Dissolution-Repackaging of Hellandite-(Ce), Mottanaite-(Ce)/Ferri-Mottanaite-(Ce)

: We investigated hellandite-group mineral phases from the Roman Region, alkali syenite ejecta, by multimethod analyses. They show a complex crystallisation history including co-precipitation of hellandite-(Ce) with brockite, resorption, sub-solidus substitution with mottanaite-(Ce), exsolution of perthite-like ferri-mottanaite-(Ce), overgrowth of an oscillatory-zoned euhedral shell of ferri-mottanaite-(Ce) and late, secondary precipitation of pyrochlore in the cribrose hellandite-(Ce) core. LREE/HREE crossover and a negative Eu anomaly in hellandite-group minerals follows f O 2 increase during magma cooling. The distinction among the hellandite-group minerals is based on the element distribution in the M1, M2, M3, M4 and T sites. Additional information on miscibility relationship among the hellandite sensu strictu , tadzhikite, mottanaite, ferri-mottanaite and ciprianiite endmembers derives from molar fraction calculation. We observed that change in composition of hellandite-group minerals mimic the ligands activity in carbothermal-hydrothermal ﬂuids related to carbonatitic compared to the associated hellandite-(Ce) and ferri-mottanaite-(Ce). The REE pattern of gadolinite, compared with hellandite-(Ce) and ferri-mottanaite-(Ce), shows a high content of HREE like that of hellandite-(Ce).


Introduction
The hellandite group comprises rare and notably complex REE-rich borosilicate minerals. The authors of [1] provided a first determination of the crystal structure by singlecrystal diffraction.  4 ]. The authors of [2] reported the crystal structure of a Th-rich hellandite-(Ce) occurring in sub-volcanic ejecta from Capranica, describing a tetrahedral distorted site, which can be occupied not only by H, as typically found for hellandite, but also by Be and Li. The authors of [3] provided the crystal chemistry of mottanaite-(Ce) and ciprianiite, both monoclinic, with the distorted tetrahedral site occupied mainly by Be and Li in mottanaite and by H in ciprianiite and hellandite. The proposed ideal formulas for the new endmembers were X Ca 4 Y (CeCa) Z [4] provided a review of the nomenclature, the crystal chemistry and the crystal structure of the minerals of the hellandite group, defining the four endmembers based on various data from the literature. They provided a new general formula for the minerals of the hellandite group as follows: X 4 Y 2 ZT 2 [B 4 Si 4 O 22 ]W 2 , where X = Na, Ca, Y, LREE 3+ at the eightfoldcoordinated M3 and M4 sites; Y = Ca, Y, HREE 3+ , Th 4+ , U 4+ at the eightfold-coordinated a Thermo Scientific™ DXR™ Raman Microscope using a 532-nm laser as an excitation source at the Department of Chemical Sciences, University of Padova. The analyses were performed using a 50× working distance objective with~2.5 cm −1 spectral resolution and 1.1 µm spatial resolution at 10 mW of power. Spectra were recorded in the range extending from 100 to 1100 cm −1 . Raman spectra were collected using an exposure time of 15 s and 20 accumulations to maximise the signal-to-noise ratio. Spectra data reduction was performed using the Thermo Scientific™ OMNIC™ 9.2.0 Spectral Software at Padua University, Padua, Italy.
Single-crystal XRD data were collected using an XcaliburE four-circle diffractometer (Agilent Technologies, Santa Clara, CA, USA) equipped with an EoS CCD area detector. The diffractometer was operated with Mo Kα radiation at X-ray tube conditions of 50 kV and 40 mA. A whole sphere of data was collected to 30 • Theta using omega scans, a scan width of 1 • , and a counting time of 30 s per frame. Intensity data were corrected for Lorentzpolarisation effects and absorption (empirical multi-scan) using the CrysAlis software (Rigaku Oxford Diffraction, Tokio, Japan). The crystal structure was determined using SHELX within the WinGX software (University of Glasgow, Glasgow, UK). Powder XRD was carried out using the X-ray diffractometer Rigaku Miniflex II, located at the DiSPUTer (Chieti, Italy). The acquisition parameters that have been used in the Xrd_Di.S.Pu.Ter.-Uda laboratory at "G. d'Annunzio" University (Chieti, Italy) are Cu-Kα (1.540598 Å) radiation generated at 30 kV and 15 mA, in an exploratory interval between 3 and 70 • 2θ, 0.1 steps, and a scan rate of 0.15 • /s. Once the diffractogram was obtained, background subtraction and indexing of peaks with semi-quantitative analysis were performed. The mineral identification was performed using the software Match 3.
For the discrimination among the different hellandite endmembers, we used a method based on the Ca content at M2, M3 and M4 sites on the dominant metal at M1 site and the content of Be and Li at T site basing on the criteria of literature and on the analysis of different chemical compositions of hellandite-group mineral endmembers [2][3][4]6]. The Ca content at site M2, M3 and M4 allows to discriminate the two main groups: hellandite sensu strictu and tadzhikite, with a Ca content at site M3, M4 less than 4 apfu and with no Ca present at site M2. Mottanaite, ciprianiite and ferri-mottanaite are distinguished from hellandite sensu strictu and tadzhikite having Ca greater than 4 apfu at sites M3 and M4, and excess of Ca at site M2. Owing the dominant metal at site M1, we distinguished tadzhikite (Ti 4+ > Al 3+ and Ti 4+ > Fe 3+ ) and ferri-mottanaite (Fe 3+ > Al 3+ and Fe 3+ > Ti 4+ ). Ciprianiite is specified by the dominance of Th and U at site M2 compared to REEs. Finally, hellandite and mottanaite, which have dominant REEs at the M2 site, are specified by the Be and Li content at the T site. Hellandite has the T site with large vacancies, while mottanaite has an ideal Be and Li content of 1.5 apfu.

