The high-pressure monazite-to-scheelite transformation in CaSeO4

Abstract The high-pressure monazite – scheelite structure transition has been observed at P >4.57 GPa in CaSeO4 by synchrotron X-ray powder diffraction. It is a first-order transition with a 4.5% volume change and is severely hindered kinetically. Scheelite-type CaSeO4 remains to a maximum experimental pressure of 42.2 GPa and no (002) reflection, specifically indicative of a subgroup transition to a fergusonite-type structure, is observed. Scheelite-type CaSeO4 remains at ambient conditions, where the tetragonal unit cell has parameters of a = 5.04801(11) c = 11.6644(5) Å and V = 297.21(3) Å3 with D calc = 4.090 g cm–3. The diffraction pattern of the recovered material was refined in space group I41/a to R p = 0.98%, w R p = 1.91%, GoF = 0.59, R Fobs = 5.04%, w R Fobs = 4.27%. The oxygen is located on the general 16f site at (0.2578(8) 0.3699(14) 0.5755(4)) and shares four identical bonds with Se (4a: ½ ½ ½) at 1.644(5) Å. The Ca (4b: 0, 0, ½) is eight-coordinated via O at 4 × 2.440(6) Å and 4 × 2.504(5) Å. This is further evidence of the dissimilarity of sulfate and selenate at high pressure and temperature conditions and the closer resemblance of the selenates to the orthophosphates, arsenates and vanadates, where this type of transition sequence has been described.


Introduction
IN the last decade considerable efforts have been made to discover new structural associations, properties and non-ambient behaviours of monazite-and scheelite-type compounds due to their technological utility. Scheelite, CaWO 4 (e.g. Grzechnik et al., 2003), is a tetragonal 8À4 coordinated mineral found in hydrothermal veins and skarns, where it is commonly associated with wolframite and other large cation refractory minerals. The scheelite structure is shared by relatively few minerals; examples include stolzite, PbWO 4 , wulfenite, PbMoO 4 , powellite, CaMoO 4 and reidite, a high pressure ZrSiO 4 (zircon) polymorph that is found in impact breccias (Reid and Ringwood, 1969;Liu, 1979;Marques et al., 2009). Raspite is a polymorph of stolzite, but it is structurally unique with no other composition sharing its topology in any conditions (Fujita et al., 1977). The scheelite structure is widely reported in synthetic compounds due to its ability to form compounds with valence variations from A 1+ B 7+ O 4 to A 4+ B 4+ O 4 . Oxides crystallizing with B = W, Mo, V, Nb, Cr, Re, I, Ge and Si have been described (e.g. Liu and Bassett, 1986) and other stoichiometries are known for chloride, fluoride and hydride systems.
In mineralogy the term monazite is used to describe a group of monoclinic lanthanide orthophosphates, Levinson modifiers are used identify the individual members of the group, so that in monazite-(La), for example, lanthanum is the dominant element at the appropriate structural site. A wide variety of minerals with the monazite structure have been described; they include arsenates and chromates with large cations such as gasparite, (Ce,REE)AsO 4 , rooseveltite, BiAsO 4 , and crocoite, PbCrO 4 . Monazite from placers and other deposits, has the potential to be a significant primary ore of rare-earth elements and thorium, though this strategic industrial process is not yet well developed on a worldwide scale. Monazites can be transformed to other structure types, or decomposed into their constituent components under appropriate conditions. A thorough treatise covering the crystal chemistry of most monazites, and transitions to and from them, can be found in Clavier et al. (2011).
It has been shown experimentally that CaSO 4 undergoes a transition from its anhydrite form to the monazite structure at high pressure (Stephens, 1964;Borg and Smith, 1975). Crichton et al. (2005) reported that monazite-type CaSeO 4 remained stable at pressures up to 11.8 GPa (though they made no claim to observe the transition at this pressure, deferring to the work of Borg and Smith (1975) for a transition pressure of above 2 GPa; see also Gracia et al., 2012;Ma et al., 2007). It has also been shown that CaSO 4 transforms to a baryte-type structure at high pressure and temperature  and this has been investigated by density functional theory (DFT) calculations (Gracia et al., 2012) and largely reproduces the experimental observations, without locating the distorted baryte, AgMnO 4 -type, intermediate. The monoclinic distortion in this structure is subtle but may nonetheless be expected from similar behaviour in perchlorate and permanganate systems, as identified by Pistorius et al. (1969). This example of CaSO 4 achieving a baryte-type structure serves as another success of the oft-used rule of thumb of going down a group in the periodic table (Ca to Ba), or increasing 'chemical pressure', as a proxy for extrinsic compression. Barium sulfate has been studied several times recently. It has been shown that baryte undergoes at least one phase transition to a structure, or structures, previously reported as tentatively triclinic (Lee et al., 2001) and orthorhombic P2 1 2 1 2 1 (Santamaría-Pérez et al., 2011) at a range of pressures (13 GPa to 27 GPa; 32 GPa by calculation, Santamaría-Pérez et al., 2011) which appear related to the state of hydrostaticity of the measurements. From the experimental work carried out under the most favourable hydrostatic environments, i.e. in helium, the transition must be located at pressures above 21.8 GPa, as it was not observed at the maximum pressure reached by the single-crystal study reported by Crichton et al. (2011). Subsequent powder data collections in helium suggest a transition pressure of 27 GPa (Santamaría-Perez et al., 2011). It is unclear whether the reported triclinic phase and the postbaryte P2 1 2 1 2 1 structure are the same. The demonstration that CaSeO 4 also forms a monazite, via rhabdophane-like intermediates, upon complete loss of water from its gypsum-type dihydrate is available in these pages (Crichton et al., 2010) and lends further credence to the structural similarity between sulfates and selenates. Indeed, CaSeO 4 is in some regard an intermediate to the anhydrite form of the sulfate and the baryte structure occurring at high pressure in CaSO 4 , or, at ambient pressure, in BaSO 4 . However, despite the fact that this same selenate form has been previously described (with caveats) as an orthorhombic P2 1 2 1 2 1 structure, it should not be confused with, and a CaSeO 4 -type designation should not be applied to, an orthorhombic post-baryte form of BaSO 4 , CeVO 4 or CaSO 4 with the same P2 1 2 1 2 1 symmetry; they are not the same (see also Santamaría-Pérez et al., 2011;Errandonea et al., 2011;Gracia et al., 2012).
High-pressure and/or high-temperature relations in various chemistries dictate that scheelites are most commonly formed from zircon-and monazite-type structures and that these may themselves transform to the fergusonite, YNbO 4type structure (Grzechnik et al., 2003 or decompose to their constituents, possibly related to underlying phase transitions in these component phases (Liu, 1979;Grzechnik et al., 2004;Crichton et al., 2009). A discrete post-fergusonite phase is not described. Scheelite-type structure have not been described for sulfates or selenates despite the facts that monazite forms are known for both, and that scheelite is a common postmonazite transition product (Snyman and Pistorius, 1963;Stephens, 1964;Borg andSmith, 1975 Crichton et al., 2005;Bradbury and Williams, 2009;Crichton et al., 2010;Clavier et al., 2011). A summary of structure-types for related chemistries can be found in Table 1. Tellurates form diverse structures of e.g. CaUO 4 -type (Hottentot and Loopstra, 1979), which do not form part of this extended general Where no distinct phase is identified, it is noted as n.d.

