Trace-Element Geochemistry of Sulﬁdes in Upper Mantle Lherzolite Xenoliths from East Antarctica

: Sulﬁdes in upper mantle lherzolite xenoliths from Cretaceous alkaline-ultramaﬁc rocks in the Jetty Peninsula (East Antarctica) were studied for their major and trace-element compositions using SEM and LA-ICP-MS applied in situ. Modal abundance of sulﬁdes is the lowest in Cpx-poor lherzolites ≤ Spl-Grt lherzolites << Cpx-rich lherzolites. Most sulﬁdes are either interstitial (i-type) or inclusions in rock-forming minerals (e-type) with minor sulﬁde phases mostly present in metasomatic veinlets and carbonate-silicate interstitial patches (m-type). The main sulﬁde assemblage is pentlandite + chalcopyrite ± pyrrhotite; minor sulﬁdes are polydymite, millerite, violarite, siegenite, and monosulﬁde solution (mss). Sulﬁde assemblages in the xenolith matrix are a product of the subsolidus re-equilibration of primary mss at temperatures below ≤ 300 ◦ C. Platinum group elements (PGE) abundances suggest that most e-type sulﬁdes are the residues of melting processes and that the i-type sulﬁdes are crystallization products of sulﬁde-bearing ﬂuids/liquids. The m-type sulﬁdes might have resulted from low-temperature metasomatism by percolating sulﬁde-carbonate-silicate ﬂuids/melts. The PGE in sulﬁde record processes are related to partial melting in mantle and intramantle melt migration. Most other trace elements initially partitioned into interstitial sulﬁde liquid and later metasomatically re-enriched residual sulﬁdes overprinting their primary signatures. The extent of element partitioning into sulﬁde liquids depends on P, T, f O 2 , and host peridotite composition. lherzolites: ( a ) three-phase roughly rounded sulﬁde inclusion (e-type) in Ol. The main body consists of Pn with only insigniﬁcant presence of Ccp along the cleavage cracks. Partly preserved pyrrhotite (Po) displays ﬂame-like relations between Pn and Po. Massive Ccp is developed inside the Pn body and at the edges of the inclusion. The scale bar is 20 µ m; ( b ) chains of multiple tiny sulﬁde inclusions (e-type) in Opx developed along the healed cracks in the host mineral. The scale bar is 50 µ m; ( c ) oval sulﬁde inclusion (e-type) in the Spl grain (resorbed edges of the grain are seen). The main volume of the inclusion consists of cleaved Pn with insigniﬁcant presence of Ccp along the cleavage cracks. Massive Ccp is developed at the edges of the inclusion. A small mss bleb is present at the edge of the inclusion. Chains of tiny sulﬁdes are located along the existing and healed cracks in Spl. The scale bar is 50 µ m; ( d ) irregular three-phase interstitial (i-type) sulﬁde grain. The main volume consists of Pn + Po often displaying ﬂame-like relations. Ccp forms lamellae along cleavage cracks in Pn. Massive Ccp is developed inside and at the edges of the sulﬁde grain. The scale bar is 50 µ m; ( e ) bi-mineral irregular interstitial (i-type) Pn + Ccp grain with strongly cleaved Pn core and massive Ccp at the edges of the grain. Ccp is sometimes exsolved along the cleavage cracks in Pn. Tiny crystals of Fe-oxides are developed around the sulﬁde grain. The scale bar is 50 µ m; ( f ), resorbed interstitial (i-type) sulﬁde resulting from the mss breakup. Sulﬁde consists of equal amounts of massive Pn and resorbed Ccp. Rare patches of Mlr are developed at the edges of sulﬁde grain. The scale bar is 20 µ m.


Introduction
Small stock-like Late Jurassic to Early Cretaceous intrusions of alkaline-ultramafic rocks (mostly alkaline picrite and ultramafic lamprophyre) in the Jetty Peninsula (Prince Charles Mountains, East Antarctica; Figure 1) contain abundant xenoliths representing samples of the subcontinental lithospheric mantle (SCLM) beneath the region [1][2][3][4][5][6]. The studied area is located on the flank of the small rift system which formed at about 650 Ma and lies on the western side of the larger Lambert-Amery rift system, which was active from the Mesozoic period to the Cenozoic period. A detailed description of the geology of the Prince Charles Mountains is given in [4][5][6][7]. The upper mantle xenoliths are mostly lherzolite with subordinate harzburgite, dunite and wherlite. Three main groups of lherzolites were identified on the basis of mineral assemblage, textural relationships between the minerals, and mineral composition: (i) clinopyroxene (Cpx)-poor spinel (Spl) lherzolites, (ii) Cpxrich Spl herzolites, and (iii) Spl-garnet (Grt) lherzolites [3][4][5]. The East Antarctic SCLM rocks have experienced multiple episodes of infiltration by silicate, carbonate, and minor sulfide melts; this has resulted in partial melting, recrystallization, and the generation of metasomatic minerals such as carbonates and sulfides [5,[7][8][9][10][11]. The deepest-sampled upper mantle rocks (Spl-Grt lherzolite) yield ambient temperatures of 1040-1180 • C with pressures up to 24 kbar that correspond to P-T conditions at depths of 75-80 km [2,4,12].
Sulfide minerals are a common accessory phase (<0.1%) in mantle rocks and are typically represented by pentlandite, pyrrhotite, and chalcopyrite [13][14][15][16]. Despite their Sulfide minerals in peridotite xenoliths from East Antarctica were heretofore almo completely unstudied. Only several sulfide grains in Spl-Grt lherzolite xenoliths had bee examined via electron microprobe and Raman spectroscopy by Solovova et al. [11]. I order to retrieve information on the origin and evolution of sulfides in mantle xenolith we conducted detailed in situ scanning electron microscopy (SEM) and laser ablation in ductively coupled plasma mass spectrometric (LA-ICP-MS) analyses of the sulfide phase in SCLM peridotite xenoliths samples from the Jetty Peninsula area. The present work based on lherzolite xenoliths that have been previously studied in detail for bulk rock an rock-forming mineral geochemistry [2][3][4]12].

Brief Characteristics of Lherzolite Xenoliths
Upper mantle xenoliths in the Jetty Peninsula area are hosted by alkaline ultramaf magmatic rocks (150 ± 5 Ma) that originated from the melting of the mantle at depths o 120-150 km [1,28,29]. The xenoliths were trapped by ascending magma at depths from 75-80 km up to the crustal-mantle boundary and were rapidly delivered to the surfac [4]. The studied lherzolite xenolith samples were collected by the authors from a singl magmatic body composed of alkaline picrite.
The Cpx-poor Spl lherzolites are medium to coarse-grained inequigranular rock composed of 78-82% olivine (Ol), 10-14% orthopyroxene (Opx), 4-7% Cpx, and 1.5-5% Spl. Low concentrations of CaO and Al2O3 (0.8-1.5 wt% and 0.5-1.6 wt%, respectively Sulfide minerals in peridotite xenoliths from East Antarctica were heretofore almost completely unstudied. Only several sulfide grains in Spl-Grt lherzolite xenoliths had been examined via electron microprobe and Raman spectroscopy by Solovova et al. [11]. In order to retrieve information on the origin and evolution of sulfides in mantle xenoliths, we conducted detailed in situ scanning electron microscopy (SEM) and laser ablation inductively coupled plasma mass spectrometric (LA-ICP-MS) analyses of the sulfide phases in SCLM peridotite xenoliths samples from the Jetty Peninsula area. The present work is based on lherzolite xenoliths that have been previously studied in detail for bulk rock and rock-forming mineral geochemistry [2][3][4]12].

Brief Characteristics of Lherzolite Xenoliths
Upper mantle xenoliths in the Jetty Peninsula area are hosted by alkaline ultramafic magmatic rocks (150 ± 5 Ma) that originated from the melting of the mantle at depths of 120-150 km [1,28,29]. The xenoliths were trapped by ascending magma at depths from 75-80 km up to the crustal-mantle boundary and were rapidly delivered to the surface [4]. The studied lherzolite xenolith samples were collected by the authors from a single magmatic body composed of alkaline picrite.
The Jetty Peninsula xenoliths contain evidence of several stages of enrichment of the SCLM rocks through melts and fluids [4,10,12]. Presence of silicate melts is evidenced by glass composition ranging from sodic basaltic melts to K-rich phonolitic melts along the grain boundaries and as interstitial patches [4,8]. Evidence of immiscible carbonate and sulfide liquids in the mantle rocks is preserved by carbonate-sulfide inclusions in silicate minerals [4] and by CO 2 -rich fluid inclusions in the silicate minerals that are described in detail in [10,11]. Multiple events of partial melting and melt extraction events followed by metasomatic re-enrichment in the East Antarctic SCLM may be related to rifting during the breakup of Gondwanaland [4][5][6][7].

