core, while Fe dominates in the rim (Fig. 3f and Fig. 8). However, the fine grained character of the segregations of Al-dominant (philipsbornite) and Fe-dominant (segnitite) alunite-supergroup members occasionally prevented reliable analysis of individual minerals. In these cases, the analyses were performed with the beam focused to a small spot, however it is accepted that even this spot was too large to sample a single homogeneous domain. Arsenic is the dominant cation at the T site, ranging from 16.5 to 32.0 wt.% As₂O₅ (1.0–1.8 apfu As), though P is elevated in some measurements on philipsbornite (≤10 wt.% P₂O₅; ≤1.0 apfu P) and slightly increased S (≤3.8 wt.% SO₃; ≤0.3 apfu S) suggests only a limited beudantite-group constituent [PbG ₃(As_0.5S_0.5O₄)₂(OH)₆]. Though philipsbornite is also rich in the phosphate-dominant plumbogummite-group constituent [PbG ₃(PO_3.5(OH)_0.5)₂(OH)₆], the composition of the segnitite rims is close to the ideal arsenate end-member. The alunite-supergroup minerals are anomalously rich in Bi (<10.9 wt.% Bi₂O₃, 0.2 apfu Bi in philipsbornite; ≤9.0 wt.% Bi₂O₃, 0.2 apfu Bi in segnitite) and Sb (≤5.7 wt.% Sb₂O₅, 0.3 apfu Sb in philipsbornite; ≤8.4 wt.% Sb₂O₅, 0.4 apfu Sb in segnitite), in which philipsbornite has an average Bi≈Sb apfu signature whereas segnitite has Sb>Bi apfu. Apart from Bi, the D site is occupied almost exclusively by Pb (16.8–29.5 wt.% PbO, 0.5–1.0 apfu Pb) accompanied (Fig. 9) with other minor divalent cations <1.2 wt.% CaO + SrO + BaO (0.1 apfu Ca + Sr + Ba). The concentrations of other trace elements such as Cu, Zn, Th, U, Si and Ce (REE) are negligible and usually below the detection limit (Table 1, Supplementary table S4). The data from EPMA yielded the average empirical formulae: philipsbornite: (Pb_0.73Bi_0.2Sr_0.03U⁴⁺_0.02REE³⁺_0.01)_Σ0.99(Al_1.6Fe³⁺_1.24Sb⁵⁺_0.17)_Σ3.01(As_1.15P_0.78S_0.05Si_0.01O_3.5(OH)_0.5)_Σ1.99(OH)₆, and segnitite: (Pb_0.83Bi_0.11Sr_0.05U⁴⁺_0.01REE³⁺_0.01)_Σ1.01(Fe³⁺_1.99Al_0.72Sb⁵⁺_0.3)_Σ3.01(As_1.57P_0.4S_0.02O_3.5(OH)_0.5)_Σ1.99(OH)₆.

Figure 8. False colour X-ray map showing the distribution of selected elements in the aggregate of a philipsbornite–segnitite series mineral.

Figure 9. Composition of alunite-supergroup minerals showing D-site occupancy in a Pb–(Sr+Ba+Ca)–Bi ternary diagram. (in atomic wt.%).

The oxidation state of antimony in the alunite-supergroup minerals was investigated by μ-XAS. There is a shift of the absorption edge and the crest of the ‘white lines' as a function of the oxidation state of Sb at both SbL ₁ and SbL ₃ edges. The absorption edge of the data set for stibnite is even lower than that for schafarzikite. As the samples contain only antimony bound to oxygen ligands, only oxidic standards were considered further. The XAS ‘white line' of the dussertite-type minerals had the same position at different sample spots and overlapped with the ‘white lines' of the Sb(V) standards. The difference in the position of the absorption edge (judged from the maxima in the second derivative of the measured curves) is < 0.1 eV (SbL ₁ line: Fig. 10). Such a difference lies within the reproducibility of the spectral position for a single sample at the beamline. These data show clearly that all Sb in the samples investigated is in the pentavalent state.

Figure 10. Comparison of the data for sample ZV-6 with the reference Sb(III) and Sb(V) compounds in the XANES region of the SbL ₁ edge.