{
For trivalent M 3+ sites and pentavalent tetrahedral sites.

{
For tetravalent M 4+ sites and tetravalent tetrahedral sites.
scheme, as they are higher coordinated 6À8 structures and therefore lack the tetrahedral sulfate and selenate groups that define these and related topologies. What then for CaSeO 4 at high load: chemical pressure via substitution of Se for S and reducing the pressure to reach a high-pressure baryte-type structure? For CaSO 4 , BaSO 4 and BaSeO 4 all form baryte and BaSeO 4 is also known as a monazite. Or, structural similitude with the other monazites and obtain a scheelite, for the monazite-scheelite transition is also described, see Clavier et al. (2011) for examples. It is unclear from systematic behaviour, now most commonly expressed in terms of the variations upon the diagrams first produced by Muller et al. (1969) (who would have CaSeO 4 as a borderline monazite-zircon), Bastide (1987) (who has similar, and scheelite at high P), or Fukunaga and Yamaoka (1979) (who would have baryte as a post-monazite), exactly what should happen. One clue is given by the only other mention of CaSeO 4 as a scheelite in the literature, as the star-graded JCPDS card number 36-0293. This card states that the compound was formed through evaporation of a solution of CaCO 3 in selenic acid. This method produced a monazite in our case: it is equivalent to 'Procedure B' of Crichton et al. (2010) and may indicate that the monaziteÀscheelite transition occurs at high temperature (through forced evaporation), and as scheelite is invariably denser than monazite, this would imply a negative Clapeyron slope. Therefore, we should cross this same transition as the pressure is increased.
To this end we have undertaken measurements to determine the post-monazite form crystallizing at high pressure in CaSeO 4 . We describe the observed transition and detail structural features of the recovered sample at ambient conditions for direct comparison of both monazite and scheelite forms under the same conditions.