Analytical Techniques
Polished rock sample mounts 25 mm in diameter were prepared from the xenolith chips for microscopic study and further SEM and LA-ICP-MS analyses. Sulfide phases were first identified and studied visually using a Carl Zeiss MicroImaging GmbH reflected light microscope equipped with the Axio Imager 2. This microscope was also used for reflected light photomicrographs. Sulfides were then analyzed for major and trace-element concentrations using the SEM and LA-ICP-MS methods. Analyses were conducted by the authors at the Department of Geochemistry and Laboratories of the Czech Geological Survey (Prague, Czech Republic) in 2021. The SEM analyses allowed the selection of grains for the LA-ICP-MS runs and to provide the necessary major-element concentrations for quantitative sulfide analysis. Major elements were analyzed using a Tescan MIRA3 GMU FEG-SEM equipped with a wavelength-dispersive spectroscopic microanalyzer (Oxford Instruments Wave, Abingdon, UK). The analyses were conducted with an accelerating voltage of 15 kV, a beam current of 8 nA, and a working distance of 15 mm. Counting times were 20 s at peak position and 10 s at the background positions. Routine sulfide analyses included Fe, Co, Ni, Cu, Zn, and S. For quantitative analyses, the following standards were used: galenite (for S), pyrite (Fe), chalcopyrite (Cu), sphalerite (Zn), pentlandite (Ni), and pure Co (Co). Analyses were processed using INCA software (Oxford Instruments).

LA-ICP-MS
The LA-ICP-MS analytical system consisted of an Agilent 7900 quadrupole (Q)-ICP-MS (Agilent Technologies Inc., Santa Clara, CA, USA) coupled with the Analyte Excite Excimer 193 nm LA system (Photon Machines, Redmond, WA, USA) equipped with a two-volume HelEx ablation cell. The laser gas system included nitrogen for the purging of the laser beam path, helium carrier gas, internal premix argon-fluoride, and flush helium gas. The Q-ICP-MS was equipped with a standard Ni sampler and skimmer cones and a quartz torch with a 2.5 mm glass injector. The laser beam diameter ("spot size") was 12-20 µm (depending on the analyzed sulfide size). The applied laser fluence was 2.83-3.14 J/cm 2 with a laser pulse rate of 10 Hz. A typical analysis consisted of 20 s of background acquisition (gas blank), 30-35 s of sample ablation, and 30-40 s of wash-out time ( Figure 2). The GLITTER 3.0 software package was used for data reduction [31,32]. The internal standardization was based on Fe concentrations determined by the SEM analysis and on the stoichiometric Fe values. Two reference materials were used for external calibration: a USGS MASS-1 standard (Fe-Cu-Zn-S pressed powder pellet; [33]) to calibrate Cr, Ni, Ga, Ge, As, Se, Mo, Cd, Ag, Sn, Sb, Te, Au, Pb, and Bi and a UQAC-FeS-1 synthetic versatile reference material from the LabMaTer (University of Quebec in Chicoutimi, Canada; [34][35][36]) to calibrate Ti, Mn, Co, Cu, Zn, and the PGE. All standards were analyzed at the beginning and at the end of each analytical session in order to monitor the sensitivity drift. A total of 29 isotopes were monitored during the analyses ( 47 Ti, 53 Cr, 55 Mn, 57 Fe, 59 Co, 61 Ni, 65 Cu, 66 Zn, 71 Ga,  74 Ge, 75 As, 77 Se, 95 Mo, 99 Ru, 103 Rh, 107 Ag, 108 Pd 111 Cd, 115 In, 118 Sn, 121 Sb, 125 Te, 185 Re,  188 Os, 193 Ir and 195 Pt 197 Au, 208 Pb, and 209 Bi). Element concentrations were calculated using the obtained isotope signal intensities.
A significant problem during routine LA-ICP-MS analysis is the occurrence of interferences. The plasma gas-related polyatomic interferences caused by argides are particularly problematic for the PGE analyses of sulfides containing Ni, Co, Zn or Cu (e.g., 99 Ru interferes with 59 [37][38][39]. Another way to avoid/reduce the influence of these interferences is to use matrix-matched standards while analyzing natural samples.
Because of LA method limitations, the majority of the analyses were conducted on sulfide phases smaller than the interaction volume of the ablation causing beam overlaps between different sulfide phases. Subsurface inclusions were also occasionally encountered during ablation. Nevertheless, we were able to analyze homogeneous sulfide phases in a few cases.
Because of LA method limitations, the majority of the analyses were conducted on sulfide phases smaller than the interaction volume of the ablation causing beam overlaps between different sulfide phases. Subsurface inclusions were also occasionally encountered during ablation. Nevertheless, we were able to analyze homogeneous sulfide phases in a few cases.

Figure 2.
Representative time-resolved LA-ICP-MS depth profiles for sulfides with a few selected isotopes shown (chalcopyrite (Ccp)-bearing pentlandite (Pn) from interstitial sulfide in the Cpx-poor Spl lerzolite sample U-1/4-3; beam diameter is 20 μm). From the left, the background count is 23 s, followed by 42 s of ablation time (which is integrated) and by 40 s of wash-out time (not shown completely).

Sulfide Petrography and Major-Element Composition
Sulfides in the Jetty Peninsula xenoliths occur mostly as interstitial grains and as globular inclusions in silicate minerals and spinel. Less frequent sulfides are present in secondary metasomatic veinlets and in silicate-carbonate patches. We grouped the sulfides into three types according to their textural occurrence (see Westner et al. [27]): sulfides enclosed in silicate minerals and spinel (e-type) occurring as spherical droplet-like grains (although some crystals are sub-to euhedral); interstitial sulfides observed along the grain boundaries of the primary mineral assemblages appearing as droplet-like, irregular, or blocky grains (i-type); and sulfides associated with metasomatic veinlets and silicate-carbonate patches (m-type). Modal abundance of sulfide minerals in the studied xenoliths is as follows: Cpx-poor lherzolites ≤ Spl-Grt lherzolites << Cpx-rich lherzolites. Major-element characteristics of the studied sulfides are given in Tables 1-4.

Sulfide Petrography and Major-Element Composition
Sulfides in the Jetty Peninsula xenoliths occur mostly as interstitial grains and as globular inclusions in silicate minerals and spinel. Less frequent sulfides are present in secondary metasomatic veinlets and in silicate-carbonate patches. We grouped the sulfides into three types according to their textural occurrence (see Westner et al. [27]): sulfides enclosed in silicate minerals and spinel (e-type) occurring as spherical dropletlike grains (although some crystals are sub-to euhedral); interstitial sulfides observed along the grain boundaries of the primary mineral assemblages appearing as droplet-like, irregular, or blocky grains (i-type); and sulfides associated with metasomatic veinlets and silicate-carbonate patches (m-type). Modal abundance of sulfide minerals in the studied xenoliths is as follows: Cpx-poor lherzolites ≤ Spl-Grt lherzolites << Cpx-rich lherzolites. Major-element characteristics of the studied sulfides are given in Tables 1-4.     Sulfide inclusions in primary minerals are represented by two major phases: pentlandite (34-40 wt% Ni, 22-28 wt% Fe, 0-0.4 wt% Pb) and chalcopyrite (32-34 wt% Cu, 27-29 wt% Fe, 0.1-0.5 wt% Ni). Chalcopyrite is developed along the cleavage cracks and/or as massive blebs inside the pentlandite matrix. Commonly fractured pyrrhotite is developed in marginal parts of the sulfide inclusions ( Figure 3a). Additionally, secondary polydymite (50-52 wt% Ni, 3.0-6.5 wt% Fe) can be developed along the sulfide bleb edges (Figure 3a). Three-and two-phase interstitial sulfides are represented by pentlandite      (38)(39)(40)(41) wt% Ni, 24-28 wt% Fe, 0.3-0.4 wt% Co) occurring inside massive pyrrhotite (58-59 wt% Fe, 1.2-1.9 wt% Ni) and as a cleaved massive species. Chalcopyrite (33-34 wt% Cu, 30-32 wt% Fe, 0.5-0.6 wt% Ni) can be developed as blocky fragments at the edges and inside the dominant pyrrhotite + pentlandite body and/or as fine exsolution lamellae in pentlandite. Grains displaying a very similar structure and composition were observed as i-type sulfides in Cpx-rich Spl lherzolites (Figure 4d). Interstitial sulfides are characterized by their massive and slightly cleaved structure and sometimes a resorbed outer part (Figure 5c,d). Compositionally, the i-type sulfides consist of a pentlandite core (40 wt% Ni, 23 wt% Fe, 0.5 wt% Co, 1.1 wt% Cu), chalcopyrite (33 wt% Cu, 30-31 wt% Fe, 0.3-0.8 wt% Ni) that is either massive or developed along cleavage cracks in the pentlandite body, rear Fe-bearing millerite (61-63 wt% Ni, 2.4-3.3 wt% Fe) along the edges of the sulfide individual, and Ni-rich mss (37-38 wt% Ni, 22-23 wt% Fe) that has been patchily developed along the edges of the sulfide bleb. Occasionally, tiny (<10 µm) Fe oxide crystals are developed next to the interstitial sulfide grains ( Figure 5c). As is the case of Cpx-poor Spl lherzolites, Spl-Grt lherzolites display skeletal sulfide crystals that are about 35 × 15 µm in size (Figure 5e) in metasomatic veinlets cross-cutting primary minerals. Skeletal sulfides are represented by two Co-rich phases corresponding to Co-rich violarite, which can be transitional to siegenite (45-53 wt% Ni, 1.1-2.7 wt% Fe, 1.8-8.6 wt% Co) and Co-millerite 34 wt% Cu, 30-32 wt% Fe, 0.5-0.6 wt% Ni) can be developed as blocky fragments at the edges and inside the dominant pyrrhotite + pentlandite body and/or as fine exsolution lamellae in pentlandite. Grains displaying a very similar structure and composition were observed as i-type sulfides in Cpx-rich Spl lherzolites ( Figure 4d). Interstitial sulfides are characterized by their massive and slightly cleaved structure and sometimes a resorbed outer part (Figure 5c,d). Compositionally, the i-type sulfides consist of a pentlandite core (

