Fedorite from Murun Alkaline Complex (Russia): Spectroscopy and Crystal Chemical Features

: Fedorite is a rare phyllosilicate, having a crystal structure characterized by SiO 4 -tetrahedral double layers located between continuous layers formed by edge-sharing (Ca,Na)-octahedra, and containing interlayer K, Na atoms and H 2 O molecules. A mineralogical-petrographic and detailed crystal-chemical study of fedorite specimens from three districts of the Murun alkaline complex was performed. The sequence of the crystallization of minerals in association with fedorite was established. The studied fedorite samples di ﬀ er in the content of interlayer potassium and water molecules. A comparative analysis based on polyhedral characteristics and deformation parameters was carried out. For the ﬁrst time, EPR, optical absorption and emission spectra were obtained for fedorite. The raspberry-red coloration of the mineral specimens could be attributed to the presence of Mn 4 + ions. Polished samples and polished thin sections with a thickness of ~40 µ m were studied by optical petrographic and mineralographic methods in transmitted and reﬂected light using an OLYMPUS BX-51 polarizing microscope. The morphogenetic features of fedorite and its paragenetic associations were investigated in order to reveal the mineral relationships and determine the sequence of mineral formation processes.


Introduction
Fedorite is a rare and complex silicate, first discovered in fenitized rocks of the Turiy alkaline complex of the Kola Peninsula, Russia [1]. In 1968, the mineral was approved by IMA and named after the Russian mathematician, crystallographer and mineralogist Evgraf Stepanovich Fedorov. Mineralogical description and chemical analysis were published in 1979 [2].
The structure of fedorite was solved for the first time in space group C by photographic method [3], but with a relatively high value of the discrepancy index (R = 9.5%).
Later, authors in [4] carried out structure refinements using an X-ray single-crystal diffractometer and a neutron single-crystal diffractometer which converged to R = 4.0% and 6.9%, respectively.
Subsequently, a second occurrence of the mineral was reported from charoite rocks of the Murun complex [5,6].
Finally, the crystal structure of the fedorite in the space group P was refined [7] and remaining uncertainties in some atom positions and ordering of cations were clarified. Samples from the Turiy and the Murun complexes were used for the study.
The ideal formula for the fedorite group was derived: A 2-3 M 7 T 16 O 38 × 2 ·nH 2 O (A = Na, K, Ba, Ca; M = Ca, Na, Mn 2+ , Fe 2+ ; T = Si, Al, Ti; X = F, Cl, OH). However, all previous studies of fedorite have been incomplete in terms of comprehensive mineral characterization, taking into account the modern methodology for the study of complex natural objects using mineralogical-petrographic, crystal-chemical and physical methods.
According to the silicate minerals hierarchy of [8], fedorite is a sheet silicate with a two-dimensional infinite tetrahedral polymerization. The tetrahedral double layer in fedorite can be considered as a 3-connected net with the designation 6 3 ( Figure S1 of Supplementary Materials). The [Si 16 O 38 ] 12        3. The sample Irk-53 (Irkutskiy district) is characterized by the greatest mineral diversity. It is a lenticular fragment (10 × 6 × 3.5 cm) consisting of the following minerals (in vol.%): aegirine (25), tinaksite (15), fedorite (10), microcline (2), quartz (33), charoite (15) (Figure 5a,b). Steacyite, galena and native copper are present in the sample in small amounts. Aegirine is represented by dark green elongated euhedral and subhedral crystals (up to 0.2 × 2.5 mm in size) with vitreous luster (Figure 5a-c). The crystals form sheaf-like and radially radiant aggregates with a diameter of up to 5 mm. Tinaksite is represented by elongated euhedral and subhedral light brown crystals (up to 0.3 × 15 mm) with a vitreous luster (Figure 6a-f). Crystals form sheaf-like aggregates up to 15 mm in length. Fedorite forms tabular subhedral pale raspberry-red crystals (up to 0.8 × 1.5 mm) with a pearly luster (Figure 6a-f). The mineral is translucent to transparent with perfect cleavage on {001}. The microcline is found in the form of single tabular anhedral greenish grains (up to 5 mm in size). Quartz is observed in the form of transparent lenticular or anhedral grains (up to 10 mm in size) with a greasy luster. Quartz is xenomorphic with respect to aegirine, tinaksite and fedorite. Charoite is a cryptocrystalline aggregate of lilac color with a silky luster. Elongated fibrous crystals (less than 1 μm thick) form aggregates that surround earlier minerals (Figures 5a,b and 6a-c). Separate charoite fibers were found in quartz in the form of inclusions that indicates a close crystallization time of these minerals. Steacyite is found in paragenesis with charoite and quartz. Euhedral transparent tetragonal yellowish-brownish crystals of steacyite reach 0.5 × 0.5 × 12 mm in size. Moreover, its grains observed in quartz are larger and have a higher degree of euhedrality than steacyite grains in charoite. Galena is found in the form of elongated subhedral and euhedral grains (up to 1.2 × 0.6 mm) in fedorite and quartz (Figure 6d-h). Small grains of native copper (up to 0.01 mm in size) were also found in quartz ( Figure 6h).

