Xenotime-(Gd), a new Gd-dominant mineral of the xenotime group from the Zimná Voda REE–U–Au quartz vein, Prakovce, Western Carpathians, Slovakia

Abstract

Xenotime-(Gd), ideally GdPO₄, is a new mineral of the xenotime group. It was discovered at the Zimná Voda REE–U–Au occurrence near Prakovce, Western Carpathians, Slovakia. It forms rare crystal domains (≤20 μm, usually ≤10 μm in size) in Gd-rich xenotime-(Y) crystals (≤100 μm in size), in association with monazite-group minerals, uraninite, fluorapatite and uranyl arsenates–phosphates. The hydrothermal REE–U–Au mineralisation occurs in a quartz–muscovite vein, hosted in Palaeozoic phyllites near exocontact with Permian granites. The density is 5.26 g/cm³, based on calculated average empirical formula and unit-cell parameters. The average chemical composition (n = 6) measured by electron microprobe is as follows (wt.%): P₂O₅ 30.1, As₂O₅ 0.5, SiO₂ 0.2, UO₂ 0.3, Y₂O₃ 15.7, (La, Ce, Pr, Nd)₂O₃ 0.5, Sm₂O₃ 5.7, Eu₂O₃ 1.4, Gd₂O₃ 29.2, Tb₂O₃ 3.9, Dy₂O₃ 10.4, Ho₂O₃ 0.4, (Er, Tm, Yb, Lu)₂O₃ 2.1, (Ca, Fe, Pb, Mn, Ba)O 0.1, total 100.5. The corresponding empirical formula calculated on the basis of 4 oxygen atoms is: (Gd_0.37Y_0.32Dy_0.13Sm_0.08Tb_0.05Eu_0.02Er_0.01Tm_0.01Nd_0.01…)_Σ1.01(P_0.98As_0.01Si_0.01)O₄. The empirical formula of the Gd-richest composition is: (Gd_0.38Y_0.31Dy_0.13Sm_0.08Tb_0.05Eu_0.02Er_0.01Nd_0.01Ho_0.01…)_Σ1.01(P_0.98As_0.01Si_0.01)O₄. The ideal formula is GdPO₄. The xenotime-type structure has been confirmed by micro-Raman spectroscopy and a Fast Fourier-Transform pattern using HRTEM. Xenotime-(Gd) is tetragonal, space group I4₁/amd, a = 6.9589(5) Å, c = 6.0518(6) Å, V = 293.07(3) ų and Z = 4. The new mineral is named as an analogue of xenotime-(Y) and xenotime-(Yb) with Gd dominant among the REE. The middle REE enrichment of xenotime-(Gd) is shared with the associated monazite-(Gd) and Gd-rich hingganite-(Y). This exotic REE signature and precipitation of Gd-bearing minerals is a product of selective complexing and enrichment in MREE in low-temperature hydrothermal fluids by alteration of uraninite, brannerite and fluorapatite on a micro-scale. The existence of xenotime-(Gd) and monazite-(Gd) is the first naturally documented dimorphism among REE phosphates. In addition, xenotime-(Gd) is only the third approved Gd-dominant mineral, after lepersonnite-(Gd) and monazite-(Gd).

Introduction

Rare earth elements (REE) represent a unique set of elements having coherent behaviour and very similar properties which are responsible for their similar geochemistry. The International Union of Pure and Applied Chemistry (IUPAC) definition of the REE group includes lanthanides (Ln), yttrium (Y) and scandium (Sc). However, due to the substantially smaller ionic radius of Sc³⁺ with respect to the rest of the group, Sc frequently enters different structural sites, and therefore Sc is commonly not included into the REE in natural geological environments. Due to the lanthanide contraction phenomenon, the REE are further divided into larger LREE (light Ln, La–Sm), medium MREE (Sm–Dy) and smaller HREE (heavy Ln, Tb–Lu and Y).

Minerals belonging to the xenotime group are anhydrous orthophosphates, orthoarsenates and orthovanadates with tetragonal symmetry and space group I4₁/amd (#141). Among phosphates, xenotime-(Y) is the most common species and contains predominantly Y³⁺ and usually low contents of other HREE as well as negligible amounts of LREE based on their preference of the xenotime-type crystal structure (cf. Ni et al., 1995). On the other hand, xenotime-(Yb) and Yb-, Dy- and Gd-rich xenotime-(Y) are rare minerals occurring in some granitic pegmatites and metamorphic-hydrothermal lithologies (Demartin et al., 1991; Förster and Rhede, 1995; Franz et al., 1996; Förster, 1998; Buck et al., 1999; Masau et al., 2000; Repina, 2011; Repina et al., 2014; Franz et al., 2015; Ondrejka et al., 2023b, 2023c).