Trace Element Characteristics of Sulfides
Results of the LA-ICP-MS sulfide analyses are summarized in Tables 5 and 6 and in a supplementary, Table S1. Due to the relatively small size of the sulfide phases that were analyzed, in most cases, the LA-ICP-MS data represent mixtures of the different proportions of two or three sulfide phases, i.e., pyrrhotite (Po), pentlandite (Pn), and chalcopyrite (Ccp). To account for these mixed spectra, results have been grouped according to their major base metal sulfide (BMS) end-member (see Hughes et al. [40]). Analyses with <5 wt.% Ni + Cu are classified as Po; 5-15 wt.% Ni or Cu are Po-Pn or Po-Ccp, respectively; 15-22 wt.% Ni or Cu are Pn-Po or Ccp-Po, respectively; and analyses with ≥22 wt.% Ni or

Trace Element Characteristics of Sulfides
Results of the LA-ICP-MS sulfide analyses are summarized in Tables 5 and 6 and in a  Supplementary, Table S1. Due to the relatively small size of the sulfide phases that were analyzed, in most cases, the LA-ICP-MS data represent mixtures of the different proportions of two or three sulfide phases, i.e., pyrrhotite (Po), pentlandite (Pn), and chalcopyrite (Ccp). To account for these mixed spectra, results have been grouped according to their major base metal sulfide (BMS) end-member (see Hughes et al. [40] Table S1. In the present section, we consider the trace-element composition of the monomineral sulfides that we were able to analyze.

Sulfides in Cpx-Poor Spl Lherzolites
The only occurrence of BMS big enough to be measured with the LA-ICP-MS is represented here by Ccp-free (<1.0% Cu) pentlandite in a big interstitial sulfide bleb (sample U-3/4-2) and low-Fe pentlandite and polydymite in a roughly oval interstitial sulfide bleb (sample U-1/4-3). Pentlandite is characterized by a high concentration of Se (511 ppm), Mo (1541 ppm), and Te (86 ppm). Concentrations of elements such as Cd, In, and Sn are very low (down to below the detection limit for Cd). The total concentration of the PGE is moderately high (204 ppm) with a very high concentration of Pd (156 ppm). The mineral is poor in the I-PGE (Ru, Rh, Os, Ir) with 0 < Os < 5 ppm and is strongly enriched in the P-PGE (Pd and Pt) relative to the I-PGE ((Pd/Ir) N = 112) (normalized to the CI chondrite; [41]). The second occurrence, low-Fe (<8% Fe) and high-Ni (>55% Ni), a variety of pentlandite occurs in a rounded interstitial sulfide bleb (Figure 3c). The low-Fe pentlandite is characterized by high concentrations of Ge (64 ppm), Se (198 ppm), Te (25 ppm), and Pb (318 ppm). Elements such as Cd, In, and Sn display very low concentrations (down to below the detection limit for In) that are similar to those observed in regular pentlandite. Concentrations of the PGE are moderate (total 68 ppm) with a very low concentration of Rh (0.60 ppm). The mineral is overall poor in the I-PGE (0 < Os < 5 ppm) and is significantly enriched in the P-PGE relative to the I-PGE ((Pd/Ir) N = 19).
Polydymite from the resorbed edge of the same sulfide bleb (Figure 3c) is characterized by elevated concentrations of Pb (290 ppm) and Ag (48 ppm) and by quite high concentrations of the PGE (total 62 ppm) with a significant contribution of Pt (23 ppm). The mineral is characterized by the low I-PGE concentrations (0 < Os < 5 ppm) and is enriched in the P-PGE relative to the I-PGE ((Pd/Ir) N = 19). Among the minor sulfide types, interstitial Ni-Cu mss (sample U-1/4-3) displays high concentrations of most trace elements. It is especially pronounced for Mo (43-44 ppm), Te (47-74 ppm), and the PGE. The high total PGE concentrations of 519-596 ppm are mostly due to very high concentrations of Pd (304-422 ppm) and Ir (69-86 ppm). In spite of the high Ir concentrations, Cu-Ni mss is poor in Os (1 ≤ Os ≤ 5 ppm) and only insignificantly enriched in the P-PGE relative to the I-PGE ((Pd/Ir) N = 2-3). Co-violarite from the Co-millerite to Co-violarite aggregates (structurally similar mineral was described by Huges et al. [40] in lherzolite xenoliths from Scotland as Co-millerite) in metasomatic veinlets displays high concentrations of Co (2.5-2.6%, as measured by the LA) and sometimes elevated concentrations of Cr (up to 890 ppm), Ag (up to 33 ppm), As (52-92 ppm), Pt (5.6-11.6 ppm), and Pb (49-57 ppm). Co-violarite has moderately low I-PGE concentrations (1 < Os < 10 ppm), and it does not show P-PGE enrichment relative to the I-PGE ((Pd/Ir) N = 0.7-1.7).

Sulfides in Cpx-Rich Spl Lherzolites
Pyrrhotite is overall characterized by the lowest concentrations of trace elements among all of the BMS that were studied (Table 5). Only the Mn (319 ppm), Se (98 ppm), Mo (33 ppm), and Zn (12 ppm) in pyrrhotite displayed concentrations similar to or even higher than those measured in other BMS (the only monomineral pyrrhotite analyzed from the i-type sulfide). The rest of the trace elements (except for Co; 77 ppm) displayed concentrations in the range of 0.5-2 ppm.
Concentrations of the PGE are low (total is 2.5 ppm). The concentrations of the I-PGE were likewise low (0 < Os < 5 ppm), and the enrichment of the P-PGE relative to the I-PGE was not high ((Pd/Ir) N = 0.2).
Pentlandite is characterized by insignificantly higher concentrations of trace elements compared to those in pyrrhotite. The only difference is the expectedly much higher concentrations of Co in pentlandite (4500-5990 ppm in e-type sulfides and 5080-8880 ppm in i-type sulfides). Pentlandite from the inclusion in spinel is characterized by very low (evidently, the lowest among all studied BMS) concentrations (often below the detection limit) of Zn, Ga, Ge, Cd, In, Sn, and Sb. Pentlandite displays moderate total concentrations of the PGE (9-52 ppm with the highest concentrations of inclusion in Spl). The mineral is characterized by very low concentrations of the I-PGE (0 ≤ Os << 5 ppm) and by a varying degree of the enrichment of the P-PGE relative to the I-PGE ((Pd/Ir) N = 0.4-8.0 in e-type sulfides and 11.6-64.8 in i-type sulfides), with the lowest degree displayed by pentlandite from the inclusion in spinel.
Chalcopyrite is characterized as elevated relative to other BMS total and individual concentrations of trace elements (Table 5)   nd, no data available; bdl, below the detection limit.

Sulfides in Spl-Grt Lherzolites
The only monomineral BMS that was analyzed in the Spl-Grt lherzolites for trace element concentrations was pentlandite from the i-type sulfide (Figure 5a). Additionally, sheer Co-violarite in metasomatic veinlets and pyrite in silicate-carbonate patches were analyzed (Figure 5e,f). Interstitial pentlandite, which can be considered as Ccp-free (<1.0% Cu), is characterized by high concentrations of Se (268 ppm) and Ag (27 ppm) and by low concentrations of Mo (2.3 ppm). Concentrations of other trace elements are generally similar to those in pentlandite from other lherzolite types (Table 5). Very high concentrations of the I-PGE (Os > 10 ppm) and enrichment in the P-PGE relative to the I-PGE ((Pd/Ir) N = 13) are typical for pentlandite.
Co-violarite from skeletal crystals displays high concentrations of Co (2.5-3.3% as measured by the LA) and sometimes elevated concentrations of Cr (371-735 ppm), As (690-740 ppm), and Sb (9-14 ppm). The mineral has moderately low concentrations of the PGE (total 8-12 ppm), and while it is characterized by a very low I-PGE content (Os is below the detection limit), it is only insignificantly enriched in the P-PGE ((Pd/Ir) N = 3-5) because concentrations of Pd are low and are generally comparable to those of Ir ( Table 6). The pyrite from silicate-carbonate patches is characterized by varying amounts of Zn (4-396 ppm) and Ge (12-149 ppm) and by very low total concentrations of the PGE (0.8-1.5 ppm), which is consistent with the very low I-PGE content (0 < Os < 0.5 ppm) and the total absence of Ir.

Discussion
A broad spectrum of sulfide mineral assemblages in the SCLM lherzolite xenoliths from the Jetty Peninsula can be explained by the presence of sulfides from different environments such as the xenolith matrix, metasomatic veinlets, and carbonate-silicate patches. Previously described occurrences of BMS in lherzolite xenoliths from other localities (e.g., 25] and references therein) are also common in the studied Antarctic xenoliths. Other types of sulfides in the lherzolites from the Jetty Peninsula are less common and are poorly studied. The only study so far targeting sulfides from East Antarctic SCLM xenoliths showed that the e-type sulfides contain assemblage of quenched sulfide melt and mss in which Ni is more abundant than Fe [11].