Hellandite-Group Mineral Phases
We have analysed four crystals of hellandite-group minerals. Crystal n. 1. Tre Croci, Vetralla (42 • 19 58 N 12 • 05 11 E). The crystal has a remarkable size (up to 7 mm) and is quite inhomogeneous at a microscale. The general structure of the crystal suggests corrosion and resorbing showing a cribrose and chessboard-like texture with vugs sometimes hosting secondary pyrochlore. A phase map of the crystal on the BSE and TIMA comparison shows microscale sub-domains with different compositions (Figure 1). Light grey domains, corresponding to the darker green in the TIMA image, are hellandite-(Ce), which make up the bulk of the crystal. Dark grey domains, corresponding to the lighter green in the TIMA image, are mottanaite-(Ce). This texture suggests an intergrowth/implication structure between two co-precipitating phases that can be interpreted in terms of diffusion-reaction processes or dissolution-repackaging. Mottanaite-(Ce) grows parallel to hellandite cleavage, according to {001} and {100}. domains, corresponding to the lighter green in the TIMA image, are mottanaite-(Ce). This texture suggests an intergrowth/implication structure between two co-precipitating phases that can be interpreted in terms of diffusion-reaction processes or dissolutionrepackaging. Mottanaite-(Ce) grows parallel to hellandite cleavage, according to {001} and {100}.  Chemical analysis was validated by single-crystal structure refinement of crystal 1. Diffraction data of a fragment from crystal 1 (0.03 0.08 0.14 mm) were collected in ϑ range  Figure 2). The a:b:c ratio calculated from unit cell parameters is 3.985:1:2.146. The crystal structure is similar to hellandite described by [4] but the unit cell parameter c is slightly smaller than that of the previous study. More remarkable, but comparable to other studies, is the mean bond length <M2-O> (M2 = Ca, Y, REE, Th, U), demonstrating that incorporation of a bit more Th and U has not much effect on the size of the M2 polyhedron (Tables 2 and 3; Figure 3a,b).       2 . Average chemical compositions of hellandite-(Ce) and ferri-mottanaite-(Ce) are reported in Table 4. Table 4. Chemical composition of crystal 2 rim and core of hellandite-(Ce) (average of 3 EMPA analyses) and ferri-mottanaite-(Ce) from RR.   Table 5. Zoning in crystal 3 has been analysed using BSE images. The variation in colour from the lighter zones to the darker zones corresponds to Ca, Ce and Th ( Figure 4).

