Platinum-group minerals and their host chromitites in Macedonian ophiolites

Platinum-group minerals (PGM) have been described in most ophiolitic chromitites around the world, but not in Macedonia, where the chromitites themselves and their host rocks remain poorly known. This paper reports, for the first time, PGM in three ophiolite occurrences in Macedonia, discusses their composition with reference to other PGM in similar contexts and the reason for the PGM/chromite association with reference to various hypotheses published in the recent literature. The compositions of the host chromitites and their environment are also presented and discussed with reference to various models.

The geology of Macedonia comprises several tectonic units, which extend into the northern and southern Balkans. The western part of Macedonia belongs to the Dinarides–Hellenides belt; the central part includes two units: the Pelagonides and the Vardar Zone (Bozovic et al., 2013). The Pelagonides are a moderate-to-high-grade metamorphic terrane of Precambrian rocks representing a thrust sheet from the distal part of the Adriatic passive margin. The Vardar Zone comprises the Western Vardar Ophiolitic Unit and the Eastern Vardar Ophiolitic Unit (Schmid et al., 2008). The highly dismembered ophiolites in the Vardar Zone are related to a back-arc basin (Pamic et al., 2002) formed by spreading during the initial stages of intra-oceanic subduction (Robertson, 2002). Among the ophiolite occurrences in Macedonia, three poorly documented complexes were studied and sampled (Fig. 1).

This massif has been the most important source of chromite of the history of the former Yugoslavia with around 1.5 Mt of ore mined between 1890 and 1945 from open pits and underground workings. The ultramafic massif is located on the western flank of the Eastern Vardar Unit (Fig. 1), emplaced onto Triassic schists and limestones and overlain by Cretaceous sediments (Fig. S1, after Jankovic, 1986, freely available online as Supplementary Material linked to this article on the GSW website of the journal, http://eurjmin.geoscienceworld.org/). The estimated thickness of the massif is 2.5 km; it is about 17 km long and 4 km wide. A summary of the geology of the complex is given by Jankovic (1986).

The massif corresponds to the mantle sequence of ophiolite stratigraphy, consisting of harzburgite and minor dunite, and dykes of pyroxenite and gabbro that are especially abundant in the northeastern part of the complex close to the basal cumulate series. Three hundred chromitite occurrences are scattered within the massif; of these, eleven were of major importance. The chromite texture is typical of ophiolite complex, e.g. nodular, orbicular, net texture, schlieren or massive ore.

Chromite occurrences seem to be irregularly distributed within the harzburgite sequence. This mantle sequence exhibits a “basal zone”, dominated by dunite, where chromite occurs as massive ore in pipe-shaped and podiform bodies. The largest and richest deposits (0.1 to 1.2 Mt) are located in this basal zone. A “middle zone” is dominated by harzburgite with minor dunite bodies and rare chromite occurrences; an “upper zone” completes the mantle series. The latter is composed of dunite with lenses of harzburgite and dykes of pyroxenite and gabbro with chromite bodies taking the form of lenses sub-parallel to the foliation of the host rock. A “mixed series” terminates the succession and has been interpreted as the basal cumulate series with dunite, pyroxenite and gabbros.

Fourteen samples of the Radusa complex were collected, mainly around the Vitina Padina and Caska ore bodies (Fig. S1a). The former ore body has been mined underground (100,000 t of ore excavated,) and three chromitite samples were collected from the main adit plus three of ultramafic rocks. For the latter, chromite veins and lenses (showing vertical layering, with average thickness 80 cm) were sampled in the open pit (seven chromitites and one peridotite).

The Lojane ophiolite complex is located in the Eastern Vardar zone. It is 11 km long and 1 to 4 km wide and consists of three units: (1) a basal mantle sequence with harzburgite and minor dunite and “fragmented bodies” and veins of pyroxenite, overlain by (2) the cumulate sequence with basal dunite, followed by lherzolite, pyroxenite and gabbro; the uppermost volcano-sedimentary sequence (3) is thought to belong to the complex (Fig. S1b, after Jankovic, 1986, available online).

The distinctive feature of this complex is that it has been intruded by numerous small tertiary granite bodies and dacitic andesite. An Sb–As sulfide mineralization located in silicified serpentinites and locally in chromitite is associated with the dacitic andesite.

Chromite mineralization occurs (1) as small bodies (irregular pods or pencil-like occurrences) in the deepest part of the harzburgite series and (2) at the transition zone between the mantle and the cumulate series; it takes the form of lenses and layers sub-parallel to the foliation. Ten major chromite deposits have been recognized, and the total chromite production from the Lojane complex is estimated at 200,000 t. Seventeen samples from five chromite occurrences (fourteen chromitites and three host rocks) were collected at the entrances to abandoned adits.

