Crystal chemistry of ferriallanite-(Ce) from Nya Bastnäs, Sweden: Chemical and spectroscopic study

A second study of ferriallanite-(Ce) from Nya Bastnäs, Sweden, extends current data by using electron probe micro-analysis (EPMA), laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analysis and brings new insights about its crystal chemistry obtained by Raman spectroscopy. The study presents the first Raman spectra for ferriallanite-(Ce) member of the allanite group (not considering the rather low-quality spectra published in preceding papers). The material does not show significant radiation damage, which is rare as allanite-group minerals often have undergone metamictisation due to significant amounts of incorporated radionuclides (U, Th). Some interior regions show pronounced zoning that correlates with variations in Raman-band positions. In spite of its significant REE content, the material is virtually non-luminescent. New additional data for allanite-(Ce) from Oßling, Germany and Domanínek, Czech Republic are also presented, which were used for comparison.

The most common compositional relationships in the allanite-ferriallanite-epidote-clinozoisite solid solution can be expressed via the following substitutional vectors: These relationships can be depicted in a ΣREE versus Al diagram for such solid solutions proposed by Petrík et al. (1995), which also allows one to estimate the Fe 2+ /Fe 3+ ratio. However, other substitutions involving Mn 2+ , Mn 3+ , Mg, Pb 2+ , V, Cr, Ti, Sn, Sc and others are possible as well (Dollase 1969;Grew et al. 1991;Bonazzi and Menchetti 2004;Ahijado et al. 2005;Miyawaki et al. 2008;Nagashima et al. 2013;Biagioni et al. 2019). The possibility of Fsubstituting for O 2in epidote-supergroup minerals has been reported as well (Peacor and Dunn 1988), but its extent tends to be generally low except from minerals of the dollaseite group (Pautov et al. 1993;Ercit 2002).
The radionuclides Th and U readily enter the crystal structure of allanite-group minerals, owing to their similar ionic radii with those of REE. The radioactive decay of Th and U causes structural damage. Even minor amounts of radionuclides may thus lead to a transformation from the crystalline to a glassy state over geological periods of time; such glassy state is referred to as metamict (Ewing 1994). Metamict minerals generally have lowered chemical resistance Fig. 1 Sketch of the crystal structure of allanite-group minerals. It consists of chains of edge-sharing M(1) octahedra and edge-sharing M(2) octahedra (red). These chains are linked by SiO 4 tetrahedra and Si 2 O 7 groups (yellow). Slightly deformed M(3) octahedra are shown turquoise. Large cavities in the framework belong to A(1) and A(2) polyhedra where elements such as Ca 2+ or REE 3+ are present in ninefold and 11-fold coordination to oxygen, respectively. Hydrogen atoms are marked "H". Sketch based on Dollase (1971) and Reissner et al. (2019), drawn with the JP-Minerals VESTA v3 software (Momma and Izumi 2011)  and are more susceptible to secondary alteration processes, compared to their crystalline counterparts. This is especially true for fluid-driven chemical alteration that may result in substantial compositional changes (Gieré and Sorensen 2004;Bonazzi et al. 2009). The accumulation of radiation damage also causes changes of physical properties, such as volume expansion that is detected from shifts of X-ray diffraction maxima to lower 2Θ values (Ewing et al. 1987;Janeczek and Eby 1993;Reissner et al. 2019). Decrease of the shortrange order may be detected using Raman spectroscopy. For instance, Raman spectra of radiation-damaged natural zircon show increased band broadening accompanied by band downshifts, depending on the self-irradiation dose. In some cases, band broadening can be used to estimate the degree of radiation damage, provided significant band broadening due to chemical variations is excluded (Nasdala et al. 1995).
The main aim of the present study was to expand our knowledge regarding crystal-chemical characteristics of allanite-group minerals. To the best of our knowledge, no detailed spectroscopic studies of ferriallanites samples have been undertaken thus far, and the only spectra of ferriallanite-(La) that have been published are of rather low quality (Varlamov et al. 2019). To provide some rationale for the relationship between Raman spectra and composition, the combination of Raman spectroscopy with EPMA was chosen as main tools. To provide basic information about the relative changes of elements distribution in the sample, LA-ICP-MS served as a supplementary method to visualise such changes in the form of qualitative maps. Also, REE-bearing minerals often show distinct luminescence that may be used as a helpful fingerprinting tool for mineral identification. The photoluminescence-based detection of allanite-(Ce) may be possible based on the Nd 3+ emission in the nearinfrared range (about 10,600-11,900 cm −1 ; see for instance Lenz et al. 2015;Zeug et al. 2017). Therefore, laser-induced photoluminescence (PL) was also included.