Sample
Micro-Raman spectra were measured on crystals 3 and 4 in the range of 100 to 1100 cm −1 . The two spectra are shown in Figure 5. The five most prominent peaks (from the strongest to the weakest) are at~880, 515, 336, 950 and 170 cm −1 . Several additional minor Raman peaks occur in the spectra. The measured Raman spectra are comparable to that of an hellandite-(Ce) from Campo Pefella (erratum for Padella), near Capranica, Viterbo, reported by RRUFF Raman database (ID R061013, [13]). After an extensive literature search, we did not find any published Raman spectra of minerals belonging to the hellandite group, and therefore our Raman spectra should represent the first published spectra for ferri-mottanaite-(Ce). Minerals 2021, 11, x FOR PEER REVIEW 11 of 26  Chondrite-normalised REE patterns for hellandite-(Ce), ferri-mottanaite-(Ce), gadolinite-(Ce) and brockite are shown in Figure 6. Chondrite-normalised REE patterns for hellandite-(Ce) and ferri-mottanaite-(Ce) show a marked crossover at the Eu-Gd level. Profiles of hellandite-(Ce) show an HREE and Y-enrichment relative to adjacent ferrimottanaite-(Ce). Brockite shows an enrichment in LREE compared to the associated hellandite-(Ce) and ferri-mottanaite-(Ce). The REE pattern of gadolinite, compared with hellandite-(Ce) and ferri-mottanaite-(Ce), shows a high content of HREE like that of hellandite-(Ce).  Chondrite-normalised REE patterns for hellandite-(Ce), ferri-mottanaite-(Ce), gadolinite-(Ce) and brockite are shown in Figure 6. Chondrite-normalised REE patterns for hellandite-(Ce) and ferri-mottanaite-(Ce) show a marked crossover at the Eu-Gd level. Profiles of hellandite-(Ce) show an HREE and Y-enrichment relative to adjacent ferri-mottanaite-(Ce). Brockite shows an enrichment in LREE compared to the associated hellandite-(Ce) and ferri-mottanaite-(Ce). The REE pattern of gadolinite, compared with hellandite-(Ce) and ferri-mottanaite-(Ce), shows a high content of HREE like that of hellandite-(Ce).

Other Associated Mineral Phases
Four mineral phases seem texturally related to the described hellandite occurrence: brockite, britholite, gadolinite and pyrochlore (Tables 6-9). Brockite and pyrochlore were found as inclusion in hellandite (in crystal 2 and in crystal 1 respectively), while britholite and gadolinite are hitherto characteristically associated with the hellandite paragenesis in syenite ejecta. We found gadolinite in association with crystal 1 and britholite in association with crystal 2. Gadolinite-like minerals have been found by [14] in association with hellandite-(Ce) from Vico Vulcano, confirming our finding.

Discussion
The nomenclature of the hellandite group is essentially based on the dominance of a specific cation in one of the sites of the complex but flexible structure of the mineral [4]. We observed two different main types of hellandite-group phases according to conventional chemical classification, as follows. Type-a crystal: discrete crystal of hellandite-(Ce) intergrowth with mottanaite-(Ce) (Figure 1); Type-b crystal: composite crystal with a core of hellandite-(Ce) exsolving vermiculations of ferri-mottanaite-(Ce) and having a euhedral shell of oscillatory zoned ferri-mottanaite-(Ce) (Figure 3d,g,h).
Type-a crystal displays chessboard-like discrete domains of hellandite-(Ce) and mottanaite-(Ce) plus crystal corrosion. These features suggest dissolution-repackaging in sub-solidus condition and disequilibrium with the initial crystallising liquid.
Type-b crystal displays a core having a perthite-like structure of ferri-mottanaite-(Ce) in hellandite-(Ce). This kind of structure is likely due to intracrystalline diffusion, which, owing to a temperature drop, creates compositional segregations in crystalline solids by exsolution. In theory, this suggests the presence of an immiscibility gap between hellandite-(Ce) and ferri-mottanaite-(Ce) at a lower temperature. It follows liquid/crystal disequilibrium with core corrosion. Instead, the next crystallisation step shows compositional equilibrium with the crystallising liquid, which is testified by the overgrowth of a euhedral shell of ferri-mottanaite-(Ce). This phase responds to compositional variation of the liquid (Ca, Ce, Th) with oscillatory zoning, according to crystallisation of other phases but preserves the ferri-mottanaite-(Ce) distinctive composition (Figure 4). This corroborates the hypothesis of an immiscible gap between hellandite-(Ce), mottanaite and ferri-mottanaite. These observation leads us to pose the challenge of calculating the molar solution among the different hellandite chemical species.
There is no experimental evidence for a sizeable solid solution among various hellanditegroup endmembers at magmatic temperatures, but we consider it a viable hypothesis based on crystal and structural chemistry relationships. In the literature, there is no information about the presence of immiscibility gaps among hellandite endmembers. The boundaries between one species and another are still somewhat uncertain, owing to the various molar fractions that can form a continuum in the chemical space (Figures 7 and 8). In addition, the determination of the molar species content is highly puzzling. To resolve this issue, we attempted an empirical approach based on several assumptions used to discriminate the different chemical species of the hellandite group.    Chemical characteristics of hellandite-group minerals. (a) Fe 3+ content at M1 site (apfu) associated with Ca content at M2, M3 and M4 sites (apfu) with the conventional minimum content for ferri-mottanaite and the saturation limit of Ca at M3, M4 sites; (b) Be + Li content (apfu) associated with Fe 3+ at M1 site/Ca 2+ at M2, M3 and M4 sites (apfu), with the separation limits based on Be and Li content. Key symbols: red stars are ferri-mottanaite from this study; orange diamonds are ferri-mottanaite from literature; yellow stars are hellandite ss from this paper; blue squares are hellandite sensu strictu from literature; green stars are mottanaite from this study; green triangles are tadzhikite from literature; violet pentagons are ciprianiite from literature; and pink hexagons are mottanaite from literature. Our immiscibility hypothesis must be tested by investigating the endmembers distribution in each specimen. The first step is to verify if the molar abundances reflect the actual chemical distribution using the ideal stoichiometric formulas. Without this calculation, it is impossible to reach any coherent thesis between observed and expected chemical features. The calculation procedure is described in the electronic excel sheet in the appendix and explained in the next paragraph.

