Search

By examining the minor and trace element contents of chromites from three intrusions—the Bushveld Complex (South Africa), the Stillwater Complex (USA), and the Great Dyke (Zimbabwe)—and comparing these chromite compositions with those of magmas from which they could have formed (komatiites and picrites) we conclude that ( i) the variations in Ti, V, Sc, and Ga contents across stratigraphy and across individual layers do not support the model of magma mixing leading to chromite-only crystallization, ( ii) the chromites from the lowest levels of the intrusions could have crystallized from komatiite liquids that were contaminated with continental crust, ( iii) the Great Dyke chromites have the highest Cr# and lowest incompatible element contents and formed from a liquid closest to komatiite, ( iv) all of the chromites, except those of the Dunite Succession of the Great Dyke have equilibrated with a liquid that also had crystallized pyroxene, ( v) the Great Dyke and Stillwater chromites show a narrower range in composition than the Bushveld chromites, and ( vi) Chromites from the western limb of the Bushveld Complex contain much higher V contents than all the other chromites. This requires either, that the oxygen fugacity ( fO2) was lower in the western Bushveld or that the chromites equilibrated with a V-rich magma. We favor a model where chromite and silicate minerals crystallized in cotectic proportions (∼2:98). The chromite, silicates, and transporting liquid are emplaced into the magma chamber. During emplacement the chromite and silicate separated due to viscous particle flow to form a massive chromite layer overlain by silicates.

We have compiled the trace element concentrations in pyrrhotite, pentlandite, chalcopyrite, and pyrite from magmatic Ni-Cu-PGE ore deposits with the aim of understanding their petrogenesis and whether these minerals can be used as indicator minerals. Among the samples, there are some of the most studied world-class Ni-Cu- (Aguablanca, Duluth, Jinchuan, Noril’sk-Talnakh-Kharaelakh, Sudbury, Voisey’s Bay, and others) and PGE-dominated (Bushveld, Lac des Iles, Stillwater, Great Dyke, and Penikat) deposits. Crustal assimilation may be constrained using As/Se and Sb/Se ratios in pentlandite. The degree of interaction between the silicate and sulfide liquids (R-factor) can be estimated by the content of highly chalcophile elements (Dsulf liq/sil liq above 1000) in sulfide minerals. The fractional crystallization of the sulfide liquid can be traced using Se/Te ratios of pentlandite. Pyrite formed by exsolution from MSS has higher Rh, Ru, Ir, and Os than co-existing pyrrhotite, whereas pyrite formed by hydrothermal alteration of pyrrhotite inherits the Rh, Ru, Ir, and Os contents of the pyrrhotite it replaced. Sulfide minerals are preserved in transported glacial cover and their trace element chemistry can be used to discriminate their source. Pentlandite from Ni-Cu deposits has much lower Rh and Pd concentrations than those from PGE-dominated deposits, pyrite from magmatic deposits has higher Co/Sb and Se/As ratios relative to pyrite from hydrothermal deposits, and chalcopyrite from magmatic deposits has much higher Ni and lower Cd concentrations than those from hydrothermal deposits.

