Atypical Mineralization Involving Pd-Pt, Au-Ag, REE, Y, Zr, Th, U, and Cl-F in the Oktyabrsky Deposit, Norilsk Complex, Russia

: Highly atypical mineralization involving Pd-Pt, Au-Ag, REE, Y, Zr, U, Th, and Cl-F-enriched minerals is found in zones with base metal sulﬁdes (BMS; ~5 vol.% to 20 vol.%) in the eastern portion of the Oktyabrsky deposit in the Norilsk complex (Russia). The overall variations in Mg# index, 100 Mg/(Mg + Fe 2+ + Mn), in host-rock minerals are 79.8 → 74.1 in olivine, 77.7 → 65.3 in orthopyroxene, 79.9 → 9.2 in clinopyroxene, and An 79.0 → An 3.7 . The span of clinopyroxene and plagioclase compositions reﬂects their protracted crystallization from early magmatic to late interstitial associations. The magnesian chromite (Mg# 43.9) trends towards Cr-bearing magnetite with progressive buildups in oxygen fugacity; ilmenite varies from early Mg-rich to late Mn-rich variants. The main BMS are chalcopyrite, pyrrhotite, troilite, and Co-bearing pentlandite, with less abundant cubanite (or isocubanite), rare bornite, Co-bearing pyrite, Cd-bearing sphalerite (or wurtzite), altaite, members of the galena-clausthalite series and nickeline. A full series of Au-Ag alloy compositions is found with minor hessite, acanthite and argentopentlandite. The uncommon assemblage includes monazite-(Ce), thorite-cofﬁnite, thorianite, uraninite, zirconolite, baddeleyite, zircon, bastnäsite-(La), and an unnamed metamict Y-dominant zirconolite-related mineral. About 20 species of PGM (plat-inum group minerals) were analyzed, including Pd-Pt tellurides, bismuthotellurides, bismuthides and stannides, Pd antimonides and plumbides, a Pd-Ag telluride, a Pt arsenide, a Pd-Ni arsenide, and unnamed Pd stannide-arsenide, Pd germanide-arsenide and Pt-Cu arseno-oxysulﬁde. The atypical assemblages are associated with Cl-rich annite with up to 7.54 wt.% Cl, Cl-rich hastingsite with up 4.06 wt.% Cl, ferro-hornblende (2.53 wt.% Cl), chlorapatite (>6 wt.% Cl) and extensive solid solutions of chlorapatite, ﬂuorapatite and hydroxylapatite, Cl-bearing members of the chlorite group (chamosite; up to 0.96 wt.% Cl), and a Cl-bearing serpentine (up to 0.79 wt.% Cl). A decoupling of Cl and F in the geochemically evolved system is evident. The complex assemblages formed late from Cl-enriched ﬂuids under subsolidus conditions of crystallization following extensive magmatic differentiation in the ore-bearing sequences. ( a ) (BSE) shows a platy grain of an unnamed (Y-rich) mineral labeled UN, which has an empirical formula (Y,Ca,REE) 2 Zr 2 (Ti,Nb) 2 Ti Fe 2+ O 14 and is associated closely with a Cl-rich annitic mica (Ann) with 10.89 wt.% K 2 O and 5.48 wt.% Cl, at the boundary with a sodic plagioclase (Pl). ( b ) (BSE) displays an irregular grain of argentopentlandite (Apn) in association with chalcopyrite (Ccp), pyrrhotite (Pyh) and magnetite (Mag). Figure 14 c shows reﬂectance spectra measured in air for grains of the zirconolite-type unnamed oxide (Y-dominant) from the Oktyabrsky ore deposit.


Introduction and Geological Context
The Norilsk complex of Permo-Triassic age hosts giant ore deposits, such as the Oktyabrsky deposit of copper, nickel and platinum group elements (PGE), developed close to the northwestern margin of the Siberian Platform in northern Russia (e.g., [1], and  [1] and references therein. The labels from DB to Ps pertain to original names of the regional faults, which are used historically in the Russian literature. The study area is shown schematically by the red square symbol.

Materials and Methods
A total of twenty-six ore-bearing samples collected in a series of boreholes ( Figure 2) were studied mineralogically. Quantitative analyses of minerals were performed at the R&D center of Norilsk Nickel at the Institute of Mining, Geology and Geotechnology of the Siberian Federal University, Krasnoyarsk, by means of scanning electron microscopy and energy-dispersive analysis (SEM-EDS) done on a Tescan Vega III SBH system (Tescan Orsay Holding, Brno, Czech Republic) equipped with an Oxford X-Act spec- Figure 1. Generalized scheme of geology of the Norilsk area, after [1] and references therein. The labels from DB to Ps pertain to original names of the regional faults, which are used historically in the Russian literature. The study area is shown schematically by the red square symbol.