Sulfides from Lherzolite Matrix
Most investigators assume that sulfide inclusions in the minerals of lherzolite matrix represent immiscible sulfide liquid that was trapped during the crystallization of peridotite minerals, whereas interstitial sulfides formed after the crystallization of the primary lherzolite minerals, i.e., during sulfide liquid migration along the cracks and mineral seams (see Solovova et al. [11], Kiseeva and Wood [15], Griffin et al. [18], Sen et al. [24], Mitchell and Keays [42]). However, Dromgoole and Pasteris [19] suggested that the often-observed lack of compositional differences between the e-and i-type sulfides could indicate that the creation of these sulfides was mainly due to the incorporation of postmetasomatic sulfurbearing and CO 2 -rich melts. The present study allows new insight into these processes.
Compositions of the studied BMS from the lherzolite matrix lie within both the mss and the mss + Liq fields of the Fe-(Ni + Co)-S system ( Figure 6). After mss (Fe-rich and Ni-poor) begins to crystallize at 1190 • C from Ni-Cu-Fe sulfide liquid, residual sulfide liquid gradually enriches in Ni relative to Fe [43][44][45][46]. Therefore, in the S-rich of the studied lherzolites system at the suggested ambient temperatures suggested for most varieties of those lherzolites (1030-1180 • C), mss of an overall pyrrhotite composition is the only possible solid phase coexisitng with the Ni-Cu-Fe sulfide liquid [21]. The maximum content of Cu (7%) would be reached at 1000 • C [21]. With further cooling, chalcopyrite starts to crystallize at 960 • C, and when the temperature decreases to 900 • C, the stability field of the solid phase (i.e., mss) precipitating from the sulfide liquid spans the entire FeS-(Ni + Co)S connection with slightly less Cu dissolving in it [21]. Therefore, the mss and possibly hightemperature pentlandite could coexist with the sulfide liquid in lower-temperature (down to 850 • C) varieties of Cpx-poor lherzolites (see Kitakaze and Komatsu [47]). Furthermore, the Cu-bearing mss underwent a subsolidus breakup of the Cu-poor mss + Cu-rich phase (Ccp) at temperature of 500-600 • C [25]. Pentlandite, a major constituent of sulfide assemblages in the lherzolite matrix, starts to crystallize from the mss at 610 • C, and it is stable in the Fe-Ni-S system below this temperature [21,48]. The remnants of the Ni-rich mss along the edges of the Ccp + Pn blebs decomposes when the temperature is below 380 • C. It forms stoichiometric NiS, which breaks-up to form millerite (NiS) and/or polydymite (NiNi 2 S 4 ) at temperatures of~300 • C [21]. With this, the transformation of e-type sulfides ends. For the i-type sulfides with stable chalcopyrite stable, the Cu-poor/free and Ni-bearing mss underwent further low-temperature transformation. A high-temperature assemblage experiences a separation of Ccp + mss break up to create Po + Pn when the temperature is below 300 • C. Flame-like pentlandite in pyrrhotite matrix (Figures 4d and 5b) becomes a product of the mss break up, likely due to homogeneous Ni nucleation at temperatures between 150 and 300 • C [20,49,50].
In summary, the BMS assemblages in the Jetty Peninsula lherzolite xenolith matrix are products of the subsolidus re-equilibration of high-temperature mss crystallized from the sulfide liquids that are mainly of FeS composition but that also contain Cu, Ni, and Co (see Bishop et al. [51]). In the studied rocks, all BMS display typical recrystallization features such as aggregates, lamellar exsolutions, flame-like structures, and low-temperature phase assemblages (Po + Pn + Ccp), suggesting that the studied sulfide assemblages are not of primary mantle origin. The rare presence of a patchy mss around the pentlanditechalcopyrite cores suggests that these assemblages have not reached full equilibrium and that subsolidus equilibration stopped at temperatures around 300 • C. Observed flame-like pentlandite in the pyrrhotite matrix suggests that the mss breakup completely finished at temperatures between 150 and 300 • C.

Sulfides from Silicate-Carbonate Patches
One line of evidence of melt incorporation into the lherzolite matrix is the existence of silicate-carbonate patches (Figure 5f,g). These patches contain pyrite and represent the loci of former immiscible silicate-carbonate-sulfide liquid. The former existence of carbonate melts was clearly distinguished by the presence of calcite, dolomite, apatite, and magnesite in some lherzolite xenoliths from the Jetty Peninsula [4]. Pyrite from silicate-carbonate patches could represent loci of former Fe-S liquid that existed immiscibly with the silicate and carbonate liquids. The quench structure of these patches suggests the very fast cooling of the liquids, while the presence of pyrite indicates the lower temperature limit for the fluid before quenching since pyrite is unstable above~740 • C [19,21]. A detailed study of these patches was, however, beyond the scope of the present investigation, and here we deal only with their sulfide component (see sections below). Minerals 2021, 11, x FOR PEER REVIEW 18 of 32

Sulfides from Metasomatic Veinlets
Another interesting feature is the presence of metasomatic veinlets cross-cutting the primary minerals in some xenoliths. Such veinlets contain minerals such as apatite, calcite, Fe-rich dolomite, ferroan magnesite, Ba-Ti-rich and Ba-Ti-poor mica, and humite-group minerals (see Kogarko et al. [9]). We found that such veinlets also contain a sulfide phase that is significantly different from the BMS in the lherzolite xenolith matrix, i.e., skeletal crystals of Co-millerite + Co-violarite (see Hughest et al. [40], and Nkoumbou et al. [46]). The formation temperature for such assemblages was just above 350 °C [52], which was close to the blocking temperatures of the subsolidus reactions in the BMS from the lherzolite marix. Such temperatures suggest that low-temperature metasomatism occurred at crustal levels. A detailed study of the veinlets was beyond the scope of the present investigation.

Trace Elements in Lherzolite Sulfides
Since multiphase sulfide assemblages as they appear in the samples studied are not of primary mantle origin, some compositional features intrinsic to the parental sulfide liquids could have been lost during the low-temperature breakup of the mss. However, trace elements often behave conservatively, keeping the compositional features of the primary material. Although the majority of information on mantle sulfides from elsewhere is derived from the PGE, we also considered the concentrations and distributions of trace

Sulfides from Metasomatic Veinlets
Another interesting feature is the presence of metasomatic veinlets cross-cutting the primary minerals in some xenoliths. Such veinlets contain minerals such as apatite, calcite, Fe-rich dolomite, ferroan magnesite, Ba-Ti-rich and Ba-Ti-poor mica, and humite-group minerals (see Kogarko et al. [9]). We found that such veinlets also contain a sulfide phase that is significantly different from the BMS in the lherzolite xenolith matrix, i.e., skeletal crystals of Co-millerite + Co-violarite (see Hughest et al. [40], and Nkoumbou et al. [46]). The formation temperature for such assemblages was just above 350 • C [52], which was close to the blocking temperatures of the subsolidus reactions in the BMS from the lherzolite marix. Such temperatures suggest that low-temperature metasomatism occurred at crustal levels. A detailed study of the veinlets was beyond the scope of the present investigation.

Trace Elements in Lherzolite Sulfides
Since multiphase sulfide assemblages as they appear in the samples studied are not of primary mantle origin, some compositional features intrinsic to the parental sulfide liquids could have been lost during the low-temperature breakup of the mss. However, trace elements often behave conservatively, keeping the compositional features of the primary material. Although the majority of information on mantle sulfides from elsewhere is derived from the PGE, we also considered the concentrations and distributions of trace elements that suggest histories of both depletion and re-enrichment in the host mantle rocks, which is consistent with observations made by Foley et al. [4], for example.

Platinum Group Elements
The PGE are highly conservative elements that may carry information on processes such as melting, sulfide precipitation, and sulfide addition via metasomatism in the upper mantle (see Alard et al. [17]). Bulk lherzolite xenoliths from East Antarctica display nonfractionated PGE patterns with concentrations of the PGE that are mostly around 0.01 × CI chondrite [53]. Such low PGE concentrations are consistent with studies by Kiseeva and Wood [15], Griffin et al. [18], Mitchell and Keays [42], and many others who show that sulfides are the main host phases for the PGE in the upper mantle rocks.
The PGE from the studied sulfides can be grouped according to the distribution pattern styles (Figure 7a  elements that suggest histories of both depletion and re-enrichment in the host mantle rocks, which is consistent with observations made by Foley et al. [4], for example.

