Crystal chemical significance of chemical zoning in dissakisite-(Ce)

Abstract

Dissakisites from Trimouns dolomite mine, France, have two kinds of single crystals: chemical-zoned and homogeneous types. Back-scattered electron microprobe (BSE) images of these dissakisites reveal both Ca–Al rich dark zones and Fe-ΣREE rich bright zones. Crystal structures of three dark and two bright zones in a chemical-zoned dissakisite and of a homogeneous zone in unzoned dissakisite were refined to individual R indices (about 3.0–5.0%) based on 1,400 observed [|F ₀| > 4σF ₀] reflections measured with MoKα X-radiation using the single crystal diffractometer. The differences in brightness between their BSE images arise from those in coupled substitutions of the elements occupying A2 and M3 sites. The main reason for these differences is that ten-coordinated A2 polyhedra and M3 octahedra are directly linked through their shared edge, which creates a great potential for making this coupled substitution. This zoning indicates that formation of the whole zoned crystal, where each zone could be grown steadily with its crystallographic axes mutually parallel to each other, may be identified as autoepitaxy.

Introduction

Dissakisite-(Ce) of an ideal formula CaCeMgAl₂Si₃O₁₂(OH)—a Mg analogue of allanite-(Ce) (e.g. Armbruster et al. 2006)—has been reported from some metamorphic occurrences (Grew et al. 1991; Enami and Zang 1988; Zakrzewski et al. 1992; De Parseval et al. 1997; Marty 2004). The crystal structure of dissakisite-(Ce) from the type locality (Grew et al. 1991) was first determined by Rouse and Peacor (1993), who found it to have the space group P2₁ /m, and demonstrated that Ca and lanthanide atoms were completely ordered in A1 and A2 sites, respectively. However, this structure refinement was not performed under compositional constraints; the chemical composition used for the final structural refinement is different from that determined by EMPA (Grew et al. 1991).

Though chemical zonings (the chemical zonations) have been generally known to reside in epidotes and allanites (e.g. Franz and Liebscher 2004; Catlos et al. 2000), the chemical zonings for dissakisite-(Ce), except for Cr-rich dissakisite-(Ce) (Yang and Enami 2003) and dissakisite-(La) (Tumiati et al. 2005), have been first observed by our present study of dissakisite-(Ce) from both Trimouns and the type locality, Balchen Mountain (Grew et al. 1991). Despite several occurrences of dissakisite, there are very few published studies on the compositional variation and crystal chemistry of this mineral.

The objectives of this paper are (1) to describe chemical compositions of each zone in dissakisite-(Ce), (2) to refine the crystal structures of their zones and (3) to explain both formation of their chemical zoning and its crystal chemistry.

Experimental methods

Sample description

The sample used in this study was dissakisite in dolomite lens from the talc-chlorite deposit of Trimouns in the French Pyrenees (De Parseval et al. 1997; Marty 2004). These dissakisites appear to be well-crystallised euhedral crystals, but have two kinds of colour: dark brown and transparent crystals (the former size: 0.5 × 0.8 × 3.0 – 1.0 × 1.5 × 5.0 mm, the latter size: 0.3 × 0.7 × 2.5 – 1.0 × 1.5 × 3.0 mm). These crystals were about 3 cm away from each other. For comparison, the crystals of light brown dissakisite-(Ce) (size: 0.05 × 1.5 – 0.07 × 2.5 mm) in the marble lenses in quartzo-feldspathic gneisses from Balchen Mountain, East Antarctica, provided by Grew et al. (1989, 1991), were analysed and are apparently equivalent to the type specimen first described by Grew et al. (1991).

Chemical composition

Chemical analyses were performed using the JEOL JAX-8621 electron microprobe equipped with three wavelength-dispersion spectrometers (WDS) at the Chemical Analysis Division, University of Tsukuba. The present dissakisites were embedded in an epoxy resin and well polished. Chemical zonings of the minerals were observed with ultra-high precision back-scattered electron (BSE) images (Fig. 1a–c). Qualitative analyses for the zones were performed with 25 kV accelerating potential and 250 nA beam current. Moreover, qualitative analyses of F for the examined zones were performed with 15 kV accelerating potential and 1000 nA beam current. However, no fluorine was detected in these zones, although Grew et al. (1991) reported F contents (0.18–0.27 wt%) of dissakisites from Balchen Mountain. Elements detected with qualitative analyses were measured by quantitative analyses. Quantitative analyses for major elements (Si, Ti, Al, Fe, Mg, Ca) of dissakisite were performed with an accelerating voltage of 25 kV, beam current of 10 nA, count times of 20 s and a beam diameter of 5 μm, while concentrations of REE were most effectively measured at 25 kV with a beam current 50 nA, count times of 20 s and a beam diameter of 5 μm. Because of the very close spacing of the X-ray lines in the L spectra of the REEs, accurate electron microprobe analysis (EMPA) of REE-bearing minerals is experimentally challenging (Reed and Buckley 1998; Nishida et al. 1999, 2003).