Materials and Methods
A total of twenty-six ore-bearing samples collected in a series of boreholes ( Figure 2) were studied mineralogically. Quantitative analyses of minerals were performed at the R&D center of Norilsk Nickel at the Institute of Mining, Geology and Geotechnology of the Siberian Federal University, Krasnoyarsk, by means of scanning electron microscopy and energy-dispersive analysis (SEM-EDS) done on a Tescan Vega III SBH system (Tescan Orsay Holding, Brno, Czech Republic) equipped with an Oxford X-Act spectrometer (Oxford Instruments Nanoanalysis, Wycombe, UK). The operating conditions were held at an accelerating voltage of 20 kV and a beam current of 1.2 nA.
Minerals 2021, 11, x FOR PEER REVIEW trometer (Oxford Instruments Nanoanalysis, Wycombe, UK). The operating con were held at an accelerating voltage of 20 kV and a beam current of 1.2 nA. The commonly used combinations of X-ray lines were used along with a standards provided by Micro-Analysis Consultants Ltd. (MAC, Cambridgeshi registration no. 11192). The K line was used for oxygen, Si (quartz standard), Ca ( tonite), K (orthoclase), Na (albite), Cu (synthetic chalcopyrite), Fe, S (pyrite and tite) and Ni, as well as Co, Ti, V and Cr; specimens of Al2O3 were used for Al, M Mg, pure Mn for Mn, sphalerite for Zn, and synthetic GaP for P. The F-and Clminerals were also analyzed using the K line, with specimens of fluorite and h standards. Furthermore, the L line and standards of pure elements were used fo Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, and Yb. The L line and synthetic Ce LaB6 were used for Ce and La. The L line was also used for Nb, Ag, Se and Sb (p ments and stibnite as standards), as well as for Zr, Cd (pure Zr and Cd), Sn (pure (PbTe), As (arsenopyrite), and Pd and Rh (pure Pd and Rh). The M line was used (pure Au), Ir (Ir), Os (Os), Pt (Pt), Bi (pure Bi), Pb (PbTe), Th (ThO2), U (pure U), a for Hf (pure Hf). The beam current was measured every 60 min using the MAC standard (registration no. 9941). The commonly used combinations of X-ray lines were used along with a set of standards provided by Micro-Analysis Consultants Ltd. (MAC, Cambridgeshire, UK; registration no. 11192). The K line was used for oxygen, Si (quartz standard), Ca (wollastonite), K (orthoclase), Na (albite), Cu (synthetic chalcopyrite), Fe, S (pyrite and pyrrhotite) and Ni, as well as Co, Ti, V and Cr; specimens of Al 2 O 3 were used for Al, MgO for Mg, pure Mn for Mn, sphalerite for Zn, and synthetic GaP for P. The F-and Cl-bearing minerals were also analyzed using the K line, with specimens of fluorite and halite as standards. Furthermore, the L line and standards of pure elements were used for Y, La, Pr, Nd, Pm, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, and Yb. The L line and synthetic CeO 2 and LaB 6 were used for Ce and La. The L line was also used for Nb, Ag, Se and Sb (pure elements and stibnite as standards), as well as for Zr, Cd (pure Zr and Cd), Sn (pure Sn), Te (PbTe), As (arsenopyrite), and Pd and Rh (pure Pd and Rh). The M line was used for Au (pure Au), Ir (Ir), Os (Os), Pt (Pt), Bi (pure Bi), Pb (PbTe), Th (ThO 2 ), U (pure U), and also for Hf (pure Hf). The beam current was measured every 60 min using the MAC cobalt standard (registration no. 9941).

Rock-Forming Minerals and the Accessory Fe-Cr-(Ti) Oxides
The overall variations in the composition of rock-forming silicates are displa Tables 1 and 2. Olivine grains vary compositionally in the range of Mg# values, Mg/(Mg + Fe 2+ + Mn), from 79.8 to 74.1. Grains of orthopyroxene display the rang 65.3. In contrast, compositions of clinopyroxene exhibit much more extensive varia the diopside-hedenbergite series, with Mg# values in the range 79.9-9.2. The un low Mg compositions formed late in the crystallization history (Table 1, Figure 4 compositions of plagioclase also cover an impressive range, from early An79.0 to with the sodic members of the series present in the "intercumulus" (interstitial) a tion (Table 2, Figure 4b).

