Three-D Mineralogical Mapping of the Kovdor Phoscorite-Carbonatite Complex , NW Russia : II . Sulfides

The world largest phoscorite-carbonatite complexes of the Kovdor (Russia) and Palabora (South Africa) alkaline-ultrabasic massifs have comparable composition, structure and metallogenic specialization, and can be considered close relatives. Distribution of rock-forming sulfides within the Kovdor phoscorite-carbonatite complex reflects gradual concentric zonation of the pipe: pyrrhotite with exsolution inclusions of pentlandite in marginal (apatite)-forsterite phoscorite, pyrrhotite with exsolution inclusions of cobaltpentlandite in intermediate low-carbonate magnetite-rich phoscorite and chalcopyrite (±pyrrhotite with exsolution inclusions of cobaltpentlandite) in axial carbonate-rich phoscorite and phoscorite-related carbonatite. Chalcopyrite (with relicts of earlier bornite and exsolution inclusions of cubanite and mackinawite) predominates in the axial carbonate-bearing phoscorite and carbonatite, where it crystallizes around grains of pyrrhotite (with inclusions of pentlandite-cobaltpentlandite and pyrite), and both of these minerals contain exsolution inclusions of sphalerite. In natural sequence of the Kovdor rocks, iron content in pyrrhotite gradually increases from Fe7S8 (pyrrhotite-4C, Imm2) to Fe9S10 (pyrrhotite-5C, C2 and P21) and Fe11S12 (pyrrhotite-6C) due to gradual decrease of crystallization temperature and oxygen fugacity. Low-temperature pyrrhotite 2C (troilite) occurs as lens-like exsolition inclusions in grains of pyrrhotite-4C (in marginal phoscorite) and pyrrhotite-5C (in axial phoscorite-related carbonatite). Within the phoscorite-carbonatite complex, Co content in pyrrhotite gradually increases from host silicate rocks and marginal forsterite-dominant phoscorite to axial carbonate-rich phoscorite and carbonatite at the expense of Ni and Fe. Probably, this dependence reflects a gradually decreasing temperature of the primary monosulfide solid solutions crystallization from the pipe margin toward its axis. The Kovdor and Loolekop phoscorite-carbonatite pipes in the Palabora massif have similar sequences of sulfide formation, and the copper specialization of the Palabora massif can be caused by higher water content in its initial melt allowing it to dissolve much larger amounts of sulfur and, correspondingly, chalcophile metals.


Introduction
Sulfur, alongside H 2 O, CO 2 and halogens, is an important volatile constituent in magmas [1][2][3].Unlike all other major volatile elements, sulfur changes its oxidation state depending on the oxygen fugacity (f O 2 ) regime.The two most abundant forms of S in silicate melts are sulfide (S 2− ) and sulfate (S 6+ ) [4][5][6].The transition between these two oxidation states is very sharp and occurs over the range of oxygen fugacity between the FMQ equilibrium and 2logf O 2 units above (FMQ + 2).However, sulfur shows radically different behavior in these two oxidation states, and, consequently, the existing sulfur species may provide important insights into the evolution of magmas and mineralizing fluids.

Materials and Methods
About 550 core samples of phoscorite, carbonatites and host rocks were taken from 108 boreholes drilled within the Kovdor phoscorite-carbonatite complex.We analyzed thin polished sections of these core samples with a petrographic microscope to estimate textural characteristics, mineral relations and grain size of sulfides.Quantitative relations between magnetic and non-magnetic pyrrhotites were determined in polished sections with a nematic liquid crystal MBBA [37].Sulfide grain sizes were estimated with the Image Tool 3.0 program [38] as a mean equivalent circular diameter.
Bulk-rock samples were analyzed by the Tananaev Institute of Chemistry of KSC RAS (Apatity) by means of inductively coupled plasma-mass spectrometry (ICP-MS) performed with an ELAN 9000 DRC-e mass spectrometer (Perkin Elmer, Waltham, MA, USA).For the analyses, the samples were dissolved in a mixture of concentrated hydrofluoric and nitric acids with distillation of silicon and further addition of hydrogen peroxide to a cooled solution to suppress hydrolysis of polyvalent metals [39].
Cation contents were calculated with the MINAL program of D. Dolivo-Dobrovolsky [40].Statistical analyses were carried out with the STATISTICA 8.0 [41] and TableCurve 2.0 [42] programs.For the statistics, resulting values of the analyses below the detection limit (see Table 1) were considered to be 10 times lower than the limit.Geostatistical studies and 3D modelling were conducted with the MICROMINE 16 program [43].Interpolation was performed with ordinary kriging.Automatic 3D geological mapping (Figure 1a) was performed by means of conversion of the rocks chemical composition to mineral composition by logical computation [24].Crystal structures of pyrrhotite samples (00-01, 36/33, 00-10-41, 01-11-91, 00-51) were studied by means of Agilent Technologies Xcalibur Eos diffractometer operated at 50 kV and 40 mA.More than a hemisphere of three-dimensional data was collected at 100 K temperature by monochromatic MoKα X-radiation with frame widths of 1 • and 5-15 s count for each frame.The crystal structures were refined in standard and non-standard settings of different space groups: P-6m2, P312, P32, P-3m, P-3, P3m, Cmce, Cca, Cc, C222 1 , C2, P2 1 for the 5C-polytype and P-6m2, Aem2, Amm2, C2/m, C/m, C2, F2/d, Imm2 for the 4C-polytype by means of the SHELX program [44] incorporated in the OLEX2 program package [45].The final models were chosen according to the following criteria: absence of violating reflections, lower means of R-factors and GOOF, absence or low number of atoms with physically unrealistic anisotropic displacement parameters (without restraints).Crystal structures were refined with a pseudomerohedral twin model with [−1 0 0 0 −1 0 0 0 1 2] twining matrix for the C2 and P2 1 models.Empirical absorption correction was applied in the CrysAlisPro program [46] using spherical harmonics implemented in the SCALE3 ABSPACK scaling algorithm.The crystal structures were visualized with the Diamond 3.2f program [47].

Sulfur Mineralization
In primary silicate rocks of the Kovdor massif, average sulfur content is comparatively low (Table 2), but it becomes higher in apo-peridotite metasomatic diopsidite and phlogopitite.Within the phoscorite-carbonatite pipe, sulfur content increases from marginal low-carbonate phoscorite to axial calcite-rich phoscorite and calcite carbonatite (Figure 1b).Besides sulfur, the Kovdor sulfuric compounds include Fe, Co, Ni, Cu, Zn, Ag, Pb and Ba (see Table 1).Many of these metals initially concentrated in rock-forming forsterite (Ni), apatite (Ba), magnetite (Fe, Co, Zn) and pyrochlore (Pb), and demonstrate unclear or even negative correlation with sulfur.Nevertheless, sulfuric compounds inherited initial distribution of these metals in earlier minerals resulting in concentric distribution corresponding to the secondary sulfides and sulfates within the phoscorite-carbonatite complex.