A rudimentary reduction of the EXAFS (extended X-ray absorption fine structure) part of the measured spectra (Supplementary figure S1) leads to similar conclusions. The coordination number of Sb in the samples is close to the coordination number in the tripuhyite standard (6) and much larger than the coordination number of Sb in the schafarzikite standard (3) (Fig. 10).

Raman spectroscopy

The Raman spectra of goethite and As-rich goethite are presented in Fig. 11. The segnitite and philipsbornite spectra (Fig. 12a) show a series of overlapping broad bands in the range of 100−1200 cm^–1. Their shapes and positions discriminate these minerals from scorodite. In the segnitite, 11 low frequency bands were resolved. The spectra show a weak broad band in the 545−645 cm^–1 and 690−1000 cm^–1 ranges (Fig. 12b). The philipsbornite spectrum in low wavenumbers is characterised by two distinct bands at 140 and 207 cm^–1 with shoulders and by a broad band in the range of 250−350 cm^–1. A series of bands in the region between 390 and 520 cm^–1 and a broad low intensity structure at 550−620 cm^–1 were also discerned (Fig. 12a).

Figure 11. Untreated Raman spectra of goethite excited by a 785 nm laser.

Figure 12. Examples of Raman spectra of (a) segnitite and philipsbornite compared to scorodite in the region of 100–1100 cm^–1 excited by 532 nm laser. (b) The results of fitting and decomposition in the stretching region.

Discussion

Compositional variations of philipsbornite–segnitite series

The philipsbornite–segnitite series in the investigated mineralisation represents a heterogeneous and multicomponent solid-solution system. These minerals have wide compositional variations of substantial elements at different crystallographic sites, particularly As–P±S (T site), Fe³⁺–Al³⁺–Sb⁵⁺ (G site) and Pb–Bi±(Sr+Ba+Ca) (D site). The chemical exchange of this complex solid-solution system is controlled by homovalent and heterovalent substitution mechanisms, documented previously by Van Wambeke (1975), Clark et al. (1986), Kolitsch et al. (1999), Mills et al. (2014), Golebiowska et al. (2016) and Pekov et al. 2016). The concentric compositional zonation of philipsbornite cores enclosed in segnitite envelopes resulted from early crystallisation of Al-dominant and phosphate-rich alunite-supergroup minerals and late crystallisation of Fe-dominant and arsenate-rich alunite-supergroup minerals, rather than from solid-state diffusion segregation (cf. Desborough et al., 2010). It is obvious that the zoning in the intermediate members is due to the strong coupling of (AsO₄)^3– with Fe³⁺ and (PO₄)^3– with Al³⁺ (cf. Rattray et al., 1996). The As–Fe and P–Al affinity and the compositional trend from As + Fe towards P + Al is also clearly recognised in other Fe arsenates (scorodite, pharmacosiderite and only slightly in arseniosiderite) (Fig. 7). This geochemical and spatial strong association and synergy is well documented especially for As and Fe in weathering products (e.g. Walker et al., 2005; Lalinská-Voleková et al., 2012). The compositional gap between Al and Fe³⁺ in the philipsbornite–segnitite series (Sejkora et al., 2011; Golebiowska et al., 2016) is not observed in our samples. We have to admit, though, that the cause therefore may lie in the fact that the beam size for the EPMA exceeded the small size of the domains in these minerals.