3.
The sample Irk-53 (Irkutskiy district) is characterized by the greatest mineral diversity. It is a lenticular fragment (10 × 6 × 3.5 cm) consisting of the following minerals (in vol.%): aegirine (25), tinaksite (15), fedorite (10), microcline (2), quartz (33), charoite (15) (Figure 5a,b). Steacyite, galena and native copper are present in the sample in small amounts. Aegirine is represented by dark green elongated euhedral and subhedral crystals (up to 0.2 × 2.5 mm in size) with vitreous luster (Figure 5a-c). The crystals form sheaf-like and radially radiant aggregates with a diameter of up to 5 mm. Tinaksite is represented by elongated euhedral and subhedral light brown crystals (up to 0.3 × 15 mm) with a vitreous luster (Figure 6a-f). Crystals form sheaf-like aggregates up to 15 mm in length. Fedorite forms tabular subhedral pale raspberry-red crystals (up to 0.8 × 1.5 mm) with a pearly luster (Figure 6a-f). The mineral is translucent to transparent with perfect cleavage on {001}. The microcline is found in the form of single tabular anhedral greenish grains (up to 5 mm in size). Quartz is observed in the form of transparent lenticular or anhedral grains (up to 10 mm in size) with a greasy luster. Quartz is xenomorphic with respect to aegirine, tinaksite and fedorite. Charoite is a cryptocrystalline aggregate of lilac color with a silky luster. Elongated fibrous crystals (less than 1 µm thick) form aggregates that surround earlier minerals (Figure 5a,b and Figure 6a-c). Separate charoite fibers were found in quartz in the form of inclusions that indicates a close crystallization time of these minerals. Steacyite is found in paragenesis with charoite and quartz. Euhedral transparent tetragonal yellowish-brownish crystals of steacyite reach 0.5 × 0.5 × 12 mm in size. Moreover, its grains observed in quartz are larger and have a higher degree of euhedrality than steacyite grains in charoite. Galena is found in the form of elongated subhedral and euhedral grains (up to 1.2 × 0.6 mm) in fedorite and quartz (Figure 6d-h). Small grains of native copper (up to 0.01 mm in size) were also found in quartz (Figure 6h).

Chemical Analysis
Chemical data of fedorite samples were obtained by means of a Jeol JXA-8200 electron microprobe (JEOL, Tokyo, Japan) equipped with a high-resolution scanning microscope, an energy-dispersive spectrometer with the Si(Li) detector (resolution-133 eV) and five wavelength-dispersive spectrometers (WDSs). The analyzed samples were polished sections of fedorite-containing rock. The conditions for the excitation and recording of analytical signals were as follows: accelerating voltage = 20 kV; probe current = 5 nA; beam diameter = 5 µm; counting time on peak = 10 s; and counting time on background = 5 s. The following standards were used: albite (Na), diopside (Mg, Ca, Si), orthoclase (K), ilmenite (Ti), Mn-garnet (Mn), pyrope (Al, Fe), celestine (Sr), barite (Ba), Nb pure (Nb), Cl-apatite (Cl) and phlogopite (F). A conversion from X-ray counts to oxide weight percentages was obtained with the ZAF data reduction system. Table 1 represents the average composition of the studied fedorite samples (Gav-43, Yak-5, Irk-53) compared with those of fedorite compositions found in the literature.

Single-Crystal X-ray Diffraction Study
Structure analyses of fedorite single crystals (samples Gav-43, Yak-5 and Irk-53) were carried out using a Bruker AXS D8 VENTURE dual-source automated diffractometer (Bruker, Berlin, Germany) equipped with a Photon 100 detector. Intensities were collected at room temperature using graphite-monochromated MoKα radiation (operating conditions-50 kV and 1 mA, crystal-to-detector distance-40 mm). Two sets of 12 frames were used for initial cell determination, whereas the entire Ewald sphere (±h, ±k, ±l) up to θmax~40 • was recorded by a combination of several ϕ and ω rotation sets, with 0.5 • scan width and 1-5 s exposure time per frame. The data collection strategies were optimized by the APEX2 program suite [14] and the reflection intensities were extracted and corrected for Lorentz-polarization by the SAINT package [15]. The CrysAlis PRO program was also used for data processing [16]. A semi-empirical absorption correction was applied by means of the SADABS software (Version 2014/4) [17]. Finally, least-squares refinements were performed using the program CRYSTALS [18]. Information concerning the data collection and refinement is listed in Table 2. The refined parameters included scale factors, atom coordinates, atomic displacement parameters and occupancies (for alkaline and alkaline earth sites and Ow positions). Occupancies for Si, O and F atoms were constrained to 1. For the refinement of the Ca and Na occupancies in the M-octahedra the restrain Ca + Na = 1 ± 0.01 was imposed. However, the authors experienced considerable difficulties in localizing interlayer anions. In contrast to the refinement carried out by [7], the occupancy of the Ow20 position was not constrained. The distribution of K atoms and oxygens of the water molecules between the identified disordered positions was guided by the results of microprobe analysis (calculation of K atoms per formula unit). The refinements reveal large displacement parameters for some K and Ow ions indicating, possibly, mixed cationic-anionic occupancy of the same structural sites.
The results of X-ray diffraction studies of the fedorite samples cannot be used to locate H atoms of water molecules in the structure due to low occupancy and disordering of the H 2 O positions.
The CIFs are deposited with the Cambridge Crystallographic Data Centre (CCDC 2016399, 2,016,400 and 2016401). Final atomic coordinates, site occupancies, equivalent/isotropic and anisotropic displacement parameters for the studied fedorite samples are reported in Tables S1-S6 of Supplementary Materials. Relevant cation-anion bond lengths are given in Tables S7 and S8 of Supplementary Materials. Crystal structures were visualized using the Vesta (Version 3.4.4) program [19].

Polyhedra Characteristics Calculation
The calculation of the characteristics of coordination polyhedra was performed to carry out a statistical analysis of structural data.
Bond-valence sum (BVS) was calculated as where Rij is the bond represent length between ions I and j, Sij is the bond valence, and R O and B are the bond-valence parameters [20]. The values R O and B for ion pairs involving oxygen were obtained in [21], whereas to calculate Sij of the cation-fluorine bonds, R O and B given in [22] were used.
An effective coordination number (ECoN) was calculated using VESTA 3.4.4. [19], which adopts ECoN defined as Σiω i The ω i is called "bond weight" of the i th bond and is determined as where l i is a bond length and l av is a weighted average bond length. See [23][24][25] et al. for details on ECoN.
The coordination numbers (CN) of polyhedral were derived from the ECoN determination as well as using the suggestion of [26] to consider solely the bonds, which have a significant contribution (>0.05 vu) to BVS of a central cation. Table 2. Selected data on single crystals, data collection and structure refinement parameters of the studied fedorite samples.