It should be noted that occurrences of “micron size grains of platinoids” are mentioned in chromitite from Lojane (Grafenauer, 1975, in Jankovic, 1986). This seems to be the first description of PGM in ophiolitic chromitite. This information seems to have passed unnoticed: it was only five years later that the first detailed description of PGM in ophiolitic chromitite was published internationally (Constantinides et al., 1980).

The Rabrovo massif corresponds to the northern extension of the Chalkidiki ophiolite belt. It is a small peridotite massif (about 0.25 km², Fig. S1c, after Jankovic, 1986, available online). The body is strongly serpentinized and tectonically deformed, partly by a granite intrusion. It is mainly composed of wehrlite, with minor dunite as lenses and layers, and rare pyroxenite and gabbro.

Chromitite bodies occur in two main zones, in dunite and in wehrlite. Some small lenses are interpreted as a tectonically dismembered primary layer originally 150 m long and up to 4 m wide. The total Cr ore production is estimated at 150,000 t. The description of the Rabrovo ophiolite (in Jankovic, 1986) suggests that in contrast with the two others described above, the mantle series is lacking, the whole massif corresponding to the basal cumulate sequence. Owing to difficult outcrop conditions, three ultramafic rocks were collected around the old working and two chromitite samples were collected from dumps at the abandoned mines.

Polished thin sections and polished sections were studied using a Leica DM4500P microscope in transmitted and reflected light. Chromite and silicates from selected chromitites and ultramafic rocks were analyzed using a CAMECA SXFive electron probe microanalyzer (EPMA) equipped with five wavelength-dispersive spectrometers under routine conditions. For each mineral, several analytical points were examined, and no zoning or significant variations were observed. Accordingly, average values are plotted on the compositional diagrams. All values given below are above the detection limit.

Chromitites were systematically examined under the microscope in reflected light, and all grains with high reflectivity were checked with an energy-dispersive spectrometer (EDS)-equipped HIROX SH-3000 scanning electron microscope (SEM). Qualitative analyses were thus obtained, enabling the PGM to be identified, and SEM images of each PGM were made, in some cases revealing complex PGM assemblages. The PGM were then analyzed using the SXFive EPMA under the following conditions: acceleration voltage 20 kV, a 20-nA beam current and 10-s counting time. The standards were pure metals, and FeS₂ for Fe and S, Cr₂O₃ for Cr, AsGa for As. The Kα lines were used for Ni, Cu, Co, Cr, S, Kβ lines for Fe, Lα lines for Ir, Ru, Rh, Pt and Lβ lines for Pd, As and Os. Interferences due to peak overlap were resolved by using an overlap correction process (CamecaPeakSight software). The calculated detection limit (in wt%) varies from one analysis to another, but is around 0.5% for Pt, 0.3% for Pd, 0.1% for Rh, 0.1% for Ru, 0.4% for Ir and 1.3% for Os. In most cases, the very small size of the grains analyzed explains the low totals obtained for most analyses. Chromium and Fe appear in all the compositions; these elements reflect excitation of the matrix. Hence, Cr and a proportion of Fe determined from knowledge of the Cr/Fe ratio of the host chromite were subtracted from the raw analytical data, and the composition was recalculated to 100%. Because of the good approximation to stoichiometry, this procedure is considered acceptable.

The dominant facies described in the literature on the Radusa and Lojane ophiolite bodies are harzburgites with minor dunite. Most of the peridotites collected appear strongly serpentinized. Pyroxenites (websterite), displaying a generally porphyroclastic texture, were also collected. They appear much less altered. Accordingly, from Radusa, we collected three websterites and a serpentinized dunite, from Lojane two clinopyroxenites and a serpentinized harzburgite and from Rabrovo a serpentinized dunite and a websterite.

Chromitite generally occurs in discontinuous layers, the chromite proportion ranging from 20 to 80%. The silicate matrix is composed of serpentine (with in rare cases relics of olivine). Some samples show local deformation occurring as centimeter-sized bands of mylonite, and frequently large undeformed, subhedral chromite crystals occur in a “matrix” of cataclased fine-grained chromite crystals (Fig. 2c). Commonly, some of the chromite crystals contain silicate inclusions (olivine, pyroxene, or amphibole, which are either isolated or form “clusters”, Fig. 2d).

For Radusa, we analyzed chromitites from the two orebodies sampled, and for Lojane, four out of the five occurrences sampled. For Rabrovo, one sample was analyzed. In terms of characteristics and texture we did not find any difference between the three massifs.

