Introduction

The ongoing transition to green energy sources has increased demand for minerals containing rare-earth elements (REEs), catalyzing both exploration and evaluation of byproduct REE recovery from existing mining-processing operations. Members of the bastnäsite [REE(CO3)F]–synchysite [CaREE(CO3)2F] group (BSG) are common REE-fluorocarbonate minerals that share crystal-structural elements (Figure 1) explaining their syntaxial growth. All structures have REE-F and CO3 layers, whereas an additional CaCO3 layer occurs in all mineral species, except bastnäsite.1,2,3 The first transmission electron microscopy (TEM) study of BSG minerals2 reports additional members and shows that all phases form a series of mixed-layer compounds, BmSn, (B, S = bastnäsite and synchysite slabs, where m and n are integers) with hexagonal/rhombohedral symmetry in which compositional variation is linked to modulation along the c* axis. Modulation is formalized as Nsat = [(m × 2) + (n × 4)] − 1 based on integration of prior and new data obtained by scanning TEM (STEM),4 where Nsat is the number of satellite diffraction maxima for all BmSn compounds with S ≠ 0. Later TEM studies have documented other polysomes and polytypes with rhombohedral or hexagonal symmetry in the bastnäsite–parisite compositional range.5,6,7,8 In contrast, single-crystal data for synchysite and parisite indicate monoclinic symmetry required to fit CO3 groups into an otherwise hexagonally stacked framework of REE-F and Ca layers.9,10 Monoclinic symmetry is also favored in recent TEM studies of parisite and synchysite with a comparison to mica polytypes.11,12

Figure 1
figure 1

Models for (a) bastnäsite and (b) synchysite on two different orientations (zone axes in square brackets) obtained using crystallographic information from Ni et al.3 and Wang et al.9 These complement models given previously by Donnay and Donnay1 and Van Landuyt and Amelinckx,2 as well as Ciobanu et al.4 but are distinct from those presented by Capitani11,12 based on data obtained by conventional transmission electron microscopy techniques. Layered structures are highlighted as overlays on the simulations. Note fit between models, simulations, and images with respect to heavier atom columns. Widths of B and S building slabs are shown; these are identical to those of Capitani.11,12

Full size image

The inability of the system CeFCO3–CaCO3 to form solid solutions was considered a factor explaining the presence of syntactic intergrowths between various BGS species, except for combinations of bastnäsite and synchysite.1 The same authors postulated periodic changes in the system conditions from which rhythmic intergrowths are formed. Ni et al.3 suggested that syntaxial growth records evolving physicochemical conditions since they result from the solid phase in equilibrium with a fluid undergoing periodic chemical changes. The first quantitative mineral-fluid stability diagrams for the system Ca-REE-C–O–H–F constructed for conditions typical of hydrothermal systems13 demonstrate the control of a(CO3)2 and aF– (a = activity) on the stability fields of coexisting bastnäsite-(Ce) and parisite-(Ce), thus, explaining the predominance of these phases with either fluorite or calcite, respectively. Preliminary petrogenetic grids predict the coexistence of bastnäsite and synchysite at high temperatures (as known from carbonatites),14 but such associations are only recently reported from hydrothermal deposits.13,15

Bastnäsite and synchysite are the most abundant REE minerals in the >10 billion ton Olympic Dam Cu–U–Au–Ag deposit, South Australia.15,16 Rare earths are not currently exploited at Olympic Dam but represent a tantalizing future opportunity given the size of the known resource. Harnessing of such opportunities can build on the rapidly evolving understanding of the deportment and behavior of elements of interest in ore and throughout the processing circuit.