Sulfur Mineralization
In primary silicate rocks of the Kovdor massif, average sulfur content is comparatively low (Table 2), but it becomes higher in apo-peridotite metasomatic diopsidite and phlogopitite.Within the phoscorite-carbonatite pipe, sulfur content increases from marginal low-carbonate phoscorite to axial calcite-rich phoscorite and calcite carbonatite (Figure 1b).Besides sulfur, the Kovdor sulfuric compounds include Fe, Co, Ni, Cu, Zn, Ag, Pb and Ba (see Table 1).Many of these metals initially concentrated in rock-forming forsterite (Ni), apatite (Ba), magnetite (Fe, Co, Zn) and pyrochlore (Pb), and demonstrate unclear or even negative correlation with sulfur.Nevertheless, sulfuric compounds inherited initial distribution of these metals in earlier minerals resulting in concentric distribution corresponding to the secondary sulfides and sulfates within the phoscorite-carbonatite complex.In the Kovdor rocks, major concentrators of sulfur are sulfides, mainly, pyrrhotite and chalcopyrite (Figure 1c), while sulfates, mainly, barite, are the products of low-temperature alteration of sulfides.Pyrrhotite is a common accessory to a rock-forming mineral of the most phoscorite and carbonatites varieties, apart from vein dolomite carbonatite and the related dolomite-magnetite-serpentine metasomatic rock.Chalcopyrite is closely associated with pyrrhotite, being predominantly concentrated in carbonate-rich phoscorite, phoscorite-related carbonatite and vein calcite carbonatite of the ore-pipe apical part (up to 20 modal %).In the apical part of the pipe axial zone, there is calcite-rich phoscorite with rock-forming chalcopyrite (up to 30 modal %).
In peridotite, rounded or irregularly shaped pyrrhotite grains (up to 120 µm in diameter) with exsolution inclusions of pentlandite fill interstices between rock-forming olivine and diopside.In foidolite, there are sporadic pyrrhotite grains in cancrinitized nepheline grains (Figure 2c) and late calcite veinlets.Apo-peridotite metasomatites, especially carbonatized and apatitized, contain up to 3 modal % of pyrrhotite that fills interstices between grains of rock-forming minerals in close association with chalcopyrite, magnetite and ilmenite (Figure 2d). 3 modal % of pyrrhotite that fills interstices between grains of rock-forming minerals in close association with chalcopyrite, magnetite and ilmenite (Figure 2d).As mentioned above, pyrrhotite is a predominant accessory to rock-forming sulfide of phoscorite and carbonatite.In marginal forsteritite and apatite-forsterite phoscorite, pyrrhotite (up to 0.2 modal %) occurs within and between forsterite grains as pseudohexagonal plate crystals (up to 500 μm in diameter) with exsolution lamellae of pentlandite inside (see Figure 2a), as well as As mentioned above, pyrrhotite is a predominant accessory to rock-forming sulfide of phoscorite and carbonatite.In marginal forsteritite and apatite-forsterite phoscorite, pyrrhotite (up to 0.2 modal %) occurs within and between forsterite grains as pseudohexagonal plate crystals (up to 500 µm in diameter) with exsolution lamellae of pentlandite inside (see Figure 2a), as well as irregularly shaped grains in close intergrowth with magnetite (Figure 2e).Content of pyrrhotite in phoscorite increases with growth of magnetite amount and reaches 8 modal % in intermediate low-carbonate magnetite-rich phoscorite.Herein, pyrrhotite usually forms close intergrowths with magnetite, and, together with valleriite, replaces earlier magnetite and fills fractures in its grains (Figure 2f).Such secondary pyrrhotite often inherits cubic inclusions of spinel from exsolved Mg-Al-rich magnetite.
Gradual transition of low-carbonate magnetite-rich phoscorite into axial calcite-rich phoscorite and phoscorite-related carbonatite is accompanied by growth of pyrrhotite content and disappearance of its connection with magnetite.In this type of rock, there are sulfide-rich areas (up to 8 m in diameter), where pyrrhotite (up to 50 modal %) and chalcopyrite cement separate grains of all other minerals (Figure 2g).Besides, pyrrhotite forms here porous rims around chalcopyrite grains (Figure 2h) and tabular to short prismatic crystals (up to 2 cm in diameter) with hexagonal prismatic {10-10} and pinacoidal {0001} faces.In selvages of vein calcite and dolomite carbonatites, pyrrhotite occurs as bronze-yellow platy crystals (up to 15 cm in diameter and 1.5 cm thick), partially replaced with pyrite and goethite (Figure 2i).
Mean equivalent circular diameter of pyrrhotite grains (Table 3) increases insignificantly from 80 µm in marginal (apatite)-forsterite phoscorite to 160 µm in intermediate low-carbonate magnetite-rich phoscorite, and then to 200 µm in axial calcite-rich phoscorite and carbonatite.This trend is similar to that of co-existing rock-forming and accessory minerals [21,26,27].3b), while cobalt and nickel are characterized by lognormal and exponential distributions correspondingly (Figure 3c,d).
Iron content in pyrrhotite (exclusive of exsolution troilite) gradually increases from earlier host silicate rocks to the latest vein carbonatites (Figure 4a), which is fully in accordance with gradual increase of pyrrhotite-5C fraction due to presence of pyrrhotite-4C in this sequence.Taking into account pyrrhotite structural data according to [48], where Fe 7 S 8 = Fe 2+  5 Fe 3+ 2 S 2− 8 , Fe 9 S 10 = Fe 2+  7 Fe 3+ 2 S 2− 10 and Fe 11 S 12 = Fe 2+ 9 Fe 3+ 2 S 2− 12 , one can see a gradual decrease of Fe 3+ /Me 2+ ratio in natural rock sequence of the Kovdor massif (Figure 4b), and from marginal (apatite)-forsterite phoscorite with pyrrhotite-4C to intermediate low-carbonate magnetite-rich phoscorite with pyrrhotite-5C and, finally, calcite-rich phoscorite and calcite carbonaite with pyrrhotite-5C and -6C (Figure 5a).Experimental details and crystallographic parameters of structurally different pyrrhotites are given in Table 4.The site occupancy factors (s.o.f.) were refined using the scattering curves for neutral atoms given in the International Tables for Crystallography [49].Due to the large scope of structural information (atom coordinates, bond-lengths, displacement parameters for samples 1-5), these data can be obtained from the CIFs, which are available as Supplementary Materials.
The pyrrhotite crystal structure was firstly described by Nils Alsen [50] in the space group P63/mmc with unit cell parameters a = 3.43, c = 5.68 Å.The description was based upon hexagonal-close packing stacked perpendicular to the c axis, where sulfur atoms approximately occupy the nodes, and iron atoms are regularly arranged in octahedral interstices of sulfur atoms [51].The vacancies between iron sites are in ordered arrangement, and confined to alternate layers of iron atoms and normal to the c axis [52,53].Formation of different pyrrhotite polytypes is caused by ordered arrangement of vacancies and/or Fe 2+ /Fe 3+ atoms that results in formation of superstructures [53].Random arrangement of vacancies leads to the formation of hexagonal 1C polytype, their partial ordering causes orthorhombic 4C modification, and fully ordered vacancy distribution produces monoclinic 5C, 6C and modulated NC structures.Up today, 4C, 5C, 6C pyrrhotite polytypes and modulated NC Experimental details and crystallographic parameters of structurally different pyrrhotites are given in Table 4.The site occupancy factors (s.o.f.) were refined using the scattering curves for neutral atoms given in the International Tables for Crystallography [49].Due to the large scope of structural information (atom coordinates, bond-lengths, displacement parameters for samples 1-5), these data can be obtained from the CIFs, which are available as Supplementary Materials.
The pyrrhotite crystal structure was firstly described by Nils Alsen [50] in the space group P6 3 /mmc with unit cell parameters a = 3.43, c = 5.68 Å.The description was based upon hexagonal-close packing stacked perpendicular to the c axis, where sulfur atoms approximately occupy the nodes, and iron atoms are regularly arranged in octahedral interstices of sulfur atoms [51].The vacancies between iron sites are in ordered arrangement, and confined to alternate layers of iron atoms and normal to the c axis [52,53].Formation of different pyrrhotite polytypes is caused by ordered arrangement of vacancies and/or Fe 2+ /Fe 3+ atoms that results in formation of superstructures [53].Random arrangement of vacancies leads to the formation of hexagonal 1C polytype, their partial ordering causes orthorhombic 4C modification, and fully ordered vacancy distribution produces monoclinic 5C, 6C and modulated NC structures.Up today, 4C, 5C, 6C pyrrhotite polytypes and modulated NC structures with N = 4.88, 5.5 etc. were reported [52,54,55].The N is defined as a number of supercell reflections along c* direction (Figure 6) situating between bright reflections of a hexagonal cell.structures with N = 4.88, 5.5 etc. were reported [52,54,55].The N is defined as a number of supercell reflections along c* direction (Figure 6) situating between bright reflections of a hexagonal cell.The crystal structure of ferrimagnetic pyrrhotite-4C, Fe 6.78 S 8 , was refined in the Imm2 space group.It differs from the previous refinements of crystal structure of stoichiometric Fe 7 S 8 in C2/c and C2 space groups [51,56] by distribution of vacancies in Fe sites (Figure 7) based on the observed differences in their chemical compositions.The Kovdor pyrrhotite-4C contains four independent iron sites with refined occupancies of 0.89, 0.75, 1 and 0.75 for Fe1, Fe2, Fe3 and Fe4 respectively.
In the crystal structure of the pyrrhotite-5C with the P2 1 symmetry, anomalously short distance 2.470 Å has been observed between Fe2 and Fe4 sites (Figure 8).There are two possible explanations for this fact: (1) one of these sites is vacant, while the second one is populated (their refined occupancies are 0.24 for the Fe2 site and 0.75 for the Fe4 site); (2) the Fe2-Fe4 interaction has at least partially bonding character.The same structural effect has been observed in [57].The crystal structure of ferrimagnetic pyrrhotite-4C, Fe6.78S8, was refined in the Imm2 space group.It differs from the previous refinements of crystal structure of stoichiometric Fe7S8 in C2/c and C2 space groups [51,56] by distribution of vacancies in Fe sites (Figure 7) based on the observed differences in their chemical compositions.The Kovdor pyrrhotite-4C contains four independent iron sites with refined occupancies of 0.89, 0.75, 1 and 0.75 for Fe1, Fe2, Fe3 and Fe4 respectively.
In the crystal structure of the pyrrhotite-5C with the P21 symmetry, anomalously short distance 2.470 Å has been observed between Fe2 and Fe4 sites (Figure 8).There are two possible explanations for this fact: (1) one of these sites is vacant, while the second one is populated (their refined occupancies are 0.24 for the Fe2 site and 0.75 for the Fe4 site); (2) the Fe2-Fe4 interaction has at least partially bonding character.The same structural effect has been observed in [57].Most samples of the Kovdor pyrrhotite-5C are crystallized in C2 space group.Their chemical composition widely varies, Fe8.84-8.99S10,which causes differences in occupancy of Fe-sites (Figure 9).From 6 to 9 iron sites (from 22 ones in C2 model) are observed.They have a different vacancy proportion ranging from 1% to 100%.Such variability of 5C polytype can explain its domination in the Kovdor massif (see Figure 3).In samples 1-5, the mean <Fe-S> distance ranges from 2.426 to 2.454 The crystal structure of ferrimagnetic pyrrhotite-4C, Fe6.78S8, was refined in the Imm2 space group.It differs from the previous refinements of crystal structure of stoichiometric Fe7S8 in C2/c and C2 space groups [51,56] by distribution of vacancies in Fe sites (Figure 7) based on the observed differences in their chemical compositions.The Kovdor pyrrhotite-4C contains four independent iron sites with refined occupancies of 0.89, 0.75, 1 and 0.75 for Fe1, Fe2, Fe3 and Fe4 respectively.