Olivine has been found (1) in serpentinized harzburgite; its forsterite (Fo) content ranges between 90.63 and 92.80 mol% and NiO between 0.24 and 0.45 wt%; (2) in the chromitite matrix (Fo 93.13–93.79, NiO 0.39–0.55 wt %); and (3) as inclusions in chromite crystals from chromitite (Fo 96.92–97.00%, NiO 0.55–0.86 wt%).

Figure 8g and h shows a laurite grain with a Cu-bearing aureole. The laurite in Fig. 8g does not exhibit significant Cu values, whereas the one in Fig. 8h bears (after correction) 7.8 wt% Cu. The chemical analysis is compatible with a structural formula assuming the substitution of Cu in laurite (Ru_0.66Cu_0.21Os_0.09Ir_0.04)S₂ (Table 3). Garuti et al. (1999) described from the Othrys ophiolite complex a composite inclusion of cuproiridsite, laurite and an unspecified Cu sulfide “that appears to have crystallized later than the PGM”.

Only one PGE alloy was found; it occurs as a monophase inclusion in sample 59-1 (Radusa, Table 3). Its recalculated composition indicates (avg. of two analyses in wt%) Ru 63.8, Os 19.7, Ir 12.2 and Pt 0.6, Rh 1.2, Co 1.1 and Cu 0.14. This composition is within the range of ruthenium compositions given by Cabri (2002). Note that the Ru/Os/Ir atomic proportions correspond to those of the coexisting laurite (Fig. 10).

Figure 8e shows a two-phase grain composed of laurite (Table 3, grain 44-1-b, point 32) and a complex Ir, Rh, Ru, Cu, Fe sulfide. This composition (Table 3, point 27) with 55.20 wt% Ir, 9.99 wt% Rh and 6.21 wt% Cu, 3.39 wt% Fe and 25.11 wt% S resembles the cuproiridsite composition given by Cabri (2002). Note that considering the small size of the grain, analyses have been recalculated by removing the laurite contribution (Ru + Os and the corresponding proportion of S) and the proportion of Fe due to the host chromite. Atomic proportions calculated on the basis of the recalculated composition are (Cu_0.516,Fe_0.320)_0.836(Ir_1.514,Rh_0.512)_2.026S_4.134 which is not so far from the ideal formula of cuproiridsite. Indeed, this mineral is commonly reported in ophiolitic chromitites. A second cuproiridsite grain, associated with a laurite grain, was also detected in sample 59-2, both however being too small to be accurately analyzed.

A second mineral from the linnaeite group was discovered in sample 59-2, Radusa (Table 3, point 2); it appears adjacent to a very small laurite crystal. The Cr/Fe ratio of the composition obtained suggests that Fe is present in the mineral composition. In the recalculated composition, we obtain (in wt%) Cu 8.37, Fe 4.39, Rh 44.83, Ir 9.34, Pt 0.76, Pd 0.58 and S 31.66. The atomic proportion calculated on the basis of seven atoms per formula unit (apfu) is: (Cu_0.545Fe_0.325)_0.870(Rh_1.802Ir_0.201Pd_0.022,Pt_0.016)_2.041S_4.083, which corresponds to cuprorhodsite.

Several laurite grains are surrounded by a Cu-bearing rim (Fig. 8g and h) and are totally enclosed in chromite crystals. Thus, the development of the Cu rim must have occurred before the trapping of the assemblage in chromite. One rim (Fig. 8g) was large enough for EPMA analyses. After correction (removing minor amounts of Os and Ru, and the corresponding portion of S due to laurite, and part of the Fe due to chromite), we obtain the following concentrations: Cu 73.69, Fe 3.46, S 22.82 (in wt%) corresponding to the following atomic proportion: Cu_0.600Fe_0.030S_0.368 (Table 3, point 12). The possible minerals corresponding to this formula could be chalcocite Cu₂S or digenite Cu₉S₅. Garuti et al. (1999) mentioned the occurrence of a Cu sulfide associated with a composite cuproiridsite-laurite association (filling the space between the PGM grains) but did not specify its composition.

Since the initial discovery of PGM in ophiolite chromitite, many publications have been devoted to the description of PGM in ophiolites all over the word, with diverse ages of emplacement and geodynamic environments. In most cases, they are composed of Ru, Os, and Ir, and minerals of the laurite–erlichmanite series are the most common. González-Jiménez et al. (2009a) studied the distribution and mode of occurrence of PGM in chromitites of 23 ophiolite occurrences, exhibiting “podiform” or “stratiform” chromitite concentrations. They found that Os-, Ir-, and Ru-rich PGM are only slightly more abundant (52%) than Pt-, Pd-, and Rh-rich PGM (48%), but when only podiform chromitites are considered, the proportion of the former reaches 72%, and 93% if one considers only primary PGM included in chromite crystals.