Rock-Forming Minerals and the Accessory Fe-Cr-(Ti) Oxides
The overall variations in the composition of rock-forming silicates are displayed in Tables 1 and 2. Olivine grains vary compositionally in the range of Mg# values, i.e., 100 Mg/(Mg + Fe 2+ + Mn), from 79.8 to 74.1. Grains of orthopyroxene display the range 78.0-65.3. In contrast, compositions of clinopyroxene exhibit much more extensive variations in the diopside-hedenbergite series, with Mg# values in the range 79.9-9.2. The unusually low Mg compositions formed late in the crystallization history (Table 1, Figure 4a). The compositions of plagioclase also cover an impressive range, from early An 79.0 to Ab 96. 3 , with the sodic members of the series present in the "intercumulus" (interstitial) association (Table 2, Figure 4b).  The samples were taken from mineralized zones in the eastern portion of the Oktyabrsky ore deposit. Labels: Wo is wollastonite; En is enstatite, and Fs is ferrosilite. The pyroxene nomenclature is after [11]. (b) displays the existence of a continuous series of compositions of plagioclase in these zones, in terms of the components albite (Ab) and anorthite (An).
Species of Cr-Ti oxides form part of the ore assemblage (Table 3). Accessory grains and aggregates of chromite and Cr-bearing magnetite, hereafter referred to as chromian spinel (Chr) for the sake of simplicity, and ilmenite (Ilm) attain 0.1-0.2 mm in size (Figures 3b and 5a,b and Table 4). The compositions of chromian spinel display a trend of Fe 3+ enrichment during crystallization, a reflection of the increasing amounts of the magnetite component (Figure 6a). They display a positive correlation of Al vs. Mg, with a coefficient R of 0.88 based on a total of 33 data-points (n = 33). Additionally, they display a gradual decrease in Mg and Al during crystallization, i.e., in the spinel component (MgAl2O4). Notable levels of Ti (up to 5% TiO2; Table 4) are characteristic of the chromian spinel, as is typical of the Norilsk complex (e.g., [12]). The ilmenite contains up to 5.42 wt.% MgO, i.e., the geikielite component. In addition, grains enriched in the pyrophanite component contain up to 5.05 wt.% MnO.  The samples were taken from mineralized zones in the eastern portion of the Oktyabrsky ore deposit. Labels: Wo is wollastonite; En is enstatite, and Fs is ferrosilite. The pyroxene nomenclature is after [11]. (b) displays the existence of a continuous series of compositions of plagioclase in these zones, in terms of the components albite (Ab) and anorthite (An).  Species of Cr-Ti oxides form part of the ore assemblage (Table 3). Accessory grains and aggregates of chromite and Cr-bearing magnetite, hereafter referred to as chromian spinel (Chr) for the sake of simplicity, and ilmenite (Ilm) attain 0.1-0.2 mm in size (Figures 3b and 5a,b and Table 4). The compositions of chromian spinel display a trend of Fe 3+ enrichment during crystallization, a reflection of the increasing amounts of the magnetite component ( Figure 6a). They display a positive correlation of Al vs. Mg, with a coefficient R of 0.88 based on a total of 33 data-points (n = 33). Additionally, they display a gradual decrease in Mg and Al during crystallization, i.e., in the spinel component (MgAl 2 O 4 ). Notable levels of Ti (up to 5% TiO 2 ; Table 4) are characteristic of the chromian spinel, as is typical of the Norilsk complex (e.g., [12]). The ilmenite contains up to 5.42 wt.% MgO, i.e., the geikielite component. In addition, grains enriched in the pyrophanite component contain up to 5.05 wt.% MnO.

Amphiboles and Micas
The amphibole compositions in the ore-bearing specimens that we have investigated consist of members of the calcic, sodic-calcic and Fe-Mg groups (Table 5, Figure 7a) The mica compositions are annitic, unusually chlorine-rich (Table 6, Figure 7b), with up to 5-7 wt.% Cl, and display a close association with platinum group minerals. Amphiboles of series 1 are generally poor in Cl, except for a potassic variety of hastingsite (composition #5 in Table 5) that contains 4.06 wt.% Cl. The variations shown in Figure 8a

Amphiboles and Micas
The amphibole compositions in the ore-bearing specimens that we have investigated consist of members of the calcic, sodic-calcic and Fe-Mg groups (Table 5, Figure 7a) The mica compositions are annitic, unusually chlorine-rich (Table 6, Figure 7b), with up to 5-7 wt.% Cl, and display a close association with platinum group minerals. Amphiboles of series 1 are generally poor in Cl, except for a potassic variety of hastingsite (composition #5 in Table 5

Compositional Variations of Apatite and Minerals of Zr, REE, Y, Th and U
Apatite is a fairly common accessory mineral in our suite of samples. Grains reach up to 0.5-1 mm in length; they exhibit various shapes, from anhedral or subhedral to nearly euhedral (Figures 9a,b and 10a). High variability in the levels of halogens and   Apatite is a fairly common accessory mineral in our suite of samples. Grains reach up to 0.5-1 mm in length; they exhibit various shapes, from anhedral or subhedral to nearly euhedral (Figures 9a,b and 10a). High variability in the levels of halogens and