In contrast to numerous minerals with REE as their essential constituents (particularly Ce and Y, more rarely La, Nd, Sm and Yb), there are currently only two Gd-dominant minerals approved by the International Mineralogical Association, Commission on New Minerals, Nomenclature and Classification (IMA-CNMNC, Pasero, 2024): lepersonnite-(Gd) CaGd₂(UO₂)₂₄(SiO₄)₄(CO₃)₈(OH)₂₄⋅48H₂O, a rare REE-uranyl carbonate from the Shinkolobwe U deposit in the DR Congo (Deliens and Piret, 1982) and monazite-(Gd) (Ondrejka et al., 2023b) which is cogenetic with the xenotime-(Gd) described here.

In this work, we describe a new Gd-dominant mineral xenotime-(Gd), discovered in a hydrothermal quartz vein with REE–U–Au mineralisation at the Zimná Voda occurrence near Prakovce, Slovakia. Xenotime-(Gd) is a new Gd-dominant member of the xenotime group and it is related to xenotime-(Y) and xenotime-(Yb) by substitution of Gd for other REEs, having the REE composition distinctly shifted towards the MREE enrichment. The new mineral and its name have been approved by the IMA-CNMNC (IMA 2023-091; Ondrejka et al., 2024) and the Levinson modifier for rare earth minerals (Levinson, 1966; Bayliss and Levinson, 1988). The symbol Xtm-Gd was given to the new mineral. The holotype specimen of xenotime-(Gd) (FIB lamella of the thin section ZV-2A4) is deposited in the collection of the Slovak National Museum, Natural History Museum, Vajanského nábrežie 2, P.O. BOX 13, 810 06 Bratislava, Slovak Republic under the catalogue number M-20412. The crystallographic information file (cif) is deposited in American Mineralogist Crystal Structure Database (AMCSD; Downs and Hall-Wallace, 2003) under the code 0021444 and is also available as Supplementary material (see below).

Occurrence

The Zimná Voda REE–U–Au occurrence was discovered in 1975, during exploration for uranium ores (Novotný and Čížek, 1979). The sample containing xenotime-(Gd) was collected by the authors in September 2017 during reconnaissance of the Zimná Voda REE–U–Au quartz vein, Prakovce, Gelnica Co., eastern Slovakia. The site is located near the main ridge of the Slovenské Rudohorie Mts., ~5.6 km to the S of the Prakovce village, 600 m to the NW of the Tri Studne elevation point (969 m a.s.l.) and 400 m NW of Trohánka bivouac shelter, at an altitude of ~950 m a.s.l., ca. 23 km WNW of Košice town at 48.7666°N, 20.9137°E.

The hydrothermal REE–U–Au vein mineralisation at the Zimná Voda occurrence is hosted in the Lower Palaeozoic metamorphic rocks of the Bystrý Potok Formation, a part of the Gelnica Group in the Gemeric tectonic unit of the Western Carpathians, which is part of the Alpine-Carpathian Mountain belt (Bajaník et al., 1983; Ivanička et al., 1989). Two quartz veins (Western and Eastern) containing REE, U and Au mineralisation were found in the area. Xenotime-(Gd) was found in the Western vein. The vein is hosted in fine-grained micaceous phyllites interbedded with fine-grained quartzites. It has an E–W strike, total length of ~90 m with an average dip of 65° to the S, and conforms to the schistosity of the host rocks. The thickness of the vein ranges from 3 to 30 cm. Along the contact, the rocks are intensively argillitised, and locally silicified, and impregnated by pyrite. Supergene alteration of pyritised rocks caused their limonitisation.

The metamorphic rocks were intruded by Hummel granites which outcrop 600 m to the SW of the investigated occurrence. These igneous rocks are leucocratic peraluminous rare-metal granites with S-type affinity and relatively high degree of magmatic fractionation, evidenced by higher P concentrations in K-feldspar and elevated contents of rare lithophile elements (Li, Rb, Cs, B, Sn, W, Nb and Ta) and F. They originated and were emplaced during the post-Variscan, Permian (260–270 Ma) extension stage (e.g. Villaseñor et al., 2021, and references therein).