In the crystal structure of the pyrrhotite-5C with the P21 symmetry, anomalously short distance 2.470 Å has been observed between Fe2 and Fe4 sites (Figure 8).There are two possible explanations for this fact: (1) one of these sites is vacant, while the second one is populated (their refined occupancies are 0.24 for the Fe2 site and 0.75 for the Fe4 site); (2) the Fe2-Fe4 interaction has at least partially bonding character.The same structural effect has been observed in [57].Most samples of the Kovdor pyrrhotite-5C are crystallized in C2 space group.Their chemical composition widely varies, Fe8.84-8.99S10,which causes differences in occupancy of Fe-sites (Figure 9).From 6 to 9 iron sites (from 22 ones in C2 model) are observed.They have a different vacancy proportion ranging from 1% to 100%.Such variability of 5C polytype can explain its domination in the Kovdor massif (see Figure 3).In samples 1-5, the mean <Fe-S> distance ranges from 2.426 to 2.454 Most samples of the Kovdor pyrrhotite-5C are crystallized in C2 space group.Their chemical composition widely varies, Fe 8.84-8.99 S 10 , which causes differences in occupancy of Fe-sites (Figure 9).From 6 to 9 iron sites (from 22 ones in C2 model) are observed.They have a different vacancy proportion ranging from 1% to 100%.Such variability of 5C polytype can explain its domination in the Kovdor massif (see Figure 3).In samples 1-5, the mean <Fe-S> distance ranges from 2.426 to 2.454 Å, and the expected elongation of bonds can be compensated by partial incorporation of Fe 3+ in low-occupied sites [48,57].
Å, and the expected elongation of bonds can be compensated by partial incorporation of Fe 3+ in low-occupied sites [48,57].Similar to nickel content in the rocks (see Table 2), the amount of Ni in pyrrhotite gradually decreases from host foidolite and diopsidite-glimmerite to marginal (apatite)-forsterite phoscorite, intermediate low-carbonate magnetite-rich phoscorite, calcite-rich phoscorite, phoscorite-related and vein calcite carbonatite, and then insignificantly increases in vein dolomite carbonatite and the related dolomite-magnetite-serpentine rock.In contrast, content of Co both in rock and pyrrhotite increases from host silicate rocks and marginal calcite-poor phoscorite to calcite-bearing phoscorite and phoscorite-related carbonatite and then slightly decreases in the vein carbonatites (Figure 4c).Correspondingly, Ni-dominant pyrrhotite is concentrated in the marginal zone of the phoscoritecarbonatite complex, Ni-Co-bearing pyrrhotite occurs in the intermediate zone, and Co-dominant pyrrhotite occupies the axial zone of the pipe (Figure 5b).
Exsolution inclusions of pentlandite in pyrrhotite occur mainly in host silicate rocks and (apatite)-forsterite phoscorite of the phoscorite-carbonatite complex marginal zone, and independent pentlandite grains (up to 120 μm in diameter) occur rarely in (аpatite)-forsterite phoscorite and dolomite-magnetite-serpentine rock.Pyrrhotite with inclusions of cobaltpentlandite dominates in the pipe axial carbonate-rich zone, while intermediate low-carbonate magnetite-rich phoscorite usually contains pyrrhotite with inclusions of both pentlandite and cobaltpentlandite with approximately equal proportions of Co and Ni-Fe.Exsolution pentlandite-cobaltpentlandite forms distinctive flame-like (up to 200 μm long, Figure 2b) round or hexangular lamellar inclusions (up to 100 μm in diameter and 10 μm thick, Figures 2a and 10b) that predominantly grow along {0001} planes of host Troilite occurs only as lens-like lamellae along (001) planes of host pyrrhotite from marginal (apatite)-forsterite phoscorite (see Figure 5a) and, rarely, phoscorite-related carbonatite (Figure 10a).The host pyrrhotite is predominantly represented by its 4C modification in marginal (apatite)-forsterite phoscorite, and by 5C modification in axial phoscorite-related carbonatite.In host silicate rocks and vein calcite and dolomite carbonatites, troilite has not been found.
Exsolution inclusions of pentlandite in pyrrhotite occur mainly in host silicate rocks and (apatite)-forsterite phoscorite of the phoscorite-carbonatite complex marginal zone, and independent pentlandite grains (up to 120 µm in diameter) occur rarely in (apatite)-forsterite phoscorite and dolomite-magnetite-serpentine rock.Pyrrhotite with inclusions of cobaltpentlandite dominates in the pipe axial carbonate-rich zone, while intermediate low-carbonate magnetite-rich phoscorite usually contains pyrrhotite with inclusions of both pentlandite and cobaltpentlandite with approximately equal proportions of Co and Ni-Fe.Exsolution pentlandite-cobaltpentlandite forms distinctive flame-like (up to 200 µm long, Figure 2b) round or hexangular lamellar inclusions (up to 100 µm in diameter and 10 µm thick, Figures 2a and 10b) that predominantly grow along {0001} planes of host pyrrhotite from boundaries of its grains and fissures.Sometimes, the oriented inclusions of pentlandite-cobaltpentlandite are found in pseudomorphs of pyrite, valleriite, carbonates (see Figure 10b) and secondary magnetite after pyrrhotite.
Minerals 2018, 8, x FOR PEER REVIEW 13 of 30 pyrrhotite from boundaries of its grains and fissures.Sometimes, the oriented inclusions of pentlandite-cobaltpentlandite are found in pseudomorphs of pyrite, valleriite, carbonates (see Figure 10b) and secondary magnetite after pyrrhotite.Chemical composition of pentlandite-cobaltpentlandite varies in all possible ranges (Table 5).Increase of Co/Ni-ratio in pyrrhotite causes logarithmical growth of this ratio in exsolution pentlandite-cobaltpentlandite (Figure 11).In the sequence of rock formation, there is a gradual increase of Co content in pentlandite-cobaltpentlandite due to Ni and Fe (see Figure 4c), which causes corresponding spatial zonation of the phoscorite-carbonatite complex (see Figure 5c).Chemical composition of pentlandite-cobaltpentlandite varies in all possible ranges (Table 5).Increase of Co/Ni-ratio in pyrrhotite causes logarithmical growth of this ratio in exsolution pentlandite-cobaltpentlandite (Figure 11).In the sequence of rock formation, there is a gradual increase of Co content in pentlandite-cobaltpentlandite due to Ni and Fe (see Figure 4c), which causes corresponding spatial zonation of the phoscorite-carbonatite complex (see Figure 5c).On the M9S8 plane of the Fe-Ni-Co-S tetrahedron (Figure 12), chemical compositions of pentlandite are concentrated within the field of the solid-solution stability at ≥200 °C [58].Chemical compositions of cobaltpentlandite correspond to the Fe4.5Ni4.5S-Co9S8trend, whereas most samples from host silicate rocks, marginal forsterite-dominant and intermediate low-carbonate magnetite-rich phoscorite are within the stability field at 300-400 °C, and most samples from axial carbonate rich phoscorite and carbonatites are within the stability field at 200-400 °C.Comparatively rare exsolution sphalerite occurs as rounded, irregularly shaped, cross-and butterfly-like inclusions (up to 100 μm, Figure 10c) in pyrrhotite from any types of rocks.Sphalerite together with chalcopyrite also form rims (up to 120 μm thick) around pyrrhotite grains in dolomite-magnetite-serpentine rock.Since chemical composition of exsolution sphalerite (Table 6) at temperature below 500 °C depends on pressure only [59][60][61], we calculate the pressure using the following equation [62]: where FeSSp is in mol %, and P is in kbar.The results (see Table 6) shows gradual pressure decrease from 4 kbar in marginal (apatite)-forsterite phoscorite to 2 kbar in axial carbonate-rich phoscorite and phoscorite-related carbonatite, and then again growth to 4 kbar in vein calcite carbonatite, and decrease to 2 kbar in vein dolomite carbonatite (i.e., decrease by 2 kbar from earlier to the latest rock within both pipe and vein series).It is necessary to note similar behavior of oxygen fugacity [26]  On the M9S8 plane of the Fe-Ni-Co-S tetrahedron (Figure 12), chemical compositions of pentlandite are concentrated within the field of the solid-solution stability at ≥200 °C [58].Chemical compositions of cobaltpentlandite correspond to the Fe4.5Ni4.5S-Co9S8trend, whereas most samples from host silicate rocks, marginal forsterite-dominant and intermediate low-carbonate magnetite-rich phoscorite are within the stability field at 300-400 °C, and most samples from axial carbonate rich phoscorite and carbonatites are within the stability field at 200-400 °C.Comparatively rare exsolution sphalerite occurs as rounded, irregularly shaped, cross-and butterfly-like inclusions (up to 100 μm, Figure 10c) in pyrrhotite from any types of rocks.Sphalerite together with chalcopyrite also form rims (up to 120 μm thick) around pyrrhotite grains in dolomite-magnetite-serpentine rock.Since chemical composition of exsolution sphalerite (Table 6) at temperature below 500 °C depends on pressure only [59][60][61], we calculate the pressure using the following equation [62]: where FeSSp is in mol %, and P is in kbar.The results (see Table 6) shows gradual pressure decrease from 4 kbar in marginal (apatite)-forsterite phoscorite to 2 kbar in axial carbonate-rich phoscorite and phoscorite-related carbonatite, and then again growth to 4 kbar in vein calcite carbonatite, and decrease to 2 kbar in vein dolomite carbonatite (i.e., decrease by 2 kbar from earlier to the latest rock within both pipe and vein series).It is necessary to note similar behavior of oxygen fugacity [26] Figure 12.Relationship between composition of the Kovdor's pentlandite-cobaltpentlandite and fields of pentlandite-cobaltpentlandite stability after [58].
Comparatively rare exsolution sphalerite occurs as rounded, irregularly shaped, cross-and butterfly-like inclusions (up to 100 µm, Figure 10c) in pyrrhotite from any types of rocks.Sphalerite together with chalcopyrite also form rims (up to 120 µm thick) around pyrrhotite grains in dolomite-magnetite-serpentine rock.Since chemical composition of exsolution sphalerite (Table 6) at temperature below 500 • C depends on pressure only [59][60][61], we calculate the pressure using the following equation [62]: FeS Sp = 20.53− 1.313P + 0.0271P It is necessarily to note that pyrrhotite-4C may become stable at about 140 • C in nature [67], while 6C and 5C superstructures can be formed at temperatures below approx.60 • C [62].Therefore, in natural rock sequence, the ratio of pyrrhotite-5C + 6C to pyrrhotite-4C (see Figure 4a,b) gradually increases due to the gradual decrease of the pyrrhotite superstructure formation temperature.These hypotheses are in good agreement with the results of calcite-dolomite and ilmenite-magnetite geothermometry for this rock sequence [26], where the average temperature of the mineral equilibration gradually decreases from 500 • C to 400 • C for carbonates and from 500 • C to 300 • C for oxides.
Table 7.Chemical composition of pyrite inclusions in pyrrhotite (mean ± SD/min-max), and temperature of their formation estimated with a Co-exchange pyrite-pyrrhotite geothermometer [66].Besides exsolution inclusions of troilite, pentlandite-cobaltpentlandite, sphalerite and pyrite, there are unit inclusions of galena, mackinawite, clausthalite, violarite, hessite, moncheite and petzite (Table 8) in pyrrhotite grains.Mackinawite is a typical mineral for axial calcite-rich phoscorite and phoscorite-related carbonatite, where it forms equant or wedge-shaped inclusions (up to 100 µm long) in pyrrhotite and, especially, chalcopyrite.Violarite is found together with pentlandite in dolomite-magnetite-serpentine rock as lens-like inclusions (up to 30 µm long and 2 µm thick) oriented along the {0001} planes of host pyrrhotite.Both these minerals are not independent from pyrrhotite.Galena forms rounded inclusions (up to 20 µm in diameter) in peripheral parts of pyrrhotite grains and pyrrhotite-chalcopyrite boundaries.In nepheline-bearing diopsidite, pyrrhotite contains tabular inclusions of moncheite (up to 15 µm in diameter and 4 µm thick), and irregularly shaped polyphase inclusions of hessite, petzite, galena and clausthalite (up to 6 µm in diameter).All these inclusions may result from pyrrhotite self-cleaning from the corresponding impurities during rock cooling.