The REE fluorocarbonates crystallized as stable phases associated with hematite, Cu–(Fe) sulfides, and uranium mineralization during a ~1590 Ma magmatic-hydrothermal event.16,17 In ores with high REE content, petrographic and compositional data15 identify a dominant bastnäsite-(Ce) and lesser synchysite-(Ce), with both phases close to stoichiometry4,15 and predictable chondrite-normalized REE fractionation patterns.15 REE fluorocarbonates are, in general, paragenetically earlier than REE phosphates, notably florencite,18 which although present throughout the deposit, are generally subordinate to REE fluorocarbonates. A final stage of REE mineral paragenesis features REE-bearing Ca–Sr-REE aluminum phosphate–sulfates of the woodhouseite–svanbergite series.19

Atomic-scale intergrowths encountered at direct contacts between bastnäsite and synchysite record chemical modulations toward intermediate members. This indicates a stepwise approach toward thermodynamic equilibrium, inferring that observed coexistence of the two species records overprinting hydrothermal events rather than co-crystallization.

Micron to nanoscale intergrowths at contacts between bastnäsite and synchysite

Understanding relationships between these minerals requires linking micron and nanoscale observations using a stepwise analytical approach.20 Small ore chips are mounted in epoxy and prepared as 1-in. polished blocks for petrographic examination using electron microscopes with µm-scale resolution to identify locations of interest from which we extract material and prepare thin foils transparent under a high-voltage (200 kV) S/TEM microscope.21

The Olympic Dam deposit formed during cycles of brecciation of host lithologies resulting in different generations of the same mineral.21 These are recognizable from assessment of textures and geochemical signatures at the µm scale in Fe oxides, sulfides, phosphates, and many other mineral groups but are especially well represented by the associations of bastnäsite, parisite, and synchysite such as those illustrated in Figure 2 and explored at the nanoscale in subsequent illustrations.2,4,5,6,7,8,11,12,22

Figure 2
figure 2

Plan and section views (a, c–h) for intergrowths between bastnäsite (Bst), parisite (Prs), and synchysite (Syn),4,15,22 with compositions shown on the plot in (b) constructed using published data.4,15 Studies using scanning/transmission electron microscopy foils obtained by focused ion beam (FIB) techniques provide a direct correlation between surface and depth at a given site of interest.20 The transect from which the material is extracted for nanoscale investigation is marked as yellow lines.

Full size image

In altered granite from the deposit margins, BSG minerals form intergrowths spanning the compositional range between bastnäsite and parisite (Figure 2a–c).4,22 Within the deposit itself, BSG grains of composition representing the two endmembers (Figure 2b) spatially coexist with one another, but sub-µm scale intergrowths are also present at their direct contacts (Figure 2d).15 Synchysite is abundant in breccias containing clasts of altered mafic lithologies and can be formed by breakdown of calcic plagioclase during interaction with REE-bearing fluids.15 As the rocks are further brecciated, earlier synchysite is marginally affected by dissolution and reprecipitation of lamellar bastnäsite (Bst-1), which is, in turn, overgrown by stubby and porous bastnäsite (Bst-2; Figure 2e). Lamellar bastnäsite (Bst-1) displays fine, rhythmic banding that extends at depth along the length of each Bst-1 lamella (Figure 2f). Likewise, banding occurs in synchysite within ~1-µm-wide domains at the contact with Bst-1 (Figure 2g–h).

Modularity of layered sequences and identity of smallest component slabs

Direct imaging of atoms in modular mineral structures can be readily achieved using the high-angle annular dark-field (HAADF) STEM technique in which the signal intensity (I) depends on the atomic number (Z) of an element (I–Z2).23 While this method can distinguish between chemically distinct atomic column types, the light elements become “invisible” in proximity to heavier elements, as is the case of C from layers positioned adjacent to REE (F) and Ca columns (Figure 1). The heaviest atoms form parallel rows of brightest dots separated by darker strips, 4.9 Å, 14 Å, and 9 Å in width (Figure 3a–e), and these are interpreted using models for bastnäsite and synchysite (Figure 1). The intervals correspond to a separation of the REE-bearing columns along the c-axis, defining the widths of the bastnäsite and synchysite slabs, B and S, respectively (Figure 3a–e).4 It can be seen that these intervals are identical on images obtained from planes with two different specimen orientations relative to the incident beam (zone axes), but that atom stacking differs in directions perpendicular to the c-axis (Figures 1 and 3a–d). Insertion of S slabs does not offset the atomic columns in a matrix of regular B slabs (Figure 3d). Likewise, compositional faults2 are recognized in irregular sequences as shown in Figure 3b.