In addition to xenotime-(Gd), the following minerals were identified in the Western vein: arsenopyrite, native bismuth, bismuthinite, brannerite, chlorite, cobaltite, fluorapatite, galena, gersdorffite, glaucodot, native gold, hingganite-(Y), kobellite, pyrite, quartz, molybdenite, monazite-(Ce), monazite-(Gd), monazite-(Nd), monazite-(Sm), muscovite, rutile, stibnite, tetrahedrite-(Fe), tintinaite, tourmaline-group minerals, uraninite and xenotime-(Y); supergene minerals are represented by arseniosiderite, goethite and other undifferentiated iron oxyhydroxides, pharmacosiderite–bariopharmacosiderite, philipsbornite–segnitite, scorodite, zeunerite, other uranyl arsenates–phosphates (most abundant are kahlerite, nováčekite, threadgoldite, rarely arsenuranospathite, autunite, chistyakovaite related mineral phases and phosphuranylite); Ondrejka et al. (2023a, 2023b, 2023c). The mineralisation probably originated by fluid-driven hydrothermal mobilization of REE, U and Au from the surrounding metamorphic rocks induced by the intrusion of the granitic rocks (Rojkovič et al., 1997; 1999). The perigranitic origin is supported by the age correlation of uraninite (246–265 Ma), the geochemical signatures and the spatial proximity of the mineralised veins with Permian rare-metal granites of the Gemeric Unit (cf. Kohút and Stein, 2005; Ferenc et al., 2021; Ondrejka et al., 2023c). Further details and regional geology can be found in Ondrejka et al. (2023c).

Experimental methods

The chemical composition of xenotime-(Gd) was studied by a JEOL JXA-8530F electron probe microanalyser (EPMA) in the wavelength-dispersive spectrometry (WDS) mode at the Earth Science Institute, Slovak Academy of Sciences in Banská Bystrica, Slovakia. An accelerating voltage of 15 kV and a probe current of 20 nA were used. The typical spot beam diameter varied from 2 to 6 μm; a more focused ≤1–3 μm beam was used only occasionally to avoid any intermediate composition in strongly heterogeneous micro-scale areas. The determination was calibrated using natural and synthetic calibrants (Table 1), and raw counts were converted to wt.% of oxides using the ZAF matrix correction. Corrections of line interferences were provided using the method by Åmli and Griffin (1975). The detection limit for all elements is typically between 0.01 and 0.15 wt.%. Element contents in the mineral formula are expressed in atoms per formula unit (apfu). The xenotime-(Gd) formula was normalised to 4 oxygen atoms.

Table 1. Conditions used for the electron microprobe analyses.

Depolarised Raman measurements were conducted on an unoriented section of xenotime-(Gd) using a Labram HR800 spectrometer (Horiba Jobin-Yvon), which was linked to an Olympus-BX41 optical microscope (Earth Science Institute, Slovak Academy of Sciences in Banská Bystrica, Slovakia). The samples were exposed to 532 nm frequency-doubled Nd-YAG and 633 nm He–Ne lasers. The system resolution was ~2 cm⁻¹.

Slicing and polishing of a lamella for electron diffraction analysis were carried out using a scanning electron microscope (SEM) coupled with a gallium-focused ion beam (FIB) source (Bauhaus University Weimar, Germany). The SEM-FIB (Helios G4 UX, ThermoFisherScientific) is equipped with a high-performance FIB source (Phoenix) that allows the polishing of TEM lamella at very low acceleration voltage or beam current. This feature is essential for obtaining undisturbed thin lamella suitable for high-resolution TEM imaging. Thin sections of samples, as used for optical light microscopic investigation and other analyses, were sputtered with an ≈ 8 nm gold layer to ensure the electric conductivity of the full sample and to reduce sample abrasion during ion beam imaging. Sites for extraction of the lamellae were selected according to previous microscopic and spectroscopic characterisation of the samples. Areas of interest were covered with an approximately 15×3×3 μm layer of platinum to further protect the sample surface against ion beam damage.

Transmission electron microscopy (TEM), high-resolution TEM (HRTEM), and scanning transmission electron microscopy (STEM) characterisations were performed using double corrected TEM/STEM JEOL JEM-ARM200CF (JEOL Ltd., Tokyo, Japan) with cold-FEG cathode operated at 200 kV located in the Centre for Nanodiagnostics of Materials, Faculty of Materials Science and Technology, Slovak University of Technology in Bratislava, Slovakia.

Images from HRTEM were recorded using a Gatan Orius SC1000 CCD Camera (Gatan Inc., Pleasanton, CA, USA) with a resolution of 2048 × 2048 pixels using the DigitalMicrograph software package (version 3.60.4435.0, Gatan Inc., Pleasanton, CA, USA). The measurement of the reflection positions and angles in Fast Fourier Transform (FFT) patterns acquired from relevant HRTEM images was done using the DIFPACK module incorporated in DigitalMicrograph software. By clicking on the spots on the FFT pattern DIFPACK locates the spots at the brightest pixel value precisely.

A high-angle annular dark field (HAADF) STEM detector with inner-collection semi-angle of 90 mrad, and a probe convergence semi-angle of 22 mrad were utilised to obtain atomic resolution HAADF STEM images. Crystal structure simulations were carried out using CrystalMaker® for Windows version 2.3.0 (Begbroke, UK).