Platinum Group Elements
The PGE are highly conservative elements that may carry information on processes such as melting, sulfide precipitation, and sulfide addition via metasomatism in the upper mantle (see Alard et al. [17]). Bulk lherzolite xenoliths from East Antarctica display nonfractionated PGE patterns with concentrations of the PGE that are mostly around 0.01 × CI chondrite [53]. Such low PGE concentrations are consistent with studies by Kiseeva and Wood [15], Griffin et al. [18], Mitchell and Keays [42], and many others who show that sulfides are the main host phases for the PGE in the upper mantle rocks.
The PGE from the studied sulfides can be grouped according to the distribution pattern styles (Figure 7a Figure 4d) and a huge sulfide grain included in Kel in Spl-Grt lherzolite (see Figure 5b). Almost unfractionated PGE distribution is observed for Ccp from an outer part of the inclusion in Spl (see Figure 4c), which is probably due to infiltration of immiscible sulfide melt from outside the Spl grain (cf.  Overall, distinguishing different sulfide populations can be done on the basis of relations between the I-PGE and the P-PGE. The residual PGE signatures are evident from (Pd/Ir)N ratio < 1, whereas (Pd/Ir)N > 1 suggests metasomatic influence [27]. Typically, the (Pd/Ir)N ratios in the e-type sulfides are much lower, and Os and Ir abundances are higher than those for the i-and m-type sulfides. This records melting processes in the e-type sulfides, while the i-and m-type sulfides have PGE patterns suggesting precipitation from Stepped distribution of the PGE is displayed by Pn (black circles), Ccp (grey circles), and Po (open circles) from a huge interstitial sulfide grain from Cpx-rich Spl lherzolite (see Figure 4d) and a huge sulfide grain included in Kel in Spl-Grt lherzolite (see Figure 5b). Almost unfractionated PGE distribution is observed for Ccp from an outer part of the inclusion in Spl (see Figure 4c), which is probably due to infiltration of immiscible sulfide melt from outside the Spl grain (cf. Figure 6c); (b) convex up PGE distributions in Pn and Ccp from sulfides of the e-type (in Spl and Ol) in Cpx-poor Spl lherzolite and in Cpx-rich Spl lherzolite (see Figures 3b and 4c); (c) almost unfractionated PGE patterns of Pn from sulfides of the i-type in Cpx-rich Spl lherzolite (see Figure 4e,f). These compositions could correspond closely to the initial stages of the mss break up; (d) insignificantly fractionated PGE patterns for Pn, Ccp, and Po from the e-and i-type sulfides in all three studied lherzolite types. The patterns are characterized by chondritic to suprachondritic PGE concentrations complicated by depletion in Pt and enrichment in Pd (see text for details); (e) strongly fractionated PGE patterns for Pn (low-Fe pentlandite is balck squares), Ccp, Po, and Pld (grey squares) from the i-type sulfides in all three studied lherzolite types. For most of these sample, an asymmetric "U"-shaped pattern due to fractionation of Ir relative to Os is typical; (f) convex up PGE patterns (complicated with a trough at Ru) from the Cu-Ni mss in Cpx-poor Spl lherzolite). Strong enrichment in Pd is observed (see text for details); (g) almost unfractionated PGE patterns (partly incomplete) for the m-type sulfides. PGE concentrations vary from chondritic (Py from silicate-carbonate patches; black diamonds) to around 10 × CI chondrite (Co-Vo from metasomatic veinlets; open diamonds).
Overall, distinguishing different sulfide populations can be done on the basis of relations between the I-PGE and the P-PGE. The residual PGE signatures are evident from (Pd/Ir) N ratio < 1, whereas (Pd/Ir) N > 1 suggests metasomatic influence [27]. Typically, the (Pd/Ir) N ratios in the e-type sulfides are much lower, and Os and Ir abundances are higher than those for the i-and m-type sulfides. This records melting processes in the e-type sulfides, while the i-and m-type sulfides have PGE patterns suggesting precipitation from silicate melts or the introduction of other forms of metasomatism [11,15,24,27]. Additionally, the relationship between Pt and Pd in residual sulfides provides information re-equilibration with the initially depleted melt portion in Pd sulfides, (inducing [Pt/Pd] N to be <1 [27]). The time-resolved laser ablation signals on sulfides mostly did not show any irregularities for the PGE. This is strong evidence that most PGE are rather evenly distributed throughout the sulfide phase than grouped in individual clusters.
Only a few of the studied e-type sulfide grains in Cpx-poor Spl lherzolite displayed residual PGE signatures ((Pd/Ir) N = 0.74-1.33, and 7.45 for Ccp) with insignificant secondary re-enrichment ([Pt/Pd] N = 0.27-0.39 and 0.09 for Ccp). These sulfides are characterized by convex up PGE patterns with Ir N over Os N (Figure 7b). On the contrary, the domination of the P-PGE over the I-PGE ((Pd/Ir) N = 3-28) suggests that almost all i-type sulfides in Cpx-poor Spl lherzolites can be considered as precipitated from melts that have migrated through the lherzolite matrix. The PGE patterns for these sulfides vary from insignificantly fractionated and similar to those observed for the e-type sulfides, but with Os N over Ir N (Figure 7d), to fractionated with the P-PGE strongly enriched over the I-PGE (Figure 7e). Interesting features are displayed by the interstitial Cu-Ni mss. They are characterized by extremely high concentrations of the PGE (519-596 ppm) where Pd strongly prevails (304-422 ppm). Anomalously high concentrations of Ir (69-86 ppm) accompanied by low concentrations of Os (1.5-5.3 ppm) suggests a state of disequilibrium and rapid solidification of the Cu-Ni mss. High Ir concentrations that are inconsistent with low Os concentrations may also suggest that sulfide liquid of such composition could be parental to the Ir-bearing sulfides of the Bowieite-Kashinite series known as orogenic peridotite massives [54]. Enrichment in Pd relative to Pt is consistent with overall metasomatic/melt origin of the Cu-Ni mss.
Sulfide inclusions in Cr-spinel and in silicate minerals comprise a population of the etype sulfides in Cpx-rich Spl lherzolite. Sulfide that is included in Cr-spinel (Figure 4c) has PGE characteristics and patterns similar to those observed for the e-type sulfides from Cpxpoor Spl lherzolite (Figure 7b). Sulfide phases from the inner part of the inclusion (Pn + Ccp) were characterized by the residual to metasomatic PGE signatures ((Pd/Ir) N = 0.36-6.39) with fingerprints of secondary re-enrichment ([Pt/Pd] N = 0.05-0.69). However, the outer part of the inclusion that is represented by chalcopyrite, displayed almost no signatures of metasomatic influence ((Pd/Ir) N = 0.82 and [Pt/Pd] N = 1.51). Chalcopyrite from the outer part of the inclusion displays an almost unfractionated PGE pattern with PGE concentrations being ≤10 × CI chondrite (Figure 7c). Chalcopyrite of this type is directly adjacent to the chain of tiny sulfide blebs healing a former crack in the spinel grain (Figure 4c). This suggests that chalcopyrite originated as interstitial mss incorporated into spinel through the cracks (in liquid or semi liquid state) but not as a product of the mss break up. On the contrary, most e-type sulfides in silicates and i-type sulfides from the lherzolite matrix are characterized by the metasomatic/melt PGE signatures ((Pd/Ir) N = 5.4-11.6) and show pronounced fingerprints of metasomatic Pd re-enrichment ([Pt/Pd] N = 0.16-0.59). They display variably fractionated PGE patterns complicated by troughs at Ir and Pt and by subto suprachondritic PGE concentrations (Figure 7d). This evidently precipitated from silicate melts migrating through the lherzolite matrix as it follows the strong domination of the P-PGE over the I-PGE reflected in ((Pd/Ir) N = 1.5-51.6). In this process, the i-type sulfides from Cpx-rich Spl lherzolits are similar to the i-type sulfides from Cpx-poor Spl lherzolites. However, two occurrences of the i-type sulfides represented by big multi-phase grains can be interpreted as carrying residual PGE signatures ((Pd/Ir) N = 0.16-0.51) with no or with only subtle evidence of the metasomatic re-introduction of Pd ([Pt/Pd] N = 0.57-1.33). These sulfides are characterized by very specific PGE patterns with a convex down I-PGE pattern followed by a deep trough at Pt (Figure 7a). Such distribution could be considered as typical for residual sulfides that are almost unaffected by enrichment processes (see Westner et al. [27]). Additionally, one group of the i-type sulfides was characterized by almost unfractionated PGE patterns ((Pd/Ir) N = 1.28-3.02; [Pt/Pd] N = 0.38-0.64) with PGE concentrations of >10 × CI chondrite (Figure 7c). In this case, such sulfides are similar to chalcopyrite from their inclusion in spinel. Such characteristics could be typical of sulfides directly precipitated from the melt (see Westner et al. [27]).
Most analyzed sulfides in the Spl-Grt lherzolites were found inside kelyphite aggregates that replaced Grt. Since sulfides from kelyphite display obvious residual PGE signatures ((Pd/Ir) N = 0.09-0.32) and were virtually unaffected by metasomatic Pd reintroduction ([Pt/Pd] N = 0.55-1.05), it is likely that intrakelyphitic sulfides were initially included in garnet and therefore are associated with the e-type sulfides with typical residual PGE patterns (Figure 7a). The absence of evidence of metasomatic influence is interesting because during kelyphite formation, some melting occurred, which resulted in metasomatizing liquid [8]. Interstitial sulfides rarely observed in Spl-Grt lherzolites (Figure 5c,d) are enriched in the P-PGE relative to the I-PGE ((Pd/Ir) N = 1.2-44.6) and display fingerprints of metasomatic Pd re-introduction ([Pt/Pd] N = 0.06-0.45), which is overall typical for most i-type sulfides from all of the considered lherzolite types. Distribution of the PGE for the i-type sulfides is characterized by the two pattern styles, i.e., almost unfractionated PGE patterns with ≥10 × CI chondrite PGE concentrations ( Figure 7d) and significantly fractionated PGE patterns often characterized by troughs at Ir and Pt and with PGE concentrations varying from just above chondritic to ≤100 × CI chondrite (Figure 7e).
In summary, most analyzed e-type sulfides maintain the residual PGE signatures and were only affected by metasomatic re-enrichment insignificantly, if at all. On the contrary, the i-type sulfides (with a few exceptions) display evidence of precipitation from the migrating silicate melts (melt-sulfides after Westner et al. [27]). Since sulfides with both residual and metasomatic PGE signatures coexist within a single xenolith sample, the PGE seem to be capable of effectively recording mantle processes related to mantle melting and intramantle melt migration (see Westner et al. [27]). Based on the observed PGE features, we interpret the e-type sulfides as residues of melting processes and the i-type sulfides as the crystallization products of sulfide-bearing (metasomatic) fluids/liquids (see Foley et al. [4], Kogarko et al. [9], Solovova et al. [11], Hughes et al. [40]).
Pyrite from silicate-carbonate patches in Spl-Grt lherzolites displays very low I-PGE content (down to complete absence of both Os and Ir), with most PGE represented by the P-PGE. This is consistent with metasomatic origin of these sulfides. Although the PGE patterns for pyrite are incomplete, there is some evidence of an unfractionated style of such patterns (Figure 7g). Co-violarite from metasomatic veinlets in both Cps-poor Spl and Spl-Grt lherzolites is characterized by unfractionated PGE patterns with suprachondritic concentrations of the elements at about 10 × CI chondrite (Figure 7g). This suggests a very simple m-sulfides origin that does not record histories of either depletion or re-enrichment.