The minor constituents of particular interest in the alunite-supergroup minerals are Sb and Bi. In supergene environments, Sb is expected to be found in its highest, pentavalent oxidation state. We assume that the alunite-supergroup minerals studied formed in such an environment by weathering of the primary base-metal mineralisation at Prakovce-Zimná Voda. This assumption agrees well with the results of the XAS which shows only Sb⁵⁺ in our samples (Fig. 10). Similarly, μ-EXAFS and XANES data of Sb-rich segnitite from Black Pine mine in Montana, USA suggested that Sb is present as Sb⁵⁺ (Mills et al., 2014). However, the associated EPMA in Mills et al. (2014) did not include data for P and Si but show a systematic deficiency at the T site (1.75–1.87 apfu As). Mills et al. (2014) rationalised this deficiency by an alternative fit to the XAS data with 85% Sb⁵⁺ and 15% Sb³⁺. Using their EXAFS data, they determined the octahedral coordination of Sb⁵⁺ and tetrahedral coordination of Sb³⁺. Doing so, they assigned Sb³⁺ to the T site, even though tetrahedral coordination of Sb³⁺ is highly unlikely because of its need to position the lone electron pair in its coordination environment. A small part of Sb³⁺ at the T site is also conceded in aliovalent Sb-rich segnitite from Berezovskoe gold deposit, Middle Urals, Russia (Pekov et al., 2016). The EPMA data from our samples do not show a significantly deficient T site (average occupancy of 2.0 apfu; Table 1) and all Sb here is considered to be pentavalent in octahedral coordination on the G site (0.1–0.4 apfu Sb⁵⁺), as previously reported in ‘antimonian’ dussertite from the Clara mine, Germany (cf. Kolitsch et al., 1999). The substantial content of pentavalent cations on the G site and the exchange of (Al,Fe)³⁺ for Sb⁵⁺ requires compensation by the deprotonation of oxygen atoms in the (AsO₃OH) tetrahedra via the 0.5Sb⁵⁺ + (AsO₄)^3– ↔ 0.5Fe³⁺ + (AsO₃OH)^2– substitution mechanism (Kolitsch et al., 1999; Pekov et al., 2016).

Furthermore, Bi in alunite-supergroup minerals occurs rarely and there are only two known Bi-dominant end-members: waylandite BiAl₃(PO₄)₂(OH)₆ (von Knorring and Mrose, 1963; Clark et al., 1986; Mills et al., 2010; Uehara and Shirose, 2013); and zaïrite, BiFe₃(PO₄)₂(OH)₆ (Van Wambeke, 1975). The arsenate-analogue of waylandite is described incompletely by Scharm et al. (1994) as ‘arsenowaylandite’ and is questionable because of inadequate data. Of these Bi-dominant alunite-supergroup mineral phases, only waylandite occurs at more than one locality. The significant Bi content in dussertite, beudantite and plumbogummite groups in the alunite supergroup has been described rarely, e.g. from a supergene zone in the Ľubietová, Podlipa deposit, Slovakia, ≤7.85 wt.% Bi₂O₃, 0.3 apfu Bi (Števko et al., 2016) and from the oxidation zone of the Berezovskoe gold deposit, Middle Urals, Russia ≤0.5 wt.% Bi₂O₃, 0.02 apfu Bi (Pekov et al., 2016). Regarding their Bi content, our samples (<10.9 wt.% Bi₂O₃; 0.3 apfu Bi) represent intermediate compositions between philipsbornite–plumbogummite–segnitite and their hypothetical ‘arsenowaylandite’–waylandite–‘arsenozaïrite’ end-members.

Raman spectroscopic identification of minerals and implications

The Raman spectra of goethite agree well with those published previously (e.g. De Faria et al., 1997; Das and Hendry, 2011). The spectra of As-rich goethite are characteristic by the presence of a band at 804–807 cm^–1, which can be assigned to the symmetric stretching vibration of the arsenate anion (Müller et al., 2010; Kloprogge and Wood, 2017). The band is strong for samples of goethite with high As content (5.4–12.9 wt.%) and is not visible when the As content is lower than 3.5 wt.% (see Fig. 11).