Compositional Variations of Apatite and Minerals of Zr, REE, Y, Th and U
Apatite is a fairly common accessory mineral in our suite of samples. Grains reach up to 0.5-1 mm in length; they exhibit various shapes, from anhedral or subhedral to nearly euhedral (Figure 9a,b and Figure 10a). High variability in the levels of halogens and hydroxyl (calculated) has been documented (Table 7; Figure 11). These variations were established in both intragranular and grain-to-grain patterns.
In addition, a single inclusion (~2 µm across) of a highly F-enriched composition was encountered, which likely represents bastnäsite-(La), although its composition somewhat deviates from the ideal formula, likely because of analytical difficulties caused by the very tiny grain-size. This grain is hosted by plagioclase (An 61      Of particular interest are the documented occurrences of an unnamed, Y-dominant, zirconolite-related mineral, hitherto unreported, which occurs as elongate and platy grains commonly associated with the Cl-rich annite (Figure 14a). Representative results of selected analyses (#8 in  failed because the mineral is metamict (I.V. Pekov, written communication), a reflection of its elevated content of Th and U. These phases at Norilsk appear to be related to stefanweissite, (Ca,REE)2Zr2(Nb,Ti)(Ti,Nb)2Fe 2+ O14, nöggerathite-(Ce), (Ce,Ca)2Zr2(Nb,Ti)(Ti,Nb)2Fe 2+ O14, and laachite, (Ca,Mn)2Zr2Nb2TiFeO14, which are all zirconolite-related minerals discovered in the Eifel paleovolcanic region, Germany [13][14][15]. The low level of radioactive elements in the relatively young complexes of that region has ensured the preservation of their crystal structure.  Figure 14c shows reflectance spectra measured in air for grains of the zirconolite-type unnamed oxide (Y-dominant) from the Oktyabrsky ore deposit.   Figure 14c shows reflectance spectra measured in air for grains of the zirconolite-type unnamed oxide (Y-dominant) from the Oktyabrsky ore deposit.
The potentially new Y-dominant mineral has a fairly low reflectance measured in air ( Figure 14c, Table 10). Attempts were made to characterize its crystal structure; these failed because the mineral is metamict (I.V. Pekov, written communication), a reflection of its elevated content of Th and U. These phases at Norilsk appear to be related to stefanweissite, (Ca,REE) 2 Zr 2 (Nb,Ti)(Ti,Nb) 2 Fe 2+ O 14 , nöggerathite-(Ce), (Ce,Ca) 2 Zr 2 (Nb,Ti)(Ti,Nb) 2 Fe 2+ O 14 , and laachite, (Ca,Mn) 2 Zr 2 Nb 2 TiFeO 14 , which are all zirconolite-related minerals discovered in the Eifel paleovolcanic region, Germany [13][14][15]. The low level of radioactive elements in the relatively young complexes of that region has ensured the preservation of their crystal structure.

Au-Ag Minerals and Variations in the System PbS-PbSe-PbTe
Argentopentlandite, the Fe-Ni-Ag sulfide (Figures 14b and 15a-c, Table 11), occurs sporadically in the samples investigated, along with other Ag-based species (hessite, acanthite, sopcheite (i.e., Ag-Pd telluride)), in grains varying from 2 µm to 50-70 µm across. Present as well are grains of Au-Ag alloys, in the range 5-15 µm; they typically accompany grains of platinum group minerals. These alloys correspond to the minerals silver (Ag-dominant) and gold (Au-dominant; Table 12), which display an extensive series of intermediate solid solutions (Figure 16).
Grains of altaite and members of the galena-clausthalite series (Table 13) typically form small inclusions: <5-10 µm, occasionally 20-25 µm, and rarely up to 50 µm, hosted by chalcopyrite, cubanite or pyrrhotite, or associated with aggregates of grains of chalcopyrite and pentlandite within silicate minerals. A close association with the Cl-rich annite (3.89 wt.% Cl) is observed. In addition, these species of chalcogenides commonly occur in mutual intergrowths or accompany Pd-based bismuthotellurides and, in some instances, are associated with hessite. The analyzed members of the system PbS-PbSe-PbTe display the common Se-for-S substitution (Figure 17), whereas the Se-for-Te (or Sfor-Te) exchanges are minor or virtually absent in these solid solutions.      Our data on argentopentlandite (analyses #3-9, Table 11; Figure 15a-c) have interesting implications. These grains (~5-10 µm to 60 µm across; e.g., Figure 14b), hosted by chalcopyrite or pyrrhotite, or occurring at their boundaries, exhibit unusually high variations in Fe (4.85-5.68), Ni (2.29-3.66) and Ag (0.50-1.00), all quoted in apfu calculated for a total of 17 apfu (Table 11). On the basis of these observations, we infer that: (1) up to 0.5 apfu of Ag at the Ag site can be replaced by another cation, namely Ni; (2) nickel, not Fe, replaces Ag in the structure, and (3) a coupled mechanism (Ag + Fe) → Ni can be inferred, which suggests that a Ni-for-Ag substitution at the Ag site is combined with a Ni-for-Fe substitution at the Fe-Ni site. The presently observed variations in Ag extend those reported previously: (Fe 5 ± 0.6 Ni 3 ± 0.4 ) Σ8 + x Ag 1 − x S 8 with 0 < x < 0.2, from Mount Windarra, Australia [16,17], and (Fe 5 ± 0.77 Ni 3 ± 0.75 ) Σ8 + x Ag 1 ± y S 8 ± z , with 0 < x < 0.30, 0 < y < 0.23, 0 < z < 0.30 from El Charcón, Spain [18]. (Table 13) typically form small inclusions: <5-10 µm, occasionally 20-25 µm, and rarely up to 50 µm, hosted by chalcopyrite, cubanite or pyrrhotite, or associated with aggregates of grains of chalcopyrite and pentlandite within silicate minerals. A close association with the Cl-rich annite (3.89 wt.% Cl) is observed. In addition, these species of chalcogenides commonly occur in mutual intergrowths or accompany Pd-based bismuthotellurides and, in some instances, are associated with hessite. The analyzed members of the system PbS-PbSe-PbTe display the common Se-for-S substitution (Figure 17), whereas the Se-for-Te (or S-for-Te) exchanges are minor or virtually absent in these solid solutions.

