Comparative thermal and compressional behaviour of natural xenotime-(Y), chernovite-(Y) and monazite-(Ce)

Abstract

ATO₄ compounds are a class of oxides which includes the rare earth element (REE) bearing phosphates and arsenates, REEPO₄ and REEAsO₄. In this study, we have investigated the isothermal high-pressure and the isobaric high-temperature behaviour of natural samples of xenotime-(Y) (ideally YPO₄), chernovite-(Y) (YAsO₄) and monazite-(Ce) (CePO₄) from the hydrothermal veins cropping out at Mt. Cervandone in the Western Italian Alps. Experimental data based on in situ X-ray diffraction (both single-crystal and powder techniques with conventional or synchrotron radiation) have allowed us to fit the unit-cell volumes and axial thermal and compressional evolution and provide a suite of refined thermo-elastic parameters. A comprehensive analysis of the role played by the crystal chemistry on the thermo-elastic response of these minerals is discussed, along with the description of the main crystal-structural deformation mechanisms for both the zircon (xenotime and chernovite) and monazite (monazite) structural types. Pressure-induced phase transitions of xenotime-(Y) and chernovite-(Y) are discussed and compared with previous literature data, whereas a change in the compressional behaviour of monazite-(Ce) at ∼18 GPa, involving an increase in the coordination number of the REE-hosting A site, is presented and discussed.

Introduction

The general formula ATO4 is commonly used in the literature to define ternary inorganic oxides (Vorres, 1962), where A and T represent two cations that can be combined with oxygen (and occasionally with other anions) into several structural types, including, but not limited to, scheelite, zircon, monazite, fergusonite, baryte, quartz, cristobalite, wolframite and rutile (Fukunaga and Yamaoka, 1979). In the context of the present study, the A site is occupied by a rare earth element (REE: lanthanides and Y), Ca, U and Th, whereas T stands for tetrahedrally-coordinated cations (As and P). In regard to the structural types, this manuscript focuses almost exclusively on the zircon and monazite structures, as demonstrated by the four minerals which are subject of this and two previous works (Pagliaro et al., 2022a, 2022b): chernovite-(Y) [nominally YAsO₄], xenotime-(Y) [nominally YPO₄], gasparite-(Ce) [nominally CeAsO₄] and monazite-(Ce) [nominally CePO₄]. The crystal structure of these minerals has been the subject of a large number of studies and reviews (e.g. Mooney, 1948; Ni et al., 1995; Boatner, 2002; Finch and Hanchar, 2003 Kolitsch and Holtstam, 2004; Clavier et al., 2011) and an overview of the monoclinic monazite-type structure and of the tetragonal zircon-type (also known, but occasionally reported as ‘xenotime-type’) is discussed in the next section.

As reported by several authors (Fukunaga and Yamaoka, 1979; Ushakov et al., 2001; Boatner, 2002; Kolitsch and Holtsam, 2004), whether the monazite or the zircon-type structure is stable (within ATO₄ phosphates and arsenates) depends on different factors. Among others, the atomic radii of either the A or T sites play a dominant role. In general, a large sized A cation promotes the crystallisation of the monazite structural type over the zircon one; on the other hand, the larger the T-site cation, the more stable is the zircon structure across the REE series. Within the REE-bearing phosphates, light REE ranging from La to Eu, with larger ionic radii than heavy REE, are hosted preferentially by the monazite-type structure, whereas heavy-REE, from Tb to Lu, including Y and Sc, fit best into the zircon-type structure (Mooney, 1948; Ni et al., 1995; Boatner, 2002; Kolitsch and Holtstam, 2004; Clavier et al., 2011). A similar behaviour has been reported for the REEAsO₄ series, although the threshold among the two structures is shifted to smaller Z numbers in the lanthanoid series: the monazite-type structure preferentially hosts REE from La to Nd, whereas the REE from Sm to Lu (as well as Y and Sc) are hosted by the tetragonal zircon-type crystal structure (e.g. Ushakov et al., 2001; Boatner, 2002).

The REE-bearing phosphates are common accessory minerals in hydrothermal alteration of granitoid rocks that can control the partitioning of REE, as well as uranium and thorium, as they tend to incorporate these elements into their crystal structures (Rapp and Watson, 1986). In addition, due to their much lower tendency to incorporate Pb, they have found a significant use in geochronological applications (Harrison et al., 2002). Due to their peculiar physical, chemical and optical properties (such as low solubility in water fluids), REE phosphates are used, or have been proposed, in several technological applications, e.g. to produce phosphors (de Sousa Filho and Serra, 2009), ceramic coatings (Morgan et al., 1995; Davis et al., 1998), and materials for the safe storage of actinides originating from radioactive waste (Oelkers and Montel, 2008; Orlova and Ojovan, 2019). Consequently, there has been significant interest in studying the behaviour of REE phosphate minerals and their synthetic counterparts under varying pressure and temperature conditions.

Recent reviews have been published by Errandonea (2017) and Strzelecki et al. (2024) for monazite- and zircon-type structures, respectively. Tables S1 and S2 (supplementary material, see below) provide a comprehensive list of the thermo-elastic parameters published in the literature for several REETO₄ compounds, along with the related bibliographic references. Concerning the response of REETO₄ compounds to high pressure (hereafter HP), as a general rule it has been postulated that, for a given structural type, the bulk modulus shifts towards lower values as the atomic radius of the A and T sites increases, which has been highlighted by several authors and corroborated by both theoretical (Zhang et al., 2008; Li et al., 2009) and experimental studies (Zhang et al., 2008; Lacomba-Perales et al., 2010; Errandonea et al., 2011). Regarding the thermal behaviour, there is general agreement on the thermal expansion coefficient of REETO₄, which shows a clear compositional trend: the thermal expansivity increases along with the ionic radii of the A cation, while it reduces if the radius of the T site increases (Subbarao et al., 1990; Perrière et al., 2007; Zhang et al., 2008; Li et al., 2009). Unfortunately, the published values of both the compressibility and thermal expansivity of the REETO₄ studied are not always internally consistent (see Tables S1 and S2), showing a certain degree of scattering, even for the same compound. In the case of thermal studies, the use of different thermal equations of state further complicates a comparative analysis. Therefore, in this study, we used the linear thermal expansion coefficient, which is commonly used in the literature, although it is not the most accurate model of the thermal elastic response (along with thermal equations of state commonly used in Earth Sciences).