Mineralogical and chemical zonations observed in massive sulfide ores from Ni-Cu-platinum group element (PGE) deposits are commonly ascribed to the fractional crystallization of monosulfide solid solution (MSS) and intermediate solid solution (ISS) from sulfide liquid. Recent studies of classic examples of zoned orebodies at Sudbury and Voisey’s Bay (Canada) demonstrated that the chemistry of magnetite crystallized from sulfide liquid was varying in response to sulfide fractional crystallization. Other classic examples of zoned Ni-Cu-PGE sulfide deposits occur in the Norilsk-Talnakh mining district (Russia), yet magnetite in these orebodies has received little attention. In this contribution, we document the chemistry of magnetite in samples from Norilsk-Talnakh, spanning the classic range of sulfide composition, from Cu poor (MSS) to Cu rich (ISS). Based on textural features and mineral associations, four types of magnetite with distinct chemical composition are identified: (1) MSS magnetite, (2) ISS magnetite, (3) reactional magnetite (at the sulfide-silicate interface), and (4) hydrothermal magnetite (resulting from sulfide-fluid interaction). Compositional variability in lithophile and chalcophile elements records sulfide fractional crystallization across MSS and ISS magnetites and sulfide interaction with silicate minerals (reactional magnetite) and fluids (hydrothermal magnetite). Estimated partition coefficients for magnetite in sulfide systems are unlike those in silicate systems. In sulfide systems, all lithophile elements are compatible and chalcophile elements tend to be incompatible with magnetite, but in silicate systems some lithophile elements are incompatible and chalcophile elements are compatible with magnetite. Finally, comparison with magnetite data from other Ni-Cu-PGE sulfide deposits pinpoints that the nature of parental silicate magma, degree of sulfide evolution, cocrystallizing phases, and alteration conditions influence magnetite composition.

The distribution of platinum-group element (PGE) and Te, As, Bi, Sb and Sn (TABS) in whole-rock samples, and in disseminated base metal sulfides (BMS) pentlandite, pyrrhotite and chalcopyrite from the Bushveld and Stillwater Complexes are reported. The samples are from: the Merensky Reef (Bushveld), the J-M Reef (Stillwater), Picket Pin deposit (Stillwater), and also barren sulfide-bearing samples, from outside the reef intervals from both intrusions. The objective of the study was to document the distribution of PGE and TABS in PGE-reef deposits, and to investigate whether TABS play a significant role during the formation of PGE-reef deposits. The whole-rock concentrations of PGE and TABS (except for As) correlate with S and PGE, and thus their distribution appear to be controlled by BMS. The distribution of As, and to a lesser extent Sb, correlate with incompatible elements and with changes in K-phologopite compositions, suggesting that these elements are controlled both by the amount of trapped liquid in cumulate rocks, and the amount of sulfides. The possible role of TABS in forming pre-nucleation clusters (nanonuggets) to enrich the reefs in PGE is considered and discarded, because the ratio of TABS/PGE

In most magmatic sulfide deposits, platinum-group elements (PGE) are found both within the structure of the base metal sulfides (BMS), pyrrhotite (Po), pentlandite (Pn), chalcopyrite (Ccp) and cubanite (Cbn) and as platinum-group minerals (PGM). Tellurium, As, Bi, Sb and Sn (TABS) are essential elements in many of these PGM. The potential role of TABS in collecting PGE, and thus forming a PGE deposit, has not been closely investigated. We have determined the concentrations of a full suite of chalcophile elements in Po, Pn, Ccp and Cbn using laser ablation-inductively coupled plasma mass spectrometry on samples from the Noril’sk-Talnakh Ni deposits. In these deposits, the Po-rich ore is thought to represent monosulfide solid solution (MSS) cumulate of the initial sulfide liquid, and the Ccp-rich ore a mixture of the fractionated sulfide liquid and intermediate solid solution (ISS). The BMS from the Po-rich ore contain lower concentrations of TABS, Pd, Pt and Au, and higher concentrations of Mo, Ru, Rh, Re, Os and Ir than BMS from the Ccp-rich ores. This observation is consistent with experimental results which show that TABS, Pd, Pt and Au are incompatible with MSS, whereas the other elements are compatible in MSS. Counter intuitively, in the Po-rich ore, the bulk of the Pd and TABS is hosted by BMS. This is because during crystallization, although only a small amount of the incompatible elements partitioned into the BMS, the fractionated liquid has migrated away; thus, the Po-rich ores represent MSS adcumulates. Therefore, as the Po-rich ores contain very little trapped liquid fraction, the BMS host the bulk of Pd and TABS. In contrast, in the Ccp-rich ore, the bulk of Au, Pd, Pt and TABS is present as PGM or electrum grains. This is because more trapped liquid is present, and as TABS Au, Pd and Pt are not compatible with ISS, they concentrated into the very last sulfide liquid, and crystallized as intergrowths of Pd-Pt-TABS PGM. The TABS then do not appear to collect Pd, Pt and Au but rather all elements are concentrated in the most fractionated sulfide liquid by crystal fractionation.