Appearance and physical properties

Xenotime-(Gd) forms rare crystal domains or zones (≤20 μm in size, usually ≤10 μm), in Gd-rich xenotime-(Y) crystals (≤100 μm in size, xenotime II, Ondrejka et al., 2023c) in close association with Gd-poor xenotime-(Y) (xenotime I, ibidem), monazite-group minerals, particularly monazite-(Gd) and monazite-(Sm), uraninite, supergene uranyl arsenates and phosphates, dispersed in quartz–muscovite gangue (Fig. 1a). The holotype specimen (FIB lamella) is positioned in the Gd-richest domain of the xenotime crystal (Fig. 1b–d). Among two xenotime-group species recognised in the Western vein at Zimná Voda occurrence, xenotime-(Y) is the most common and occurs in all of 6 samples studied (ZV-1 to ZV-6). Xenotime-(Gd) was found only in the ZV-2 sample and unlike monazite-(Gd) with occurrence in multiple thin sections, xenotime-(Gd) was found only in the thin section ZV-2A4. The lustre, hardness, cleavage and parting of xenotime-(Gd) could not be determined, nor could density be measured due to an insufficient quantity of mineral for physical measurement. We assume, however, that these properties are similar to those observed in other xenotime-group minerals. The density of 5.26 g/cm³ was calculated on the basis of the average empirical formula and calculated unit-cell volume (see below). The density of the ideal xenotime-structured formula GdPO₄ is 5.66 g cm^–3 (Rodriguez-Liviano et al., 2013) to 5.77 g cm^–3 (Ni et al., 1995) calculated from the unit-cell volume of synthetic GdPO₄ due to the significant presence of lighter elements, mostly Y. In comparison, the density of natural xenotime-(Y) varies between 4.4 and 5.1 g cm^–3 (Anthony et al., 2024). Optical properties could not be determined because of the small sample size, but they are probably similar to other xenotime-group minerals. The value of n = 1.87 calculated from the Gladstone-Dale relationship (Mandarino, 1979, 1981) using an empirical formula and calculated density of the holotype specimen is slightly higher than the refractive index of natural xenotime-(Y), which is up to 1.83 (Anthony et al., 2024) due to enrichment in heavier REE.

Figure 1. Back-scattered electron images (BSEI) of (a) a large aggregate of monazite-group minerals (MGM) with xenotime-(Gd) (Xtm-Gd), xenotime-(Y) (Xtm-Y), uraninite (Urn) and uranyl arsenates-phosphates (U-As-P) in quartz-muscovite gangue (Qz + Ms). The red dashed area indicates the Gd-richest (Gd-apfu-dominant) composition Gd₂O₃>29 wt.%; Y₂O₃<16 wt.%; black dashed area indicates Gd-rich (Gd-apfu-dominant) Gd₂O₃=28–29 wt.%. The position of the holotype FIB lamella (#M-20412) is documented herein. (b) The spot positions of electron microprobe analyses in Table 2. (c) The FIB lamella (holotype specimen) positioned in the Gd-richest domain of the xenotime crystal. (d) The extraction of the FIB lamella from the sample. Mineral symbols from Warr (2021). Note: Images (b), (c) and (d) are rotated 90 degrees to the view in (a).

Chemical composition

Chemical analyses were carried-out on crystals in the ZV-2A4 thin section and are positioned within and in close vicinity of the FIB lamella holotype specimen (Fig. 1b). The representative chemical composition is shown in Table 2. The xenotime-(Gd) μm-scale domains are chemically relatively homogeneous, with no distinct variations in the Y/Gd mass ratio (0.47–0.50). However, those Gd-dominant domains are always a part of a larger and heterogeneous Gd-rich to Gd-medium-rich xenotime-(Y) with Y/Gd = 0.56–2.57. The analysed domains have a composition with Gd>Y>Dy > Sm > Tb (28.5–29.3 wt.% Gd₂O₃, 0.36–0.38 apfu Gd; 15.2–16.1 wt.% Y₂O₃, 0.31–0.33 apfu Y; 10.1–10.8 wt.% Dy₂O₃, 0.13 apfu Dy; 5.3–6.2 wt.% Sm₂O₃, 0.07–0.08 apfu Sm; 3.7–4.0 wt.% Tb₂O₃, 0.05 apfu Tb). The average chemical composition of xenotime-(Gd) calculated from six point electron-microprobe analyses is as follows (wt.%): %): P₂O₅ 30.1, As₂O₅ 0.5, SiO₂ 0.2, UO₂ 0.3, Y₂O₃ 15.7, (La, Ce, Pr, Nd)₂O₃ 0.5, Sm₂O₃ 5.7, Eu₂O₃ 1.4, Gd₂O₃ 29.2, Tb₂O₃ 3.9, Dy₂O₃ 10.4, Ho₂O₃ 0.4, (Er, Tm, Yb, Lu)₂O₃ 2.1, (Ca, Fe, Pb, Mn, Ba)O 0.1, total 100.5. The corresponding empirical formula calculated on the basis of 4 oxygen atoms is: (Gd_0.37Y_0.32Dy_0.13Sm_0.08Tb_0.05Eu_0.02Er_0.01Tm_0.01Nd_0.01…)_Σ1.01(P_0.98As_0.01Si_0.01)O₄. The empirical formula of xenotime-(Gd) in its Gd-richest spot is: (Gd_0.38Y_0.31Dy_0.13Sm_0.08Tb_0.05Eu_0.02Er_0.01Nd_0.01Ho_0.01…)_Σ1.01(P_0.98As_0.01Si_0.01)O₄ which leads to the end-member formula GdPO₄ requiring (in wt.%): Gd 62.35, P 12.28, and O 25.37, or Gd₂O₃ 71.86, P₂O₅ 28.14, total 100.00 wt.%.