The relative ratio of the A and T ionic radii not only affects the structure types adopted by a given compound at ambient conditions, but also its pressure stability field and the structural type of the high-pressure polymorphs. Such a relationship is well described by the so-called ‘Bastide diagram’ (Bastide, 1987), for which one of the most recent graphical representations is reported in Lopez-Solano et al. (2010) (Fig. 1, modified). The zircon-type compounds, xenotime-(Y) and chernovite-(Y) in this study, may transform at high-pressure into monazite-type or scheelite-type polymorphs. Whether a zircon → scheelite or a zircon → monazite → scheelite transformation occurs depends on the reciprocal relations among the ionic radii of the three atoms involved. A large T cation promotes a zircon-to-scheelite phase transition, whereas a small T cation favours an intermediate monazite-type polymorph. For the A-site cation, the larger it is, the more likely it is that a monazite polymorph will form over the scheelite one, ‘shifting’ the stability field to higher pressures. All the studied zircon-type ATO₄ silicates show a zircon-to-scheelite phase transition, with high-pressure Raman and Density Functional Theory calculations suggesting the occurrence of a high-pressure lower symmetry polymorph preserving the zircon structural configuration before the reconstructive phase transition to reidite (Stangarone et al., 2019). The zircon-to-scheelite phase transition has been described for YAsO₄ (∼8 GPa) and YCrO₄ (∼4.2 GPa) (Errandonea et al., 2011), as well as YVO₄ (above ∼7.5 GPa) (Jayaraman et al., 1987; Wang et al., 2004; Manjón et al., 2010). A comprehensive description of the zircon-to-monazite phase transition in phosphates, including YPO₄ (xenotime) is reported in Hay et al. (2013). The relations between monazite and its HP-polymorphs again depend upon the reciprocal relations among the A, T and oxygen ionic radii. The monazite-to-post-baryte phase transition (space group P2₁2₁2₁) has been described for REEPO₄ and REEVO₄ at increasing pressures with decreasing REE atomic radius (Lacomba-Perales et al., 2010; Ruiz-Fuertes et al., 2016; Errandonea, 2017). On the other hand, monazite-type LaVO₄, PrVO₄ and NdVO₄, under compression, undergo a phase transition to a monoclinic BaWO₄-II-type structure (Errandonea et al., 2016; Panchal et al., 2017; Marqueño et al., 2021). Eventually, the monazite-to-scheelite transition has been described for the high-pressure polymorphs of YPO₄ and several other REE-free compounds, as SrCrO₄ and CaSeO₄ (Crichton et al., 2012; Gleissner et al., 2016).

Figure 1. The so-called ‘Bastide diagram’ showing the relationships among structural types as a function of the atomic radii of cations at the A site (r _A), T site (r _T) and oxygen (r _O), within the ATO₄ family. The fields corresponding to the SrUO₄ and BaWO₄-II structures are labelled as orthorhombic (Cmca, Pbcm, Pnma) and monoclinic 14, respectively (2, 10, 12, 14 refer to the space group numbers). The post-baryte field is not reported (modified after López-Solano et al., 2010).

This study, following the research conducted by Pagliaro et al. (2022a, 2022b), which includes a detailed crystal chemistry description, focuses on the high-pressure and temperature behaviour of four REETO₄ mineralogical species from the same locality (Mt. Cervandone, Piedmont, Italy).

Based on experimental single-crystal or powder X-ray diffraction data collected in situ (high-P or high-T) at synchrotron beamlines or in conventional diffraction laboratories, a comparative analysis of the elastic behaviour and structural deformation mechanisms as a function of the crystal chemistry and structure type has been performed. The adoption of up-to-date experimental techniques and crystallographic methods has allowed us to describe a P-induced structural re-arrangement peculiar of the monazite structure type, previously reported by Pagliaro et al. (2022b) for gasparite-(Ce) and here confirmed also for monazite-(Ce). Such an intermediate structural configuration, implying an increase in the number of oxygen atoms bonded to the A site (from 9 to 10), before the occurrence of the phase transition to the post-baryte-type polymorph at higher pressures, has not been reported in earlier literature for monazite-type phosphates.

Crystal structure description

Zircon-type crystal structure

The first studies concerning the crystal structure of zircon date back to the early 20^th century and were carried out independently by Vegard (1916, 1926), Binks (1926), Hassel (1926) and Wyckoff and Hendricks (1928), in the framework of the pioneering works about the silicate’s structure determination, later gathered by Bragg (1929) in his Atomic Arrangement in Silicates. After the first studies on zircon, its structural type has been described in several REE-bearing compounds, including xenotime-(Y) (Vegard, 1927) and the synthetic counterpart of chernovite-(Y), YAsO₄ (Strada and Schwendimann, 1934).

The zircon-type structure is characterised by a tetragonal I-centred lattice (space group I4₁/amd). The tetragonal zircon-type structure is constructed by infinite chains of polyhedra, developed along the [001] direction (Fig. 2a and 2d), as the result of the connection, along the polyhedral edges, between the eightfold coordinated A-site dodecahedron (AO₈ or REEO₈) and the TO₄ tetrahedra (Fig. 2c). The AO₈ polyhedron displays two independent A–O atomic distances (Fig. 2c), whereas the TO₄ is an undistorted tetrahedron defined by a single T–O bond distance. Each chain is in contact with four others on the (001) plane, through connecting edges along an AO₈ unit and the surrounding four (Fig. 2b). The atomic coordinates of the A- and T-sites are placed in special, fixed positions, both characterised by a 2m point symmetry. The oxygen atom is also at a special position (m), being its y and z coordinates the sole refinable parameters.