Other Trace Elements
A number of studies that have been conducted on non-PGE trace elements so trace elements have shown that at upper mantle conditions, the extent of element partitioning into sulfide liquid depends on pressure, temperature, oxygen fugacity, and host rock composition [15,[55][56][57][58][59][60]. Experimental studies have shown that at upper mantle conditions such as those suggested for the Jetty Peninsula lherzolite suite [2,4,12], partial melting depletes the elements in rocks that strongly partition into sulfide melts (P-PGE, Cu, S, Se, Te, Ni, Co, Ag, Pb, Bi, ±Sn, Sb, and As), whereas elements such as the I-PGE, Mo, Zn, Mn, and V tend to be retained in mantle residue [17,58,[60][61][62]. The rest of the trace elements behave transitionally, either partitioning to sulfide melts or retaining in the residue. The behavior of most transitional elements depends especially strongly on the oxygen fugacity. Li and Audetat [58] and Steenstra et al. [60] showed that increasing f O 2 typically decreases trace element partitioning into sulfide liquid; therefore, increasing their retention in mantle residue.
We studied the behavior of chalcophile (Cu, Zn, Ga, Ge, As, Se, Ag, Cd, In, Sn, Sb, Te, Pb, and Bi) and a few siderophile (Mn, Ni, Co, and Mo) and lithophile (Ti and Cr) trace elements in different sulfide phases. The observed features were compared with those acquired from experimental work for the system "mss-sulfide liquid-mafic silicate liquid" [58]. The experiments were performed at 15-30 kb and 1175-1300 • C, i.e., at the conditions close to those suggested for the Cpx-rich Spl and Spl-Grt lherzolites from East Antarctica [4].
Features of trace element distributions in the studied sulfides are depicted in CI chondrite-normalized spider-diagrams (Figure 8a-g). Distribution patterns of non-PGE trace elements look very similar for all sulfide types except for the m-type sulfides. Generally, siderophile elements and a part of the chalcophile elements display almost unfractionated patterns (often complicated, however, by peaks at Cu and Ag), with concentrations varying from chondritic to~10 × CI chondrite. The right part of all spider diagrams in Figure 8 shows significant steady depletion in the chalcophile elements such as In, Sn, Ge, Ga, and often Zn and in lithophile Ti and Cr (down to ≥0.01 × CI chondrite).
Minerals 2021, 11, x FOR PEER REVIEW 23 of 32 acquired from experimental work for the system "mss-sulfide liquid-mafic silicate liquid" [58]. The experiments were performed at 15-30 kb and 1175-1300 °C, i.e., at the conditions close to those suggested for the Cpx-rich Spl and Spl-Grt lherzolites from East Antarctica [4].
Features of trace element distributions in the studied sulfides are depicted in CI chondrite-normalized spider-diagrams (Figure 8a-g). Distribution patterns of non-PGE trace elements look very similar for all sulfide types except for the m-type sulfides. Generally, siderophile elements and a part of the chalcophile elements display almost unfractionated patterns (often complicated, however, by peaks at Cu and Ag), with concentrations varying from chondritic to ~10 × CI chondrite. The right part of all spider diagrams in Figure  8 shows significant steady depletion in the chalcophile elements such as In, Sn, Ge, Ga, and often Zn and in lithophile Ti and Cr (down to ≥0.01 × CI chondrite).  Base metal sulfides of the e-type are characterized by higher concentrations of both Ti and Cr than the BMS of the i-type sulfides (Table 5). This suggests "conservative" behavior of these lithophile elements during mantle melting and their preferable retention in the residue. Although Li and Audetat [58] did not include Ti and Cr in their experimental compositions, they showed that V (a proxy for Cr) would preferably be retained in the residue. An interesting observation was made for chalcopyrite from the sulfide inclusion in Cr-spinel (Figure 4c). Chalcopyrite developed along the edges of the inclusion displays strongly elevated concentrations of Cr (5659 ppm), whereas chalcopyrite from lamellae in the same inclusion does not show significant enrichment in Cr (823 ppm). Therefore, Cu-rich liquid incorporated into the Cr-spinel interior through the cracks and represented now by massive chalcopyrite around the Pn-Ccp core seems to interact with  [41]). Diagrams are grouped according to the sulfide types identified on the basis of the PGE systematics (see Figure 7). (a) Unmetasomatized residual sulfides of the e-and i-types from Cpx-rich Spl lherzolites and Spl-Grt lherzolites. Peaks at Cu and Ag are observed for sulfides displaying mixed spectra, i.e., Pn-Ccp (see text for more details); (b) residual sulfides of the e-type in Cpx-poor Spl lherzolite and Cpx-rich Spl lherzolite. Peaks at Cu and Ag are observed for sulfides displaying mixed spectra (Pn-Ccp). Monomineral Pn does not display a peak at Cu, but it is still enriched in Ag; (c) melt-like (see Westner et al. [27]) sulfides of the i-and e-types that display unfractionated PGE patterns; all such melt-like sulfides display mixed spectra; (d) melt-like sulfides of the i-and e-types from all lherzolite types (most analyzed sulfides belong with this group) that display mixed but not significantly fractionated PGE patterns; (e) melt-like sulfides of the i-type that display the P-PGE as being significantly over the I-PGE in the PGE patterns. Sheer Pn and Pld do not display a peak at Cu, but they are still enriched in Ag; (f) Cu-Ni mss in Cpx-poor Spl lherzolite (no data on Ti and Cr are available); (g) partly incomplete patterns for m-type sulfides. Py displays generally unfractionated patterns complicated by troughs at Ni, Co, Cu, and Pb and subchondritic concentrations of Ti and Cr. Co-Vo is enriched in siderophile elements and is depleted in lithophile elements. Markers are as seen in Figure 7. Grey lines are for Pn-Ccp (see text for more details).
Base metal sulfides of the e-type are characterized by higher concentrations of both Ti and Cr than the BMS of the i-type sulfides (Table 5). This suggests "conservative" behavior of these lithophile elements during mantle melting and their preferable retention in the residue. Although Li and Audetat [58] did not include Ti and Cr in their experimental compositions, they showed that V (a proxy for Cr) would preferably be retained in the residue. An interesting observation was made for chalcopyrite from the sulfide inclusion in Cr-spinel (Figure 4c). Chalcopyrite developed along the edges of the inclusion displays strongly elevated concentrations of Cr (5659 ppm), whereas chalcopyrite from lamellae in the same inclusion does not show significant enrichment in Cr (823 ppm). Therefore, Cu-rich liquid incorporated into the Cr-spinel interior through the cracks and represented now by massive chalcopyrite around the Pn-Ccp core seems to interact with the host Cr-spinel. Very low concentrations of Mn are typical for all sulfides from Cpx-poor Spl lherzolites: these are below the detection limits in the e-type sulfides and are 3-24 ppm for the i-type sulfides. A different picture was observed for Cpx-rich Spl lherzolites and Spl-Grt lherzolites. Here, the e-type sulfides contain more Mn than i-type sulfides. It therefore seems that at higher P-T conditions, Mn is preferably retained in the residue. This is consistent with experimental data suggesting that Mn behaves as a siderophile element and is retained in mantle residue. However, pentlandite from the sulfide inclusion in Cr-spinel (Cpx-rich Spl lherzolite) displays low concentrations of Mn (32-49 ppm) that are comparable to those in sulfides of the i-type. However, chalcopyrite that is developed along the marginal parts of the sulfide inclusion in Cr-spinel displays strongly elevated concentrations of Mn (2122 ppm). Once again, as in the case of Cr, chalcopyrite from the exsolution lamellae in the same inclusion is much poorer in Mn (497 ppm). This observation points once more to the interaction of Cu-rich sulfide material with the host mineral. Cobalt is a dominant siderophile trace element in most studied sulfide types. Because Co behaves coherently with Ni, pentlandite is expectedly the main concentrator of Co (>5000 ppm). There is no significant difference between the Co concentrations in pentlandite from sulfides of the e-and i-types ( Table 5 and Table S1). Sulfides of both types were characterized by low Mo concentrations (mostly 1-10 × CI chondrite). For two analyses that have very high Mo concentrations in interstital pentlandite (1541-1561 ppm; Table 5 and Table S1), Mo forms zones of enrichment within a pentlandite body (Figure 9), probably due to the presence of molybdenite solid solution. Molybdenum is more concentrated in chalcopyrite from sulfides of the i-type than from sulfides of the e-type, whereas there are no differences in the concentration of the element in the pentlandite from sulfide of different types (Table 5 and  Table S1). Thus, we suggest preferential partitioning of Mo to sulfide liquid relative to the residue. According to experimental studies, Mo mostly behaves as a transitional element with some tendencies to be retained in the residue. Therefore, the observed behavior of Mo in chalcopyrite could be due to certain P-T-f O 2 conditions that are typical for each lherzolite type. the host Cr-spinel. Very low concentrations of Mn are typical for all sulfides from Cpxpoor Spl lherzolites: these are below the detection limits in the e-type sulfides and are 3-24 ppm for the i-type sulfides. A different picture was observed for Cpx-rich Spl lherzolites and Spl-Grt lherzolites. Here, the e-type sulfides contain more Mn than i-type sulfides. It therefore seems that at higher P-T conditions, Mn is preferably retained in the residue. This is consistent with experimental data suggesting that Mn behaves as a siderophile element and is retained in mantle residue. However, pentlandite from the sulfide inclusion in Cr-spinel (Cpx-rich Spl lherzolite) displays low concentrations of Mn (32-49 ppm) that are comparable to those in sulfides of the i-type. However, chalcopyrite that is developed along the marginal parts of the sulfide inclusion in Cr-spinel displays strongly elevated concentrations of Mn (2122 ppm). Once again, as in the case of Cr, chalcopyrite from the exsolution lamellae in the same inclusion is much poorer in Mn (497 ppm). This observation points once more to the interaction of Cu-rich sulfide material with the host mineral. Cobalt is a dominant siderophile trace element in most studied sulfide types. Because Co behaves coherently with Ni, pentlandite is expectedly the main concentrator of Co (> 5000 ppm). There is no significant difference between the Co concentrations in pentlandite from sulfides of the e-and i-types (Tables 5 and S1). Sulfides of both types were characterized by low Mo concentrations (mostly 1-10 × CI chondrite). For two analyses that have very high Mo concentrations in interstital pentlandite (1541-1561 ppm; Tables 5 and S1), Mo forms zones of enrichment within a pentlandite body (Figure 9), probably due to the presence of molybdenite solid solution. Molybdenum is more concentrated in chalcopyrite from sulfides of the i-type than from sulfides of the e-type, whereas there are no differences in the concentration of the element in the pentlandite from sulfide of different types (Tables 5 and S1). Thus, we suggest preferential partitioning of Mo to sulfide liquid relative to the residue. According to experimental studies, Mo mostly behaves as a transitional element with some tendencies to be retained in the residue. Therefore, the observed behavior of Mo in chalcopyrite could be due to certain P-T-fO2 conditions that are typical for each lherzolite type.  