Base metal Sulfides and Nickeline
As noted, base metal sulfide mineralization typically develops in the form of large, primary grains with droplet-like or angular shapes, or interstitial disseminations (Figure 3a). It is composed of varying proportions of chalcopyrite, pyrrhotite, troilite, and pentlandite, with lesser amounts of cubanite and rare bornite.
The grains of chalcopyrite-type minerals, analyzed in all of the ore-bearing specimens, display notable variations in Cu and Fe. The overall ranges observed for a total n of 119 data-points are: Cu 0.84-1.10 (mean 0.99), Fe 0.95-1.18 (1.02), and S 1.90-2.06 (2.00), calculated for a total of 4 apfu; the Cu/Fe ratio is in the range 0.75-1.15 (mean 0.97). In the absence of structural data, we cannot specify whether or not the chalcopy-

Base Metal Sulfides and Nickeline
As noted, base metal sulfide mineralization typically develops in the form of large, primary grains with droplet-like or angular shapes, or interstitial disseminations (Figure 3a). It is composed of varying proportions of chalcopyrite, pyrrhotite, troilite, and pentlandite, with lesser amounts of cubanite and rare bornite.
Several hundreds of PGM grains were analyzed. They are dominantly located at the boundaries of grains of base metal sulfides (BMS) with silicate minerals (in many cases with the Cl-enriched annite), or are hosted entirely by chalcopyrite or, less commonly, by other BMS. The Pd-Pt tellurides, bismuthotellurides, Pd-Pt stannides and sperrylite are
Several hundreds of PGM grains were analyzed. They are dominantly located at the boundaries of grains of base metal sulfides (BMS) with silicate minerals (in many cases with the Cl-enriched annite), or are hosted entirely by chalcopyrite or, less commonly, by other BMS. The Pd-Pt tellurides, bismuthotellurides, Pd-Pt stannides and sperrylite are most abundant in the ore-bearing zones. The other species of PGM (#30-50 in Table 3) are subordinate or rare.
The bulk of the PGM grains are characteristically tiny, ranging from ≤1-2 to 10-15 µm for most grains (on the order of 80%). We estimate that about 10-15% of the examined grains exceed 20 µm, whereas ≤5% of the grains attain or exceed 0.1 mm in the longest dimension.
Characteristic  Figures 25 and 26) that also has a pyrite-type structure [25]. other hand, the possibility of a paolovite structure tending towards a formula Pd2(Sn,As) cannot be excluded. Insufficient grain sizes precluded a structural study. The second compound, unnamed Pd11Ge3As2 (#11, 12 in Table 20), forms minute grains (10 µm across) associated with majakite, at the boundary of a chalcopyrite grain. Alternatively, this arsenide-germanide could represent Pd2(Ge,As) as an As-enriched variant of palladogermanide (Pd2Ge), discovered at the Marathon deposit, Coldwell complex, Ontario, Canada [27,28].
The third phase is the most uncommon and may well represent the first documented example of a Pt-rich arseno-oxysulfide. Its composition is: Pt 34.20, Cu 30.84, Fe 1.29, As 18.57, S 5.23, and O 9.97, for a total of 100.10 wt.%; this can satisfactorily be recalculated to a formula of (Pt0.82Fe0.11)Cu2.26As1.15S0.76O2.90, based on a total of 8 apfu. The simplified formula is PtCu2AsSO3. As can be seen in Figure 29, this phase formed as a result of the replacement of the precursor sperrylite, which has a notably "pure" and stoichiometric composition (Pt1.01As1.99); it is also associated with an intermediate member of the kotulskite-sobolevskite solid solution: Pd1.01(Te0.50Bi0.49). Note that the observed texture ( Figure 29) clearly points to the late formation of the entire grain of intergrown PGM, deposited along the boundary, presumably after the associated grain of Cu-bearing troilite: (Fe0.97Cu0.01-0.02)S1.00.    Note: The atomic proportions are based on a total of 2 atoms per formula unit, apfu.         Analysis #25 represents results of the test (i.e., repeat analysis) performed using a mode of wavelength-dispersive spectrometry (WDS). The presence of essential amounts of Au was thus confirmed.          Note: Atomic proportions in compositions of sperrylite (Spy) and majakite (Mjk) are based on a total of 3 atoms per formula unit (apfu), and for unnamed Pd11Ge3As2 they are based on a total of 16 apfu.     2 Sn are based on a total of 3 atoms per formula unit (apfu). Compositions of unnamed Pd 6 Sn 2 As were recalculated on the basis of a total of 9 apfu. Analysis #25 represents results of the test (i.e., repeat analysis) performed using a mode of wavelength-dispersive spectrometry (WDS). The presence of essential amounts of Au was thus confirmed.  Three of the analyzed compounds are noteworthy. The first is unnamed Pd 6 Sn 2 As (Table 18), which commonly occurs in direct contact or adjacent to grains of the Cl-rich annite (Figures 27b and 28a). A phase having this formula was previously reported in the Norilsk complex [26]. The likely existence of a separate site for As seems reasonable because of the known differences in the crystal-chemical properties of Sn and As. On the other hand, the possibility of a paolovite structure tending towards a formula Pd 2 (Sn,As) cannot be excluded. Insufficient grain sizes precluded a structural study.
The third phase is the most uncommon and may well represent the first documented example of a Pt-rich arseno-oxysulfide. Its composition is: Pt 34.20, Cu 30.84, Fe 1.29, As 18.57, S 5.23, and O 9.97, for a total of 100.10 wt.%; this can satisfactorily be recalculated to a formula of (Pt 0.82 Fe 0.11 )Cu 2.26 As 1.15 S 0.76 O 2.90 , based on a total of 8 apfu. The simplified formula is PtCu 2 AsSO 3 . As can be seen in Figure 29, this phase formed as a result of the replacement of the precursor sperrylite, which has a notably "pure" and stoichiometric composition (Pt 1.01 As 1.99 ); it is also associated with an intermediate member of the kotulskite-sobolevskite solid solution: Pd 1.01 (Te 0.50 Bi 0.49 ). Note that the observed texture ( Figure 29) clearly points to the late formation of the entire grain of intergrown PGM, deposited along the boundary, presumably after the associated grain of Cu-bearing troilite: (Fe 0.97 Cu 0.01-0.02 )S 1.00 .