Figure 2. Crystal structure of the zircon-type materials viewed (a) along the [010] and (b) [001] directions and showing (c) the chains running along the c directions and the bond distances configuration among the AO₈ polyhedron and (d) a side view of the overall crystal structure. Structure drawings have been made using the software Vesta3 (Momma and Izumi, 2011).

Monazite-type crystal structure

Parrish (1939), within the first crystallographic studies on monazites, identified its correct space group as P2₁/n. The first description of the monazite-type structure has been reported by Mooney (1948), who investigated the La, Ce, Pr and Nd phosphates as part of the Manhattan project and described the REE atomic site in eightfold coordination. The crystal structure of monazite with the REE site in ninefold coordination has been proposed by Ueda (1953, 1967), but with non-reliable average P–O bond lengths of ∼1.6 Å. The structure was later described correctly by Beall et al. (1981), Mullica et al. (1984) and Ni et al. (1995), whereas an exhaustive review of the monazite-structure type has been carried out by Boatner (2002) and then by Clavier et al. (2011). The monazite-type structure can be described as made by infinite chains running along the [001] direction (c-axis), comprising the alternation of the REE-coordination polyhedra and the T-hosting tetrahedra (Fig. 3).

Figure 3. Crystal structure of monazite, viewed along (a) the [100] and (b) [010] directions; a chain-like unit is highlighted in blue; (c) coordination polyhedron of the REE-bearing A site, with 9 independent A–O bonds; and (d) general view of the monazite structure. Structure drawings have been made using the software Vesta3 (Momma and Izumi, 2011).

The REE-polyhedron coordination environment has nine corners (oxygen ligands, REEO₉, Fig. 3c). According to Mullica et al. (1984), the REEO₉ polyhedron can be described as an equatorial pentagon (sharing vertices with five TO₄ tetrahedra of five adjacent chains in correspondence of the O1_b, O2_b, O2_c, O3_b and O4_b oxygens), interpenetrated by a tetrahedron (made by the O1_a, O2_a, O3_a and O4_a oxygen atoms, see Fig. 3c), which is along the [001] direction in contact with two subsequent TO₄ tetrahedra, leading to the formation of the infinite chain units (Fig. 3a,b). The REE–O2_a bond length is significantly longer than the other REE–O bonds, contributing to a significant distortion of the REEO₉ polyhedron (Beall et al., 1981; Ni et al., 1995; Clavier et al., 2011), which can be considered as 8+1 coordinated.

Samples and experimental methods

The mineral samples of monazite-(Ce), xenotime-(Y) and chernovite-(Y), investigated in this study, originate from the same locality at Mt. Cervandone, Piedmont, Italian western Alps, where they are found as accessory phases in alpine-type fissures within hydrothermal quartz veins (Graeser and Albertini, 1995) that cross-cut pegmatitic dykes (Guastoni et al., 2006). The latter, intruded in leucocratic gneisses, are enriched in REE and have a strong NYF (Niobium-Yttrium-Fluorine enrichment) signature (Černý, 1991a, 1991b; Černý and Ercit, 2005). An overview of the geological background of the source rocks is described in Pagliaro et al. (2022a), along with a detailed chemical analysis of the samples of this study by means of an electron microprobe operating in WDS mode. The experimental chemical formulas of the investigated minerals are summarised in Table 1.

Table 1. Average (and range of the measured) chemical composition (expressed in oxide wt.% and in atoms per formula unit (apfu) calculated on the basis of 4 oxygen atoms) of the chernovite-(Y), xenotime-(Y) and monazite-(Ce) samples under investigation

* (Ca,Th)-poor chernovite in the text and Ch₁₀ sample in Pagliaro et al. (2022a); **(Ca,Th)-enriched chernovite in the text and Ch₁₃ in Pagliaro et al. (2022a); ***Xen₁₄ sample in Pagliaro et al. (2022a); ****Mon₁₄ sample in Pagliaro et al. (2022a)

b.d.l. – below detection limit

In situ high-pressure experiments

In situ high-pressure single-crystal synchrotron X-ray diffraction experiments have been conducted on chernovite-(Y), xenotime-(Y) and monazite-(Ce) at the P02.2 beamline (PETRA-III synchrotron at DESY, Hamburg, Germany) and ID15B beamline (European Synchrotron Radiation Facility, Grenoble, France) using different classes of P-transmitting fluids as shown in Table 2. For all the experiments, the crystals were loaded in membrane-driven diamond anvil cells (DAC), equipped with Boehler-Almax designed diamonds/seats. Metallic foils (steel or rhenium) were pre-indented to ca. 40–70 μm and then drilled by spark-erosion to obtain P-chambers. Ruby spheres were employed as pressure calibrants (pressure uncertainty ±0.05 GPa; Mao et al., 1986; Chervin et al., 2001). All data collections were based on a ω-rotation with 0.5° per step and 0.5 s (monazite and chernovite) or 1 s (xenotime) of exposure time per frame. At ID15B, X-ray diffraction (XRD) data were collected using an Eiger2 9M CdTe detector positioned at 179 mm from the sample with a monochromatic 30.2 keV (λ = 0.4099 Å) beam, whereas XRD patterns at P02.2 were collected on a Perkin Elmer XRD1621 detector at 373 mm from the sample and a monochromatic incident beam with E = 42.67 keV (λ = 0.2906 Å). Further details on the beamlines setups are presented in Merlini and Hanfland (2013) and Poreba et al. (2022) for ID15B and Rothkirch et al. (2013), Liermann et al. (2015) and Bykova et al. (2019) for P02.2. Indexing of the X-ray diffraction peaks, unit-cell refinements and intensity data reductions were performed using the CrysAlisPro package (Rigaku Oxford Diffraction, 2020). Absorption effects, due to the DAC components, were corrected using the semi-empirical ABSPACK routine, implemented in CrysAlisPro.