Gav-43
Yak  The volume of the coordination polyhedron (Vp) is the sum of the volumes of general tetrahedral with three of the vertices in ligands and the fourth one in the volume center of the polyhedron. The volume center is the point with coordinates which considered as the simple arithmetic averages of the coordinates of all the ligands, three of which lie on the same face of the coordination polyhedron. The centroid of a polyhedron is defined as the point having the smallest deviation of distances to the polyhedron's vertices [27].
With the introduction of these concepts, for further analysis, the following parameters are calculated: The average distance from the volume center to the ligands (r v ); The distance of the central atom to the volume center (∆v); The average distance from the centroid to the ligands (r s ); The distance of the central atom to the centroid (∆); The volume of the sphere fitted to the positions of ligands (Vs); The volume eccentricity calculated as ECCv = 1 -[(r s -∆)/r s ] 3 ; The volume sphericity calculated as SPHv = 1-3σ rs /r s , where σ rs is a standard deviation of distances from ligands to the centroid.
The eccentricity is a numerical characteristic of a conical section, pointing out the degree of its deviation from the circle. The circle has zero eccentricity, ellipses have an eccentricity of less than 1. The sphericity is a measure of how spherical a fitted sphere is. It is less than 1 for nonspherical objects.
A computer program (IVTON) which calculates these parameters is available [28].
To better understand interrelations involved in polyhedral distortion, it is necessary to describe it quantitatively. The used distortion parameters are the following: The volume distortion (υ) [29] is calculated relative to an ideal polyhedron with the same coordination number and inscribed inside the sphere with the radius r s . υ = (Vi -Vp)/Vi where Vi is the volume of the ideal polyhedron, and Vp the volume of the coordination polyhedron. The calculation was performed in IVTON [28].
Bond length distortion was calculated as BLD = 100/n· (Ii/Io) 2 /4, where Io is the center to vertex distance for an undistorted tetrahedron whose volume is equal to that of the distorted tetrahedron with bond length Ii [31]; Octahedral angle variance was calculated as OAV = Octahedral quadratic elongation was calculated as OQE = 6 i=1 (Ii/Io) 2 /6, where Io is the center to vertex distance for an undistorted octahedron whose volume is equal to that of the distorted octahedron with bond length Ii [31].
Bond length and edge length distortion (BLD and ELD) are a measure of the dispersion of the individual bond lengths and edge length, i.e., a large value indicates dispersed bonds, while low ones indicate that the bonds are closely grouped around an average value. The bond angle variance (TAV and OAV) is equal to 0 for a regular polyhedron and is >0 for a distorted polyhedron. The quadratic elongation (TQE and OQE) is dimensionless and equal to 1 for a regular polyhedron while it is >1 for a distorted polyhedron.
The polyhedral geometry characteristics and distortion parameters were calculated for studied fedorite from Murun massif and compared with those calculated for fedorite and structurally related minerals from the literature (Tables S9-S13 of Supplementary Materials). Further details are given in the "Structural features" section below.

Spectroscopy Study
The infrared (IR) absorption spectra of fedorite (Gav-43 and Irk-53 samples) were measured by the Fourier-Transform Infrared method (FTIR) using an FT-801 (Simex, Novosibirsk, Russia) spectrometer. Powdered samples were mixed with preliminary dried KBr, pelletized, and measured with a resolution of 1 cm -1 . A total of 32 scans were collected in the wavenumber range 480 to 4000 cm -1 . The FTIR spectrum of an analogous pellet of the pure KBr was used as a reference.
The electron paramagnetic resonance (EPR) spectra were registered using an X-band spectrometer RE1306 (KBST, Yartsevo, Russia) operated at a microwave frequency of 9380 MHz. The spectrometer was equipped with a quartz cryostat suitable for temperatures of 77 and 295 K.
Diffuse optical absorption spectra of both samples were obtained by a Perkin-Elmer Lambda 950 UV/VIS/NIR spectrophotometer at 300 K using an integrated sphere as proposed in [32,33].
The sample Gav-43 has relatively large grains that are cleaved to a thin plate with a thickness of 0.1 mm. The transmittance of the single crystal plate was measured.
The Ce 3+ related photoluminescence excitation (PLE) spectra were measured with a Perkin-Elmer LS-55. The PLE spectra were corrected for the varying intensity of exciting light due. Luminescence corresponding manganese ions was excited with a 405 nm solid-state laser having 50 mW power. The samples were placed in vacuum cryostat at 77 K. Luminescence was registered using a grating monochromator MDR-2 and a photomodule Hamamatsu H6780-04.
It was not possible to obtain the spectroscopy data for the sample Yak-5 since the mineral grains contain intergrowths with associating minerals and are not suitable for analysis.