Table 2. Representative and average EPMA data and mineral formula of xenotime-(Gd). Spot numbers correspond to positions on Fig. 1.

*Note: ΣA = sum of cations at the A site (Th+U+Al+REE+Fe+Pb+Mn+Ca+Ba+Na); ΣB = sum of cations at the B site (S+P+As+Si).

In general, the element distribution shows enrichment towards the MREE, depletion of HREE+Y and negligible LREE abundances. The chondrite-normalised patterns exhibit conspicuous maxima at Gd and Tb (MREE hump; for further details, see Ondrejka et al., 2023c). The concentration of Th was usually below the detection limit of the EPMA; U attains only 0.4 wt.% UO₂ (0.004 apfu U). Arsenic is always present (≤0.7 wt.% As₂O₅, 0.014 apfu As) and attests to limited PAs_–1 (chernovite) substitution. Satisfactory analytical totals (99 − 101 wt.%), excellent stoichiometry calculated on an anhydrous basis and Raman spectra (see below) indicate that a potential role of a tetrahedral array of (OH)^– groups is negligible, if any. Other trace elements (Ca, Sr, Fe, Pb, Si, F and S) have negligible concentrations or are below detection limits (Table 2).

Crystallography

Xenotime-(Gd) is isostructural with other xenotime-group minerals; its tetragonal (space group I4₁/amd) structure consists of isolated PO₄ tetrahedra separated by intervening (REE)O₈ polyhedra. A smaller REE-containing polyhedron in the xenotime structure with the 8-fold coordination compared to a monoclinic monazite structure with the 9-fold coordination accommodates the smaller cations; therefore, the xenotime structure prefers HREE over MREE and LREE (e.g. Miyawaki and Nakai, 1993; Ni et al., 1995). Consequently, xenotime crystals with MREE enrichment are rare.

To determine the crystal structure of the mineral, HRTEM and STEM characterisation on focused ion beam (FIB) lamella was used. This procedure enables measurement of the same phases that were previously analysed by EPMA. A single-crystal X-ray diffraction study of xenotime-(Gd) was not carried out due to the small size of the crystals. Powder X-ray diffraction data could also not be obtained, but because the xenotime structure of synthetic GdPO₄ is known (tetragonal, space group I4₁/amd, a = 6.9648(4) Å and c = 6.1050(5) Å; Rodriguez-Liviano et al., 2013; a = 6.9612(7) Å and c = 6.1026(10) Å (Clavier et al., 2018), a pattern was calculated using the cif (Table 3). The xenotime structure can be described as [001]-oriented chains formed of alternating phosphate tetrahedra and REE polyhedra (Vegard, 1927). Each PO₄ tetrahedron is isolated from other PO₄ tetrahedra and shares two of its edges with the REEO₈ polyhedra. In synthetic GdPO₄, the smaller P⁵⁺ cations are tetrahedrally coordinated with a P−O bond length of 1.554(4) Å and two different OPO angles [101.6(3)° and 113.6(2)°]. The PO₄ tetrahedra are connected to Gd³⁺ cations with coordination number 8 and two unique Gd−O bond lengths [2.343(4) and 2.397(4) Å] (Rodriguez-Liviano et al., 2013). Therefore, a HRTEM with FFT techniques was used to determine the unit-cell parameters of the crystal examined.

Table 3. Calculated powder X-ray diffraction pattern (λ = 1.5406 Å) for xenotime-(Gd) compared to synthetic xenotime-(Gd) (Rodriguez-Liviano et al., 2013).*

* The data were calculated with the program Diamond, v 4.0, for the composition and unit-cell parameters of the natural sample investigated in this work.