Fig. 1

Back-scattered electron images of dissakisite: a chemical zoning of dark brown dissakisite from Trimouns, b chemical homogeneity of light brown and transparent dissakisite from Trimouns, c chemical zoning of light brown dissakisite from Balchen Mountain. The crystals (a) and (c) show complex internal zoning, as documented in different shades of gray with straight boundary segments. The homogeneous core lies in the crystal of dissakisite from Trimouns (b), but the thin-rim fringed by chemical zoning was observed

The modified methods devised through careful measurement of these X-ray lines have improved the quality of the REE data collected by EMPA (Reed and Buckley 1998; Nishida et al. 1999, 2003). X-ray intensities of La, Ce and Nd were measured at the Lα₁ X-ray line, while those of Pr, Sm and Gd were measured at the Lβ₁ lines. Standards for REE were synthetic Ca–Al silicate glasses each containing REE, obtained from P&H Development Ltd. The chemical compositions of the standards are given in Table 1 of Hoshino et al. (2005). All data were corrected with a ZAF matrix-correction program. The above-mentioned method for REE analysis has been applied successfully (Hoshino et al. 2005, 2006). Representative chemical compositions of the chemical zones are presented in Table 1. The occurrences of the present dissakisites from Trimouns and Balchen Mountain indicate that they are not associated with Fe³⁺-bearing minerals (Grew et al. 1989, 1991; De Parseval et al. 1997; Marty 2004). Furthermore, assignment of all Fe to both Fe²⁺ and Fe³⁺ led to no reasonable chemical formulae for the present dissakisites. Therefore, chemical formulae of dissakisite-(Ce) were calculated on 12 O + 1 OH and all Fe were assumed to be Fe²⁺ (Grew et al. 1991; De Parseval et al. 1997).

Table 1 Electron microprobe analyses of dissakisites from Trimouns (Tri 1–5: chemical zoned, Tri 6: homogeneous) and Balchen Mountain (Bal 1–5: chemical zoned)

Crystal structure analysis

Using the stereomicroscope, each zone of the examined crystals from Trimouns on the BSE images (Fig. 1a, b) was completely scooped out from the polished samples with a hard metal needle with constant reference to the BSE photographs, carefully selected for the structure refinements (Fig. 1a, b). The chemical-zoned dissakisite from Balchen Mountain was too small to perform the structure analysis for a single crystal (Fig. 1c). Single-crystal X-ray-diffraction intensities for dissakisites from Trimouns were obtained with an Enraf-Nonius CAD4 automated diffractometer with graphite-monochromated MoKα radiation. For each zoned crystal, 25 reflections in the range 20° ≤ 2θ ≤ 30° were centred, and the unit-cell dimensions (Table 2) were refined from the resultant setting angles using the least-squares technique. Intensity data were collected in the ω-2θ scan-mode at an affixed scan-rate of 1.0° 2θ/min. The Psi-scan data were measured on five reflections at 10° increments above the diffraction vector, and an absorption correction was applied using the semi-empirical method (North et al. 1968). The data were corrected for Lorentz, polarisation and background effects, averaged and reduced to structure factors; about 1,400 unique reflections were observed [F ₀ > 4σ(F ₀)].

Table 2 Details on data collection and refinement of dissakisite from Trimouns

Results

Chemical composition

Both chemical-zoned and homogeneous dissakisites are distinctly observed on BSE images (Fig. 1a–c). The former crystals display a series of chemically distinct zones that reflect overgrowth of different types of dissakisite. Moreover, the chemical zonings in the examined dissakisites are characterised by straight boundary limits. Our present observation confirms that there are several chemical zones with straight boundary limits in all the seven crystals of dissakisite-(Ce) from Balchen Mountain provided by Grew et al. (1991). As an example, one of these seven crystals is shown in Fig. 1c; the chemical compositions were determined (Table 1). The present chemical zonings (Fig. 1a, c) certainly represent the magmatic zoning evidenced by marking several different levels of composition with straight boundary limits (Poitrasson 2002; Gieré and Sorensen 2004). In the BSE image of dissakisites from Trimouns, dark brown crystals that can be seen with the naked eye have chemical zoning while transparent ones are homogeneous (Fig. 1a, b).