Chalcopyrite and Products of Its Alteration
In natural sequence of the Kovdor rocks, content of copper gradually increases from host silicate rocks to earlier (apatite)-forsterite phoscorite, intermediate low-carbonate magnetite-rich phoscorite, and then to the latest calcite-rich phoscorite and carbonatites (see Table 2).On this reason, chalcopyrite content and grain size increase with growth of carbonate amount in the rock (up to 15 modal % and 1 cm in diameter, Figure 14a).In fact, this mineral is absent in peridotite and, rarer, in non-altered foidolite.In hydrothermally altered ijolite-urtite, it occurs as inclusions (up to 40 μm in diameter) in newly formed cancrinite (see Figure 2c) and calcite.Apo-peridotite diopsidite and phlogopitite contain irregularly shaped chalcopyrite grains (up to 1 mm in diameter) associated with pyrrhotite, magnetite and ilmenite in interstices of rock-forming silicates and apatite (see Figure 2d).

Chalcopyrite and Products of Its Alteration
In natural sequence of the Kovdor rocks, content of copper gradually increases from host silicate rocks to earlier (apatite)-forsterite phoscorite, intermediate low-carbonate magnetite-rich phoscorite, and then to the latest calcite-rich phoscorite and carbonatites (see Table 2).On this reason, chalcopyrite content and grain size increase with growth of carbonate amount in the rock (up to 15 modal % and 1 cm in diameter, Figure 14a).In fact, this mineral is absent in peridotite and, rarer, in non-altered foidolite.In hydrothermally altered ijolite-urtite, it occurs as inclusions (up to 40 µm in diameter) in newly formed cancrinite (see Figure 2c) and calcite.Apo-peridotite diopsidite and phlogopitite contain irregularly shaped chalcopyrite grains (up to 1 mm in diameter) associated with pyrrhotite, magnetite and ilmenite in interstices of rock-forming silicates and apatite (see Figure 2d).Marginal (apatite)-forsterite phoscorite is usually free from chalcopyrite.In intermediate low-carbonate phoscorite, chalcopyrite, together with pyrrhotite, fills interstices between grains of earlier forsterite, phlogopite, hydroxylapatite and magnetite (Figure 14b).Chalcopyrite usually occurs in marginal parts of pyrrhotite-chalcopyrite segregations or forms irregularly shaped gulf-like inclusions growing inside pyrrhotite grains from their margins.It also fills cleavage fractures in phlogopite, impregnates serpentine pseudomorphs after forsterite, accompanies late dolomite veinlets and segregations.In voids of vein dolomite carbonatite, late chalcopyrite forms druses of well-shaped tetrahedral crystals (up to 1 mm in diameter) in association with pyrite, anatase and titanite [11].
In calcite-rich phoscorite and carbonatite, chalcopyrite forms irregularly shaped chalcopyrite grains (up to 1 cm in diameter, see Figure 14a) partially replaced and rimmed by fine-grained pyrrhotite (see Figure 2h), as well as inclusions in magnetite, calcite and dolomite.Сhalcopyrite grains sometimes contain irregularly shaped relicts (up to 15 μm in diameter, Figure 14c) of high bornite (cubic, a = 5.47 Å) accompanied by newly formed covellite [11].
Chemical composition of chalcopyrite varies insignificantly (Table 10), and its averaged formula corresponds to a theoretical one.Nevertheless, like pyrrhotite, chalcopyrite contains different exsolution inclusions (Table 11); therefore, we can assume that its initial composition is more complex.In particular, chalcopyrite grains usually carry sharply bounded cross-or star-like exsolution inclusions of sphalerite (up to 100 μm in diameter), as well as its gradually bounded rims around chalcopyrite grains and veinlets within them (up to 20 μm thick, Figure 14d).In axial carbonate-rich phoscorite and phoscorite-related carbonatite, there are exsolution inclusions of Marginal (apatite)-forsterite phoscorite is usually free from chalcopyrite.In intermediate low-carbonate phoscorite, chalcopyrite, together with pyrrhotite, fills interstices between grains of earlier forsterite, phlogopite, hydroxylapatite and magnetite (Figure 14b).Chalcopyrite usually occurs in marginal parts of pyrrhotite-chalcopyrite segregations or forms irregularly shaped gulf-like inclusions growing inside pyrrhotite grains from their margins.It also fills cleavage fractures in phlogopite, impregnates serpentine pseudomorphs after forsterite, accompanies late dolomite veinlets and segregations.In voids of vein dolomite carbonatite, late chalcopyrite forms druses of well-shaped tetrahedral crystals (up to 1 mm in diameter) in association with pyrite, anatase and titanite [11].
In calcite-rich phoscorite and carbonatite, chalcopyrite forms irregularly shaped chalcopyrite grains (up to 1 cm in diameter, see Figure 14a) partially replaced and rimmed by fine-grained pyrrhotite (see Figure 2h), as well as inclusions in magnetite, calcite and dolomite.Chalcopyrite grains sometimes contain irregularly shaped relicts (up to 15 µm in diameter, Figure 14c) of high bornite (cubic, a = 5.47 Å) accompanied by newly formed covellite [11].
Chemical composition of chalcopyrite varies insignificantly (Table 10), and its averaged formula corresponds to a theoretical one.Nevertheless, like pyrrhotite, chalcopyrite contains different exsolution inclusions (Table 11); therefore, we can assume that its initial composition is more complex.In particular, chalcopyrite grains usually carry sharply bounded cross-or star-like exsolution inclusions of sphalerite (up to 100 µm in diameter), as well as its gradually bounded rims around chalcopyrite grains and veinlets within them (up to 20 µm thick, Figure 14d).In axial carbonate-rich phoscorite and phoscorite-related carbonatite, there are exsolution inclusions of cubanite in mackinawite-bearing chalcopyrite (Figure 14e).
Besides, chalcopyrite may contain fine impregnation of volynskite, tsumoite and native silver.Volynskite and tsumoite were found in calcite-magnetite phoscorite as thin intergrowth (up to 5 µm in diameter) within chalcopyrite inclusions in magnetite.Native silver forms thin (<1 µm thick) veinlets and xenomorphic inclusions (up to 3 µm in diameter) in chalcopyrite grains, especially, at the contacts between chalcopyrite grains and pyrrhotite, sphalerite.