A thin lamella of a xenotime-(Gd) single crystal was tilted in the electron microscope to obtain a low-index zone axis parallel to the primary electron beam. At this setting, HRTEM resulted in the recording of a defect-free single crystalline structure (Fig. 2a). The corresponding FFT pattern was acquired from the entire recorded area and the interplanar distances obtained by measuring the reflection positions and the angles between individual reflections were determined (Fig. 2b). Using this method, a set of 25 reflections were measured using FFT and with refinement gave the following unit-cell parameters: a = 6.9589(5) Å, c = 6.0518(6) Å, V = 293.07(3) ų and Z = 4.

Figure 2. (a) High-resolution TEM image of the crystal lattice viewed along the [11$\bar{1}$] direction. (b) Relevant FFT pattern showing the [11$\bar{1}$] zone axis.

The measured angles and interplanar distances were compared with relevant data obtained using a standard stereographic projection of the tetragonal crystal of xenotime-(Gd) (Rodriguez-Liviano et al., 2013). Based on this, and also using the rules for indexing point electron diffractions, Miller indices were assigned to all reflections in the FFT pattern (Fig. 2b).

The experimentally obtained d-spacings of 0.5001 nm and 0.457 nm are in good agreement with that of 0.49015 nm and 0.45664 nm, respectively, reported for interplanar distances of ($\bar{1}10$) and ($\bar{1}$0$\bar{1}$) planes of tetragonal xenotime-(Gd) (Rodriguez-Liviano et al., 2013). Moreover, the measured angle of 62.90° between these reflections corresponds well with the 62.23° valid for the angle between ($\bar{1}10$) and ($\bar{1}0\bar{1}$) planes according to standard stereographic projection of xenotime-(Gd). By evaluation of the FFT pattern, it was determined that the zone axis of the examined xenotime-(Gd) single crystal is [$11\bar{1}$].

A Xenotime-(Gd) single crystal oriented in the [$11\bar{1}$] direction was also characterised using atomic-scale resolution imaging. Atomic-resolution HAADF STEM (Z contrast) and BF STEM images display the distribution of atomic columns in the tetragonal crystal structure of xenotime-(Gd) oriented along the [$11\bar{1}$] zone axis (Fig. 3). Bright spots in HAADF STEM image (Fig. 3a) represent Gd atomic columns, which is concordant with a superimposed tetragonal crystal structure representation of xenotime oriented in the [$11\bar{1}$] direction. Gadolinium atomic columns appear dark in the simultaneously recorded atomic resolution BF STEM image (Fig. 3b).

Figure 3. (a) Atomic-resolution HAADF STEM image of a xenotime-(Gd) single crystal observed along the [11$\bar{1}$] direction. The superimposed atomic structural model (bottom right) viewed along the [11$\bar{1}$] direction identifies the Gd atomic columns (red) in a xenotime-(Gd) single crystal. (b) A simultaneously recorded atomic-resolution BF STEM image of a xenotime-(Gd) single crystal observed from the [11$\bar{1}$] direction. Dark spots represent Gd atomic columns.

It can be assumed that the unit-cell parameters of the observed mineral correspond to the weighted average of the corresponding end-members unit-cell parameters. The unit-cell parameters of synthetic REEPO₄ with a xenotime structure display a near-ideal linear trend of increasing unit-cell parameters with increasing ionic radii of REE cations. Moreover, the content of other possible substituents (U, Th, Ca, Si and As), which could alter the linear trend, is very low in the xenotime-(Gd) sample and, therefore, the influence of these elements is negligible for the cell-parameter calculation. The following are the calculated mean unit-cell parameters for studied xenotime-(Gd) from the weighted sum of end-members: a = 6.9319(4) Å, c = 6.0697(5) Å, V = 291.66(1) ų and Z = 4.

The refined and calculated unit-cell parameters are in very good agreement. Moreover, the empirical unit-cell parameters obtained from the FFT are slightly lower than the unit-cell parameters of synthetic analogues of xenotime-(Gd) of Rodriguez-Liviano et al. (2013) and Clavier et al. (2018), which reflects the presence of other REE with smaller ionic radii (Table 4).

Table 4. Unit-cell parameters for natural xenotime-(Y) and xenotime-(Gd) compared to the published data for synthetic xenotime compounds.