The origin of the PGM in ophiolitic chromitite has been widely debated. Gijbels et al. (1974) and Naldrett & Cabri (1976) invoked the possibilities of incorporation of the PGE at high temperature, followed by exsolution on cooling. It is now accepted that the morphology of the PGM and their diversity is not compatible with the exsolution process. Constantinides et al. (1980) and Talkington et al. (1984) then suggested that the PGM were trapped during chromite growth.

Augé (1988) proposed that the PGM crystallized before, or contemporaneously with, chromite and then were trapped, PGM playing the role of a nucleus. The accumulation of chromite crystals to form pods or layers will play the role of a “mechanical” collector of the PGM formed in the magma, thereby explaining the relative enrichment of PGE in chromitite. This has been demonstrated for the Tiébaghi massif (New Caledonia ophiolite) where PGM have also been found as inclusions in disseminated chromite from coexisting dunite, “precursory” to podiform chromitite.

All the PGM discovered in the Macedonian chromitites are enclosed in chromite crystals, and among the 45 grains discovered, 44 grains are laurite (41 single grains and 3 composite grains) and 1 is a ruthenium grain. The composite grains are composed of a laurite and a cuproiridsite (2 cases) or a laurite and a cuprorhodsite (1 case). According to González-Jiménez et al. (2009a), cuproiridsite represents 1% of the PGM included in unaltered podiform chromitite, whereas cuprorhodsite is not mentioned, Rh taking the form of hollingworthite (PtAsS). Except for the absence of PGM in fractures and silicate matrix, the distribution of PGM in podiform ophiolite chromitite from Macedonia is in good agreement with the statistical distribution established by González-Jiménez et al. (2009a). The composition of the Macedonian laurite is in perfect agreement with literature data (Johan, 1986; Augé & Johan, 1988; Garuti et al., 1999, compilation in Kapsiotis et al., 2009; González-Jiménez et al., 2009a). Similarly, Ru alloys together with Ir and Os are common in ophiolitic chromitites as is cuproiridsite. Cuprorhodsite seems to be much less common. The presence of Cu minerals associated with the PGM has also been described (Augé, 1988;,Garuti et al., 1999). The Cu sulfides or Cu–Fe sulfides, together with Ni sulfides or Ni–Fe sulfides very commonly, occur as inclusions in chromite crystals, with or without attached PGM. The Cu–S aureole observed for the Radusa and Lojane samples suggests that Cu–S crystallization followed the Ru–Os–S crystallization. Copper was present in the PGE–S system as indicated by presence of Cu-bearing PGM and also the presence of significant amounts of Cu in solid solution in laurite (see Table 3, point 29).

Recent developments in the application of Re–Os isotopic analysis to PGM-bearing chromitites, including direct Os isotope measurement on individual PGM grains (Malitch et al., 2003; Malitch, 2004; Ahmed et al., 2006; Shi et al., 2007; Marchesi et al., 2011; González-Jiménez et al., 2012a and b, 2013), bring new insight to our understanding of the origin of PGM in ophiolite chromitite. In a recent paper, González-Jiménez et al. (2014a) confirmed that in situ analysis of the Os isotopes in individual grains of PGM in ophiolitic chromitites reveals very heterogeneous compositions, even at the scale of single grains. They suggested that the genesis of PGM with heterogeneous Os-isotope compositions probably reflects the crystallization of chromite by the mixing/mingling of multiple batches of melts that have sampled different mantle sources. This mixing mechanism may produce “mixed” Os-isotope compositions even within single PGM. They proposed three mechanisms to explain the origins of PGM in ophiolitic chromitites. The first suggests that not all PGM now hosted in the ophiolitic chromitites were necessarily formed by direct precipitation from melts. Residual phases such as laurite or Os–Ir alloys could be produced by the breakdown of pre-existing PGE-bearing base-metal sulfides in the mantle. The PGM already present in the mantle could also be incorporated as a solid or a solid/melt mush into the parental melts of chromitites. This hypothesis is still very speculative. The second suggests that the PGM were introduced into pre-existing chromite once they were formed. Chromitites may be transported deep into the mantle by subduction processes, and possibly incorporated into the convecting mantle. Again, this model needs to be constrained. They could be infiltrated and veined by metasomatic fluids/melts that would promote the precipitation of PGM. The third hypothesis involves the partial or complete destruction of the PGM, and their recrystallization, during polyphase metamorphism or recycling of the chromitites in the deeper mantle.