Evidence for Extensive Differentiation in the Ore-Bearing Sequences
The associations of ore-forming elements, which are atypical owing to differences in their geochemical nature (cf. Pd-Pt vs. REE, Y, U, Th), are attributed to their incompatible behavior and mutual accumulation at late stages of crystallization in a system that had attained a high degree of magmatic differentiation. Extensive variations are observed in the compositions of all rock-forming minerals in the eastern portion of the ore deposit (Tables 1 and 2). The overall variations in the ore-bearing sequences are as follows: olivine Mg# 79.8 → 74.1, orthopyroxene Mg# 77.7 → 65.3, clinopyroxene 79.9 → 9.2, and plagioclase An 79.0 → to 3.7 (Ab 96.3). The oxide mineralization also displays

Evidence for Extensive Differentiation in the Ore-Bearing Sequences
The associations of ore-forming elements, which are atypical owing to differences in their geochemical nature (cf. Pd-Pt vs. REE, Y, U, Th), are attributed to their incompatible behavior and mutual accumulation at late stages of crystallization in a system that had Our textural observations of argentopentlandite (e.g., Figure 14b) imply its crystallization from pockets of a residual sulfide melt at late stages. Our results (Figure 15a-c) show that Ni can substitute up to a half of the Ag site, and is incorporated via a coupled mechanism (Ag + Fe) → Ni, in which the Ni-for-Ag substitution at the Ag site is combined with the Ni-for-Fe substitution at the Fe,Ni site. On the basis of a structural analogy with pentlandite, one can expect that levels of f S 2 play an important role in controlling the Ni:Fe ratio in argentopentlandite. Another example of coupled substution, involving Rh and Co, was reported for pentlandite at Bolshoy Khailyk, western Sayans, Russia: Rh 3+ + Co 3+ + → 3 Fe 2+ [39]. Related schemes of coupled substitution may well be common in pentlandite-type phases in other deposits.
The Cl-enriched species coexist with assemblages rich in REE, Y, Zr, U, Th, Au-Ag and Pd-Pt. Note, for example, the documented associations of monazite-(Ce) with chlorapatite (Figure 12a), thorite and uraninite, of the unnamed zirconolite-related phase with grains of Cl-rich annite (Figures 13a,b and 14a), and of the platinum group minerals (sobolevskite-kotulskite, mertieite, Pd-Sn-(As) alloy and sperrylite) with the Cl-rich annite (Figures 22b, 24b, 27a,b and 28a,b). In addition, the veinlet-like grains of sperrylite display an intimate association with a Cl-rich apatite (4.17 wt.% Cl). These findings point to the existence of high levels of Cl in the fluid medium during the crystallization of the orebearing sequences.
The compositional variations (Tables 5 and 6 and Figure 7a-c) show a clear relationship between the Cl-enrichment and corresponding enrichments in K and Fe in compositions of the amphiboles and micas. These relationships are consistent with crystallochemical concepts [40]. The positive Fe vs. Cl correlation (Figure 8c) and the observed relationships of Cl with K indicate that Cl efficiently accumulated in a fluid phase associated with portions of late intersitial melts rich in Fe and K.