Mineral Chemistry
There is a general agreement in the literature that the endmember classification of hellandite group is based on the occupancy of Ca at M2, M3 and M4 sites, the dominant metal at the M1 site (Fe 3+ , Al 3+ and Ti 4+ ) and T site (Be and Li). In Figure 7, conventional Our immiscibility hypothesis must be tested by investigating the endmembers distribution in each specimen. The first step is to verify if the molar abundances reflect the actual chemical distribution using the ideal stoichiometric formulas. Without this calculation, it is impossible to reach any coherent thesis between observed and expected chemical features. The calculation procedure is described in the electronic excel sheet in the appendix and explained in the next paragraph.

Mineral Chemistry
There is a general agreement in the literature that the endmember classification of hellandite group is based on the occupancy of Ca at M2, M3 and M4 sites, the dominant metal at the M1 site (Fe 3+ , Al 3+ and Ti 4+ ) and T site (Be and Li). In Figure 7, conventional separation limits are highlighted by red lines. M2 site Ca saturation marked by Ca in M3-M4 sites accompanied by Fe 3+ increase in M1 site, produces a progressive trend where hellandite sensu strictu and tadzhikite have Fe 3+ < 0.5 apfu and Ca M2,M3,M4 ≤ 4 apfu. In contrast, mottanaite and ciprianiite have Ca M2,M3,M4 > 4 apfu and Fe 3+ <0.5 apfu and, finally, ferri-mottanaite has Fe 3+ > 0.5 apfu and Ca M2,M3,M4 > 4 apfu.
Data plotting suggests that these limits are not an insurmountable barrier and that a particular uncertainty is associated with the recalculation of the various chemical species in a.p.f.u. In fact, in Figure 7a, the hellandite ss sample from [2] does not plot neatly on the left of the Ca saturation limit. In addition, three samples of ferri-mottanaite from this study plot on beneath the conventional minimum content of Fe 3+ for ferri-mottanaite (Figure 7a) despite the samples having all the other chemical features to be a ferri-mottanaite (i.e., high Be). Another discriminant factor is provided by the T site's occupancy (Be + Li) (Figure 7b). This paper and literature data about hellandite-group minerals suggest that a limit of Be + Li 0.8 apfu marks the divide between hellandite ss and tadzhikite vs. ciprianiite, mottanaite and ferri-mottanaite. Tadzhikite is then distinguished from hellandite ss by the content of Ti 4+ , representing the dominant element in the M1 site. Although, the schemes work in a general sense they require further reflection and detailed understanding.
Ternary diagrams in Figure 8 confirm the ability of the M1 and M2 sites to define different hellandite endmembers in the frame of a generalised chemical variation trend. Ca in M2 coupled with Fe 3+ , Ti 4+ and Al 3+ in M1 defines the mottanaite, ferri-mottanaite and ciprianiite trend in contrast with hellandite ss and tadzhikite. From this point of view, we have just two completely separated hellandite subgroups. Over time, the number of hellandite chemical species has grown based on this criterion, generating a nomenclature that could be overabundant.
We calculated the molar solution calculation, estimating the proportion of Ti and Fe/Fe + Al + Ti at the M1 site, respectively. The proportion of Ce + Y, Th + U, REE/REE + Th + U at M2 site, respectively. The values obtained are proportioned to Al/3 and proportional assigned to Fe + Al + Ti. These values are normalised to 1. Be + Li is used to calculate ferri-mottanaite, mottanaite and ciprianiite, assuming an ideal content of Be + Li of 1.5, 1.5 and 0.5, respectively. The T-site vacancies are proportioned after dividing by 2 to hellandite and tadhzikite. This complex calculation was empirically confirmed by a correspondence of the dominant molar fraction in the solid hellandite-group molar solution with the chemical classification. A limit of this calculation, which is the best we have imagined, is underestimating ciprianiite. Owing to the theoretical lower content of Be + Li in ciprianiite for mottanaite and ferri-mottanaite samples, which may be chemically classified as ciprianiite, these show a low molar fraction of ciprianiite. Results of this calculation are shown in Table 10. Molar fractions are helpful to investigate the presence of immiscibility gaps, which are relevant to explain the observed textural features of immiscibility among hellandite, mottanaite and ferri-mottanaite. In Figure 1, we note that hellandite and mottanaite form discrete adjacent domains, which indicate probably a sub-solidus substitution, and, in Figure 3d, vermicular exsolution of ferri-mottanaite in hellandite-(Ce). This seems confirmed in Figure 9, which shows a large, molar solid solution gap among hel sensu strictu/Σothers and Σothers/hel sensu strictu ratios ranging from 58 to 100 and from 42 to 0, respectively. Our discussions about the different molar fractions are strongly related to the observed textural data. For this reason, they are not extended to literature data for which we do not know the textural occurrence. In our case, we observe that the first crystallising hellandite contains up to 73% of other endmembers, up to 24% of mottanaite and up to 23% of ferri-mottanaite, still classifying as hellandite ss. Mottanaite contains up to 34-36% of ferri-mottanaite and up to 9% of hellandite. Ferri-mottanaite contains 37% of mottanaite and 11-12% of hellandite. Complete molar proportions are shown in Figure 10.  Figure  10.