The study of Te, As, Bi, Sb and Se (TABS) has increased over the past years due to their use in the development of low‐carbon energy technologies. However, there is a scarcity of mass fraction values of TABS in geological reference materials. This underlines the difficulty in undertaking routine determinations of these elements. The mass fractions of TABS were determined in geological reference materials using hydride generation‐atomic fluorescence spectrometry (HG‐AFS), calibrated with standard solutions. Comparisons with literature values were used to validate the method. Samples from the GeoPT proficiency test were also analysed. For most elements, there are no assigned or even provisional values for many of the GeoPT and reference materials because of the wide range of results reported. For mass fractions above the quantification limit of the method, our results are in good agreement with the median of GeoPT results. Thus, we propose GeoPT median values as informational values for these elements. In contrast, at mass fractions < 0.5 µg g−1 median values of Se from GeoPT are systematically higher than our results. Our Se results are in agreement with the reference materials down to 0.02 µg g−1, which suggest that many of the results for Se reported in GeoPT testing are too high.

The distribution of highly siderophile elements is used in the study of a wide variety of geological topics, from planet formation and evolution to the formation of ore deposits. Under mantle and crustal conditions, these elements behave as highly chalcophile elements, and pentlandite (Pn) is an important host for most of these elements. Therefore, understanding how Pn forms is important to understanding the processes that control these elements. The classic model for the formation of Pn is that below 650 °C, the high-temperature sulfides—monosulfide solid solution (MSS) and intermediate solid solution (ISS)—are no longer stable and exsolve into pyrrhotite (Po), Pn, and chalcopyrite (Ccp). However, Pn has been shown to be the main host of Pd in many ore deposits, and given that Pd is incompatible with both MSS and ISS, this observation is inconsistent with the exsolution model. Furthermore, experimental work has shown that Pn can form by peritectic reaction between MSS and fractionated sulfide liquid. To date, this type of Pn has not been reported in natural samples. In our study of chalcophile-element concentrations in Pn from iconic magmatic Ni–Cu–platinum-group element deposits, we observed three textures of Pn: contact Pn in between Po and Ccp, granular Pn included within Ccp or Po, and flame Pn included within Po. The contact Pn shows zonation in Mo, Rh, Ru, Re, Os, and Ir, with these elements being enriched toward the Po contact and depleted toward the Ccp contact. In some cases, Pd displays a zonation antithetical to that of these elements. In this contribution, we propose that the contact Pn formed via the peritectic reaction described above, and inherited Mo, Ru, Rh, Re, Os, and Ir from the MSS, whereas Pd was contributed from the fractionated sulfide liquid. We expect that this type of Pn should be present wherever MSS and fractionated sulfide liquid remained in contact.