Table 2. Details pertaining to the in situ high-pressure and high-temperature experiments of this study

* SC-XRD: single-crystal X-ray diffraction; PXRD: powder X-ray diffraction

** He: helium (Klotz et al., 2009); Ne: neon (Klotz et al., 2009); m.e.w.: methanol:ethanol:water = 16:3:1 (Angel et al., 2007)

*** See the text for further details

^# Institute of Mineralogy and Petrography, University of Innsbruck, Austria

Based on the experimental intensity single-crystal XRD data, the structure refinements were performed using the Jana2020 software (Petříček et al., 2023), starting from the structural models reported by Pagliaro et al. (2022a) for the mineral samples from the same locality. The site occupancy factors of the A (lanthanide-bearing) and tetrahedral sites were fixed according to the average chemical composition obtained from EPMA-WDS analysis (Table 1), disregarding the elements with a concentration lower than 0.03 atoms per formula unit and assuming a full occupancy for both the sites. In addition, for monazite-(Ce), the atomic displacement parameters (ADP) of the oxygen atoms were refined as isotropic. All the refinements converged with no significant correlations among the refined variables. Refined structural models are deposited as crystallographic information files (cifs) and are available as Supplementary material (see below).

For a chernovite-(Y) sample with a slightly larger amount of Ca and Th replacing Y and REE, an in situ high-pressure powder XRD experiment was conducted at the P02.2 beamline of PETRA III (Hamburg, Germany) with a wavelength of λ = 0.2906 Å (42.67 keV) and a Debye-Scherrer geometry. The sample was loaded into a DAC equipped with Boehler-Almax designed diamonds of 400 μm culet size along with the pressure-transmitting medium (see Table 2 for details) and ruby spheres for pressure determination (P – uncertainty ±0.05 GPa; Mao et al., 1986; Chervin et al., 2001). The data collection strategy at any pressure point consisted of a 30° rotation along ω, for an exposure time of 60 s. The X-ray diffraction signals captured by the Perkin Elmer XRD1621 flat panel detector have been finalised and integrated by means of the Dioptas software (Prescher and Prakapenka, 2015), in order to remove the background noise due to DAC components and extract the 2θ-intensity pattern for any experimental dataset. The unit-cell parameters were determined by fitting the powder XRD data by means of the Rietveld full-profile method using the GSAS-II software (Toby and Von Dreele, 2013): the unit-cell parameters, crystallite size, individual scale factor and profile parameters (pseudo-Voigt function) have been refined. Moreover, the background signal has been interpolated through a Chebychev polynomial function, with 4 to 15 terms.

In situ high-temperature experiments

In situ high-temperature single-crystal X-ray diffraction data were collected at the Institute of Mineralogy and Petrography of the University of Innsbruck, using a Stoe IPDS II diffractometer equipped with a Heatstream HT device, providing a continuous flow of hot N₂. The primary X-ray beam was generated by an X-ray tube (Mo-anode), operating at 50 kV and 40 mA. A plane graphite monochromator and a multiple pinhole collimator (0.5 mm) were used to guide the beam onto the sample. The image-plate detector was placed at a distance of 100 mm. The temperature calibration had been conducted previously using phase transitions of KNO₃, Ag₂SO₄, K₂SO₄ and K₂CrO₄ powders into glass capillaries. The samples, single crystals of monazite-(Ce), xenotime-(Y) and chernovite-Y (ca. 50–120 μm³) have been inserted into SiO₂ glass capillaries (0.1 mm in diameter). For all the samples, the data collection consisted in a 180° ω-rotation with a step size of 1° and variable exposure times. The temperature accuracy is ≤ 5°C. Further details about the experimental setting are reported in Krüger and Breil (2009). Data collection and reduction has been performed using X-Area (Stoe and Cie, 2008). The indexed cell parameters were always compatible with either the unit cells of chernovite-(Y), xenotime-(Y) or monazite-(Ce).

Structure refinements were performed based on the single-crystal XRD data adopting the same procedure reported previously for the high-pressure data, and refined structural models are available as cifs (supplementary materials, see below). The thermal evolution of significant structural parameters (i.e. A–O bonds, A-coordination polyhedral and T-coordination polyhedral volumes) has been determined from the refined structure models, by means of the tools implemented in the VESTA3 software (Momma and Izumi, 2011).

In situ high-temperature X-ray powder diffraction experiments were performed on the chernovite sample relatively enriched in Ca and Th at the MCX beamline of the Elettra synchrotron (Basovizza, Trieste Italy), with a wavelength of λ = 0.7293 Å (17 keV) and a Debye-Scherrer setting. The sample, ground to powder in an agate mortar, was loaded in a SiO₂ glass capillary (0.3 mm as outer diameter). For any experimental point, the data collection strategy consisted of a 2θ-scan between 8° and 60°. A step size of 0.008° was applied and an equivalent counting time for 1 s/step used. The X-ray diffraction effects were collected by the high-resolution scintillator detector available at the beamline. During the data collection, the sample was spun at a rate of 1000 rotations per minute along the φ-axis. The sample was heated by an air blower, operating between 30 and 1000°C. Further details concerning the experimental setting are reported in Rebuffi et al. (2014) and Lausi et al. (2015).