Figure 3
figure 3

(a–e) High-resolution high-angle annular dark-field scanning transmission electron microscope images of bastnäsite (Bst), parisite (Prs), and synchysite (Syn) from both types of assemblages shown in Figure 2.4,15,22 Both regular (a, c–e) and irregular layer sequences (b) can be interpreted using the widths of the B and S layers and their respective combinations.4 Comparable approaches to interpretation of BSG mineral intergrowths are shown in prior publications.2,4,5,6,7,8,11,12,22 (f) Intensity profile (yellow lines on c–e) for the three minerals showing signal maxima for Ce-(F) and minima for Ca. Note: F, compositional fault.

Full size image

TEM studies of continuous series of one-dimensional structures have shown that compositional changes are directly linked to layer combinations.24,25,26 Once the building modules have been defined, HAADF STEM imaging is a powerful tool for tracking stacking sequences, as, for example, shown for Bi-(Pb) chalcogenides.27,28,29 This also applies to REE fluorocarbonates,4 and despite the lack of detail for the lighter, CO3 layers, the widths of the B and S slabs are identical with those reported in conventional TEM studies.11,12

The simplest BS combination, 14 Å in width, defines parisite, CaREE2(CO3)3F2 (Figure 3a, e). Slab modularity is confirmed by intensity profiles along stacking sequences of bastnäsite, synchysite, and parisite (Figure 3f), which display intensity maxima matching the Ce(+ F) columns and minima for the Ca columns; the latter are visible as fainter atoms through the middle of the S slabs in synchysite and parisite.

Slab stacking counted across intergrowths expresses rhythmic compositional changes

Average compositions of BGS minerals within rhythmically banded intervals that are below the resolution of the electron microprobe beam (<1 µm) can be calculated in terms of B/S ratios within the group by directly counting slab stacks.1,2,4 The sequence illustrated as Figure 44 shows B/S ratios in the range 1.05–2.27, with a single outlier (6.8), indicating a gradual transition from parisite to unnamed B2S. However, of the two species, only parisite displays intervals of regular BS stacking (up to 21), whereas various polytype combinations account for close-to-stoichiometric B2S compositions4,6 along the ~1-µm-long profile in parisite-B2S lamella bordered by bastnäsite (Figure 2c).

Figure 4
figure 4

Overview of intergrowths along profiles in parisite-B2S (Prs, a),4,22 bastnäsite (Bst, b, c), and synchysite (Syn; d, e).15 Intervals selected for slab stack counting are numbered and B/S ratios1,2,4 are shown in yellow. (f) Ranges of B/S ratios ranges for selected intervals and BSG members. Rgt, röntgenite.