Zinc does not show any preference in partitioning into a sulfide of any particular type, and Zn amounts are more dependent on the xenolith type. Concentrations of Zn are much lower in sulfides from Cpx-poor Spl lherzolite than in sulfides from Cpx-rich Spl lherzolite and Spl-Grt lherzolite. Since Cpx-poor Spl lherzolite was subjected to intensive melting and depletion [4], it seems that Zn was significantly removed from the rocks. The much higher concentrations of Zn in Cpx-rich Spl lherzolite and in Spl-Grt lherzolite could be due to metasomatic re-enrichment, which intensively affected these two lherzolite types [4]. Concentrations of Ga, Ge, and In are typically low in the studied sulfides (often below the detection limit), and no overall partitioning tendencies could be revealed at such a low concentration of the elements. However, chalcopyrite is richer in Ga than in other BMS, and concentrations of Ga are higher in the chalcopyrite from the e-type sulfides than those from the mineral of the i-type sulfides ( Table 5 and Table S1). Since chalcopyrite is often a product of mss breakup and since Ga (like Cu) accumulates mostly in chalcopyrite, the amount of Ga in chalcopyrite would reflect that amount in the parental mss. Therefore, Ga would rather be retained in mantle residue during partial melting. Concentrations of Ge in the studied BMS are generally similar for the both e-type and i-type sulfides, with no partitioning preferences observed. Concentrations of Cd in pyrrhotite are higher than in pentlandite, and pyrrhotite from the e-type sulfides is richer in Cd (5.4-7.1 ppm) than the mineral from sulfides of the e-type (<1 ppm). It therefore seems that Cd, similar to Ga, would preferably be retained in the residue during partial melting. In spite of very low concentrations (Table 5 and Table S1), Sn and Sb show signatures of more preferable accumulation in e-type sulfides. Arsenic displays a slight tendency to partition more preferably into sulfide liquid that is more enriched in the i-type sulfides than in e-type sulfides although experimental data suggest that As would rather be retained in residue during partial melting. Selenium in the BSM does not show any partitioning preferences for any sulfide type. Silver displays much higher concentrations in e-type sulfides than in i-type sulfides ( Table 5 and Table S1) in Cpx-poor Spl lherzolites, suggesting Ag being retained in the mantle residue during partial melting. For other lherzolites, however, there is no difference in the Ag concentrations in different sulfide types. Since the redox state does not significantly differ between the types of the studied lherzolites [4,30], it seems that the behavior of Ag would be rather sensitive to P-T conditions. The element would preferably be retained in residue at lower P-T but would partition to sulfide liquid at a higher P-T. Such behavior of Ag seems to be consistent with experimental data. Lead displays very similar behavior, suggesting that the partitioning of the element strongly depends on the P-T conditions. Although experimental studies show that while Pb is preferably retained in the residue (which is, in part, consistent with our observations for lower P-T), overall, the behavior of Pb is more complicated than that of Ag and requires more detailed study. Concentrations of Te are low but are higher than those of Ga, Ge, Cd, In, Sn, and Sb (Table 5 and Table S1). The element displays a pronounced tendency to preferably partition into the e-type sulfides and is retained in the residue during partial melting. Typically, low concentrations of Bi are characteristic for the both e-type and i-type sulfides. No tendencies in the distribution of the element were observed, but that was most likely due to low Bi concentration in most of the studied sulfides.
Interstitial Cu-Ni mss from Cpx-poor Spl lherzolite displays overall high concentrations of most trace elements (from <10 up to~1000 × CI chondrite). The mss was not analyzed for Ti and Cr, but it displays a significant range of Mn concentrations (Table 5). Since Mn concentration directly correlates with the concentration of Fe, we suggest the presence of tiny Mn-Fe oxides within the mss matrix. Concentrations of Co (2560-3150 ppm) are lower in the Cu-Ni mss than those in pentlandite but are higher than in the BMS of the e-or i-types. This points to disequilibrium for the Cu-Ni mss that had not yet started to break up into the individual BMS. Concentrations of Mo are higher than in any other sulfide phase except for the i-type sulfides in Cpx-poor Spl lherzolite (Tables 5 and S1). These observations are consistent with the suggestion that at the P-T-f O 2 conditions typical for mantle rocks parental to the studied xenolith samples, Mo would rather partition to the sulfide liquid than be retained in the residue. The Cu-Ni mss contains very low amounts of Zn (below the detection limit), which is generally consistent with our observations that the BMS from Cpx-poor Spl lherzolites are significantly depleted in Zn. This can also suggest that the Cu-Ni mss was locally generated and did not migrate from a remote mantle domain. The concentration of elements such as Ge, Ga, As, Cd, In, Sb, Sn, and Bi are very low in the Cu-Ni mss (often below the detection limit), suggesting either their preferential retention in the mantle residue during partial melting or their low overall concentration in the host rock. Cu-Ni mss contains very high amounts of the elements such as Se, Ag, Te, and Pb ( Table 5), suggesting that they partition into sulfide liquid at certain P-T-f O 2 conditions. Trace element CI chondrite-normalized patterns (Figure 8f) are similar to those observed for the BMS (at higher trace element concentrations than for the Cu-Ni mss, however). If the Cu-Ni mss precipitated directly from the melt, then the primary trace element (other than the PGE) composition of the studied BMS (both e-and i-types) was significantly affected by the fractionation between sulfide phases due to the influence of melt fluids (as it can be judged from trace element chondrite-normalized patterns).
M-type sulfides typically contain moderate amounts of Ti and Cr with Co-violarite from Spl-Grt lherzolite displaying higher amounts of these elements (Table 6). On the contrary, Co-violarite from Cpx-poor Spl lherzolite contains the lowest amounts of Ti and Cr. Such differences suggest different Co-violarite origin conditions than those of the two xenolith types derived from different SCLM levels [4].
Interestingly enough, although Mn is supposed to behave coherently with Fe, both the highest and the lowest Mn amounts were measured in Fe-rich sulfides such as pyrite (Table 6). Among all sulfides, Co-violarite, by stoichiometry, is the richest in Co (up to 15 wt%), as it is a major element in this mineral. Pyrite contains only 1-113 ppm Co, which is consistent with Ni-poor nature of pyrite. Sulfides of the m-type typically contain Mo amounts that are similar to those of the i-type BMS. Concentrations of Zn are very variable in the m-type sulfides (Table 6), likely due to the very mobile and volatile nature of the element. Concentrations of Ga are low in all m-type sulfides, with the highest concentrations measured in Co-violarite (0.84-3.59 ppm). Sulfides of the m-type display the highest overall concentrations of volatile elements such as Ge and As among all of the studied sulfides (Table 6). This points to the potential participation of these two elements in the metasomatic re-enrichment of the BMS, probably at the crustal horizons. Among the m-type sulfides, pyrite contains the lowest amounts of Se (below 40 ppm), whereas Co-violarite displays elevated concentrations of Se (up to 273 ppm). Concentrations of Ag are low in pyrite and Co-violarite from Spl-Grt lherzolites (0.5-1.9 ppm), but they are high in Co-violarite from Cpx-poor Spl lherzolite (28-48 ppm). As shown for the BMS, Ag is very sensitive to P-T conditions. Therefore, different Ag concentrations suggest different conditions for the generation of metasomatic veinlets in Cpx-poor Spl lherzolites and in Spl-Grt lherzolites. Concentrations of Cd and Sb are generally low in most m-type sulfides except for Co-violarite from Cpx-poor lherzolite, which display non-typically high amounts of Cd (10-12 ppm) and Sb (1-8 ppm). Concentrations of In are often below or just slightly above the detection limit that is similar to the BMS from lherzolite matrix. Concentrations of Sn in m-type sulfides are generally similar to those observed in the BMS from the lherzolite matrix. Tellurium is unevenly distributed among the m-type sulfides, with concentrations varying from below the detection limit in some pyrite and Co-violarite from Spl-Grt lherzolite to 34 ppm in Co-volarite from Cpx-poor Spl lherzolite. Similarly, concentrations of Pb vary from being very low in pyrite and Co-violarite from Spl-Grt lherzolite (0.1-2.7 ppm) to being very high (>60 ppm) in Co-violarite from Cpx-poor Spl lherzolite. Such distributions of Te and Pb are supposed to be due to the P-T-(f O 2 ) conditions for the generation of the certain m-type sulfides. Bismuth is very poor in all m-type sulfides, with concentrations similar to those for the studied BMS. Co-violarite is enriched in siderophile elements and depleted in lithophile elements (Figure 8g). Trace element patterns for m-sulfides (Figure 8g) are different from those for BSM, suggesting a different origin of the m-type sulfides.
All things considered, we can say that unlike in the case of the PGE, only a few other trace elements (Ti, Cr, Cd, ±Mo) in the BMS from lherzolite xenoliths display pronounced preferences to either partition to the sulfide liquid or to be retained in the residue during partial melting. Observed partitioning is mostly consistent with experimental studies [58,60] and PGE features. Most analyzed trace elements (other than the PGE) however, display dual behavior, i.e., in some lherzolites they preferably partition to sulfide liquid, while in others they would be retained in the residue. Such dual behavior seems to depend strongly on the P-T-f O 2 conditions during partial melting, which is consistent with the features observed during the experimental runs [58,60]. Mostly similar trace element (other than the PGE) composition of the e-type and i-type sulfides accompanied by the dual behavior of most analyzed trace elements in the studied lherzolite xenoliths could point to the remobilization of interstitial sulfides and their aggressive influence on the sulfides enclosed in minerals. The presence of the chains of tiny sulfide droplets along existing or healed fractures in primary xenolith minerals (Figure 4b,c) might be the result of such remobilization. Unlike most other trace elements, the PGE show different concentrations in sulfides of the e-and i-types, likely due to much more conservative behavior of the PGE preserving their primary features from being affected by later metasomatic processes. Sulfides of the e-type display residual PGE characteristics, whereas the i-type sulfides mainly display fingerprints of the secondary metasomatic re-enrichment. The i-type sulfides carry residual PGE signatures in only a few cases. All of the conducted observations show that the suggestion of Dromgoole and Pasteris [19] regarding the lack of the compositional differences between the e-and i-type sulfides because of the incorporation of postmetasomatic S-bearing and CO 2 -rich melts is close to reality. Such a lack of compositional differences was due to the later metasomatic re-enrichment of the e-type sulfides by the melts/fluids responsible for the generation of i-type sulfides. The few inconsistencies that exist between the experimental data and features observed in natural systems (lherzolite xenoliths) can be related to difficulties in precisely modeling the natural systems while preparing the experimental runs. Therefore, a complex approach including both the study of the natural systems and the experimental work at the upper mantle P-T-f O 2 conditions is necessary in order to have a comprehensive view of the features of trace element behavior in silicate-sulfide systems.
Among the m-type sulfides, the trace element characteristics of the Co-violarite from two different lherzolite types, i.e., from Cpx-poor Spl lherzolite (32601-9b) and Spl-Grt lherzolite (DN-4), provided an opportunity to compare features of the metasomatic fluids/melts affecting the SCLM rocks derived from different upper mantle levels. It is notable that Co-violarite from Cpx-poor Spl lherzolite has much higher concentrations of most trace elements (other than the PGE) than the mineral from Spl-Grt lherzolite ( Table 6). While a sample (xenolith) of Cpx-poor Spl lherzolite was most likely trapped from the shallowest upper mantle horizons bordering the crust (as we can judge from the low ambient temperature (852 • C) and the composition of the xenolith), a sample of Spl-Grt lherzolite was trapped from much deeper upper mantle horizons (T = 1125 • C; P = 21.5 kb; [4]). Since all of the studied xenoliths were collected from the same magmatic body and because they were delivered to the surface relatively quickly, it is unlikely that there was much variability in the interaction of the individual xenoliths with the transporting melt. Since the Cpx-poor Spl lherzolite xenolith 32601-9b might have been trapped from the uppermost mantle horizons bordering the lower crust, the influence of crustal fluids that are rich in elements such as Te, Pb, Ag, Se, and Zn might have been strong at those horizons. On the other hand, the metasomatic alteration experienced by the Spl-Grt lherzolite xenolith DN-4 might have been due to the migration of the intramantle fluids (silicate melt-related?) which were relatively depleted in "crustal" trace elements and instead enriched in elements such as As, Ge, Sn, and Sb.