Discussion
The Kovdor phoscorite-carbonatite complex has gradual concentric zonation in terms of the rocks modal and chemical composition, as well as grain size, chemical composition, crystallochemical features and properties of rock-forming and accessory minerals [11,21,22,24,[26][27][28][29].This zonation mainly results from gradual change in chemical composition of the residual phoscorite melt from silicate-rich to carbonate-rich during crystallization from the pipe margins to its axial zone [22].
Geochemically, nickel behaves like Mg and Fe 2+ , and readily substitutes them in most crystalline phases including olivine [71].Besides, there is a strong negative dependence of Ni activity on FeO tot content in silicate melt [72,73].However, even in a high-magnesian system, like the Kovdor peridotite and forsteritite, nickel can remain in the residual silicate melt (SM) because its partition coefficient D Ol/SM depends significantly on sulfur content [74].In fact, at fixed P-T conditions, the partition coefficients of chalcophile elements M between monosulfide solid solution (MSS), (Fe,Ni,Co) 1-x S, and silicate melt are controlled by the ratio f S 2 /f O 2 , which, in turn, depends on FeO tot content in sulfide-saturated silicate melt [75][76][77] As a result, the MSS-SM partition coefficients for Ni, Cu and Co substantially increase with FeO tot decrease in silicate melt (Figure 15a), and we can conclude that crystallization of Mg-rich (apatite)-forsterite phoscorite can produce Fe-Ni-Co-Cu-rich sulfide melt.mainly results from gradual change in chemical composition of the residual phoscorite melt from silicate-rich to carbonate-rich during crystallization from the pipe margins to its axial zone [22].Geochemically, nickel behaves like Mg and Fe 2+ , and readily substitutes them in most crystalline phases including olivine [71].Besides, there is a strong negative dependence of Ni activity on FeOtot content in silicate melt [72,73].However, even in a high-magnesian system, like the Kovdor peridotite and forsteritite, nickel can remain in the residual silicate melt (SM) because its partition coefficient D Ol/SM depends significantly on sulfur content [74].In fact, at fixed P-T conditions, the partition coefficients of chalcophile elements M between monosulfide solid solution (MSS), (Fe,Ni,Co)1-xS, and silicate melt are controlled by the ratio fS2/fO2, which, in turn, depends on FeOtot content in sulfide-saturated silicate melt [75][76][77]: (Fe,M)O(SM) + 0.5S2 = (Fe,M)S(MSS) + 0.5O2,