Crystal Chemical Structure
B4Si4 dominates the hellandite structure, which is not easily influenced by the occupancy of the other sites. This study confirms that for hellandite sensu strictu, the cell parameters a, b, c and V positively correlate with the increase of CaO wt% [2,4]. Ciprianiite, mottanaite and ferri-mottanaite do not show a clear trend and show cell parameters variation despite a similar CaO wt% content ( Figure 11).
The authors of [4] observed that plotting the mean bond length <T-O> (Å) vs. site scattering (ss) at the T site have a strong correlation predictive of Be + Li abundance in the T site. Figure 12a shows that our hellandite-(Ce) has a very high ss at the T site (epfu). We refined a <T-O> of 1.74 Å compared to 1.633 Å, which is the mean bond distance of <Be-O> [15], suggesting that a larger cation, such as Li, occupies the T site as well. The radius of Li + is larger than Be 2+ (r Li = 0.59 Å, r Be = 0.27 Å in tetrahedral coordination, [16]). Assuming Be and Li on T, we may then conclude that the crystal has a higher Be/Li ratio than crystals from other studies. However, these T site considerations are subtle because Be and Li are weak scatterers. Cautious discussion is needed, and there is still a discrepancy between ss at the T site of XRD and chemical analysis. Furthermore, this conclusion is not satisfactory as the higher Be/Li ratio is not confirmed by comparing our data with those of the literature. In fact, to satisfy the correlation between T site features and Be content, more than double of Be would be necessary to explain very high ss at the T site (epfu) (Figure 12a-c). This is unrealistic as the uncertainties of the chemical analysis of Be (based on five analyses) are small and do not indicate higher Be concentration. We