The quantitative EMPA of each zone demonstrated that Ce predominates La and all other REE in the analysed dissakisites; therefore, it can safely be said that the minerals from Trimouns (Tri 1–6) and Balchen Mountain (Bal 1–5) correspond to dissakisite-(Ce) (Table 1). These Tri- and Bal-dissakisites are LREE-enriched with Ce₂O₃ up to 11.18 and 10.8 wt%, La₂O₃ up to 6.61 and 9.9 wt% and are abundant in other REE elements (Nd₂O₃ up to 6.60 and 2.62 wt%, Pr₂O₃ up to 1.86 and 1.18 wt%, Sm₂O₃ up to 1.20 and 0.35 wt%, Gd₂O₃ up to 0.70 and 1.30 wt%). The BSE images of the chemical-zoned crystals from Trimouns represent three dark zones (Tri 1–3) rich in Ca, Al and Mg and two bright ones (Tri 4,5) rich in Fe and ΣREE. In addition, BSE images of zoned crystals from Balchen Mountain show three dark zones (Bal 1–3) rich in Ca and Al and two bright ones (Bal 4,5) rich in Fe and ΣREE (Table 1; Fig. 1a–c). It is widely known that most allanites preferentially incorporate minor radioactive elements of Th and U and often occur in the metamict state (e.g., Deer et al. 1986). The present dissakisites from Trimouns and Balchen Mountain are not associated with radioactive minerals, and their ThO₂ contents are below the detection limit. Therefore, these crystals which have not suffered radiation damage are suitable samples for crystal structure refinement.

Structure refinement

The SHELXL-97 (Sheldrick 1997) program was used throughout this work. The structure of dissakisite from Trimouns was refined in the space group P2₁ /m, with the structural parameters provided by Rouse and Peacor (1993). Although the unweighted and weighted R factors supplied by Rouse and Peacor (1993) are 2.4 and 5.7%, respectively, their structural data are almost consistent with those in this study (Tables 3, 4 and 5). In the final stage of refinement, a three-dimensional difference-Fourier map was calculated, but the hydrogen atom position could not be determined.

Table 3 Atomic coordinates and anisotropic displacement parameters (Ų) for Tri 1

Table 4 Selected interatomic distances (Å) for Tri 1

Table 5 Bond-valence calculation for Tri 1

Refinements have been carried out with Si assigned to only tetrahedral sites, because it was routinely fixed on three tetrahedral sites of epidote group minerals (Peacor and Dunn 1988; Rouse and Peacor 1993). Form factors for Ca and (ΣREE + Ca) were applied, where ΣREE correspond to La, Ce, Pr, Nd, Sm and Gd. On the basis of the representative EMPA values (Table 1), occupancies of Ca in A1 site and (ΣREE + Ca) in A2 site were assigned to each site (Table 6). Form factors for M1, M2 and M3 were modelled with Al³⁺, Al³⁺ and (Al³⁺ + Fe²⁺ + Mg²⁺) respectively, based on the considerations of charge balance, similar refinements for other epidote group minerals (e.g., Dollase 1971; Kvick et al. 1988; Rouse and Peacor 1993), and known interatomic distances in dissakisite (Rouse and Peacor 1993). In particular, A1³⁺ has a strong preference for M2 in epidote structures, whereas, clarifying any ambiguity in the distribution of other cations through refinement is difficult because of the similarity in respective form factors of Al³⁺ and Mg²⁺. Therefore, the occupancies of M1 and M3 sites were assigned on the basis of the representative EMPA values (Tables 1 and 6). The six zones of Tri 1–Tri 6 did not differ greatly in the refined structure data from one another. Therefore, the structural data for Tri 1 (minimum brightness) are given as follows: final atomic coordinates and anisotropic displacement parameters are presented in Table 3; selected interatomic distances are provided in Table 4 and bond valences analysis is given in Table 5. Assigned and refined site occupancies of dissakisites from Trimouns are shown in Table 6. The present dissakisites incorporate some Fe²⁺ into A2 site (Table 6). Based on electron-microprobe analytical data of allanite, Catlos et al. (2000) have suggested that some Fe²⁺ has been incorporated into the A sites, potentially substituted for Ca²⁺. The incorporations of Fe²⁺ into A2 site were structurally confirmed by structural refinements for the present dissakisites. However the present chemical-zoned dissakisites from Balchen Mountain (Bal 1–5) were too small to perform structure analysis for a single crystal. Therefore, the site occupancies for dissakisite from Balchen Mountain (Bal 1–5) were correspondingly assigned on the basis of the representative EMPA values, although Si = 3 were assumed (Table 6). Data for atomic coordinates, anisotropic displacement parameters, selected interatomic distances and bond valence analysis of Tri 2, Tri 3, Tri 4, Tri 5 and Tri 6 can be obtained from the first author (M. Hoshino).