The Raman spectra of segnitite and philipsbornite show broad peaks in the range of 850−900 cm^–1 however these are missing in our (and published) scorodite data (Myneni et al., 1998; Filippi et al., 2007; Culka et al., 2016). Our spectra of segnitite and philipsbornite are almost identical and only deconvolution revealed minor differences, as could be expected from their structural and chemical similarity within the dussertite group. The philipsbornite spectra also agree well with the RRUFF database (R060683, Lafuente et al., 2015). That, however, is contradictory to a difference found in the spectrum of philipsbornite in Frost et al. (2013). As these authors provide the only interpretation published to date, we adopted their peak assignment of segnitite and applied it to both segnitite and philipsbornite, relying on the accuracy of our and RRUFF records (see Fig. 12). The segnitite band centred at 863 cm^–1 is due to ν ₁ symmetric stretching vibrations of (AsO₄)³⁻ and 807 cm^–1 is attributed to (AsO₄)³⁻ ν ₃ antisymmetric stretching (Frost et al., 2005). In dussertite, peaks near 700 and 750 cm^–1 are related to the As–OH stretching mode (Frost et al., 2011), possibly analogous to bands at 699 and 756 cm^–1 in our segnitite spectra. The 320−500 cm^–1 wavenumber range contains bands assigned to ν ₂ and ν ₄ (AsO₄)³⁻ ion-bending modes. The philipsbornite broad band at 856 cm^–1 can be assigned to ν ₁ symmetric stretching of (AsO₄)³⁻, the bands at 696 and 753 cm^–1 to As–OH stretching mode. The series of bands in the range of 370−550 cm^–1 belong to bending modes ν ₂ and ν ₄ within the (AsO₄)³⁻ ion. According to Frost et al. (2013) symmetric stretching vibrations might overlap with antisymmetric bands. The multiple overlapping bands indicate loss of degeneration due to strong distortion of the (AsO₄)³⁻ tetrahedra which might be enhanced by substitutions in both cation and anion positions. The peaks of lower intensity centred at 945 and 941 cm^–1 are assigned to symmetric stretching modes of (PO₄)^2– and the peak at 991 cm^–1 in segnitite to analogical modes of (SO₄)^2– units. This suggests a mixed composition of measured phases within the philipsbornite–segnitite series. Our calculation of vibrational Raman activity on the base of group theory shows that the cation bonding in octahedral position (9d Wyckoff position) should not be observed in Raman spectra.

Evolution of the oxidation zone and formation of supergene minerals

Mineralogical and textural observations suggest that the oxidation zone evolved through several stages (Fig. 13):

Figure 13. Summary scheme of the oxidation zone main stages.

(I) Oxidation–hydration weathering of uraninite. The primary uraninite I was converted to uraninite II and ‘gummite’ under near-neutral conditions. Texturally, this stage is marked by an increase of porosity and fracturing. The products of this stage have been described in more detail by Ondrejka et al. (2023b).

(II) Initial acidic stage. Massive oxidation of sulfides modified and buffered the composition of the weathering fluids. Here, the sulfide oxidation led to an increased activity of H⁺, formation of acid-rock drainage, and attack of the pre-existing uraninite, brannerite, fluorapatite and muscovite. The acidic solutions were supplied with arsenate, phosphate, sulfate, uranyl, Fe(III), Al(III), Pb(II), Sb(V), As(V), Bi(III), and other elements released into the solution.

This stage is characterised by the formation of diverse autunite-type minerals (kahlerite, nováčekite, arsenuranospathite, chistyakovaite, threadgoldite and autunite) accompanied by abundant scorodite, hydrous ferric arsenates and locally abundant philipsbornite–segnitite-series minerals that represent the sole sink of the remobilised Sb and Bi.

(III) Advanced acidic-neutral stage. Phosphuranylite forms as a direct replacement of uraninite in the vein portions without sulfide remnants. Both arsenopyrite and scorodite are replaced by pharmacosiderite and later bariopharmacosiderite and arseniosiderite. These minerals precipitate under near-neutral conditions, after the acidity-generating capacity of the sulfides has been exhausted.

(IV) Mature stage. The latest goethite and other Fe oxides form after most of the As was already fixed in Fe arsenates. The Fe oxides form the youngest veinlets that crosscut the older supergene minerals or replace pharmacosiderite.