Samples and preparation
We have investigated a 6 cm long specimen of ferriallanite-(Ce) layer in amphibole matrix originating from a mine dump near Nya Bastnäs, Skinnskatteberg area, Sweden (Fig. 2). Two other samples were studied as references for comparison, metamict allanite-(Ce) from a pegmatite near Domanínek in Czech Republic (Čermák 2013) and acicular, well-crystallized allanite-(Ce) of hydrothermal origin from Oßling, Germany (Nasdala 1993). The Bastnäs specimen was cut into several polished sections (100 μm thickness) and attached to a glass slide. Some fragments were heated in air to check for possible annealing of radiation damage. For this, the fragments were placed in a Pt crucible and heated to 280 °C at a rate of 30 °C/min. At the end of a 96-h run, the furnace was switched off and the sample was allowed to cool to below 50 °C before taken out of the oven. Reference samples were also embedded in epoxy and polished. For EPMA analysis, samples were coated with carbon.

LA-ICP-MS
LA-ICP-MS analyses were performed by using a pulsed Nd:YAG laser system LSX-213 G2 + working at 213 nm wavelength and using a pulse duration of ∼4 ns. The ablation system is equipped with a 2-volume ablation cell (HelEx) designed to enable rapid eluting of the ablationgenerated aerosol in a large format cell. Helium was used as a carrier gas with a flow rate of 0.9 l/min. The aerosol was transported from the ablation cell through a 1 m long FEP (Fluorinated Ethylene Propylene) transport tube (2 mm in diameter) to the ICP discharge of the quadrupole mass spectrometer ICP-MS Agilent 7900. The mass spectrometer operated at the forwarded power of 1550 W and Ar gas flow rate of 15 l/min (outer plasma gas). The analysis was performed by a set of continuous line scans using laser pulses covering a defined area of the sample to create qualitative maps. Qualitative maps were obtained with the spot diameter of 4 µm, fluence of 10 J·cm −2 , scan speed of 4 µm/s and repetition rate of 20 Hz. The following isotopes were monitored during the LA-ICP-MS scan measurements with the total integration time of 1 s: 24 Mg, 27 Al, 29 Si, 31 P, 43,44 Ca, 55 Mn, 57 Fe, 88 Sr, 89 Y, 90 Zr, 139-159 Ln, 232 Th, 238 U. Data were processed in ILaps by Faltusová et al. (2022).

Raman and photoluminescence spectroscopy
Raman and PL spectra, including line scans, were obtained at room temperature by means of a Horiba Jobin Yvon LabRam-HR Evolution. The system was equipped with an Olympus BX41 optical microscope, a Si-based, Peltiercooled charge-coupled device detector, and a diffraction grating with 1800 grooves per millimetre. Spectra were excited with the 473 nm emission of a diode laser (10.5 mW at the sample surface). Attempts to obtain PL spectra were also made with 532, 633 and 785 nm laser excitation. The instrument was calibrated using the Rayleigh line, resulting in a wavenumber accuracy of 0.5 cm -1 . An Olympus 100 × objective (numerical aperture 0.90) was used. Spectra were obtained in a range 110-4000 cm −1 in confocal mode; the lateral resolution was better than 1 μm. The spectral resolution was better than 1.2 cm -1 . Line scans were done using an automated, software-controlled x-y stage. All spectra were obtained in areas about 10 μm away from EPMA points (to avoid unwanted effects of the electron-beam impact). Band fitting was done after appropriate background correction, assuming Lorentzian-Gaussian band shapes.

General description
Ferriallanite-(Ce) from Nya Bastnäs shows irregular zoning, which is even visible by the unaided eye. When observed in plane-polarised transmitted light, a wide range of colour sectors (ranging from light green to dark brown) with intensive pleochroism are seen. In cross-polarised transmitted light, these sectors have high second-order interference colours that are, however, widely masked by body colour (Fig. 3).