Experimental
The starting material was prepared from a batch of synthetic gypsum-type CaSeO 4 .2H 2 O by heating overnight in air at 600ºC. This produced a fine powder, which was checked to be pure monazite-type CaSeO 4 on a Miniflex-II lab diffractometer. The sample was subsequently loaded, along with a ruby for pressure measurement (Forman et al, 1972), in a membrane-driven diamond-anvil cell equipped with 300/250 mm culet diamonds and a steel gasket with an initial hole diameter of 150 mm and thickness of 36 mm. The pressure-transmitting medium was helium. High pressure measurements were made at beamline ID09A at the ESRF, using a focussed X-ray beam from a bent Laue monochromator operating at 30 keV. Two-dimensional diffraction patterns were collected on a mar555 flat-panel detector at a distance of 310 mm from the sample and were corrected and integrated to conventional one-dimensional 2y-intensity datasets using fit2d software (Hammersley et al., 1996), after integration parameters were established using a silicon powder standard, Hammersley et al. (1994). Diffraction datasets were fitted using Jana2006, Petricek et al. (2006).

Measurements at high pressure
The results of refinements of the monazite-type phase up to 4.2 GPa are shown in Fig. 1. The unit cell compresses smoothly, with the b axis (green) and c axis (blue) showing identical compressibility and the a axis (red) being fractionally more compressible. These can be quantified as axial moduli after fitting an equation-of-state (EoS) formulism to the cube of the cell-length measurements and this results in K a = 57.1(2) GPa, K b = 65.2(12) GPa and K c = 67.8(10) GPa. The b angle reduces linearly at a rate of À7.60(3)10 À4 º GPa À1 . A similar fit of a second-order BirchÀMurnaghan EoS to the volumetric data results in V 0 /Z = 78.328(5) Å 3 and K T = 69.1(7) GPa. This value is approximately 60% of the bulk modulus of monazite, CePO 4 , at 109 GPa (Huang et al., 2010), which is one third more dense with D = 5.23 g cm À3 .
Comparing this with models of predicted bulk moduli (e.g. Hazen and Finger, 1982), using the average CaÀO bond length at ambient of 2.513 Å , a bulk modulus of~77 GPa is obtained, which is no better than a fair approximation.
Upon further increase in pressure, Fig. 2, we observe the growth of a small peak (at 5.2º, asterisked in Fig. 2). It was not, however, until several GPa (and hours) later that the peaks became sufficiently distinct from the monazite to produce definite data on the nature of this transition and even at the highest pressure, monazite-type peaks were still observed (though they could not be identified without prior information). This sluggish transition is typical of these oxide materials, though admittedly this is the most extreme case we have encountered in our exploration of these materials to date. We point out that this situation would probably be improved by heating; however, this is unavailable to a Heloaded cell. As we increased the pressure further, up to 42 GPa, we were particular to the observation of a weak low-angle peak that would correspond to the extinct (002) of scheelite, which is indicative of the transition to the subgroup (I2/a) fergusonite phase (e.g. Grzechnik et al., 2003). This test proved negative and we decompressed to ambient and recovered the sample for further comparisons with monazite at room conditions. A Le Bail fit (Le Bail et al., 1988) of the recovered-to-ambient sample converged to R p = 4.73%, wR p = 8.47%, with a GoF = 0.94 using 4243 reflections, 12 parameters (5 profile, 5 background, a and c) and one constraint (b = a) in P1 with metric tetragonal cell parameters. It was evident from this fit that some monazite phase was also present. Nevertheless, as a test of data robustness, the symmetry was assessed (I-centred lattices and the I4 1 /a space group offered little degradation of the quality of fit parameters, and had the highest ratio of reflections extinct/ generated of about Ý) and the structure was solved using the charge-flipping method, with the input of the chemistry and Z = 4 into an otherwise default Superflip run (Palatinus and Chapuis, 2007). The Expo suite of programs (Altomare et al., 1999) was also used to solve the structure directly from the same set of extracted intensities, with a final refined R = 6.07% (groups/parameters = 5.50). This was used as an initial model for a The onset of the transition can be seen by the appearance of the small peak at~5.2º, which is asterisked. The small peaks visible at higher angles are due to He, the pressure-transmitting medium, and are marked 'He'.
Rietveld refinement, which converged rapidly (Fig. 3) and produced the parameters listed in Tables 2 to 5 and the structure shown in Fig. 4.