Table 6 Assigned and refined site occupancies for dissakisites from Trimouns and occupants assigned on a basis of representative EMPA values of those from Balchen Mountain

Discussion

Crystal chemistry of dissakisite-(Ce)

Chemical zoning of the present dissakisites (Fig. 1a, c), marking several different levels of composition with straight limits, certainly represents magmatic zoning (Poitrasson 2002; Gieré and Sorensen 2004). All of the seven examined dissakisite crystals (see Fig. 1c for one of the seven crystals) from Balchen Mountain have tiny chemical zones with straight boundary limits, which were not described by Grew et al. (1991). Thus, it is difficult to collect a chemically homogeneous single crystal of dissakisite-(Ce) from Balchen Mountain available for crystal structure analysis (Rouse and Peacor 1993) because the crystals show complex chemical zoning in the order of 10 μm (Fig. 1c). Furthermore, the chemical composition dictated by the structural refinement of dissakisite from Balchen Mountain (Rouse and Peacor 1993) differs considerably from that analysed by EMPA (Grew et al. 1991). It is crystallo-chemically significant that the crystal structure of the present dissakisite from Trimouns was refined under the constraints of accurate chemical composition by EMPA (Tables 1 and 6). The crystal structure of dissakisite-(La) has been determined by Lavina et al. (2006). Its lattice constants (a = 8.959 Å, b = 5.7226 Å, c = 10.232 Å, β = 115.19° for untreated DISS3) are larger than those of dissakisite-(Ce): a = 8.890–8.900 Å, b = 5.6716–5.6938 Å, c = 10.090–10.097 Å, β = 114.705–114.838° for Tri 1–Tri 6 in this study and a = 8.905 Å, b = 5.684 Å, c = 10.113 Å, β = 114.62° in Rouse and Peacor (1993). The expansion of lattice constants for dissakisite-(La) is interpreted to be caused by metamictisation (Lavina et al. 2006); atomic displacement parameters of A1, A2, O8 and O9 are very large for metamict dissakisite-(La). Although atomic displacement parameters for A1 and A2 sites of metamict dissakisite-(La) are nearly equal to those of the present dissakisite-(Ce) from Trimouns, atomic displacement parameters for O8 and O9 for the former dissakisite are significantly larger than those for the latter. Thus, this difference suggests that the present dissakisite-(Ce) from Trimouns is non-metamict.

The bond valence sum was calculated to establish the credibility of the refined crystal structure. The calculation was done from the process parameters provided by Brese and O’Keeffe (1991). Bond valences of O4 and O10 are undersaturated in comparison with the ideal bond valence, 2.00, for O atom (Table 5). The bond valence sums for O4 and O10 are 1.63–1.68 and 1.28–1.32 v.u., respectively, implying that the O10 atom is the donor oxygen for the H atom (Table 5). Although H atom position was not determined on the difference-Fourier map, inspection of the bond valence sums reveals that the H atom is uniquely adjacent to the O10 atom and involved in a hydrogen bond with O4. From the viewpoint of bond valence sums (Brown 1976; Hawthorne 1992), the chemical bonds between O10 and H in the present dissakisites are consistent with those in ferriallanite-(Ce) (Kartashov et al. 2002) and allanite-(Ce) (Hoshino et al. 2005).

Mechanism of chemical zoning in dissakisite-(Ce)

The charge-coupled substitution of REE³⁺ and Mg²⁺ for Ca²⁺ and Al³⁺ respectively relates dissakisite to clinozoisite (Gieré and Sorensen 2004) (Fig. 2a), and one of the most noticeable differences between these two minerals is that the ideal dissakisite contains only Mg²⁺ in the M3 site, whereas the endmember clinozoisite contains Al³⁺ (Ercit 2002). The differences in brightness between BSE images of dissakisites from Trimouns and Balchen Mountain cannot be explained by simple coupled substitutions, but by the following ones (Fig. 2b, c): in Trimouns dissakisite Ca²⁺(A2) + (Mg²⁺ + Al³⁺)(M3) ↔ ΣREE³⁺(A2) + 2Fe²⁺(M3), where the former creates darker BSE images than the latter; moreover, in Balchen Mountain dissakisite Ca²⁺(A2) + Al³⁺(M1) ↔ ΣREE³⁺(A2) + Fe²⁺(M1), the former presents darker BSE images than the latter (Table 6; Fig. 2b, c). A crucial difference in the occupancy site of those elements participating in the above coupled substitutions has arisen between A2 and M3 sites in Trimouns dissakisite and between A2 and M1 ones in Balchen Mountain (Table 6). Furthermore, close observation of these zones under the polarising microscope reveals that they have simultaneous extinction with one another (Fig. 1a, c). Both chemical zoning of the present dissakisite-(Ce) and the simultaneous extinction of these zones with one another demonstrate that formation of the whole zoned crystal, grown jointly with mutually parallel axes in the open system, may be identified as autoepitaxy (Bonev 1972).