MO(SM) + FeS(MMS) = MS(MSS) + FeO(SM).
As a result, the MSS-SM partition coefficients for Ni, Cu and Co substantially increase with FeOtot decrease in silicate melt (Figure 15a), and we can conclude that crystallization of Mg-rich (apatite)-forsterite phoscorite can produce Fe-Ni-Co-Cu-rich sulfide melt.In earlier low-carbonate phoscorite, pyrrhotite is formed at the final stage of magnetite crystallization, because comparatively low sulfur content (see Table 2) still requires melt saturation in FeS complexes.For this reason, ferrimagnetic pyrrhotite-4C (or the corresponding MSS) first closely associates with magnetite forming gulf-like ingrowths inside its grains.Then, with transition to carbonate-rich phoscorite and carbonatite, iron-rich pyrrhotite-5-6C (or the corresponding MSS) crystallizes independently of magnetite.Increase of mean equivalent circular diameter of pyrrhotite grains reflects a decrease of the rock crystallization rate from the pipe wall to its axis.This trend is similar to that of co-existing rock-forming and accessory minerals, in particular, forsterite, magnetite and baddeleyite [22,26,27].
Our data on crystal structure of the Kovdor pyrrhotites are consistent with numerous previous studies of the structure [48,51,[54][55][56][57]78], which demonstrate different space groups for the most common 4C, 5C and 6C polytypes.Such structural diversity is caused by different P-T conditions of crystallization, amount of incorporated Fe 3+ or different mechanisms of vacancies distribution in the structure.According to the structural complexity classification [79,80], 4C polytype is "simple", while 5C and 6C modifications are "intermediate" (Figure 16).Relatively low-temperature formation of pyrrhotite-5C in comparison with pyrrhotite-4C is in good agreement with a common tendency of structural complexity growth with crystallization temperature decrease.From this point of view, use of the term 'pyrrhotite-nC' for 5C and 6C polytypes is justified by the absence of significant differences between their complexity.In earlier low-carbonate phoscorite, pyrrhotite is formed at the final stage of magnetite crystallization, because comparatively low sulfur content (see Table 2) still requires melt saturation in FeS complexes.For this reason, ferrimagnetic pyrrhotite-4C (or the corresponding MSS) first closely associates with magnetite forming gulf-like ingrowths inside its grains.Then, with transition to carbonate-rich phoscorite and carbonatite, iron-rich pyrrhotite-5-6C (or the corresponding MSS) crystallizes independently of magnetite.Increase of mean equivalent circular diameter of pyrrhotite grains reflects a decrease of the rock crystallization rate from the pipe wall to its axis.This trend is similar to that of co-existing rock-forming and accessory minerals, in particular, forsterite, magnetite and baddeleyite [22,26,27].
Our data on crystal structure of the Kovdor pyrrhotites are consistent with numerous previous studies of the structure [48,51,[54][55][56][57]78], which demonstrate different space groups for the most common 4C, 5C and 6C polytypes.Such structural diversity is caused by different P-T conditions of crystallization, amount of incorporated Fe 3+ or different mechanisms of vacancies distribution in the structure.According to the structural complexity classification [79,80], 4C polytype is "simple", while 5C and 6C modifications are "intermediate" (Figure 16).Relatively low-temperature formation of pyrrhotite-5C in comparison with pyrrhotite-4C is in good agreement with a common tendency of structural complexity growth with crystallization temperature decrease.From this point of view, use of the term 'pyrrhotite-nC' for 5C and 6C polytypes is justified by the absence of significant differences between their complexity.Decrease of monosulfide crystallization temperature causes growth of D MSS/SM for Cu and Co (Figure 15b), while Ni partition coefficient remains the same or even decreases [77,81], which enables fractionation of Ni and Cu-Co between low-and high-temperature sulfides respectively.Therefore, we believe that concentrically zoned distribution of pyrrhotite with different Fe, Ni and Co contents is a response to a gradual decrease of crystallization temperature from the ore-pipe marginal forsterite-dominant zone to its axial carbonate-rich zone.Exsolution of MSS or the corresponding higher-temperature pyrrhotite modifications produced lamella or flames of pentlandite in the ore-pipe marginal zone, and cobaltpentlandite in the pipe axial zone.
For the same reason, copper is concentrated in the pipe axial carbonate-rich zone.Carbonate melt appears to contain a significant amount of water [82] and copper as stable hydrosulfide complexes [83].Their destruction by bornite and chalcopyrite crystallization after carbonates caused intensive hydrothermal alteration of associated minerals, and formation of secondary pyrite, valleriite and other phases.
It is interesting to compare sulfide mineralization of the phoscorite-carbonatite complexes of the Kovdor (Russia) and Palabora (South Africa) alkaline-ultrabasic massifs that have comparable composition, structure and metallogenic specialization [12].However, the Loolekop phoscorite-carbonatite pipe in the Palabora massif is the largest carbonatite-hosted copper deposit in the world 850 Mt @ 0.5% Cu [84], and the Kovdor phoscorite-carbonatite complex has comparatively low copper content.What caused such a significant enrichment of the Loolekop phoscorite and carbonatites in copper?
The Loolekop phoscorite-carbonatite pipe is a close relative to the Kovdor one.The Loolekop pipe (1.4 × 0.8 km) is situated in the eastern part of the Proterozoic central-type Palabora complex of apo-peridotite serpentinite (central stock), shonkinite (outer ring intrusion) and the related diopsidite and phlogopitite between them.It has a clear concentric zonation with marginal low-carbonate phoscorite, intermediate "banded" zone of interlayered carbonate-rich phoscorite and phoscorite-related calcite carbonatite and axial stockwork of "transgressive" calcite and dolomite carbonatite [85][86][87].
All these rocks are rich in pyrrhotite with exsolution inclusions of pentlandite-cobaltpentlandite [87], and contain also sufficient amount of copper sulfides: dominant bornite and chalcocite in phoscorite and banded carbonatite and dominant chalcopyrite with relicts of bornite and exsolution inclusions of cubanite in transgressive carbonatite [85,86,88,89].Minor copper minerals include covellite, valleriite, tetrahedrite, and native copper [12,85,90].As a result, copper content increases from <0.3 wt % in marginal phoscorite to 0.3-0.9wt % in intermediate banded carbonatite and about 1 wt % in axial transgressive carbonatite [85,86].Decrease of monosulfide crystallization temperature causes growth of D MSS/SM for Cu and Co (Figure 15b), while Ni partition coefficient remains the same or even decreases [77,81], which enables fractionation of Ni and Cu-Co between low-and high-temperature sulfides respectively.Therefore, we believe that concentrically zoned distribution of pyrrhotite with different Fe, Ni and Co contents is a response to a gradual decrease of crystallization temperature from the ore-pipe marginal forsterite-dominant zone to its axial carbonate-rich zone.Exsolution of MSS or the corresponding higher-temperature pyrrhotite modifications produced lamella or flames of pentlandite in the ore-pipe marginal zone, and cobaltpentlandite in the pipe axial zone.
For the same reason, copper is concentrated in the pipe axial carbonate-rich zone.Carbonate melt appears to contain a significant amount of water [82] and copper as stable hydrosulfide complexes [83].Their destruction by bornite and chalcopyrite crystallization after carbonates caused intensive hydrothermal alteration of associated minerals, and formation of secondary pyrite, valleriite and other phases.
It is interesting to compare sulfide mineralization of the phoscorite-carbonatite complexes of the Kovdor (Russia) and Palabora (South Africa) alkaline-ultrabasic massifs that have comparable composition, structure and metallogenic specialization [12].However, the Loolekop phoscorite-carbonatite pipe in the Palabora massif is the largest carbonatite-hosted copper deposit in the world 850 Mt @ 0.5% Cu [84], and the Kovdor phoscorite-carbonatite complex has comparatively low copper content.What caused such a significant enrichment of the Loolekop phoscorite and carbonatites in copper?
The Loolekop phoscorite-carbonatite pipe is a close relative to the Kovdor one.The Loolekop pipe (1.4 km × 0.8 km) is situated in the eastern part of the Proterozoic central-type Palabora complex of apo-peridotite serpentinite (central stock), shonkinite (outer ring intrusion) and the related diopsidite and phlogopitite between them.It has a clear concentric zonation with marginal low-carbonate phoscorite, intermediate "banded" zone of interlayered carbonate-rich phoscorite and phoscorite-related calcite carbonatite and axial stockwork of "transgressive" calcite and dolomite carbonatite [85][86][87].
In phoscorite and phoscorite-related "banded" carbonatite, sulfides fill interstices and thin fractures in magnetite-apatite-forsterite-calcite aggregate, and form irregularly shaped segregations with pyrrhotite in core and copper-bearing sulfides in marginal zone, while in the late "transgressive" carbonatite, they form subparallel vertical lenses and veinlets (usually, up to 3 cm thick) within a vertical ore-zone with cross-section about 200 m × 600 m [85,87].In both rocks, pyrrhotite and pentlandite-cobaltpentlandite are the earliest sulfides, high bornite is the next one, and chalcopyrite and other minerals follow them.
Thus, the Kovdor and Loolekop phoscorite-carbonatite pipes have similar sequences of sulfide formation (Figure 17a), from earlier pyrrhotite to intermediate copper sulfides and late pyrite and valleriite.There are only differences in positions of argentopentlandite and mackinawite that are common products of chalcopyrite exsolution in the Kovdor complex, and substitute pentlandite in the Loolekop pipe [12,87,89].It is necessary to note that secondary djerfisherite is widely spread in the Kovdor complex due to higher alkalinity of this massif.In phoscorite and phoscorite-related "banded" carbonatite, sulfides fill interstices and thin fractures in magnetite-apatite-forsterite-calcite aggregate, and form irregularly shaped segregations with pyrrhotite in core and copper-bearing sulfides in marginal zone, while in the late "transgressive" carbonatite, they form subparallel vertical lenses and veinlets (usually, up to 3 cm thick) within a vertical ore-zone with cross-section about 200 × 600 m [85,87].In both rocks, pyrrhotite and pentlandite-cobaltpentlandite are the earliest sulfides, high bornite is the next one, and chalcopyrite and other minerals follow them.
Thus, the Kovdor and Loolekop phoscorite-carbonatite pipes have similar sequences of sulfide formation (Figure 17a), from earlier pyrrhotite to intermediate copper sulfides and late pyrite and valleriite.There are only differences in positions of argentopentlandite and mackinawite that are common products of chalcopyrite exsolution in the Kovdor complex, and substitute pentlandite in the Loolekop pipe [12,87,89].It is necessary to note that secondary djerfisherite is widely spread in the Kovdor complex due to higher alkalinity of this massif.The sequences of sulfide formation in the phoscorite-carbonatite pipes are in a good accordance with the upper limits of their stability [62] that linearly decrease from 1100 °C for high bornite to 103 °C for chalcocite (Figure 17b).For the Kovdor complex, these relations can be supplemented with temperature of pyrite-pyrrhotite equilibration ranged from 170 °C in intermediate low-carbonate magnetite-rich phoscorite and dolomite carbonatite to 300 °C in axial carbonate-rich phoscorite and phoscorite-related carbonatite, and temperatures of the pyrrhotite superstructure formation from 140 °C for pyrrhotite-4C to 60 °C for pyrrhotite-5C [62].
One of the most likely reasons for sulfide specialization in the Loolekop phoscorite-carbonatite pipe in comparison with the Kovdor one is insignificant differences in oxygen fugacity of initial phoscorite melts (due to higher water content), because the larger oxygen fugacity causes more intensive copper extraction by melt from surrounding rocks [5,[91][92][93].In fact, even a slight increase in oxidation state (to ≥FMQ + 1) can shift the balance between sulfide and sulfate components dissolved in melt towards more soluble sulfates, allowing magmas to dissolve much larger amounts of sulfur than reduced melts [94].Higher water content in the Palabora's magma is confirmed, in particular, by much deeper alteration of peridotite and phoscorite into serpentinite, much abundant pegmatites, much richer valleriite mineralization, etc. [85,90,95].