Chemical Composition
The crystal-chemical formulas were calculated on the basis of (Si + Al) = 16 atoms per formula unit (apfu) and assuming H 2 O content consistent with the refined occupancies of the Ow sites at the interlayer level (Tables S1, S3, S5 of the Supplementary Materials). As noted earlier in [7], in fedorite some of the F ions may be substituted by Cl. Some of the analyzed samples contain a slightly reduced amount of fluorine and chlorine (Yak-5 and Irk-53 sample). Accordingly, the calculated atomic proportion of OH group, substituting F and Cl in the crystal structure, for the studied fedorite samples ranges from 0.01 to 0.45 apfu. The presence of OH groups and H 2 O molecules was confirmed by infrared (IR) spectroscopy. The crystal-chemical formulas of the studied samples of fedorite and those from literature are represented in Table 3. All Ca was located at the M site followed by Mn, Fe, Mg and Na atoms until full occupancy. Remaining Na, as well as K, Sr, Ba, having large ionic radii, were assigned to interlayer A site. Considering that the total negative charge in the ideal formula is 78 valence units (v.u.) it can be noted that our samples have a deficiency of positive charge, which is compensated by the substitution of part of the bridging oxygen atoms (O unsh ) of the Si-tetrahedral layers by OHgroups. It should be noted that the manganese content is given by assuming it in the form of Mn 2+ ; however, we discuss the state of manganese in the "Spectroscopy" section, below.  [7]. While the content of sodium in the samples analyzed in [7] is higher (Table 1). It is noteworthy that aluminum and chlorine concentrations in the studied sample are low in comparison with the fedorite from the Turiy complex, whereas fluorine content is reduced in comparison with the Murun sample analyzed by [7]. In addition, our samples taken from different districts of massif slightly differ in their content of manganese, chlorine and strontium (Table 1).
Structure refinements of fedorite crystals converged to good values of the agreement factors (2.34 ≤ R ≤ 3.10% and 2.92 ≤ wR ≤ 3.52%, Table 2). Our structural models are generally consistent with those proposed by [7]. Small differences can be observed in bond lengths and angles when comparing the tetrahedra and octahedra in the three structures (Tables S7 and S8 of the Supplementary Materials). However, refinement results here provide some geometric details including modelling of the interlayer ions site disorder, speciation and occupancies.
To describe the fedorite crystal structure (Figure 7) it is more convenient to consider tetrahedral and octahedral sheets and interlayer contents separately. To describe the fedorite crystal structure (Figure 7) it is more convenient to consider tetrahedral and octahedral sheets and interlayer contents separately. In the fedorite crystal structure, there are eight tetrahedrally coordinated symmetrically

Si-Tetrahedra (T-Sheets)
The sheets of tetrahedra (T 2 ) have six tetrahedra whose vertices point in one direction toward a sheet of octahedra (O), and two (Si1 and Si6) whose vertices point in the other direction. The T 2 is related to the T 2 layer by the center of symmetry via the bridging oxygen. Projections along the a, b and c axes of the single and double layers are shown in Figure S2a-c of Supplementary Materials, respectively. The size of the six-membered rings, which formed a single layer, is about 6.34 × 3.24 and 5.16 × 5.08 Å (as viewed along the c axis, Figure S2c Figure S2b of Supplementary Materials). There are further six-and eight-membered rings with a size of~5.49 × 5.34 and~5.80 × 5.66 Å, respectively. These values are quite close to those found in other silicates from the Murun charoite-bearing rocks (see, for example, [32,[34][35][36]).

M-Octahedra (O-Sheets)
The O layers are effectively sandwiched between two symmetrically equivalent T 2 layers (T 2 T 2 ). The O sheets are joint to the T 2 layers through apical atoms of oxygen from the SiO 4 tetrahedra of the 6-membered rings. In the octahedral sheet of the fedorite crystal structure, the symmetrically independent crystallographic cation sites are four positions containing different amounts of Ca and Na ( Figure S3  In all of the three analyzed crystals, the <M-O> bond lengths and the octahedral volumes are very close. This similarity most possible is due to the very similar composition of the octahedral sites and closed values of Na and Ca ionic radii (1.02 Å and 1.00 Å for Na and Ca with CN = 6; [38]).

Interlayer Cations (A) and H 2 O Molecules
The interlayer cavities are occupied by alkaline (K and Na) and minor alkali-earth (Sr and Ba) cations and water molecules. The six-membered Si-O rings host alkaline ions which are split over two different positions; here they are labelled Na5 and K1 (Na5-K1~0.57-0.89 Å, Table S8 and Figure S4 of Supplementary Materials). According to the refinement, the occupancies are~47-52% (Na5) and 24-27% (K1) in different samples.
Another common feature for the three samples under study is the presence of a water molecule at Ow20 position. In contrast to the structural models of [7], where the occupancy of this position was initially set equal to 1, in our models its occupancies were refined. They are equal to 53%, 85% and 0.52% in the crystal structural model of Gav-43, Yak-5 and Irk-53 samples, respectively. The Ow20 atom is bonded strongly to the Na5 cation. On the opposite side Na5 is bonded to F atom, and with six oxygens of tetrahedral ring it forms an eight-vertex coordination polyhedron with <Na5-O,F> 2.544-2.552 Å (Table S8 of Supplementary Materials). In the structural model of the sample Gav-43, K atoms are placed at positions K2, K3, K4, while position K2 is split into K2 1 and K2 2 , having an occupancy up to 0.17 and distances to oxygen atoms in the range 2.79-3.14 Å (Tables S1 and S8 of Supplementary Materials). Four water molecules (Ow21-Ow24) have occupancies equal to 0.20-0.32. The short distances between some of these positions (1-2.5 Å, Table S8 of Supplementary Materials) ensure the statistical distribution of corresponding oxygen atoms.
In the structure of the sample Yak-5, a large degree of disordering of the structural sites assigned to the positions of K is noted. Six K sites (split K2, K3 and K4) are occupied by 14-21%. While the oxygen of the water molecule has a Ow21 position with an occupancy of 11%.
Finally, in the sample Irk-53, all interlayer positions, except for the aforementioned Ow20, are filled with cations according to the results of the microprobe analysis. Seven positions of cations (other than those Na5 and K1 mentioned yet) are populated by potassium by 8-17% (Tables S5 and S8 of the Supplementary Materials).