Whatever the mechanism involved, it remains true that the presence of PGM cannot be dissociated from the processes of chromitite formation, described as multiple injections of magma depositing chromite (of probably various origins, and containing – or not containing – PGM crystallized from various sources and levels in “mini” chambers to form pods and layers). Thus, the model developed some years ago by Lago et al. (1982) probably contains the basic mechanisms that have never since been questioned. This model is perfectly compatible with the various mechanisms suggested by González-Jiménez et al. (2014a and b) to explain the characteristics of the PGM and the associated chromitites and can easily be applied to the Macedonian examples. The contributions of “hydrous fluids” to the origin of chromite deposits, while often mentioned (Johan et al., 1982, 1983; Melcher et al., 1997), have never been fully accepted.

With a high Mg number (around 0.6), a high Cr/Cr + Al + Fe³⁺ ratio, and very low TiO₂ and Fe₂O₃ contents, the Macedonian chromitites are typical of podiform chromitite (Barnes & Roeder, 2001). Moreover, there are very few variations from one occurrence to another and from one complex to another. With a high Cr content, their composition is very similar to that of chromitite from Greek ophiolites (Pindos, Economou-Eliopoulos et al., 1999; Vourinos, Augé, 1988).

It is generally accepted that such a composition reflects crystallization from “boninitic melt in an island arc environment” (Akmaz et al., 2013), as opposed to a mid-oceanic ridge setting. With a specific enrichment in Os, Ir and Ru relative to Rh, Pt and Pd, it also corresponds to the Type I defined by González-Jiménez et al. (2014b).

The strong alteration undergone by harzburgite and dunite has totally obliterated their texture and chemistry. On the other hand, all the other facies sampled in the three complexes, pyroxenite (with in some cases minor olivine, totally serpentinized), websterite (more common) and clinopyroxenite appear well preserved. Two types of texture are represented: a typical adcumulate magmatic texture (Fig. 2b) and a porphyroclastic texture (Fig. 2a). In most cases, this latter texture is the result of mantle deformation. In this context, pyroxenite occurs either as dykes (which are rather common in the upper part of mantle series) or cumulates. Adcumulate textures are relatively undeformed and can be found either in dykes in the mantle series or in the cumulate sequence. The porphyroblastic texture has completely obliterated the magmatic texture of the pyroxenites, but their relatively high proportion within the complexes suggests that they correspond to basal cumulates that have been affected by mantle deformation. In spite of their deformation, the mineral composition shows wide variations (more pronounced at Radusa) compatible with differentiation trends marked by a decrease in the Mg number in both pyroxenes, accompanied by a decrease in Cr₂O₃ and an increase in Al₂O₃. Similarly, in the Radusa disseminated spinel, we see a clear increase in Fe³⁺, with a similar increase in TiO₂ at a low Mg number. At Lojane (only two samples), clinopyroxene seems less differentiated, and disseminated spinels have low TiO₂, low Fe³⁺ and a high Mg number.

Obviously, Radusa, and to a lesser extent Rabrovo and Lojane, are examples of differentiated pyroxenite cumulates that have been affected by mantle deformation. It remains clear, however, that all the chromitites occurrences sampled are typical podiform mantle chromitite and do not belong to the cumulate evolution of the complexes.

Platinum-group minerals have been found as inclusions in chromitite from three ophiolite bodies in Macedonia. Their composition is typical for this type of context, dominated by the laurite–erlichmanite solid solution, with rare Ru–Os–Ir alloy, cuprorhodsite and cuproiridsite. One of the characteristics of the Macedonian PGM is a relative Cu enrichment, with Cu-PGM, but also, the presence of Cu in solid solution in laurite, and the occurrence of Cu sulfides rimming the PGM trapped in chromite crystals, suggests that Cu was present in the PGE–S system. Similarly, the host podiform chromitite exhibits textures and compositions typical of ophiolitic environments. Samples collected in the mantle sequence of the Lojane and Radusa bodies are strongly altered, except for some pyroxenite dykes. Basal cumulate pyroxenite displays an unusual porphyroblastic texture typical of mantle deformation. At the scale of the ophiolite body, the pyroxenites exhibit a differentiation trend compatible with a fractional crystallization process, confirming that they are deformed cumulates.

We thank Guillaume Wille for the EPMA analyses obtained with the CamecaSXFive at the “Plateforme analytique régionale” (BRGM – University – CNRS, Orléans), and Prof. Uysal and an anonymous reviewer for their critical comments. J.V. Guy-Bray edited the English text. We dedicate this work to the memory of Zdenek Johan, who taught us to share his passion for PGE mineralization.