Full size image

Counting of slab stacking in selected intervals across the entire width of one of the rhythmically banded bastnäsite lamellae and the band in synchysite show B/S ratios ≥1 and ≤0.5, respectively (Figure 4b–e). Interval selection was dictated by an appreciation of intergrowth density with lengths of disordered stacking intergrowths varying gradually from >200 nm to <100 nm, and then ≤50 nm away from the contact with neighboring lamellae. These intervals are interspersed with regular sequences of B or S slabs, respectively, representing the host mineral, and with block width at least the same size as those of the selected intergrowths. Variation of B/S ratio within the wider (~4.6 µm) bastnäsite is greater, with two intervals with B/S ratios far exceeding 1 (4.3 and 3.25), well above the mean (1.3) of the other five intervals (Figure 4c). In the narrower (~0.6 µm) synchysite band, B/S ratios span a smaller interval, 0.27–0.58 (Figure 4e). An appreciation of compositional variation within the series is obtained by plotting measured B/S ratios of these intervals next to unnamed members of the series positioned between bastnäsite (B/S ➔ ∞) and parisite (B/S = 1), followed by röntgenite, Ca2Ce3(CO3)5F3 (BS2; B/S = 0.5), the only named mineral between parisite and synchysite (B/S = 0) (Figure 4f). Ignoring the outlier, the trend from parisite to B2S across profile 1 is averaged by a middle composition (B/S = 1.55), showing that the system reaches toward chemical equilibrium by short-range intergrowths.4,22 Within the Olympic Dam deposit,15 intergrowths of S slabs in bastnäsite have higher oscillations than those of B slabs in synchysite, but in each case, the oscillations are compositionally close to the member in the series next to that of the host (e.g., parisite and röntgenite for bastnäsite and synchysite, respectively). Therefore, the assembly of intergrowths and unpatterned host blocks is interpreted to record rhythmic compositional variation from either of the two endmembers in the series. The difference in the trends of the sets of B/S ratios for the selected profiles in bastnäsite and synchysite (downward and upward, respectively), relative to the same system boundary, is the larger, enclosing synchysite grain in each case.

Disordered stacking sequences record transition to thermodynamic equilibrium

Reading B and S slabs across intergrowths intervals (Figure 5) shows subtle chemical modulations of the same, or close composition, by different stack combinations, in agreement with established polytypic modularity within the bastnäsite–synchysite group.1,2,4,5,6,7,8,11,12 Ordered and disordered stacking sequences in the compositional range parisite-B2S modulate compositions across intervals ranging in lengths from ~13 to 54 nm (Figure 5a).4

Figure 5
figure 5

Polytypic modularity within the bastnäsite–synchysite group1,2,4,5,6,7,8,11,12 shown as chemical modulations of the same, or close composition by different stack combinations of B and S slabs read across intergrowth intervals. Longest intervals from each profile in Figure 4c and 4e are subdivided and assessed in terms of stacking sequences and B/S ratios. Explicit reading of B and S slabs in each sub-interval is given between //. Note: F, compositional fault.

Full size image

The longest intervals displaying intergrowths in bastnäsite and synchysite from Figure 4 were subdivided and assessed in terms of stacking sequences and their individual B/S ratios (Figure 5b–c). In these cases, elimination of B- or S-only blocks of limited width (e.g., 7B slabs) demonstrates that the B/S ranges are much narrower and are represented by mean B/S ratios of 1.025 and 0.548, much closer to parisite and röntgenite, respectively (compared with 1.63 and 0.27 shown in Figure 4 for the same profiles 2 and 3). Nonetheless, even though sequences of BS slabs or BS2 slabs are present up to several repeats, the explicit reading of the stacking sequences shows a high degree of randomness, or in other words, the stacking sequences are disordered. Notwithstanding the disorder, the data reflect clear transitions from the endmembers in the series to the next intermediate phase (i.e., bastnäsite to parisite, and synchysite to röntgenite), supporting the thermodynamic calculations for coexisting bastnäsite–parisite pairs.13

The examples used here to illustrate features of REE-fluorocarbonate mineralogy at Olympic Dam are indicative of chemical oscillations leading to self-patterning, known from a range of geological environments and widely reported in geochemical systems that are close to thermodynamic equilibrium.30 Constraints on the type of chemical oscillator producing the patterning in a closed or open system would require modeling of still larger data sets. Further interrogation of these types of data to extract the periodicity of the (integer) number of B and S slabs and their variations in width would be a basis for defining rhythmical fluctuations at the interface between a growth layer and REE-bearing fluids, whether in an open or closed system as previously suggested by Ni et al.3