Conclusions
1.Primary silicate rocks of the Kovdor massif (peridotite, foidolite-melilitolite and forsteritite) are free of sulfides due to sulfur fractionation in fluid phase.For the same reason, hydrothermally altered parts of these rocks, including diopsidite, phlogopitite and skarn-like rocks, carry rich The sequences of sulfide formation in the phoscorite-carbonatite pipes are in a good accordance with the upper limits of their stability [62] that linearly decrease from 1100 • C for high bornite to 103 • C for chalcocite (Figure 17b).For the Kovdor complex, these relations can be supplemented with temperature of pyrite-pyrrhotite equilibration ranged from 170 • C in intermediate low-carbonate magnetite-rich phoscorite and dolomite carbonatite to 300 • C in axial carbonate-rich phoscorite and phoscorite-related carbonatite, and temperatures of the pyrrhotite superstructure formation from 140 • C for pyrrhotite-4C to 60 • C for pyrrhotite-5C [62].
One of the most likely reasons for sulfide specialization in the Loolekop phoscorite-carbonatite pipe in comparison with the Kovdor one is insignificant differences in oxygen fugacity of initial phoscorite melts (due to higher water content), because the larger oxygen fugacity causes more intensive copper extraction by melt from surrounding rocks [5,[91][92][93].In fact, even a slight increase in oxidation state (to ≥FMQ + 1) can shift the balance between sulfide and sulfate components dissolved in melt towards more soluble sulfates, allowing magmas to dissolve much larger amounts of sulfur than reduced melts [94].Higher water content in the Palabora's magma is confirmed, in particular, by much deeper alteration of peridotite and phoscorite into serpentinite, much abundant pegmatites, much richer valleriite mineralization, etc. [85,90,95].
Both pyrrhotite and chalcopyrite fill interstices between the main rock-forming minerals, including carbonates, and form late veinlets and irregularly shaped segregations with inclusions of surrounding minerals.Usually, chalcopyrite (with relicts of earlier bornite and exsolution inclusions of cubanite and mackinawite) crystallizes around grains of pyrrhotite (with inclusions of pentlandite-cobaltpentlandite and pyrite), and both these minerals contain common exsolution inclusions of sphalerite.Temperature of pyrite-pyrrhotite equilibration ranges from 100 • C to 400 • C, and pressure of pyrrhotite-sphalerite equilibration reaches 10 kbar; 4.
For the most part, pyrrhotite corresponds to its non-magnetic 5C polytype.Ferrimagnetic pyrrhotite-4C forms individual grains, marginal zones of non-magnetic pyrrhotite crystals and thin lens-like inclusions in pyrrhotite-5C.Low-temperature pyrrhotite 2C (troilite) occurs as lens-like exsolution inclusions in grains of pyrrhotite-4C and -5C.In natural sequence of the Kovdor rocks, iron content in pyrrhotite gradually increases from Fe 7 S 8 to Fe 9 S 10 and Fe 11 S 12 in accordance with gradual decrease of crystallization temperature and oxygen fugacity.