Results and discussion

Compressional behaviour of the REETO₄ minerals

The evolution of the unit-cell parameters of the investigated samples with pressure is reported in Table S3 and shown in Figs 4 and 5. For monazite-(Ce), the evolution of unit-cell parameters vs. P shows no evidence of phase transformation in the entire P-range investigated. Both the zircon-type minerals, on the other hand, show the occurrence of a phase transition. Chernovite-(Y), at pressures higher than ∼10 GPa is no longer stable, undergoing a phase transition to several single-crystal fragments, for which, the position of the peaks in the XRD pattern is compatible with a tetragonal scheelite-type structure. Otherwise, at pressures exceeding ∼17 GPa, xenotime-(Y) undergoes a single-crystal to single-crystal phase transition towards a monazite-type structure. For any pressure ramp, the elastic behaviour of the mineral studied has been described by fitting an isothermal Birch-Murnaghan EoS (BM-EoS), truncated at the second or third order, to the P–V data (a comprehensive description of the BM-EoS formalism can be found in Angel, 2000) using the EoS-Fit7_GUI software (Gonzalez-Platas et al., 2016). For monazite, similarly to what was observed for the isostructural gasparite-(Ce) by Pagliaro et al. (2022b), a change in the compressional behaviour was detected at ∼18 GPa. The parameters refined by fitting the experimental data by BM-EoS are reported in Table 3, among them: the bulk modulus K_V = β_V ^–1 = –V*(∂P/∂V) and its pressure derivative K_V’ = (∂K_V/∂P)_T.

Figure 4. High-pressure evolution of the unit-cell parameters (normalised to ambient conditions values) of (a) the investigated (Ca,Th)-poor and (b) (Ca,Th)-enriched chernovite-(Y) samples and (c) of their respective normalised unit-cell volumes with the refined Birch-Murnaghan equations of state. Empty symbols refer to data collected in decompression.

Figure 5. High-pressure evolution of the unit-cell parameters (normalised to ambient-conditions values) of (a) xenotime-(Y) and (c) monazite-(Ce), (b) the normalised unit-cell volumes of the ambient-pressure and high-pressure polymorphs of xenotime-(Y) and (d) of the monoclinic β angle of monazite-(Ce).

Table 3. Refined equations of state parameters from the fit to the experimental high-pressure and high-temperature unit-cell volume data (see text for further details)

* From Pagliaro et al. (2022b).

α_V: volume thermal expansion coefficient at ambient conditions calculated on the basis of the Holland-Powell EoS.

a ₀ and a ₁ are two refinable variables in the Holland-Powell EoS (Holland and Powell, 1998).

The evolution of significant structural parameters (i.e., A–O bonds, A-coordination polyhedral and T-coordination polyhedral volumes) with pressure has been determined, based on the refined structure models, by using the VESTA3 software (Momma and Izumi, 2011). The corresponding values are reported in Table S4.

The results of this study confirm that arsenates are always more compressible than the isostructural phosphates (Table 3), in agreement with the observed relationship that the larger the ionic radii of the A and T sites, the higher the bulk compressibility (e.g. Zhang et al., 2008; Li et al., 2009; Lacomba-Perales et al., 2010; Errandonea et al., 2011). The refined bulk moduli apparently suggest that monazite-structure type minerals are more compressible than those with a zircon-structure type (Table 3). However, internally consistent theoretical data (Zhang et al., 2008; Li et al., 2009) show that there is a clear increase in compressibility along the lanthanoid series from Lu to La, with a discontinuity in the form of a stiffening when transforming from the zircon to the monazite structure type. Therefore, it can be concluded that, for the investigated minerals, the softening induced by the larger A site has a stronger impact than the stiffening induced by the monazite structure type.

The two minerals investigated, with the same zircon structure type, i.e. chernovite-(Y) and xenotime-(Y), undergo different phase transitions paths. Chernovite-(Y) experiences an irreversible transition, from a single-crystal to several crystal fragments, towards a scheelite-type polymorph between ca. 10.5 and 11 GPa. The same phase transition occurs, for synthetic YAsO₄ in a powder XRD experiment, in a broader pressure range between 8 and 12 GPa (Errandonea et al., 2011). The apparent discrepancy between these two studies may be ascribed to: (1) a different kinetics of the phase transition between a single crystal and a polycrystalline material; and (2) slightly different chemical compositions, as a higher phosphorous content in the presently investigated mineral decreases the average radius of the T site; or a combination of both. In general, the same phase transition has already been observed to occur in other ATO₄ compounds, where the pressure of transition increases with the decrease of the A and T atoms ionic radii (Wang et al., 2004; Zhang et al., 2009; Lacomba-Perales et al., 2010; Errandonea et al., 2011). The relationship between the ionic radii of the A and T atoms and the type of structure stable at ambient and high-pressure conditions is well known and was first described by Fukunaga and Yamaoka (1979) and Bastide (1987), and later discussed in other publications, e.g. in Lopez-Solano et al. (2010). As suggested by the Bastide diagram (Fig. 1) and already reported in the literature (Tatsi et al., 2008; Zhang et al., 2009; Lacomba-Perales et al., 2010; Musselman, 2017; Heuser et al., 2018), xenotime-(Y) and isomorphous phosphates undergo a single-crystal to single-crystal phase transition towards a high-pressure polymorph, xenotime-(Y)-II, showing a monazite-type structure. The transition observed in our experiments occurs at a pressure (> 17 GPa) consistent with those reported for synthetic YPO₄ compounds (Zhang et al., 2008; Lacomba-Perales et al., 2010). The reversibility of this phase transition is confirmed by this single-crystal study, even though with a significant hysteresis, as the tetragonal polymorph is recovered in decompression only between 6.3 and 1.3 GPa.