Raman spectroscopy

The Raman spectra of xenotime-(Gd) and Gd-poor xenotime-(Y) have been documented previously in Ondrejka et al. (2023c). These spectra are influenced significantly by the photoluminescence (PL) of REE³⁺, masking Raman signals in the low-frequency range (100–750 cm^–1) (see Fig. 4). Despite the intense PL, specific bands can be discernible in xenotime-(Y) at 650, 580 and 484 cm^–1, and in xenotime-(Gd) at 640, 576–577 and 483 cm^–1. The Raman modes become clearly distinguishable between 800 and 1200 cm^–1 in both spectra excited by 532 nm and 633 nm lasers. Xenotime-(Y) exhibits distinctive bands centred at 999, 1024 and 1057 cm^–1, while the spectra of xenotime-(Gd) display lines at 994–996, 1018 and 1052 cm^–1, along with an additional, less intense band at 965 cm^–1, and minor broad features at approximately 870, 900 and 1150 cm^–1.

Figure 4. Depolarised Raman spectra captured for both natural xenotime-(Y) and xenotime-(Gd) (Prakovce, Zimná Voda ZV-2 sample) using two distinct laser excitations: 532 nm (black line, upper) and 633 nm (grey line, lower). The spectra are annotated with labelled sections indicating regions of photoluminescence (PL) and Raman bands. The intensity scale bar is provided in arbitrary units (a. u.).

The positions of the Raman bands for xenotime-(Gd) align with previously reported data on xenotime-type orthophosphates (Begun et al., 1981; Lenz et al., 2015; Švecová et al., 2016; Yahiaoui et al., 2017; Lösch et al., 2019; Clavier et al., 2018). Generally, lattice vibration modes manifest below 300 cm^–1 (Begun et al., 1981; Lösch et al., 2019) or up to 400 cm^–1 (Clavier et al., 2018) encompassing translations and rotation modes of the entire (PO₄)^3– unit. According to this literature, bands in the spectral range of 950–1100 cm^–1 relate to the P–O symmetric and antisymmetric stretching modes of (PO₄)^3– (Table 5). The frequencies observed for xenotime-(Gd) are lower than those for xenotime-(Y), aligning with the shifting positions of ν ₁ and ν ₃ observed in experimental studies on REE substitutions in xenotime-type orthophosphates (Begun et al., 1981; Clavier et al., 2018). Peaks at 965 cm^–1, 900 and 1150 cm^–1 tentatively associate with Si and As substitution in an anion position in natural xenotime, necessitating further detailed examination. The strong PL is caused by the presence of Er³⁺ and Ho³⁺ (under 633 nm excitation), and by Er³⁺, Eu³⁺, Ho³⁺ and Sm³⁺ (under 532 nm excitation) (Fig. 4). Notably, Sm³⁺ probably contributes to the broadening of the 1017 cm^–1 band, and the faint, broad band around 1150 cm^–1 is possibly attributed to this luminescent centre (Lenz et al., 2015). Despite the Raman signal being overshadowed by PL, we deduce the absence of structural H₂O or (OH)^– in xenotime-(Gd) because of the absence of bands within the stretching modes range of these groups after 633 nm excitation.

Table 5. Raman bands for xenotime-(Gd).

Notes: ν1 and ν3 – symmetric and antisymmetric stretching; ν2 and ν4 – symmetric and antisymmetric bending. Lattice modes: T – translation, R – rotation; REE photoluminescence bands not included.

Discussion

Xenotime-(Gd) is a new lanthanide orthophosphate mineral of the xenotime group with a natural unique Gd-dominance over the other REE cations. It represents only the third Gd-dominant mineral, described in Nature, and approved by IMA-CNMNC (Pasero, 2024), after lepersonnite-(Gd), CaGd₂(UO₂)₂₄(SiO₄)₄(CO₃)₈(OH)₂₄⋅48H₂O, from the Shinkolobwe uranium deposit, DR Congo (Deliens and Piret, 1982) and cogenetic dimorphous monazite-(Gd) from the Zimná Voda occurrence near Prakovce, Slovakia (Ondrejka et al., 2023b). However, xenotime-(Gd) contains a significantly higher content of Gd: ≤29.3 wt.% Gd₂O₃, in comparison to lepersonnite-(Gd) (only 2.1 wt.% Gd₂O₃; Deliens and Piret, 1982) and monazite-(Gd) ≤23.4 wt.% Gd₂O₃ (Ondrejka et al., 2023b). Consequently, xenotime-(Gd) contains the highest Gd concentration among approved minerals. We emphasise the use of appropriate and precise analytical methods in these data, as there are other published data that document the possible presence of other Gd-dominant ‘minerals’, unfortunately without suitable analytical method or measurements. The presence of Gd-dominant ‘monazite' with 42.5 wt.% Gd₂O₃ and an empirical formula (Gd_0.55Y_0.25Dy_0.1Sm_0.05Nd_0.05Th₀Ca₀)(PO₄) is noted from alkali-feldspar syenite pegmatite at the Myan Gyi mine, near Mogok, Myanmar (Kartashov, web data at mineralienatlas.de; mindat.org); however, with no further information. The impressive REE accumulations in lignite coals from the Russian Far East Pavlovka deposit contain fine-grained authigenic unknown Gd- and Dy-dominant minerals (Seredin, 1996). Several nano- to micro-sized particles of Gd–Ti–Zr oxides with dominant Gd occupancy and 9−57 wt.% Gd₂O₃ were identified in the lunar regolith from Mare Crisium (Bogatikov et al., 2004; Mokhov et al., 2011). In addition, an unnamed Gd-dominant mineral (Gd > Ce > La) close to the Gd₂Ti₄O₁₁ formula was noted from fumaroles in the active Kudriavy Volcano, Kuril Islands, Russia (Bogatikov et al., 2004). However, all the above-mentioned findings of possible Gd-dominant minerals need to be taken and interpreted with caution, because the data were obtained using semiquantitative EDX analysis without appropriate analytical details available. Furthermore, other analytical data (e.g, XRD, micro-Raman) are lacking.