Crystal chemistry
BSE imaging revealed detailed zonation patterns that reflect variations in chemical composition. Textural relations indicate that interior regions with primary growth zoning were replaced by patchily zoned domains. These domains were chosen for more detailed examination (see lines 1 and 2 in Fig. 4). Differences is BSE intensity reflect mainly the variation of Ca (9.5-16.1 wt% of CaO) and REE (14.6-27.1 wt% of REE 2 O 3 ), which are accompanied by changes in total Fe oxide (15.4-19.1 wt%) and Al 2 O 3 (11.9-17.4 wt%). Results of chemical analyses are summarised in Table 2 (  The presence of U and Th was below the detection limits of EPMA (0.09 wt% and 0.07 wt%, respectively).
Data obtained by LA-ICP-MS correlate with EPMA results, especially in REE-rich and REE-poor zones. The edge of the BSE-zoned area (marked with white lines in Fig. 6) has slightly increased contents of Y and U. Yttrium and Th are found in elevated concentrations also in chaotically zoned interior regions surrounding the primary growth zonation. Other elements detected did not show significant spatial variations.

Spectroscopy
Raman line scans across REE-rich and REE-poor primary growth zones revealed minor variations in spectral positions of Raman bands, not exceeding 13 cm -1 ; there are also accompanying minor variations of band shapes and relative intensities. Changes in band positions are particularly noticed in the spectral ranges 800-1100 cm -1 and 500-600 cm -1 . An example for accompanying changes in band shapes is the triplet of bands around 430 cm -1 . It should be noted that, in spite of ferriallanite-(Ce) contains OHgroups, no significant O-H stretching signal was detected.
The Raman spectrum of the fragment of ferriallanite-(Ce) that was annealed at 280 °C for 4 days is largely similar to the spectrum of the untreated counterpart (Fig. 7). Both spectra are somewhat similar to the Raman spectrum of fresh allanite-(Ce) from Oßling, Germany. In contrast, metamict allanite-(Ce) from Domanínek, Czech Republic, yielded a notably different Raman spectrum that is typical of glassy silicates, with a weak band around 950 cm -1 (Fig. 7).
Untreated and annealed fragments of the sample were also checked for possible luminescence using blue, green, red and near infrared laser excitation. No significant emission was detected.