Bedrock in arctic and subarctic regions is covered by glacial deposits making the discovery of new mineral deposits difficult. Indicator mineral methods using glacial sediment have thus been developed for mineral exploration in such drift-covered areas. However, sulfide indicator minerals have been under utilized because it was thought that they would not survive glacial transport and post-depositional oxidation during soil formation. In this contribution we show that the 0.25–1 mm non-ferromagnetic heavy mineral concentrates of Quaternary till and esker samples from the Churchill province in northern Quebec, Canada, contain thousands of pyrite and chalcopyrite grains and a few sulfarsenide grains. Accordingly, sulfide minerals do survive glacial and glaciofluvial transport, even in the relatively oxidizing environment of the eskers, and their presence indicates the potential presence of mineralized bedrock up ice. The study area is therefore ideal to test the use of sulfide mineral chemistry for mineral assessment and vectoring. The composition of the pyrite and chalcopyrite grains recovered from the glacial deposits have been determined by LA-ICP-MS and compared with known values for sulfides in magmatic and hydrothermal deposits. Although some elements (e.g. Ag, Cu, Zn, Pb, W, Ba, La, and Yb) are enriched in narrow rims on some sulfide grains, indicating their limited mobility during oxidation, most elements have not been mobilized and reflect initial sulfide compositions in bedrock sources. The binary diagram Co/Sb versus Se/As shows that most of the pyrite grains in surficial sediments are of magmatic origin although some are from hydrothermal sources. The hydrothermal pyrites are enriched in hydrothermal pathfinders (Au, Hg, Ag, Tl, Pb, Zn, Cu, and Mo). The ternary diagram Se-Cd-Ni shows that chalcopyrites from both magmatic and hydrothermal deposits are present in glacial sediments. The high Cd/Zn ratios of the hydrothermal chalcopyrites are indicative of a high crystallization temperature, typical of metamorphosed VMS or SEDEX deposits. Integrated maps combining bedrock geology, glacial transport directions, sample locations, and sulfide grain compositions and populations can be used to delineate target sectors for mineral exploration. Here sulfides have been transported over ~100 km roughly towards north, from sources in the Rachel-Laporte Zone and the Labrador Trough, where metasedimentary/metavolcanic rocks and mafic/ultramafic intrusive rocks are favorable hosts for hydrothermal and magmatic mineralization, respectively.

The composition of the magmas from which the chromites that form the massive chromite layers of the Stillwater, Great Dyke and Bushveld Complexes are of interest both to understand the economic importance of the resources in the layers (Cr and PGE), but also in understanding how these layers form. Magmas that have been suggested as parental to the intrusions are boninites or crustally contaminated komatiites. Another magma that could be considered in recognition of the continental setting of the Bushveld and Great Dyke is picrite associated with continental flood basalts. In order to investigate whether any of these magmas are suitable parental magmas for the chromites we have determined major and trace elements in komatiites of low metamorphic grade, boninites and chromites from low-Ti and high-Ti picrites of the Emeishan Provence.In order to test whether the chromites are in equilibrium with volcanic magmas we first modelled the major and minor element composition of the chromites that should have crystallized from the komatiite, boninites and picrite liquids using SpinMelt v2. The compositions are approximately correct. In terms of major and minor elements none of the chromites from the layered intrusions match boninite chromites. The Great Dyke chromites are similar to chromites from komatiites. The chromites the Bushveld are slightly more evolved with higher Ti contents and lower Cr# and resemble the chromites from the low-Ti picrites of Emeishan. The Stillwater chromites have similar Ti contents to the Emeishan low-Ti picrites, but have lower Cr#. Their compositions resemble chromite compositions reported from the North Atlantic Igneous Provence.Hafnium, Ta, Cu, Sn, Sc, Ti, Mn, Ni, Co, Mn, Ga, V and Zn were determined by LA-ICP-MS. To compare the composition of the chromites an estimate of their partition coefficients into chromite was made based on the concentrations of elements in komatiite chromite divided by element in komatiite. The elements were then arranged in order of compatibility and the chromites normalized to the median komatiite chromite. Podiform chromites from boninites are depleted in most elements and none of the layered intrusions chromites resemble them. The chromites from the Great Dyke have essentially flat patterns close to 1 times komatiite, but with negative Cu anomaly and a slight positive Sn anomaly. The Bushveld and Stillwater chromites are richer in Al, Ga, V and Ti than the komatiite chromite and are depleted in Cu. The patterns resemble the chromites form the low Ti-picrites form Sn to Zn, but differ from picrites from Hf to Cu. The picrites are enriched in Hf, Ta and Cu.The chromite compositions suggest that boninite magmas are not involved in forming the chromites from layered intrusions. The Great Dyke chromites appear to have a komatiitic affinity. The Bushveld and Stillwater chromites appear to have a low-Ti picrite affinity.