Fate and uptake of Sb and Bi in the oxidation zone

The main carriers of Sb and Bi in the oxidation zone are surprisingly the arsenates of the philipsbornite–segnitite series. Other supergene minerals, particularly Fe oxides that are usually eager to adsorb Sb, have negligible concentrations of these elements (Fig. 6b). Antimony and Bi are rather exceptional components of the philipsbornite–segnitite alunite-supergroup mineral phases and suggest unusual conditions during formation from the Sb–Bi precursors (in our case, stibnite, gudmundite, berthierite, tetrahedrite-(Fe), gersdorffite and tintinaite–kobellite). The acidic conditions and sizeable As supply during the initial acidic stage created conditions conducive for the precipitation of the alunite-supergroup mineral phases. Philipsbornite is a stable phase under acidic conditions (pH < 4; Leverett et al., 2003) and segnitite could be synthesised by hydrothermal methods in the pH range 1.0–1.5 (Mills, 2007). Furthermore, Sb-rich segnitite from Black Pine Mine, Montana, USA appears to have formed under oxidising conditions and pH < 2 (Mills et al., 2014). The alunite-supergroup mineral phases removed Sb and Bi from the acidic solutions and locked them up in their crystal structures. Doing so, they depleted the solutions in Sb and Bi to the extent that these elements were not available for the secondary minerals that formed later. Apart from philipsbornite–segnitite, scorodite is also abundant. This mineral, however, does not tolerate Sb in its structure because of the different size and coordination of As⁵⁺ and Sb⁵⁺ (Kossoff et al., 2015).

Variations in the uranyl minerals in response to environmental conditions

The formation of the autunite-group minerals conforms to the strongly acidic environment of the initial acidic stage. Uranyl arsenates prevail over phosphates, and both probably precipitate from an aqueous solution rich in uranyl complexes at a high activity of As(V) and comparatively lower activity of P(V). In contrast, the less abundant phosphuranylite was formed under near-neutral conditions once the sulfides vanished and ceased supplying acidity into the weathering fluids (cf. Krivovichev and Plášil, 2013; Plášil, 2014). The change in the assemblages of the uranyl minerals indicates, together with pharmacosiderite–bariopharmacosiderite and arseniosiderite (cf. Drahota and Filippi, 2009 and references therein), the gradual change from strongly acidic to mildly acidic or near-neutral conditions.

Conclusions

In this work, we investigated weathering of hypogene minerals within the oxidation zone at the Prakovce-Zimná Voda REE–U–Au quartz vein, Gemeric Unit, Western Carpathians, Slovakia. These minerals weather and develop alteration rims or are totally replaced by a variety of supergene phases rich in Fe, As, U, Pb, Sb, Bi, S, P, Ca and Ba. This supergene mineral assemblage represents a typical stable association under the surface conditions of the oxidised zone in the uranium deposits. Uranyl arsenates and minor phosphates of autunite-type group together with scorodite and philipsbornite–segnitite-series minerals formed by oxidising fluids during decomposition and leaching of primary hypogene uraninite, brannerite and base-metal sulfides, such as arsenopyrite, gersdorffite, pyrite, galena, minerals of the stibnite–bismuthinite series and tintinaite–kobellite. A progressive change of pH from acidic to near-neutral due to the gradual consumption of sulfides resulted in the formation of late phosphuranylite, pharmacosiderite and arseniosiderite. Goethite represents the latest hydrous ferric oxide.

Using the μ-XAS technique, we found that the oxidation state of Sb in philipsbornite–segnitite-series minerals is essentially pentavalent, and together with the presence of Fe oxides and arsenates (Fe³⁺) and uranyl minerals (U⁶⁺) suggests oxidative conditions during weathering. Our study also indicates that in the supergene environment in the quartz vein rich in Fe and As, accompanied by elevated concentrations of U, Pb, Sb, Bi, S, P, Ca and Ba at oxidising conditions, various hydrous ferric arsenates are dominant secondary minerals. Uranium was incorporated into uranyl arsenates–phosphates, whereas Sb, Bi and Pb were taken up into philipsbornite–segnitite. Antimony and Bi are rather exceptional within this mineral supergroup and suggest unusual conditions (i.e. very low pH) during their formation. Phosphate binds preferentially with Al whereas arsenate has higher affinity to Fe³⁺ in most of the minerals studied, however, the strongest P–Al and As–Fe synergy is documented in the alunite-supergroup minerals. The source of Fe, As, Pb, Bi, Sb, and S was probably the base-metal sulfides (particularly arsenopyrite and pyrite) and sulfosalts (minerals of the bismuthinite–stibnite and kobellite–tintinaite series), whereas U was released from uraninite and brannerite, P from fluorapatite and Al, K and Ba from muscovite or from feldspars in the host rocks.