Crystal Chemical Structure
B 4 Si 4 dominates the hellandite structure, which is not easily influenced by the occupancy of the other sites. This study confirms that for hellandite sensu strictu, the cell parameters a, b, c and V positively correlate with the increase of CaO wt% [2,4]. Ciprianiite, mottanaite and ferri-mottanaite do not show a clear trend and show cell parameters variation despite a similar CaO wt% content ( Figure 11). The authors of [4] observed that plotting the mean bond length <T-O> (Å) vs. site scattering (ss) at the T site have a strong correlation predictive of Be + Li abundance in the T site. Figure 12a shows that our hellandite-(Ce) has a very high ss at the T site (epfu). We refined a <T-O> of 1.74 Å compared to 1.633 Å, which is the mean bond distance of <Be-O> [15], suggesting that a larger cation, such as Li, occupies the T site as well. The radius of Li + is larger than Be 2+ (r Li = 0.59 Å, r Be = 0.27 Å in tetrahedral coordination, [16]). Assuming Be and Li on T, we may then conclude that the crystal has a higher Be/Li ratio than crystals from other studies. However, these T site considerations are subtle because Be and Li are weak scatterers. Cautious discussion is needed, and there is still a discrepancy between ss at the T site of XRD and chemical analysis. Furthermore, this conclusion is not satisfactory as the higher Be/Li ratio is not confirmed by comparing our data with those of the literature. In fact, to satisfy the correlation between T site features and Be content, more than double of Be would be necessary to explain very high ss at the T site (epfu) (Figure 12a-c). This is unrealistic as the uncertainties of the chemical analysis of Be (based on five analyses) are small and do not indicate higher Be concentration. We conclude that there is another possible explanation for this structural peculiarity of crystal 1. conclude that there is another possible explanation for this structural peculiarity of crystal 1. Figure 11. Structural parameters variation with CaO wt% content, following [2]. Key symbols are the same as in Figure 7.  . Structural parameters variation with CaO wt% content, following [2]. Key symbols are the same as in Figure 7. According to Figure 12d, there is a slight inverse correlation between the Th-U content and the mean bond length <M2-O> (Å). Hellandite-(Ce) has a Th + U content similar to the other hellandites [4] but a bit shorter bond < M2-O> (Å). However, ciprianiite and ferri-mottanaite, despite having a higher content of Th + U compared to hellandites, have a slightly shorter < M2-O> (Å). We deduce that the Th + U content does not affect the According to Figure 12d, there is a slight inverse correlation between the Th-U content and the mean bond length <M2-O> (Å). Hellandite-(Ce) has a Th + U content similar to the other hellandites [4] but a bit shorter bond <M2-O> (Å). However, ciprianiite and ferri-mottanaite, despite having a higher content of Th + U compared to hellandites, have a slightly shorter <M2-O> (Å). We deduce that the Th + U content does not affect the <M2-O> (Å) site parameters very much.

Genetic Conditions
Crystal texture and crystallisation hierarchy indicate a complex evolution and crystal composition variation during the late stage of syenite crystallisation and a possible contribution of REE-rich carbothermal/hydrothermal fluids. There is a strict link between late stage carbonatitic fluids and secondary mineralisation in syenites [11]. At sub-magmatic temperatures, the progressively oxidising conditions and the increasing activity of P 2 O 4 2− and SiO 4 2− ligands favour the precipitation of brockite with hellandite ss (crystal n.2 in Figure 3c-f).
A hypothesis given in the literature [17,18] considers that hellandite crystal as a substitution product of britholite. This type of substitution is possible by introducing B in the system at a late magmatic stage. Alkali syenites contain several minerals that could be precursors of hellandite, and other B-and Be-bearing minerals (e.g., vicanite, gadolinite) such as silicates (thorite, stillwellite, perrierite, allanite) and phosphates (monazite, britholite, brockite). Boro-silicates genesis was likely controlled by introducing exogenous boron in the original mineral composition leading to the replacement associations. We suggest that britholite is not suitable to explain some of the feature of hellandite-(Ce) crystal 1. From a chemical point of view, we observe a peculiar behaviour of REE. Figure 6 displays an intense concentration of HREE in hellandite-(Ce). In our opinion, the best candidate to explain the structural and chemical features of hellandite is gadolinite. It is monoclinic, has high HREE and a large ss at T site. This explains the extreme value of ss at T site and HREE content compared to the hellandites in literature. It is possible to write a simple equation to pass from gadolinite-(Ce) to hellandite-(Ce): Our observations suggest that hellandite-(Ce) and brockite may become unstable at lower temperature and higher fO2, which explains the resorption of hellandite-(Ce) and dissolution of brockite.
REE patterns comparison among minerals may be complex due to crystallisation conditions and co-precipitating phases with propensity to select specific REE. Average REE content does not change notably in ferri-mottanaite-(Ce) compared to hellandite-