The association of platinum-group elements (PGE) and the chalcophile elements Te, As, Bi, Sb and Sn (TABS) has been documented in several magmatic sulfide deposits. These groups of elements are either hosted within sulfide minerals, or combine to form discrete platinum-group minerals (PGM) associated with sulfide minerals. However, the concentration of TABS in parental magmas from which magmatic sulfide deposits formed was still missing. This study presents the distribution of TABS and Se in B-1, B-2 and B-3 rocks of the Marginal Zone of the Bushveld Complex. These rocks have been proposed as representative of the parental liquids from which the Bushveld Complex crystallized, thus allowing us to assess the concentration of Se and TABS in the liquids from which some of the largest PGE deposits in the world have formed. Concentrations of As and Sb in the initial Bushveld liquid (B-1) are significantly higher than in primary magmas, whereas the Se and TABS of later magmas (B-2 and B-3) are similar to primary magmas. We attribute the difference due upper crustal contamination of the B-1 magma, whereas the B-2 and B-3 magmas were most likely contaminated with a plagioclase-rich residuum formed upon the partial melting of the upper crust. Moreover, we modeled the concentrations of the TABS in the Merensky Reef using a mixture of two of the magma types present in the Marginal Zone (the B-1 and B-2) as the initial silicate liquid. The modeled concentrations closely resemble the measured values obtained for a section across the Merensky Reef at the Impala mine. This supports the B-1 and B-2 mixture as an appropriate initial liquid for the crystallization of the Merensky Reef. The modeling also shows that the distributions of Se, Te and Bi across the Merensky Reef are controlled by the sulfide liquid component. In contrast, As and Sb distributions are influenced both by the amount of silicate melt component in the cumulates and the sulfide liquid component. This is because Se, Te and Bi are moderately to strongly chalcophile elements, but As and Sb are only slightly chalcophile elements. Consequently, the effect of crustal contamination for elements with high partition coefficients between sulfide and silicate liquid (Te, Bi and Se) is obscured by the interaction of sulfides with a large volume of silicate magma. Therefore, the concentrations of these elements are higher in samples with greater proportions of sulfide minerals. In contrast, for elements with lower partition coefficients (As and Sb), the whole-rock concentrations are not upgraded by the presence of sulfide minerals, and thus the effect of crustal contamination can be more readily assessed.

We have compiled the trace element concentrations in pyrrhotite, pentlandite, chalcopyrite, and pyrite from magmatic Ni-Cu-PGE ore deposits with the aim of understanding their petrogenesis and whether these minerals can be used as indicator minerals. Among the samples, there are some of the most studied world-class Ni-Cu- (Aguablanca, Duluth, Jinchuan, Noril’sk-Talnakh-Kharaelakh, Sudbury, Voisey’s Bay, and others) and PGE-dominated (Bushveld, Lac des Iles, Stillwater, Great Dyke, and Penikat) deposits. Crustal assimilation may be constrained using As/Se and Sb/Se ratios in pentlandite. The degree of interaction between the silicate and sulfide liquids (R-factor) can be estimated by the content of highly chalcophile elements (Dsulf liq/sil liq above 1000) in sulfide minerals. The fractional crystallization of the sulfide liquid can be traced using Se/Te ratios of pentlandite. Pyrite formed by exsolution from MSS has higher Rh, Ru, Ir, and Os than co-existing pyrrhotite, whereas pyrite formed by hydrothermal alteration of pyrrhotite inherits the Rh, Ru, Ir, and Os contents of the pyrrhotite it replaced. Sulfide minerals are preserved in transported glacial cover and their trace element chemistry can be used to discriminate their source. Pentlandite from Ni-Cu deposits has much lower Rh and Pd concentrations than those from PGE-dominated deposits, pyrite from magmatic deposits has higher Co/Sb and Se/As ratios relative to pyrite from hydrothermal deposits, and chalcopyrite from magmatic deposits has much higher Ni and lower Cd concentrations than those from hydrothermal deposits.