A close association of Pd-Ag tellurides with a Cl-rich amphibole, i.e., the Cl-dominant analogue of ferro-pargasite (up to 4.5 wt.% Cl; 1.2 Cl apfu), was previously reported from the Lukkulaisvaara layered intrusion, Kola Peninsula, Russia. Pods of plagioclase-bearing orthopyroxenite rich in Cu-Fe-Ni sulfide mineralization and rich in Pd, Pt and Ag probably formed by the in situ crystallization of isolated volumes of H 2 O-saturated melt at Lukkulaisvaara [41,42]. Inclusions of lukkulaisvaaraite, Pd 14 Ag 2 Te 9 [43], hosted by grains of ferro-chloro-pargasite, clearly formed in a Cl-rich environment [41]. Another occurrence of ferro-chloro-pargasite (4.1 wt.% Cl), reported at Mount Poaz in the Monchepluton layered complex, Kola Peninsula, Russia, formed as a result of an autometasomatic reaction involving a hydrous fluid with plagioclase and pyroxene [30]. This rare species of amphibole, containing up to 6.34 wt.% Cl and 1.74 Cl apfu, was also described in the Tudor gabbroic complex, north of Madoc, Ontario, Canada [44]. Interestingly, notable amounts of Cl are present in shoshonite-type (K-enriched) silicate melt inclusions hosted by PGE alloys [45].
A strong Cl-enrichment in the medium also accounts for the abundance of Cl-rich apatite in the ore zones. Extensive solid solutions involving the components chlorapatite, fluorapatite and hydroxylapatite are documented in the F-Cl-OH diagram ( Figure 11). These variations extend to the end member chlorapatite and involve compositionally heterogeneous, cryptically zoned grains, as well as grain-to-grain variations present in the same sample (Table 7, Figure 10). These characteristics are consistent with data on apatite compositions from the Monchepluton layered complex, in which the degassing of the crystallizing melt caused a decoupling of Cl and F [30]. Fluorine mostly remained in the melt; in contrast, Cl was partitioned efficiently into an H 2 O-bearing fluid phase. Consequently, the early-stage apatite incorporated combinations of OH and F, with a low content of Cl. At a late stage, chlorapatite crystallized from a Cl-rich fluid, and ferro-chloro-pargasite formed at Poaz [30]. The geochemical behavior of Cl was traced on the basis of findings related to apatite from layered intrusions. The maximum Cl recorded in compositions of apatite, close to the level of the Pd-Pt horizon, likely corresponds to the first appearance and the massive crystallization of plagioclase primocrysts at Kivakka, Karelia, Russia [31,46]. Similarly, a relationship exists between the stratigraphic levels of the crystallization of chlorapatite and zones of relative Pd-Pt enrichments at Kläppsjö, Sweden [47]. Especially high contents of Cl are present in a compound of Pb-Cl-(OH) composition associated with PGM in the Merensky Reef, Bushveld complex, South Africa [48]. Occurrences of bowlesite, PtSnS [49], are present in the Reef. In layered intrusions, chlorapatite typically tends to occur in lower units of ultramafic cumulates [50][51][52]. However, no definite relationships have been found at Monchepluton, Russia, between extents of magnesium enrichment and levels of Cl in the apatite [30].
Our study reveals the presence of apatite grains of varying compositions, which can range from fluorapatite to chlorapatite in the same sample (Table 7). These variations are attributed to the process of the degassing of the crystallizing melt, with the decoupling of Cl and F. The presence of micrometric inclusion of a La-dominant carbonate-fluoride phase, related to bastnäsite (hosted by plagioclase), is consistent with the inferred decoupling of F from Cl. The likely source of Cl at Norilsk is thus presumably intrinsic and related to magmatic differentiation. This inference is consistent with the exotic occurrences of Pd-Bi chlorides documented in the ore of this complex [53].