Structural Relationship between Fedorite, Lalondeite, Martinite and Ellingsenite, Comparison of Polyhedral Characteristics and Deformation of Coordination Polyhedra
Lalondeite, (Na,Ca) 6 (Ca,Na) 3 Si 16 O 38 (F,OH) 2 ·3H 2 O, having a OT 2 T 2 O module, similar to fedorite, has been found at Mont Saint-Hilaire (Quebec, Canada) [39]. In the M-sheet of the structure of the lalondeite one site is occupied by Na + , two positions are filled by Ca 2+ , and the fourth site is mixed. In the cavities of the structure, two disordered Na and three H 2 O molecules are accommodated.
Martinite, (Na, ,Ca) 12 13 ]·6H 2 O, from Aris complex (Namibia), described by [41]. In martinite and ellingsenite crystal structures, T 2 -layers are not condensed, allowing the X layer to form. Figure S5 (Table S9 of Supplementary Materials). In the martinite crystal structure, there are two [6]-coordinated Ca 2+ ions and two [6]-coordinated Na + ions. In ellingsenite, the O layer is composed of Na-, two Ca-octahedrally coordinated sites and one cation site that is occupied by 50% Ca 2+ and 50% Na + . The interlayer component (X) in martinite houses poorly ordered Na polyhedra and H 2 O molecules. The bonding between polyhedra in the X component and the adjacent T 2 sheets is weak. It confirms the lower-than-ideal bond-valence sums of X-atoms. The spacing between the T layers in ellingsenite is about 2.3 Å larger than that in martinite. According to [41], a part of the silica is located within the Na-H 2 O interlayers. However, this is an assumption that requires verification.
Despite the fact that there are large and low-valence cations (K and Na) and H 2 O molecules in the large cavities between the T 2 sheets of fedorite and lalondeite, their tetrahedral layers remain condensed. Presumably, that can be explained by the presence of Na polyhedral of low coordination in the X layers of martinite and ellingsenite.
In Tables S10 and S12 of Supplementary Materials, selected calculated geometrical parameters related to the coordination polyhedra of the three structures under study and structurally related minerals are listed. The polyhedral and the sphere volumes (Vp and Vs) illustrate the space demand of the various cations. Vp and vs. for Si are very similar in all structures (2.10-2.17 Å 3 and 17.00-17.81 Å 3 , respectively). Depending on the amount of B in tetrahedral sites of martinite, a steady decrease in Vp and vs. (up to 1.66 and 13.57 Å 3 , respectively) with a decreasing of bond valence sum (BVS) and average distance from the central atom to the ligands (up to 2.96 vu, 1.480 Å, respectively, for "Si1" position) are observed (see Table S10 of supplementary material). The average distance from the central atom to the ligands of two Si sites whose vertices point in the direction opposite to the O-sheets-"Si1" and "Si6"-are slightly shorter and bond valence sums (BVS) are slightly higher (Table S10 of supplementary material). The Vp and volume eccentricity (ECCv) also increases compared to ECCvs calculated for other Si tetrahedra.
The volume distortion (υ) of two Si sites whose vertices point in the direction opposite to the O-sheets ("Si1" and "Si6") is lower than the ones for Si, whose vertices point in one direction toward a sheet of octahedra (O). The same is valid for TAV, BLD and ELD parameters (Table S11 of supplementary material). Figure 8a,b and Figure S6 in the Supplementary Materials show a relationship between the distortion parameters for the T1-T8 tetrahedral site of the minerals listed in Table S11 of Supplementary Material. The volume distortion (υ) of two Si sites whose vertices point in the direction opposite to the Osheets ("Si1" and "Si6") is lower than the ones for Si, whose vertices point in one direction toward a sheet of octahedra (O). The same is valid for TAV, BLD and ELD parameters (Table S11 of supplementary material). Figure 8a,b and Figure S6 in the Supplementary Materials show a relationship between the distortion parameters for the T1-T8 tetrahedral site of the minerals listed in Table S11 of Supplementary Material.   [7], lalondeite [39], martinite [40] and ellingsenite [41]. Data were taken from Table S11 Figure 9a,b shows a good correlation between the volume sphericity (SPHv) vs. sphere volume (Vs) and polyhedral volume (Vp). The Vp and vs. of Na5 in fedorite and related polyhedra positions in fedorite models of [7], martinite and ellingsenite are much larger (Table S12 of Supplementary Materials, Figure 9b). An exception is lalondeite, in which a different geometry of the arrangement of the M-octahedra and the absence of 8-coordinated Na-polyhedron in the cavity are noted.
value of volume sphericity (SPHv), and SPHvs of M2, M3 and M4 do not differ significantly. In particular, Figure 9a,b shows a good correlation between the volume sphericity (SPHv) vs. sphere volume (Vs) and polyhedral volume (Vp). The Vp and vs. of Na5 in fedorite and related polyhedra positions in fedorite models of [7], martinite and ellingsenite are much larger (Table S12 of Supplementary Materials, Figure 9b). An exception is lalondeite, in which a different geometry of the arrangement of the M-octahedra and the absence of 8-coordinated Na-polyhedron in the cavity are noted.  [7], lalondeite (Lal) [39], martinite (Mar) [40] and ellingsenite (Ell) [41]. Data were taken from Table S12 of Supplementary Materials. The octahedra volume distortion in fedorite follows the trend υ(M1) > υ(M2) > υ(M4) > υ(M3). We observe that the investigated fedorite samples have the same trend in terms of BLD, ELD, OAV and OQE parameters ( Figure 10 and Figure S7 of Supplementary Materials).
Comparison of distortion parameters for M1-M4 in fedorite with those in lalondeite, martinite and ellingsenite shows significantly larger differences than those in the T-sheet.
Comparison of distortion parameters for M1-M4 in fedorite with those in lalondeite, martinite and ellingsenite shows significantly larger differences than those in the T-sheet.