Outlook

Bastnäsite-(Ce) and synchysite-(Ce) are stable phases in the Olympic Dam deposit and are overwhelmingly stoichiometric in composition.15 Micron to nanoscale observations allow, however, for documentation of a hitherto unrecognized complexity at the direct contact between the two minerals. Likewise, atomic-scale intergrowths, compositionally constrained in the range of bastnäsite–parisite and with both ordered and disordered domains, are intimately intergrown with molybdenite, and form during an early stage of Fe-metasomatism on the outskirts of the deposit.4,22 This implies that thermodynamic equilibrium among REE fluorocarbonates can be attained via atomic-scale intergrowths within compositional ranges that are controlled by host lithologies and the physicochemical parameters of individual stages of mineralization. Minerals that form mixed-layer compound series can readily accommodate superimposed stages of mineralization via nanoscale (re)-equilibration. The comparable range of B/S ratios in domains with the densest intergrowths nevertheless indicates that the system undergoes chemical oscillation and ordering as it strives toward thermodynamic equilibrium.

Studies of the composition and fine structures of bastnäsite-synchysite minerals from Olympic Dam4,15,22 carry implications for understanding the genesis and evolution of breccia-type deposits and for constraining the stability of phases in the bastnäsite–synchysite group. There is currently considerable interest in assessing the mineralogy and geochemistry of REE at Olympic Dam, one of Australia’s largest potential, yet currently unexploited, REE resources, and for identifying pathways for future recovery of REEs. The observed atomic-scale complexity of bastnäsite–synchysite boundary domains in poorly described elsewhere and the consequences which the observed atomic-scale features may have for downstream processing and REE recovery are unknown at the present time.

Variations in REE abundances (Figure 6) are indicators of time–space evolution in a deposit.15,31 Distinct textural types of BSG endmembers from Olympic Dam record a wide range of HREE/MREE versus LREE ratios (Figure 6a).15 Among these, stubby bastnäsite and synchysite display subtle variation in chondrite-normalized REE fractionation patterns15 also identifiable from the REE distributions on STEM maps (Figure 6b–c). A comparison of various data sets for bastnäsite and synchysite15,31,32,33,34,35,36 shows mean La/Ce and La/Nd ratios in BSG minerals from Olympic Dam span a wider range of values relative to other deposit types (Figure 6d).

Figure 6
figure 6

Rare-earth element (REE) distributions in bastnäsite (Bst) and synchysite (Syn) from Olympic Dam15 [plots in (a, b) and scanning transmission electron microscopy maps in c] and in deposits worldwide.31,32,33,34,35,36 Stubby bastnäsite, richer in Ce and La than synchysite (b, c), displays a gradational trend, toward synchysite (a),15 which is attributable to the nanoscale intergrowths (Figures 4, 5). (d) mean La/Ce and La/Nd ratios in BSG minerals from Olympic Dam compared to examples in other deposit types. Note: HAADF, high-angle annular dark field.

Full size image

REE fluorocarbonates are among the dominant hosts for REE in many of the world’s largest REE deposits,37 including Bayan Obo (Inner Mongolia, China),31 Phalaborwa (South Africa),38 and Mountain Pass (California, USA).39 Fluorocarbonates are typically light-REE enriched and, therefore, may not be the main hosts for the more valuable heavy rare earths (e.g., Al-Ali et al.40). Although coexisting bastnäsite–parisite appears relatively widespread globally,41 identifying deposits in which bastnäsite and synchysite coexist is more problematic. Williams-Jones and Wood14 review REE mineral assemblages reported up to 1992 yet remarkably few bastnäsite–synchysite assemblages. Further, they acknowledge ambiguity in the literature about whether these or other assemblages necessarily represent equilibrium. Later literature also reveals few, if any, well-documented examples. Rare cases of reported coexisting bastnäsite–synchysite are interpreted to represent nonequilibrium crystallization or replacement (e.g., “Bastnäsite overgrows parisite and synchysite”41 or “overgrowth of bastnäsite”12). Syntaxial intergrowths of the Ca-REE-fluorocarbonates are reported42 but unfortunately not illustrated to allow assessment of whether these are analogous to those shown here from Olympic Dam.