Fig. 2

Diagrams illustrating the potential combinations of elements for several zones of dissakisites from Trimouns and Balchen Mountain. The present dissakisite are related to clinozoisite via the coupled substitution. a Ca²⁺ + Al³⁺ ↔ REE³⁺ + Mg²⁺, reported by Gieré and Sorensen (2004). The coupled substitutions of elements of dissakisite from Trimouns and Balchen Mountain plotted are b Ca²⁺(A₂) + (Mg²⁺ + Al³⁺) (M3) ↔ REE³⁺(A2) + 2Fe²⁺(M3) (refined by the structure analysis) and c Ca²⁺(A2) + Al³⁺(M1) ↔ REE³⁺(A2) + Fe²⁺(M1) (assigned on a basis of the representative EMPA values). Units are atoms per formula unit (a.p.f.u.)

The variation in each crystal structure of zoned dissakisite-(Ce) from Trimouns may be checked by comparing the volumes of two A-, three M- and three Si-polyhedra, calculated using the IVTON computer program (Balić-Žunić and Vickovic 1996) (Table 7). As a result, the volumes of A1, (M1, M2) and (Si1–Si3) sites occupied by Ca, Al and Si cations, respectively, were similar among six refined crystals. On the other hand, the sum of volumes of A2 and M3 sites occupied by (ΣREE and Ca) and (Mg, Al and Fe) atoms, respectively, bears a linear relationship to the length of the b axis, albeit unchangeable against the a and c axes (Fig. 3). Lattice parameters of dissakisite-(La) in the Ulten zone are plotted far away from those of the present dissakisite-(Ce) in Trimouns (Fig. 3), probably because of the expansion of lattice parameters caused by strong metamictisation (Lavina et al. 2006). Moreover, lattice parameters of dissakisite-(Ce) from Balchen Mountain determined by Rouse and Peacor (1993) are also plotted apart from those of the present dissakisite-(Ce) (Fig. 3) because the values of A2 and M3 site-occupancies refined to the former dissakisite-(Ce) (Rouse and Peacor 1993) differ from the EMPA ones obtained by the present study (Tables 1 and 6). Although Bonazzi and Menchetti (1995) showed that the presence of REE³⁺ decreases the unit-cell parameters, c and β, of epidote group minerals and increases both b values of their unit cells and volumes (see Fig. 2 in Gieré and Sorensen (2004)), these relationships have not been observed in the present dissakisites. Higher ΣREE and Fe²⁺ contents at A2 and M3 sites in a bright zone of the present dissakisite-(Ce) from Trimouns promote expansion of the corresponding polyhedra because of the coupled substitution of Ca²⁺(A2) + (Mg²⁺ + Al³⁺)(M3) ↔ ΣREE³⁺(A2) + 2Fe²⁺(M3): sum of A2 and M3 polyhedral volumes in the bright zone, occupied by more enriched ΣREE and Fe²⁺, are larger than those in the dark zone (Table 7). Furthermore, the volume expansion of A2 and M3 polyhedra exerts a potent influence on the b axis of dissakisite-(Ce) (Fig. 3): ten-coordinated A2 polyhedra share an edge with M3 octahedra (Fig. 4), and so the cations occupying the other polyhedra adjacent to these two polyhedra can easily mediate the charge difference in the above coupled substitution between A2 and M3 polyhedral cations. Therefore, this type of chemical zoning may be promoted by the wide diversity of chemical composition, even in a single crystal in thermodynamic equilibrium. Great tolerance of the dissakisite-(Ce) structure for this diversity can bring on autoepitaxy in the present chemical zoning.

Table 7 Volumes (ų) of coordination polyhedra for dissakisites

Fig. 3

Lattice parameters versus the sum of the volumes of A2 and M3 sites. Volumes of the polyhedra are calculated by IVTON program (Balič-Žunič and Vickovic 1996)

Fig. 4

Projection of crystal structure of dissakisite. Each ten-coordinated A2 polyhedron shares one edge with M3 octahedra. Moreover, the A2 polyhedra form linear single chains with one bridging oxygen atoms per polyhedron, resulting in lengthening the b axis