Spectroscopy
The infrared spectra of the fedorite samples Gav-43 and Irk-53 are given in Figure 11. The group of peaks observed at 1119, 1030, 792, 615 cm -1 may be assigned to the asymmetric and symmetric Si-O stretching modes of the SiO4 tetrahedra. The relationship between the peaks at 1119 and 1030 cm -1 is different in the Gav-43 and Irk-53 samples. This difference could be explained by the inequality of disordering of the interlayer ions and the partitioning of cations between the octahedral positions in the Gav-43 and Irk-53 samples. The absorption band at 1627 cm -1 is assigned to O-H bending. The spectra of the samples Gav-43 and Irk-53 differ in the 2800-4000 cm -1 spectral region (see the inset of Figure 11). In both samples, peaks at 3177 and 3428 cm −1 are attributed to stretching modes of H2O in different structural positions. However, intensities of these peaks in the samples Gav-43 and Irk-53 are different. It means that the distribution of H2O molecules on interlayer positions is different. The sharp lines at 3551, 3635, 3656 cm -1 could be assigned to OH-modes. The intensities are also different in the Gav-43 and Irk-53 samples. The Irk-53 sample contains a higher concentration of OHions than the Gav-43.
The infrared spectra features found in this study are in agreement with the previous ones reported for structurally related minerals of similar chemical composition. The positions of the bands (cm -1 ) in the IR spectra of studied fedorite samples, lalondeite, martinite and ellingsenite are summarized in Table S14 of Supplementary Materials.