Characteristic Features of Pd-Pt Mineralization
The zones of noble metal mineralization investigated in the eastern part of the Oktyabrsky deposit bear some notable similarities to the classic deposits at Norilsk. For instance, the major species of PGM, i.e., Pd-(Ag)-Pt tellurides, bismuthotellurides and Pd-Pt stannides (Table 3), are generally common in the massive or high-grade deposits rich in sulfides of the Norilsk complex (cf. [2,26]). Furthermore, a close association with Au-Ag alloys is present, as is very typical at Norilsk.
On the other hand, there are certain prominent and distinctive characteristics, based on our findings. (1) The mean dimension of PGM grains is micrometer-sized (≤10 µm across) in the ore assemblages examined. (2) The majority of the PGM exhibit evidence of late crystallization, commonly taking place after the deposition of grains of the associa-ted base metal sulfides (BMS). As a result, these PGM phases typically occur at the boundary of the BMS with silicate minerals (Figures 20a,b, 22a,b, 23a,b, 27a,b and 29). (3) The documented association of PGM with the Cl-enriched species is important, including representatives of the series of chlorine-enriched micas and amphiboles (Figures 22b, 24b, 27a,b and 28a,b). (4) The atypical association of PGM with various species enriched in REE, Y, Zr, Th, U, and Cl-F is a prominent feature. (5) Species of Pd-Pt-Cu stannides, i.e., taimyrite, cabriite and tatyanaite, which are very abundant in the massive sulfide ores [2,[54][55][56][57], are virtually absent in our specimens. (6) The occurrence of a Ge-bearing phase of PGM, i.e., the unnamed Pd 11 Ge 3 As 2 or Pd 2 (Ge,As), is a distinctive feature. The Ge-based species of PGM occur very rarely worldwide, with a type locality known in the Marathon Cu-Pd deposit of the Coldwell alkaline complex in Canada [27,28]. They are also found in the Yoko-Dovyren pluton, Baikal region, Russia [58]. (7) The appearance of an arseno-oxysulfide phase, PtCu 2 AsSO 3 hitherto unreported, was a result of a local buildup of f O 2 during a late fluid and subsolidus reaction with the precursor sperrylite ( Figure 29). Copper was likely remobilized and contributed by the fluid, and was likely released from the associated chalcopyrite, whereas the S was contributed by the associated troilite.
Several PGE-based oxides have been described in different settings and complexes, including Pb-, Sb-, Bi-, and Tl-bearing oxides, which reflect the compositions of the precursor PGM grains [59][60][61][62][63][64]. However, at least some of these grains could be composed of cryptic mixtures. No bonding was found to occur between platinum and oxygen in a Pt-Fe oxide grain studied by X-ray absorption spectroscopy [65], and so this likely corresponds to a mixture rather than a single mineral species.
The textural relations (Figures 20a,b, 22a,b, 23a,b, 27a,b and 29) are indicative of the late crystallization of the complex mineralization of Pd-Pt-Au-Au and associated species enriched in REE, Y, Zr, U, and Th (Figures 12a,b, 13a,b and 14a) from batches of Cl-enriched fluid under submagmatic conditions of crystallization. The low temperatures of the deposition of the PGM are consistent with experimental results. The values of the upper limit of stability are low for synthetic analogues of sopcheite Ag 4 Pd 3 Te 4 (383 • C), froodite PdBi 2 (∼480 • C), michenerite Pd-Te-Bi (501 • C), sobolevskite PdBi and kotulskite PdTe (∼750 • C) [66][67][68][69]. In addition, synthetic specimens of kotulskite, merenskyite PdTe 2 , atokite Pd 3 Sn and paolovite Pd 2 Sn are stable at 400 • C in the systems Pd-Sn and Pd-Te [70]. It is likely that a high-temperature crystallization (>800 • C) of sperrylite can proceed directly from a silicate melt [71], as was proposed in the case of this PGM in a low-sulfide horizon of the Noril sk 1 intrusion, Russia [10]. However, this mode of origin appears highly unlikely in our case. Indeed, veinlets or rim-like grains of sperrylite have formed in the specimens examined, such as those shown in Figures 5b and 29. These conclusively point to the late formation of the entire ore assemblage in the presence of Cl and other volatiles in the growth medium at a low temperature during a submagmatic stage of crystallization. This inference is corroborated by the occurrence of hydrothermally formed sperrylite in the Imandra layered complex, Kola Peninsula, Russia [72]. In addition, sperrylite is the main species of PGM in the Copper Cliff South mine associa-ted with the Sudbury structure, Ontario, Canada, at which the existence of Cl-bearing fluids is inferred [73].

Conclusions
1. The compositional variations described here document the extensive differentiation of the primary melt attained in ore-bearing zones close to the eastern contact of the Oktyabrsky ore deposit. As a result, a great variety of ore constituents (e.g., Pd-Pt, Au-Ag, REE, Y, Zr, U and Th), all rather incompatible, accumulated in that zone. Involved as well were chalcogens (S, Se, Te, O), semimetals (Sb, As, Ge) and metals (Bi, Pb, Sn), which can form ligands with the PGE in a Cl-rich fluid.
2. The newly recognized style of complex mineralization enriched in the incompatible constituents represents a variant of ore at Norilsk, the potential value of which may be enhanced by the complexity and diversity of the ore constituents.
3. We suppose that the abundance of Cl and other volatiles in the fluid-enriched system has promoted the remobilization and deposition of polycomponent mineralization in zones of deuteric alteration, likely controlled by faults, or in ore zones of contact type, or possibly even in detached zones hosted by wallrocks in the exocontact of the Norilsk complex. One can expect such zones to resemble the Kirrakkajuppura PGE mineralization of the Penikat complex in Finland, which has huge grades of Pd and Pt that developed in virtually BMS-free rocks [60,61].