Despite a relative scattering between the published values of bulk compressibilities of zircon- and monazite-type phosphates and arsenates, determined on the basis of experimental and theoretical studies, the elastic parameters refined for the mineral species of this study (Table 3) are in agreement with those reported in the literature (Table S1). In the zircon-type minerals chernovite-(Y) and xenotime-(Y), the bulk compression is significantly anisotropic. Indeed, the structure is approximately two times more compressible within the (001) plane than along [001] (Figs 4 and 5, Table S5), which corresponds to the direction of the polyhedral chains evolution. Such a behaviour is strongly controlled by the high tetragonal symmetry, which limits the intra-chain deformation, as confirmed by the behaviour of the two independent A–O bond distances, where the A–O_a bonds, oriented parallel to [001] (Fig. 2), are the less compressible in both the minerals (Table S4). The bulk compression is therefore mainly accommodated by the AO₈ coordination polyhedron, as suggested by its bulk modulus, refined by modelling the polyhedral volume data obtained using Vesta3 with a II-BM equation of state by means of the EoS-Fit7-GUI software (Table 4). For both chernovite-(Y) and xenotime-(Y), bulk moduli values of the A-polyhedra are much lower than those obtained for the AsO₄ and PO₄ tetrahedra, which behave as quasi-rigid units (Table 4). The same conclusion can be drawn for monazite-(Ce) (Table 4) and gasparite-(Ce) (Pagliaro et al., 2022b), which, in addition, show that the AsO₄ tetrahedra are slightly more compressible than the PO₄ ones. As observed previously by Pagliaro et al. (2022a), the structural features (and responses to external T and P stimuli) of these compounds are strongly controlled by the crystal chemistry, in particular of the T sites. Moreover, not only do PO₄ and AsO₄ tetrahedra have a different compressional behaviour, but the nature of the prevailing T site controls the size (i.e. the average ionic radius) at ambient conditions of the A polyhedron (Pagliaro et al., 2022a). This leads, given a similar chemical composition of the A site, to slightly larger compressibility for the more ‘expanded’ A-polyhedra of the arsenates among isostructural minerals. The control exerted by the crystal chemistry on the elastic response is also pointed out by the chernovite sample relatively enriched in Ca and Th studied here, which shows a slightly lower bulk modulus value (Table 3). The lack of structure refinements prevents an unambiguous derivation of the structural mechanisms responsible for this behaviour, but it can be suggested that the more ‘expanded’ A-polyhedron (due to the higher content of the larger Th and Ca ions, as reported in Table 1) is coherent with the previous observations that a larger average ionic radius of the A site generates a softer polyhedron.

Table 4. Refined equation-of-state parameters pertaining to the A- and T-sites coordination polyhedra from the high-pressure (BM2 EoS) and high-temperature (Holland-Powell EoS, only A-site coordination polyhedron) experiments

* Due to the high uncertainties, these data should be considered as a qualitative estimation of the TO₄ units rigid behaviour

** From Pagliaro et al. (2022b)

The description of the elastic anisotropy in monazite-(Ce) is less straightforward, as the monoclinic symmetry does not allow us to rely on the axial compressibilities alone, given the variation of the β angle with pressure. Therefore, the finite Eulerian unit-strain tensor of monazite-(Ce) between ambient-P and 18.39 GPa has been calculated using the Win_Strain software (Angel, 2011) and with a geometric setting with X//a* and Y//b. The results show that the principal axes of maximum and minimum unit-strain do not correspond with any of the crystallographic axes, as described by the tensor values in the following matrix:

\begin{equation*}\left( {\begin{array}{*{20}{c}} {{\varepsilon _1}} \\ {{\varepsilon _2}} \\ {{\varepsilon _3}} \end{array}} \right)\left( {\begin{array}{*{20}{c}} {158.2\left( 2 \right)^\circ }&{90^\circ }&{56.2\left( 2 \right)^\circ } \\ {90^\circ }&{180^\circ }&{90^\circ } \\ {68.2\left( 2 \right)^\circ }&{90^\circ }&{33.8\left( 2 \right)^\circ } \end{array}} \right) \cdot \left( {\begin{array}{*{20}{c}} a \\ b \\ c \end{array}} \right)\end{equation*}