Beside xenotime-(Gd), xenotime-(Y) from the Zimná Voda REE–U–Au occurrence is also enriched in Gd (11.6 to 27.9 wt.% Gd₂O₃, 0.13 to 0.35 apfu Gd). Moreover, inclusions of xenotime-(Gd), monazite-(Gd) and monazite-(Sm) (≤50 μm in size) with up to 29.8 wt.% Gd₂O₃ (≤0.38 apfu Gd) have been found recently in spessartine garnet in the Otov rare-element granitic metapegmatite (beryl–columbite-phosphate subtype), Czech Republic (Vrtiška, Ondrejka and Uher, unpubl. data). Rare occurrences of Gd, Dy-rich and Yb-dominant compositions in the xenotime-group minerals have also been described from some magmatic and metamorphic-hydrothermal systems (Demartin et al., 1991; Förster and Rhede, 1995; Förster, 1998; Buck et al., 1999; Masau et al., 2000; Franz et al., 2015). In addition to Prakovce-Zimná Voda, the highest published Gd₂O₃ content in xenotime-(Y) attains 25 wt.% (0.31 apfu Gd), from the quartz vein of Au–REE mineralisation in the Subpolar Urals, Russia (Repina, 2010, 2011).

Such unusual enrichment of Gd in xenotime-(Gd) as well as in some other REE minerals [e.g. monazite-(Gd), hingganite-(Y)] and the progressive development of a MREE enhanced signature could be a result of local selective complexing of REE during granite to pegmatite sequence solidification in aqueous F–Cl–Li–CO₂-bearing fluids (Masau et al., 2002; Twardak et al., 2023). Moreover, hydrothermal xenotime-(Gd) and associated Gd-rich minerals precipitated in response to the alteration of MREE-selective, but nominally REE-free uraninite, brannerite and fluorapatite precursors by low-temperature hydrothermal fluids. Further details and description of Gd-MREE-enriched minerals in the Prakovce-Zimná Voda occurrence can be found in Ondrejka et al. (2023c).

The existence of GdPO₄ and mixed Gd_xLn_1–xPO₄ dimorphism has been experimentally replicated and the kinetic-thermodynamic (meta)stability of the monazite and zircon structure types under various P-T conditions has been described in the literature (e.g. Mullica et al., 1990; Celebi and Kolis, 2002; Kolitsch and Holtstam, 2004; Rodriguez-Liviano et al., 2013; Hay et al., 2013; Hefferman et al., 2016; Meng et al., 2016; Li et al., 2018; Muñoz and Rodríguez-Hernández, 2018; and references therein). There is a general consensus that the structural boundary between monazite and xenotime usually lies between Gd and Tb (e.g. Ni et al., 1995; Gratz and Heinrich, 1998; Hay et al., 2013; Rodriguez-Liviano et al., 2013). The investigation of natural GdPO₄ orthophosphates from Prakovce, Zimná Voda (Ondrejka et al., 2023b, 2023c) confirms the possibility of (1) substantial Gd incorporation into REE-selective structures and (2) the stabilisation of Gd-dominant orthophosphate with a zircon/xenotime-type structure by substitution of smaller HREE+Y cations in the A site (Mullica et al., 1986, 1990; Gratz and Heinrich, 1998; Clavier et al., 2011; Rodriguez-Liviano et al., 2013; Ondrejka et al., 2023c). The stabilisation of the monazite- versus xenotime-type structure of the natural REEPO₄ species depends on the average REE³⁺ ionic radius (Kolitsch and Holtstam, 2004). In addition, the xenotime-(Gd) – monazite-(Gd) pair represents the first naturally occurring dimorphism among the known REE phosphates.