The concentrations of Te, As, Bi, Sb and Se (TABS+) in magmas from large igneous provinces (LIPs) are of interest because these elements are important in the formation of platinum-group minerals (PGM) found in magmatic Ni Cu platinum-group element (PGE) deposits. Furthermore, the TABS+ are volatile and hence they may be lost in degassing and have a role to play in the mass extinctions associated with LIPs events because these are mostly toxic elements. However, the concentrations of TABS+ in magmas from LIPs are not well documented due in part to the analytical difficulties and in part to the numerous processes that affect their distribution. We have determined the concentration of TABS+ and PGE in rocks from the Parana Magmatic Province (PMP) in order to assess the effects of fractional crystallization, sulfide segregation and magma degassing. Decrease in the Rh, Ru and Ir concentrations with Mg and Cr indicate that these elements behaved as compatible elements during crystal fractionation. Based on changes in the Cu/Pd sulfide saturation occurred in some cases resulting into a decrease in Pd and Pt. Arsenic and Sb behave as incompatible elements, whereas Se, Te and Bi show variable behaviour. Most lavas from the PMP display negative As, Se and Te anomalies on mantle-normalized patters coupled with Te/Cu, Se/Cu and As/Th and Sb/Th ratios lower than mantle values, which we attribute to loss of the TABS+ during degassing.

The Jaguar Ni sulfide deposit (mineral reserves of 58.6 Mt at 0.95 wt% Ni) occurs along a regional fault zone in the Carajas Mineral Province in the Amazonian Craton, Brazil. This study presents the first detailed description of the Jaguar deposit and provides a basic framework to constrain the origin of the Ni sulfide mineralization. Jaguar nickel sulfide orebodies are associated with ductile-brittle hydrothermal alteration zones developed over felsic subvolcanic or granitic country rocks. Prominent features of the Jaguar deposit include: i. its location along regional scale deep fault systems, ii. extensive zones of biotite-chlorite alteration overprinted by magnetite-apatite-quartz alteration zones, iii. sulfide mineralization occurring as steeply dipping orebodies closely associated with magnetite-apatite rich bodies, iv. sulfides that are disseminated or form the matrix of breccias consisting of pyrite and millerite, with minor pentlandite, chalcopyrite and pyrrhotite, v. pyrite, magnetite and apatite occur mostly as euhedral to subhedral crystals enclosed in irregular aggregates of fine-grained anhedral millerite, vi. sulfide mineralization with high Ni (Ni/S ~ 0.15–0.30) and Co (Co/S ~ 0.005–0.015) contents, high Ni/Cu ratios (~5–25) and low Zn and PGE contents, vii. metal association consisting of LREE, Fe, U, P, Pb, Bi, Te, Ni, and Co, and viii. mineralized zones with high F contents and a positive correlation of F with S, Ni and P2O5. The geological setting and characteristics of the Jaguar deposit are similar to the iron-oxide-copper-gold (IOCG) deposits in the Carajas Mineral Province, indicating that they all belong to a regional hydrothermal system. The existence of nickel-rich members of IOCG type mineralizations in the Carajas Mineral Province was previously suggested for other deposits (e.g., GT34, Castanha) and interpreted as deeper portions of the IOCG mineral system. The close association of nickel sulfide mineralization in the Jaguar deposit with magnetite-apatite-rich breccias may be interpreted as an iron-oxide-apatite (IOA) member within the regional hydrothermal system. In contrast to most hydrothermal nickel deposits in the world, the Jaguar deposit is neither hosted by mafic-ultramafic rocks nor associated with black shales. Therefore, the Jaguar deposit is interpreted as an uncommon hydrothermal Ni deposit associated with the IOCG mineral system in the Carajas Mineral Province. Our results provide new insights for worldwide exploration for Ni sulfide deposits in geological environments different from those traditionally considered prospective.