The analysis of the finite Eulerian unit-strain tensor allowed the determination of the mean compressibility values along the axes of the strain ellipsoid (with ε ₁>ε ₂>ε ₃): ε ₁ = 0.003030(11) GPa^–1, ε ₂ = 0.00200(2) GPa^–1, ε ₃ = 0.0014(12) GPa^–1. The directions of minimum and maximum compressibility are within the (010) plane and the anisotropic scheme is ε ₁:ε ₂:ε ₃=2.16:1.43:1. A comparison with the finite Eulerian unit-strain tensor reported for gasparite-(Ce) by Pagliaro et al. (2022b) points out that these isostructural minerals share a similar elastic anisotropy. The shared compressional behaviour extends to the structural deformation mechanisms acting on the atomic scale: the nine independent A–O bond distances in monazite-(Ce) have different compressional evolutions, as shown in Fig. 6 (see also Table S4). A moderate scattering of the monazite-(Ce) unit-cell parameters can be observed between 10 and 18 GPa, which implied that we should fit the V–P data using a Birch-Murnaghan equation of state truncated to the II-order (Table 3). On the basis of the available experimental data and structure refinements, it is not possible to unambiguously detect a change in the compressional behaviour at P < 18 GPa. On the contrary, a change in the response to compression clearly occurs at pressures exceeding ∼18.4 GPa, as evidenced by the significant deviation in the high-pressure evolution of the monoclinic β angle (Fig. 5; Table S3), similar to gasparite-(Ce) (Pagliaro et al., 2022b). This behaviour has already been reported for synthetic monazite-type LaPO₄ and CePO₄ (Huang et al., 2010), but not in another high-pressure investigation of CePO₄ (Errandonea et al., 2018). The structure refinements performed in this study allows us to draw a relationship between the change in the elastic behaviour and the structural re-configuration. Figure 6 and Table S4 show that, at lower pressures, the O_3c atom is too far from the A site to be considered to effectively belong to its coordination sphere (Fig. 3). However, the A–O_3c interatomic distance shows a significant shortening under compression (Fig. 6; Table S4), so that, at a pressure consistent with the change in the compressional behaviour, the O_3c atom is close enough to the A site to enter its coordination sphere, as suggested by the individual bond valences calculated for selected structure refinements and reported in Table S6. As a consequence, the coordination number of the A site would increase from 9 (8+1) to 10 (8+2), even though with different contributions from the individual bonds, given the longer bond distances of the of A–O2_a and A–O3_c (Tables S4 and S6). As the same behaviour has already been independently described for gasparite-(Ce) (Pagliaro et al., 2022b) and is analogous to what is described to occur at 3.25 GPa in the monazite-type crocoite (PbCrO₄, Bandiello et al., 2012; Errandonea and Kumar, 2014), we are inclined to believe that this is an intrinsic feature of the monazite structure type. Therefore, the known transition to a post-baryte-type polymorph at P higher than 26 GPa (Ruiz-Fuertes et al., 2016) is accomplished through an intermediate structural configuration, still preserving the monazite symmetry and atomic arrangement, but characterised by an increase from 9 to 10 in the number of the oxygen atoms bonded to the lanthanide-bearing site. This intermediate structural configuration is accomplished, with no clear discontinuity, by a smooth approach of the O3_c atom to the A site (where O3_c corresponds to O3_a and O3_b in the coordination sphere of two further A atoms). It is worth noting, that the same configuration, with a 10-fold coordinated A-site, is also shared by the monazite-type high-pressure polymorph of CaSO₄ (Crichton et al., 2005). The higher pressure at which the change of the elastic behaviour occurs in monazite-(Ce) (ca. 18.4 GPa, vs. ca. 15 GPa in gasparite-(Ce), Pagliaro et al., 2022b) is consistent with the higher P-stability expected for the isostructural phosphate with smaller average atomic radii.

Figure 6. (a) Bulk moduli as a function of the A-site atomic radii for several REETO₄ (T = As,P,V) minerals, after Li et al. (2009) and Zhang et al. (2008). (b) High-pressure evolution of the A–O interatomic bond distances in monazite-(Ce).

Despite being affected by larger experimental uncertainties, the same deformation mechanisms described for monazite-(Ce) and gasparite-(Ce) can be derived by the analysis of the refined structural models of the monazite-type high-pressure xenotime-(Y)-II polymorph (Table S4). In this case, the possible occurrence of a change in the compressional behaviour, according to the previous description, should be verified by investigations performed up to higher pressure values, though interatomic A–O3_c bond distances of ∼2.8–2.9 Å (Table S4) can suggest a 10-fold coordinated A site. A striking and anomalous feature shown by the experimental results of this study concerns the much larger refined compressibility if compared to what is reported in other studies for the same polymorph (K_V = 146(5) GPa in this study; K_V = 206 and 266 GPa in Zhang et al. (2009) and Lacomba-Perales et al. (2010), respectively). The relatively high number of experimental data of this study leads us to the conclusion that the refined bulk modulus value we obtained is very reliable, implying a similar compressibility between the ambient and high-pressure polymorphs of the investigated xenotime-(Y). It is worth mentioning that a similar compressibility can also be found between the zircon- and monazite-type polymorphs of LaVO₄, with K_{V 0} = 93(2) GPa (Yuan et al., 2015) and 95(5) GPa (Errandonea et al., 2016), respectively. The thorough re-investigation of the zircon-to-monazite phase transition in xenotime and of the elastic behaviour of the xenotime-II polymorphs appear, therefore, mandatory for a comprehensive understanding.

Thermal behaviour of the REETO₄ minerals

The thermal unit-cell parameters evolution from the three in situ single-crystal high-T ramps on monazite-(Ce), chernovite-(Y) and xenotime-(Y) and from the powder ramp on chernovite-(Y), relatively enriched in Ca and Th, are reported in Table S7 and shown in Fig. 7. The thermo-elastic behaviour has been modelled according to the isobaric equation of state modified from Pawley et al. (1996) and Holland and Powell (1998) and implemented in the EoS-Fit7_GUI (Gonzalez-Platas et al., 2016). The linear thermal expansion coefficients have also been refined using the TEV software (Langreiter and Kahlenberg, 2015). The refined parameters and the calculated thermal expansion coefficients at ambient conditions α_V = 1/V*(∂V/∂T)_P are reported in Table 3.

Figure 7. High-temperature evolution of the unit-cell parameters (normalised to ambient-conditions values) of (a) (Ca,Th)-poor and (b) (Ca,Th)-enriched chernovite-(Y), (c) xenotime-(Y) and (d) monazite-(Ce).

A comparison of the linear thermal expansion coefficients refined in this study for xenotime-(Y) and monazite-(Ce) and those already published for the same compounds reveal a good agreement with the literature data (see Table S2 and references therein). However, a significant discrepancy is observed between the values refined for chernovite-(Y) in this study and those reported previously by experimental investigations of YAsO₄ (Kahle, 1970; Schopper, 1972; Reddy et al., 1988). As none of the cited references provide experimental unit-cell parameters, nor structure refinements, it is not possible to discuss such a discrepancy, but it is worth noting that the refined values for both the (Ca,Th)-poor and (Ca,Th)-enriched chernovites of this study are self-consistent and diverge from the literature data. With this in mind, future investigation of other natural or synthetic YAsO₄-type zircon compounds seems imperative.