Spectroscopy
The infrared spectra of the fedorite samples Gav-43 and Irk-53 are given in Figure 11 correspond to Mn 2+ ions (I = 5/2). Manganese ion substitutes a cation of calcium and surrounded by two or one fluorine ions. The intensity of the EPR signal of the sample Irk-53 is about three times higher than in the sample Gav-43. This is in good agreement with the average chemical composition shown in Table 1. Electron microprobe analysis revealed that the concentration of Mn in the sample Irk-53 is higher than in the sample Gav-43.  The sample Gav-43 demonstrates a wide photoluminescence peak at about 390 nm under 337 nm excitation (Figure 13). This luminescence has 70 ns decay time constant. The excitation spectrum monitored at 390 nm is given by a black solid curve. In the spectrum, five bands peaked at 217, 235, 300, 316, 337 nm appear. Luminescence having a close peak at 410 nm under laser or X-ray excitation was observed by Yarovoy et al. [42]. The authors assumed, based on alkali-earth fluoride emission data, that this luminescence was due to 5d-4f transitions in Eu 2+ ions substituting Ca 2+ ions. However, analysis of the excitation spectrum was not performed by [42]. The excitation spectrum in Figure 13 does not contain a europium typical "staircase" structure. Eu 2+ related emission decay time constant is usually about 400-1000 ns that is longer than the one observed in the fedorite. Furthermore, Eu 2+ centers have strong EPR signal, but it is not found in the studied samples. Therefore, the observed in our samples UV luminescence peaked at 390 nm cannot be related to Eu 2+ ions. However, this The infrared spectra features found in this study are in agreement with the previous ones reported for structurally related minerals of similar chemical composition. The positions of the bands (cm -1 ) in the IR spectra of studied fedorite samples, lalondeite, martinite and ellingsenite are summarized in Table S14 of Supplementary Materials.
In both fedorite samples, EPR signals were observed at room and 77 K temperatures ( Figure 12). The EPR spectra exist in a sextet structure with 90 Gauss splitting with g-factor of 2.001. These bands correspond to Mn 2+ ions (I = 5/2). Manganese ion substitutes a cation of calcium and surrounded by two or one fluorine ions. The intensity of the EPR signal of the sample Irk-53 is about three times higher than in the sample Gav-43. This is in good agreement with the average chemical composition shown in Table 1. Electron microprobe analysis revealed that the concentration of Mn in the sample Irk-53 is higher than in the sample Gav-43.
Minerals 2020, 10, x FOR PEER REVIEW 17 of 23 correspond to Mn 2+ ions (I = 5/2). Manganese ion substitutes a cation of calcium and surrounded by two or one fluorine ions. The intensity of the EPR signal of the sample Irk-53 is about three times higher than in the sample Gav-43. This is in good agreement with the average chemical composition shown in Table 1. Electron microprobe analysis revealed that the concentration of Mn in the sample Irk-53 is higher than in the sample Gav-43.  The sample Gav-43 demonstrates a wide photoluminescence peak at about 390 nm under 337 nm excitation ( Figure 13). This luminescence has 70 ns decay time constant. The excitation spectrum monitored at 390 nm is given by a black solid curve. In the spectrum, five bands peaked at 217, 235, 300, 316, 337 nm appear. Luminescence having a close peak at 410 nm under laser or X-ray excitation was observed by Yarovoy et al. [42]. The authors assumed, based on alkali-earth fluoride emission data, that this luminescence was due to 5d-4f transitions in Eu 2+ ions substituting Ca 2+ ions. However, analysis of the excitation spectrum was not performed by [42]. The excitation spectrum in Figure 13 does not contain a europium typical "staircase" structure. Eu 2+ related emission decay time constant is usually about 400-1000 ns that is longer than the one observed in the fedorite. Furthermore, Eu 2+ The sample Gav-43 demonstrates a wide photoluminescence peak at about 390 nm under 337 nm excitation ( Figure 13). This luminescence has 70 ns decay time constant. The excitation spectrum monitored at 390 nm is given by a black solid curve. In the spectrum, five bands peaked at 217, 235, 300, 316, 337 nm appear. Luminescence having a close peak at 410 nm under laser or X-ray excitation was observed by Yarovoy et al. [42]. The authors assumed, based on alkali-earth fluoride emission data, that this luminescence was due to 5d-4f transitions in Eu 2+ ions substituting Ca 2+ ions. However, analysis of the excitation spectrum was not performed by [42]. The excitation spectrum in Figure 13 does not contain a europium typical "staircase" structure. Eu 2+ related emission decay time constant is usually about 400-1000 ns that is longer than the one observed in the fedorite. Furthermore, Eu 2+ centers have strong EPR signal, but it is not found in the studied samples. Therefore, the observed in our samples UV luminescence peaked at 390 nm cannot be related to Eu 2+ ions. However, this luminescence is attributed to 5d-4f transition in Ce 3+ ion, where Ce 3+ ion substitutes Ca 2+ ion. The charge compensation mechanism is discussed below. A similar band was observed in agrellite having almost equivalent ligand field [32,43]. Contrary to the information in the [42], the luminescence of Eu 2+ peaked at 410 nm was not been observed in agrellite by [43]. Five peaks in the excitation spectrum correspond to transitions from the ground 4f state of the Ce 3+ ion to the crystal field split 5d state. Ce 3+ ions have a low symmetry point group, therefore, the 5d state is split into five levels [32,44]. In the sample Irk-53 Ce 3+ related luminescence is absent.
Minerals 2020, 10, x FOR PEER REVIEW 18 of 23 luminescence is attributed to 5d-4f transition in Ce 3+ ion, where Ce 3+ ion substitutes Ca 2+ ion. The charge compensation mechanism is discussed below. A similar band was observed in agrellite having almost equivalent ligand field [32,43]. Contrary to the information in the [42], the luminescence of Eu 2+ peaked at 410 nm was not been observed in agrellite by [43]. Five peaks in the excitation spectrum correspond to transitions from the ground 4f state of the Ce 3+ ion to the crystal field split 5d state. Ce 3+ ions have a low symmetry point group, therefore, the 5d state is split into five levels [32,44].
In the sample Irk-53 Ce 3+ related luminescence is absent. Optical absorption spectra in near-infrared, visible and ultraviolet spectral regions are given in Figure 14. The full range spectrum of a thin crystal plate is given in the inset of Figure 14. The strong intense peak at 255 nm and the fundamental absorption edge at 230 nm are found. The absorption band peak at 535 nm causes the sample Gav-43 to be a pale raspberry-red to bright mauve. In the sample Irk-53, this band has a lower intensity and it has not pure color. Under 405 nm excitation, both samples demonstrate a luminescence peak at 600 nm at room temperature ( Figure 15). This luminescence could be attributed Mn 2+ emission due to spin-forbidden transition 4 T1 → 6 A1, where Mn 2+ is located in the octahedral crystal field [45]. Yarovoy et al. [42] observed an emission peak at 530 nm. This luminescence band was attributed to Mn 2+ centers. However, in the Optical absorption spectra in near-infrared, visible and ultraviolet spectral regions are given in Figure 14. The full range spectrum of a thin crystal plate is given in the inset of Figure 14. The strong intense peak at 255 nm and the fundamental absorption edge at 230 nm are found. luminescence is attributed to 5d-4f transition in Ce 3+ ion, where Ce 3+ ion substitutes Ca 2+ ion. The charge compensation mechanism is discussed below. A similar band was observed in agrellite having almost equivalent ligand field [32,43]. Contrary to the information in the [42], the luminescence of Eu 2+ peaked at 410 nm was not been observed in agrellite by [43]. Five peaks in the excitation spectrum correspond to transitions from the ground 4f state of the Ce 3+ ion to the crystal field split 5d state. Ce 3+ ions have a low symmetry point group, therefore, the 5d state is split into five levels [32,44].
In the sample Irk-53 Ce 3+ related luminescence is absent. Optical absorption spectra in near-infrared, visible and ultraviolet spectral regions are given in Figure 14. The full range spectrum of a thin crystal plate is given in the inset of Figure 14. The strong intense peak at 255 nm and the fundamental absorption edge at 230 nm are found. The absorption band peak at 535 nm causes the sample Gav-43 to be a pale raspberry-red to bright mauve. In the sample Irk-53, this band has a lower intensity and it has not pure color. Under 405 nm excitation, both samples demonstrate a luminescence peak at 600 nm at room temperature ( Figure 15). This luminescence could be attributed Mn 2+ emission due to spin-forbidden transition 4 T1 Figure 14. Diffuse optical absorption spectra of the fedorite samples Gav-43 (dashed red curve) and Irk-53 (solid black curve). In the inset, the absorption spectrum of the crystal plate of the sample Gav-43 measured in the transmission regime is given. The absorption band peak at 535 nm causes the sample Gav-43 to be a pale raspberry-red to bright mauve. In the sample Irk-53, this band has a lower intensity and it has not pure color. Under 405 nm excitation, both samples demonstrate a luminescence peak at 600 nm at room temperature ( Figure 15). This luminescence could be attributed Mn 2+ emission due to spin-forbidden transition 4 T 1 → 6 A 1 , where Mn 2+ is located in the octahedral crystal field [45]. Yarovoy et al. [42] observed an emission peak at 530 nm. This luminescence band was attributed to Mn 2+ centers. However, in the later research of this group [46][47][48], other emission peaks at 570 and 600 nm related to Mn 2+ emission were observed. Such inconsistent results may be due to a lack of photomultiplier sensitivity correction in the earlier article [42]. A wide luminescence peak at 700 nm with long decay time about 600 µs appears in the sample Gav-43 at 77 K, but this luminescence is quenched at room temperature. The excitation spectrum of this emission coincides with the optical absorption spectrum. It has a peak at 535 nm and wide structureless plateau from 300 to 410 nm. Therefore, we can conclude that the absorption band at 535 nm and luminescence at 700 nm are related to one center. This corresponds to the luminescence of Mn 4+ centers in silicates and fluoride-containing materials [49][50][51]. According to the Tanabe-Sugano diagram [52], Mn 4+ could demonstrate broad absorption bands in blue and green spectral range in the octahedral crystal field. The emission is attributed to spin-forbidden 2 E → 4 A 2 transition and the spin-allowed 4 T 2 → 4 A 2 transition. A nonhomogeneous broadening of the spectrum can be expected due to positional disordering in the fedorite crystal structure. later research of this group [46][47][48], other emission peaks at 570 and 600 nm related to Mn 2+ emission were observed. Such inconsistent results may be due to a lack of photomultiplier sensitivity correction in the earlier article [42]. A wide luminescence peak at 700 nm with long decay time about 600 μs appears in the sample Gav-43 at 77 K, but this luminescence is quenched at room temperature. The excitation spectrum of this emission coincides with the optical absorption spectrum. It has a peak at 535 nm and wide structureless plateau from 300 to 410 nm. Therefore, we can conclude that the absorption band at 535 nm and luminescence at 700 nm are related to one center. This corresponds to the luminescence of Mn 4+ centers in silicates and fluoride-containing materials [49][50][51]. According to the Tanabe-Sugano diagram [52], Mn 4+ could demonstrate broad absorption bands in blue and green spectral range in the octahedral crystal field. The emission is attributed to spin-forbidden 2 E → 4 A2 transition and the spin-allowed 4 T2 → 4 A2 transition. A nonhomogeneous broadening of the spectrum can be expected due to positional disordering in the fedorite crystal structure.
On the other hand, the absorption band at 535 nm could be attributed to Mn 3+ centers [53]. However, Mn 3+ ions have not luminescence at 700 nm region. Therefore, the pale raspberry-red coloration of the sample Gav-43 could be due to the oxidation of Mn 2+ ions to Mn 4+ .
Mn 4+ ions can be stabilized in the crystal structure only in the octahedral coordination [45]. Therefore, the charge compensation may be reached as a result of heterovalent substitution Ce 3+ +Ca 2+ → Na + +Mn 4+ .
In [42,46,47], far-red emission peaks at 740 and 800 nm due to Fe 3+ centers were observed. The samples studied in our work do not contain trivalent iron ions. All iron ions in the studied fedorite samples are divalent and they have not emission. That fact is confirmed by the absence of a Fe 3+related EPR signal. Therefore, no emission from Fe 3+ centers is found in our samples.