The Neoarchean (ca. 2.75 Ga) Luanga Complex, located in the Carajas Mineral Province in Brazil, is a medium-size layered intrusion consisting, from base to top, of ultramafic cumulates (Ultramafic Zone), interlayered ultramafic and mafic cumulates (Transition Zone) and mafic cumulates (Mafic Zone). Chromitite layers in the Luanga Complex occur in the upper portion of interlayered harzburgite and orthopyroxenite of the Transition Zone and associated with the lowermost norites of the Mafic Zone. The stratigraphic interval that hosts chromitites (∼150 meters thick) consists of several cyclic units interpreted as the result of successive influxes of primitive parental magma. The compositions of chromite in chromitites from the Transition Zone (Lower Group Chromitites) have distinctively higher Cr# (100Cr/(Cr + Al + Fe3+)) compared with chromite in chromitites from the Mafic Zone (Upper Group Chromitites). Chromitites hosted by noritic rocks are preceded by a thin layer of harzburgite located 15–20 cm below each chromitite layer. Lower Cr# in chromitites hosted by noritic rocks are interpreted as the result of increased Al2O3 activity caused by new magma influxes. Electron microprobe analyses on line transverses through 35 chromite crystals indicate that they are rimmed and/or extensively zoned. The composition of chromite in chromitites changes abruptly in the outer rim, becoming enriched in Fe3+ and Fe2+ at the expense of Mg, Cr, Al, thus moving toward the magnetite apex on the spinel prism. This outer rim, characterized by higher reflectance, is probably related to the metamorphic replacement of the primary mineralogy of the Luanga Complex. Zoned chromite crystals indicate an extensive exchange between divalent (Mg, Fe2+) cations and minor to none exchange between trivalent cations (Cr3+, Al3+ and Fe3+). This Mg-Fe zoning is interpreted as the result of subsolidus exchange of Fe2+ and Mg between chromite and coexisting silicates during slow cooling of the intrusion. A remarkable feature of chromitites from Luanga Complex is the occurrence of abundant silicate inclusions within chromite crystals. These inclusions show an adjacent inner rim with higher Cr# and lower Mg# (100 Mg/(Mg + Fe2+)) and Al# (100Al/(Cr + Al + Fe3+)). This compositional shift is possibly due to crystallization from a progressively more fractionated liquid trapped in the chromite crystal. Significant modification of primary cumulus composition of chromite, as indicated in our study for the Luanga Complex, is likely to be common in non-massive chromitites and the rule for disseminated chromites in mafic intrusions.

The Luanga Complex is part of the Serra Leste Magmatic Suite, a cluster of PGE-mineralized mafic–ultramafic intrusions located in the northeastern portion of the Carajas Mineral Province. The Luanga Complex is a medium-sized layered intrusion consisting of three main zones: i. the lower Ultramafic Zone comprising ultramafic adcumulates (peridotite), ii. the Transition Zone comprising interlayered ultramafic and mafic cumulates (harzburgite, orthopyroxenite and norite) and iii. the upper Mafic Zone comprising a monotonous sequence of mafic cumulates (norite) with minor orthopyroxenite layers. Several PGE-mineralized zones occur in the Transition Zone but the bulk of the PGE resources are hosted within a 10–50 meter thick interval of disseminated sulfides at the contact of the Ultramafic and Transition Zones. The compositional range of cumulus olivine (Fo 78.9–86.4 ) is comparable to those reported for layered intrusions originated from moderate primitive parental magmas. Mantle normalized alteration-resistant trace element patterns of noritic rocks are fractionated, as indicated by relative enrichment in LREE and Th, with negative Nb and Ta anomalies, suggesting assimilation of older continental crust. Ni contents in olivine in the Luanga Complex (up to 7500 ppm) stand among the highest values reported in layered intrusions globally. The highest Ni contents in olivine in the Luanga Complex occur in distinctively PGE enriched (Pt + Pd > 1 ppm) intervals of the Transition Zone, in both sulfide-poor and sulfide bearing (1–3 vol.%) rocks. The origin of the PGE- and Ni-rich parental magma of the Luanga Complex is discussed considering the upgrading of magmas through dissolution of previously formed Ni-rich sulfide melts. Our results suggest that high Ni contents in olivine and/or orthopyroxene provide an additional exploration tool for Ni–PGE deposits, particularly useful for target selection in large magmatic provinces.