Discussion
This is the fourth Ca-containing oxide scheelite reported after the tungstate, scheelite proper   Table 1. We would then propose that there should be significant opportunity for solid solution in mixed Ca(Cr,Se)O 4 chemistries and that, furthermore, Se might be used to stabilize a Cr-rich monazite and, conversely, Cr may stabilize a Se-containing zircon. Such compositions remain to be described. Given the list of similarities between these structures and chemistries, it becomes clear that it is in polyhedral contacts, in the packing of these subunits, where any contrast between the monazites and the scheelites must be significant. In the scheelite structure, each Ca-polyhedron edge-shares to its four adjacent and symmetrically identical units and each Se-tetrahedron is fully corner-shared to these. In the monazite structure there is only edge-sharing from one CaO 8 polyhedron to its adjacent CaO 8 unit and a SeO 4 tetrahedron; the remaining liaisons are cornershared via a bridging apical oxygen. The space occupied by the polyhedral units in the unit cell The refinement parameters are as follows: R obs = 5.04%, wR obs = 4.27%, R all = 5.46%, wR all = 4.38%. The profile contained 4306 reflections, with a GoF = 0.59, an R p = 0.98% and wR p = 1.91%. The number of parameters refined were the above (eight) plus 15 background, five profile and overall scale. The minor monazite content was ignored, but not excluded and the Compton background from diamonds was not subtracted; y min was set to 0. This represents a reduction of void space of~4%. Indeed, the volume change upon transition was estimated from the degree of volumetric compression (Fig. 5) at the onset pressure of 4.57 GPa to be~4.5%. Refinement of the structures was not prudent in the strongly mixed region, except from a few points obtained on decompression (those at less than 20 GPa). Nonetheless, the extracted unit-cell values of the scheelite cell exhibit smooth compression. Fitting of a third-order EoS to the axial and volumetric data highlights a 45% difference in the compressibility of the a and c axes, at K a = 97.6(12) GPa, K a ' = 5.38(13) and K c = 67.2(17) GPa, K c ' = 4.22(17). Comparison of volumetric data of this scheelite, where fitted V 0 /Z = 74.3103(13) Å 3 , K T = 84.2(5) GPa and K' = 5.00(5), with others shows that it is most similar to CaMoO 4 powellite, at 82.5(7) GPa , which also has the more comparable average CaÀO distance of any scheelite at 2.458 Å , cf.~2.472 Å here. Both these values are higher than that obtained for the tungstate scheelite, where K T = 74(6) GPa and K' = 3.89(10), Grzechnik et al. (2003), though again the average bond length is similar, CaÀO8 = 2.459 Å . These values are somewhat surprisingly at odds with those for scheelite-type CaCrO 4 , where K T has been estimated at 125.1Ô6.2 GPa by Long et al. (2006). Therefore, CaSeO 4 displays a transition sequence that can be described as closely resembling those of the orthophosphates. We have previously shown that CaSeO 4 has rhabdophane-like intermediates on partial dehydration, on full dehydration it is a monazite and at high pressure it forms a scheelite. Calcium sulfate also shows both rhabdophane-like structures (g-CaSO 4 , bassanite) and monazite forms; i.e. structures with topologies that are common to natural orthophosphates; but, in contrast, at high load it reverts to the baryte-type structure. Consequently, the monaziteÀscheelite rule-of-thumb breaks down but the anhydriteÀbaryte does not, indicating that sulfates (selenates) are particularly sensitive to our description in systematic trends. Other than those structures mentioned above this same monazitescheelite transition sequence might be expected for both the large divalent cation selenates SrSeO 4 and PbSeO 4 , which are monazites (Effenberger and Pertlik, 1986), especially when Sr and Pb tungstates and molybdates are already described as scheelites. However, despite the similarities of the selenates with CaCrO 4, SrCrO 4 and PbCrO 4 (Effenberger and Pertlik, 1986;Pistorius and Pistorius, 1962), only full structural descriptions exist for scheelite-type chromates with M 3+ lanthanide chemistries (i.e. post monazites, with nominal pentavalent, rather than hexavalent Cr, as above). A similar situation occurs for the smaller transition metal selenates that crystallize in the CrVO 4 -type structure, also known as NiSO 4 -type, when M = Mg, Ni, Cu, Cr; as, only nominal V 5+ scheelites are described, and it therefore seems unlikely that these will produce scheelite-type structures at high load. Other transition-metal selenates crystallize in the CuSO 4 -type (also known as ZnSO 4 -type) and these are proto-CrVO 4 structures. The almost ubiquitous prevalence for pentavalent chemistries in scheelite-type Cr-and V-containing systems might therefore be a limiting condition for further exploration in that direction, in spite of the similarity between the structures of CaCrO 4 and CaSeO 4 .

Conclusion
We have compressed calcium selenate from its monazite structure up to and through the kinetically limited transition to the scheelite form. The recovered high pressure phase is identified as of scheelite type and the density difference is some 4.5% at the onset of the transition and 5% greater than that of monazite at 3.880 g cm À3 in standard conditions. The cell parameters obtained here are identical to those of JCPDS 36-0293 to 1:10k, confirming that scheelite-type CaSeO 4 can be formed by both wet chemical and high-pressure routes.