A comparative analysis of the thermo-elastic behaviour of the investigated minerals shows that monazite-(Ce) is much more expansible than the two zircon-type compounds (Table 3). In the latter, the [001] direction, i.e. the least compressible at high-P, is found to be the most expansible at high-T. The thermo-elastic anisotropy of monazite-(Ce) cannot be directly described based on the unit-cell parameters behaviour, given its monoclinic symmetry. The thermal expansion of monazite-(Ce) has been modelled with the TEV software (Langreiter and Kahlenberg, 2015) and at the temperature of 400°C (i.e. ΔT ∼ 380°C) its relationship with the unit-cell axes, with a geometric setting with X//a* and Y//b, is described by the following matrix:

\begin{equation*}\left( {\begin{array}{*{20}{c}} {{\alpha _1}} \\ {{\alpha _2}} \\ {{\alpha _3}} \end{array}} \right)\,\angle \,\left( {\begin{array}{*{20}{c}} {109.02^\circ }&{90^\circ }&{5.47^\circ } \\ {19.02^\circ }&{90^\circ }&{84.53^\circ } \\ {90^\circ }&{0^\circ }&{90^\circ } \end{array}} \right)\, \cdot \,\left( {\begin{array}{*{20}{c}} a \\ b \\ c \end{array}} \right)\,\end{equation*}

The mean thermal expansivity along the axes of the unit-strain ellipsoid, determined at 400°C, is: α₁ = 11.56·10^–6 K^–1, α₂ = 9.93·10^–6 K^–1 and α₃ = 7.39·10^–6 K^–1, leading to the anisotropic scheme α₁: α₂: α₃ = 1.56:1.34:1.

The structure refinements (thermal evolution of selected structural parameters is reported in Table S8) showed that for the zircon-type minerals xenotime-(Y) and chernovite-(Y), the coordination polyhedron hosting the lanthanides ions (A-polyhedron) has a paramount role in accommodating the bulk thermal expansion, its refined thermal expansivity being almost the double of the value referred for the unit-cell volume (Tables 3, 4), whereas the two independent A–O bond distances show a comparable behaviour, different to what is shown at high-P (Table S8). In monazite-(Ce) as well, the A-polyhedron plays a significant role in accommodating the thermal expansion, but of a lesser magnitude with respect to the tetragonal minerals (Tables 3, 4), given the larger degrees of freedom for structure deformation induced by the monoclinic symmetry. In all the cases, it is worth noting that the tetrahedra, being either PO₄ or AsO₄, appear to behave as rigid units in the temperature range investigated (Table 4).

As a general observation based on the experimental data of this study, xenotime-(Y) appears more expansible with temperature than chernovite-(Y). Even though the discrepancy between our data and the previously published thermal expansion behaviour of other YAsO₄ compounds (Kahle, 1970; Schopper, 1972; Reddy et al., 1988) suggests caution in this regards. However, our observation is consistent with the results reported by Li et al. (2009) for APO₄ and AAsO₄ (A = lanthanides) and based on theoretical calculations of lattice energies, where phosphates always show a larger expansibility than isostructural arsenates sharing the same A cation for both the zircon and monazite structural types.

Concluding remarks

The experimental data reported in this study, along with those published by Pagliaro et al. (2022b) on the high-pressure behaviour of gasparite-(Ce), provide a suite of thermo-elastic parameters for natural lanthanide-bearing phosphates and arsenates cropping out in the hydrothermal veins of the Mt. Cervandone.

Consistently with the previous scientific literature, the two zircon-type minerals undergo different P-induced phase transitions. Chernovite-(Y), above ∼10 GPa converts to a scheelite-type polymorph, whereas xenotime-(Y), above ∼17 GPa, transforms into a monazite-type polymorph by a reconstructive phase transition, single-crystal to single-crystal in character.

The monazite-type phosphates and arsenates do not undergo any phase transitions in the explored pressure range of this study. However, the analysis of the compressional evolution of the unit-cell parameters and of the crystal structures of monazite-(Ce) (this study) and gasparite-(Ce) from the same locality (Pagliaro et al., 2022b), highlighted a change in the compressional behaviour occurring at ∼18 and 15 GPa, respectively. This is related, for both the minerals, to a structural re-arrangement involving the smooth approach of a tenth oxygen atom into the coordination sphere of the lanthanide-bearing A site, which increases its coordination from 9 (8+1) to 10 (8+2) before the expected phase transition to a high-pressure polymorph. A general conclusion from both the present experiments and previous studies is that, despite the numerous papers publishing high-quality data on these crystalline compounds relevant to Earth and Materials sciences, there are still unexplored regions, whose physical-chemical features should be described by the adoption of up-to-date experimental techniques, facilities and crystallographic methods.

The analysis of the elastic and structural response with pressure and temperature of the investigated minerals of this study, combined with those on gasparite-(Ce) (Pagliaro et al., 2022b), lead us to conclude that their compressional and thermal behaviours are not, as commonly observed, induced by the same mechanisms opposite in sign. For example, phosphates are less compressible than the isostructural arsenates, but at high temperature they are more expansible. In the same way, the zircon-type structure is less compressible along the c crystallographic direction, but at high-T along the same direction the higher expansivity is shown. This behaviour can be explained by the significant control exerted on these compounds by the chemical strain. As described by Pagliaro et al. (2022a), the chemical nature of the T cations (P or As) has a paramount role in controlling not only the TO₄ tetrahedra, but also the other structural parameters: the substitution of P by As not only expands the tetrahedron, but the A coordination polyhedron too (at the same chemical composition of the A site). This study shows that the chemically more ‘expanded’ structures of the arsenates are more compressible than the chemically more ‘compressed’ structures of phosphates, but in response to a thermal perturbation they show the opposite behaviour being less expansible. Such an observation is confirmed by the slightly different behaviours of the two investigated chernovite samples. The sample relatively enriched in Th and Ca, due to the larger ionic radii of these cations with respect to the dominant Y in the A site, show a chemical expansion at ambient conditions (as described in Pagliaro et al., 2022a) that is reflected by a slightly larger compressibility at high pressure and a slightly lower expansivity at high temperature.