The association of platinum-group elements (PGE) and the chalcophile elements Te, As, Bi, Sb and Sn (TABS) has been extensively documented in several magmatic sulfide deposits over the past years. However, understanding the roles of TABS during the formation of magmatic sulfide deposits partially depends on constraining the concentration of TABS on the liquids from which they crystallized. This study presents the distribution of TABS (apart from Sn) in rocks of the Marginal Zone of the Bushveld Complex. These rocks record the composition of the parental magmas from which the Bushveld Complex have crystallized. The major and trace elements of the marginal rocks have been modelled as mixtures of komatiite, continental crust and a plagioclase-rich residuum. Similar mixtures are required to model the TABS in the marginal rocks, with the continental crust component contributing a large part of the As, Sb and Bi budgets in these melts. The concentrations of the TABS in the Merensky Reef can be modelled as a mixture of two of the magma types present in the Marginal Zone (the B-1 and B-2). The modelling also reveals that the distributions of Se, Te and Bi in the reef are essentially controlled by the presence of sulfide minerals, whereas As and Sb distributions are controlled by both sulfide minerals and melt component. This is because Se, Te and Bi are moderately to strongly chalcophile elements, but As and Sb are only slightly chalcophile elements. Thus, whole-rock As and Sb concentrations are not upgraded by the formation of the sulfide minerals, and may still be used to trace crustal contamination.

The Luanga Complex, located in the eastern portion of the Carajas Mineral Province, is part of a cluster of PGE-mineralized layered intrusions, grouped into what is known as the Serra Leste magmatic suite. The Luanga deposit, the largest PGE deposit in South America, has two distinct styles of PGE mineralization. The first type, termed as Sulfide Zone, consists of a 10–50 m thick interval with disseminated base metal sulfides (pentlandite > pyrrhotite >>> chalcopyrite) located along the upper contact of the intrusion's Ultramafic Zone. The Sulfide Zone extends along the entire length of the intrusion (~3 km) and hosts the bulk of PGE resources of the Luanga Complex (i.e., 142 Mt at 1.24 ppm Pt + Pd + Au and 0.11% Ni). The second type of PGE mineralization, termed as low-S-high-Pt-Pd Zones, consists of 2–10 m thick stratabound PGE mineralization within a sequence of interlayered ultramafic and mafic cumulates located above the Sulfide Zone. Host rocks of the low-S-high-Pt-Pd Zones consist mainly of sulfide- and chromite-free harzburgite and orthopyroxenite. These mineralized rocks do not show any distinctive texture or change in modal composition. The Sulfide Zone and low-S-high-Pt-Pd Zons have distinct PGE distribution. The Sulfide Zone has Pt/Pd ratios of 0.52 and a positive correlation between PGE and S. The low-S-high-Pt-Pd Zones have Pt/Pd ratios of 1.2 and depletion in IPGE relative to primitive mantle. The platinum-group minerals (PGM) observed in the Sulfide Zone are predominantly Pt-Pd-bismuthtellurides, stanides and arsenides, mainly enclosed within sulfide minerals. In contrast, the PGM observed in low-S-high-Pt-Pd Zones are mainly Pt-arsenides, stannides and antimonides, mostly enclosed within alteration silicates. Differences in texture, geochemistry and PGM assemblage between these mineralization styles suggest that they originated from distinct geological processes. The Sulfide Zone was formed by a major event of segregation of an immiscible sulfide liquid, whereas the low-S-high-Pt-Pd Zones formed by a sulfide liquid saturation followed by sulfur loss during post-magmatic alteration. The identification of PGE-rich layers in rocks without sulfides or chromite at the Carajas Mineral Province is important as these may have been overlooked during previous exploration programs.