Conclusions
The studied sulfides from the Jetty Peninsula SCLM lherzolite xenoliths are developed as inclusions in rock-forming minerals, interstitial grains between such minerals, tiny blebs along the fracture surfaces in primary minerals, and individuals in carbonate-silicate patches and metasomatic veinlets cross-cutting the lherzolite primary mineral assemblages. Sulfide modal abundance in the xenoliths is Cpx-poor lherzolites ≤ Spl-Grt lherzolites << Cpx-rich lherzolites, with major mineral assemblages of pentlandite + chalcopyrite ± pyrrhotite. Minor and rare sulfides are polydymite, millerite, violarite, siegenite, and the mss. A portion of the primary immiscible sulfide liquid formed during the partial melting of mantle material was trapped in the solid-liquid residuum represented by the lherzolite and eventually ended up as the BMS. The multiphase sulfide assemblages in the lherzolite xenolith matrix are not of primary mantle origin, but a product of subsolidus re-equilibration of the primary high-temperature mss. The major-element composition of the sulfide assemblages indicates that subsolidus low-temperature equilibration was incomplete and likely stopped at temperatures ≤300 • C.
More conservative PGE preserve primary features showing that e-type sulfides represent either residues from the melting processes or resulted from the remobilization of sulfide liquid incorporated in the host minerals along the cracks. The i-type sulfides are crystallization products of intramantle sulfide-bearing fluids/melts. The i-type sulfides very rarely display residual PGE signatures. Only a few non-PGE trace elements record residual characteristics in e-type sulfides. The most mobile trace elements that partitioned into interstitial sulfide liquid could metasomatically re-enrich residual sulfides of the e-type by overprinting their residual signatures. Such dual behavior of the trace elements seems to strongly depend on P-T-f O 2 conditions during and after partial melting. The i-type sulfides record the trace element characteristics of crystallization from the intramantle sulfide-bearing melts/fluids. Sulfide-bearing metasomatic veinlets and silicate-carbonate patches might have resulted from low-temperature metasomatism at the basis of the Earth's crust and localized at the intramantle metasomatism due to percolating immiscible sulfidecarbonate-silicate fluids/melts. A complex approach including both a study of the natural systems and experimental work at the upper mantle P-T-f O 2 condition is necessary in order to have a comprehensive view of trace element behavior in silicate-sulfide systems.