Discussion
Results of chemical analyses of allanite-group minerals were converted to mineral formulae using normalization of the sum of metal cations to 8 and total charge to 25. This approach allows one to estimate the Fe 3+ /Fe 2+ ratio, under the assumption that there are no A-and M-site vacancies and no "oxy-allanite" component present (Ercit 2002). Analytical totals close to 100 wt% and the virtual absence of elements that are characteristic for altered allanite-group minerals and common in radiation damaged specimens (Na, Pb, P, Ti; Ueda 1955;Deer et al. 1986), we may assume that our sample did not experience significant chemical alteration. This, however, appears to be in contrast to the irregular zoning exhibited by a major part of the sample.
Our sample is much poorer in Mg and REEs and richer in Al, compared to ferriallanite-(Ce) from the Bastnäs deposit studied by Holtstam et al. (2003), which may indicate considerable chemical variability of this mineral. Ferriallanite-(Ce) yielded Raman spectra that are somewhat similar to the fairly unsharp spectra of ferriallanite-(La) (Varlamov et al. 2019); however, our spectra show much more pronounced and well-defined bands. Spectra depend on chemical compositions of the sample, especially its REEs contents (Fig. 8). Stretching motions in sorosilicates for Si 2 O 7 and SiO 4 complexes occur as intense bands above 700 cm −1 . These bands belong to the antisymmetric and symmetric stretching vibration of Si-O b -Si (O b -bridging oxygens) and of Si-O nb (O nb -non-bridging oxygens) and were observed in ferriallanite-(Ce) from 830 to 1100 cm -1 Fig. 5 Chemical composition of ferriallanite-(Ce) and its variations along the two arrows shown in Fig. 4 with all points (total 51 analyses). a Plot of Ca against Al. b Plot of REE against Al. Radial lines emanating from the clinozoisite composition allow to estimate the portion of Fe 3+ in total Fe (central y axis; Petrík et al. 1995). The grey levels of the symbols visualise the observed BSE intensities (b) (a) (compare Lazarev 1972;Le Cléac'h andGillet 1990 or Makreski et al. 2007). According to Chukanov et al. (2014), bands above 1039 cm -1 are related to stretching vibrations of the Si-O b -Si depending on the value of the Si-O b -Si angle, where with the increasing angle, the frequency of corresponding vibrations increases as well. Varlamov et al. (2019) discussed that the force characteristics rather than the size of a cation in the A(2) site determine this angle; however, after a thorough examination of the crystal-chemistry of the epidote supergroup minerals, the change in Si1-O9-Si2 angle is related to the M3 octahedral volume rather than the A2-O bond strength (see Bonazzi and Menchetti 1995). This is especially noticeable in REE-bearing epidote supergroup minerals, where the Si1-O9-Si2 angle has lower values due to the entry of larger divalent cations at M3 rather than trivalent cations in A2, and is consistent with our observation of shifts in Raman bands in the 300-500 cm -1 region that are assumed to be assigned to external modes of M-O interactions (Makreski et al. 2007;see Fig. 9). Bands in the range 550-700 cm −1 were described by Wang et al. (1994) as symmetric stretching of Si-O b -Si bonds.
Annealing of ferriallanite-(Ce) at 280 °C was mainly done to check for possible annealing-induced spectral changes that, if present, might indicate structural reconstitution of radiation damage. According to Reissner et al. (2019), structural recovery of radiation-damaged allanite-(Ce) is already significant at 280 °C. Annealing at higher temperatures is expected to be more efficient for structural reconstitution; however, potential dehydrogenation and Fe 2+ oxidation (Janeczek and Eby 1993;Bonazzi and Menchetti 1994;Bonazzi et al. 2009;Reissner et al. 2019) might affect Raman spectra as well and hence bias interpretation. Inconspicuous differences between spectra of untreated and annealed samples (Fig. 7) suggest that radiation damage must be insignificant which is also evidenced by a comparison with the Raman spectrum of a metamict allanite-(Ce) sample from Domanínek. The observation Fig. 6 Trace-element distribution patterns obtained by LA-ICP-MS from strongly zoned ferriallanite-(Ce). As a visual guide, the edge of the BSEzoned area is marked by a white line. Note that BSE intensity and La and Ce content correlate that Raman bands of ferriallanite-(Ce) are generally broader than bands of non-radiation-damaged allanite-(Ce) from Oßling ( Fig. 7) is assigned to "chemical band broadening" due to more elevated contents of cation substitutions (compare Ruschel et al. 2012). The virtual absence of luminescence, in particular of emissions related to REEs, in spite of significant concentrations of REE 3+ (Table 2), is explained by quenching, most likely due to elevated concentrations of Fe 2+ (Pascoal et al. 1999).
As demonstrated in a study of the migration of U and Pb in U-rich minerals (Holá et al. 2021), LA-ICP-MS imaging of individual isotopes can be advantageous in elucidating the evolution of geological processes. This is particularly beneficial for the detection of isotopes of trace elements with sub-ppm concentrations (Chew et al. 2021), which EPMA is unable to detect. Unfortunately, due to the very small dimensions of the investigated areas of interest and the measured signal-to-noise ratio, our results are merely semi-quantitative. The U and Th content analysed by LA-ICP-MS in ferriallanite-(Ce) from Bastnäs region was reported by Holtstam et al. (2003) as 2.6 ± 0.2 ppm for U and 0.25 ± 0.03 ppm for Th (in total ~ 3 ppm), which on our estimate according to the low signal intensity (counts per second) should correspond to the same values. Differences in signal intensity in and out of the zonation were observed; however, it is important to highlight the fact that due to a different orientation of the ferriallanite-(Ce) crystal in the studied profile, the material could be ablated differently (Tsibidis et al. 2020). Fig. 7 Raman spectra obtained from unannealed and annealed chips of the Bastnäs (Sweden) ferriallanite-(Ce), in comparison with spectra of a wellcrystallized (Oßling, Germany) and a metamict (Domanínek, Czech Republic) allanite-(Ce) reference. Note that spectra of the unannealed and annealed Bastnäs chips are widely similar, which excludes significant annealing-induced structural reconstitution and hence supports the generally low level of radiation damage in this sample

Intensity (a.u.)
Raman shift (cm -1 ) 400 700 1000 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.