Conclusion
Three different mineral associations with fedorite found out on the Murun massif were considered in the article.
The (1) is associated with brookite-anatase-feldspar-quartz rocks, while the (2) and (3) are joined with charoite mineralization. The mineral relationships and the form of their occurrences allowed us to determine the following sequence of the crystallization: aegirine-apatite-tinaksitefedorite-microcline-steacyite-quartz-charoite.
The study of melt and fluid inclusions in the minerals of charoite rocks, carried out by [54,55], showed that the crystallization of the studied associations occurred in a heterophasic system with the participation of carbonate-silicate and carbonate melts, gaseous and water-salt fluids. Moreover, early rock-forming minerals were formed at high temperatures (more than 800 °C), and the final On the other hand, the absorption band at 535 nm could be attributed to Mn 3+ centers [53]. However, Mn 3+ ions have not luminescence at 700 nm region. Therefore, the pale raspberry-red coloration of the sample Gav-43 could be due to the oxidation of Mn 2+ ions to Mn 4+ .
Mn 4+ ions can be stabilized in the crystal structure only in the octahedral coordination [45]. Therefore, the charge compensation may be reached as a result of heterovalent substitution Ce 3+ +Ca 2+ → Na + +Mn 4+ .
In [42,46,47], far-red emission peaks at 740 and 800 nm due to Fe 3+ centers were observed. The samples studied in our work do not contain trivalent iron ions. All iron ions in the studied fedorite samples are divalent and they have not emission. That fact is confirmed by the absence of a Fe 3+ -related EPR signal. Therefore, no emission from Fe 3+ centers is found in our samples.

Conclusion
Three different mineral associations with fedorite found out on the Murun massif were considered in the article.
The mineral relationships and the form of their occurrences allowed us to determine the following sequence of the crystallization: aegirine-apatite-tinaksite-fedorite-microcline-steacyite-quartz-charoite. The study of melt and fluid inclusions in the minerals of charoite rocks, carried out by [54,55], showed that the crystallization of the studied associations occurred in a heterophasic system with the participation of carbonate-silicate and carbonate melts, gaseous and water-salt fluids. Moreover, early rock-forming minerals were formed at high temperatures (more than 800 • C), and the final stages of crystallization took place at lower temperatures (600-450 • C). For brookite-anatase-quartz veins of the Murun massif, it has been proven that mineral associations with brookite, rutile, and sulfides crystallized at temperatures of 280-210 • C, while mineral parageneses of pyrite and galena formed within 255 and 210 • C [55]. Taking into account these data, we can conclude that fedorite from brookite-anatase-feldspar-quartz rocks of the Gavrilovskaya zone crystallized at lower temperatures compared to the fedorite from charoite rocks. This assumption is confirmed by low-temperature mineral paragenesis with rutile and sulfides, as well as the presence of a larger amount of water in the crystal structure of fedorite from the Gavrilovskaya zone. Previously, the conclusion about the low-temperature formation was made for other minerals of the reyerite-gyrolite group (e.g., [56]), and, specifically, for structurally close to fedorite phases, considered in this paper: lalondeite [39], martinite [40] and ellingsenite [41].
The crystal structure of fedorite is characterized by a higher regularity in the geometry of coordination polyhedra, but also by a higher disordering of interlayer ions. Finally, for the first time, EPR, optical absorption and emission spectra of Mn 2+ ions are obtained for fedorite. Some samples containing Mn and Ce have pale raspberry-red color due to presence of Mn 4+ ions.