5.
High complexity of crystal structure of pyrrhotite-5C (I G,total = 429.75bits per unit cell), which is the most common within the Kovdor massif, confirms the lowest temperature of its formation in comparison with other polytypes.Wide dissemination of this modification is associated with structural stability (variations in composition Fe 8.84-8.99 S 10 ) due to different vacancies ordering mechanisms resulting in two structural modifications: P2 1 and C2.

6.
Within the phoscorite-carbonatite complex, content of Co in pyrrhotite gradually increases from host silicate rocks and marginal forsterite-dominant phoscorite to axial carbonate-rich phoscorite and carbonatite due to Ni and Fe.This dependence probably reflects a gradual decrease of crystallyzation temperature in primary monosulfide solid solutions from the pipe margin toward its axis.7.

Figure 1 .
Figure 1.Distribution of rock types (a), sulfur content (b) and probability of pyrrhotite and chalcopyrite presence (c) within the Kovdor phoscorite-carbonatite complex.

Figure 1 .
Figure 1.Distribution of rock types (a), sulfur content (b) and probability of pyrrhotite and chalcopyrite presence (c) within the Kovdor phoscorite-carbonatite complex.

Figure 3 .
Figure 3. Frequency histograms of the total metals (a), Fe (b), Ni (c) and Co (d) contents in pyrrhotite and troilite of the Kovdor phoscorite-carbonatite complex.

Figure 4 .
Figure 4. Mean contents of Fe, Ni and Co (at.%, mean ±95% confidence interval) in pyrrhotite ((a) without the troilite account) and exsolution pentlandite-cobaltpentlandite (c), as well as ratio Fe 3+ /Me 2+ in pyrrhotite (b), in the order of the rock formation sequence.

Figure 3 .
Figure 3. Frequency histograms of the total metals (a), Fe (b), Ni (c) and Co (d) contents in pyrrhotite and troilite of the Kovdor phoscorite-carbonatite complex.

Figure 3 .
Figure 3. Frequency histograms of the total metals (a), Fe (b), Ni (c) and Co (d) contents in pyrrhotite and troilite of the Kovdor phoscorite-carbonatite complex.

Figure 4 .
Figure 4. Mean contents of Fe, Ni and Co (at.%, mean ±95% confidence interval) in pyrrhotite ((a) without the troilite account) and exsolution pentlandite-cobaltpentlandite (c), as well as ratio Fe 3+ /Me 2+ in pyrrhotite (b), in the order of the rock formation sequence.

Figure 4 .
Figure 4. Mean contents of Fe, Ni and Co (at.%, mean ±95% confidence interval) in pyrrhotite ((a) without the troilite account) and exsolution pentlandite-cobaltpentlandite (c), as well as ratio Fe 3+ /Me 2+ in pyrrhotite (b), in the order of the rock formation sequence.

FinalFigure 6 .
Figure 6.Reconstructed sections of reciprocal space obtained for (h0l) and (hk0) sections for 1 (a,c) and 4 pyrrhotite samples (b,d) and enlarged fragments of these sections (e-g).For sample 1, there was used the transformation matrix [010 001 100].White arrows and numbers indicate reflections and its indices.On corresponding schemas, large dark red circles and small unfilled circles belong to the

Figure 6 .
Figure 6.Reconstructed sections of reciprocal space obtained for (h0l) and (hk0) sections for 1 (a,c) and 4 pyrrhotite samples (b,d) and enlarged fragments of these sections (e-g).For sample 1, there was used the transformation matrix [010 001 100].White arrows and numbers indicate reflections and its indices.On corresponding schemas, large dark red circles and small unfilled circles belong to the hexagonal cell (a = 3.43, c = 5.68 Å) and supercell respectively; black and red arrows indicate subcell and supercell vectors respectively.

Figure 15 .
Figure 15.Effect of FeO tot content in SM ((a) T = 1200 • C and P = 1.5 GPa) and melt temperature ((b) FeO tot = 3.7 wt %) on partition coefficients of Ni, Co and Cu between MSS and SM [77].

Figure 17 .
Figure 17.Order of sulfide formation in the Kovdor and Loolekop phoscorite-carbonatite pipes (a) and the corresponding stability temperature [62] for the Kovdor's sequence (b).Mineral abbreviations are shown in Section 2.

Figure 17 .
Figure 17.Order of sulfide formation in the Kovdor and Loolekop phoscorite-carbonatite pipes (a) and the corresponding stability temperature [62] for the Kovdor's sequence (b).Mineral abbreviations are shown in Section 2.

Table 2 .
Median contents of sulfide-forming elements in rocks of the Kovdor massif.

Table 3 .
Grain size and chemical composition of the pyrrhotite group minerals (mean ± SD/min-max).

Table 4 .
Structural data for studied pyrrhotites.
2, where FeS Sp is in mol %, and P is in kbar.The results (see Table6) shows gradual pressure decrease from 4 kbar in marginal (apatite)-forsterite phoscorite to 2 kbar in axial carbonate-rich phoscorite and