Geochemistry of komatiites and basalts in Archean greenstone belts of Russian Karelia with emphasis on platinum-group elements

Archean komatiites form an important host rock for Ni-Cu sulfide deposits in the world. In this study, we examined Archean komatiites and komatiitic basalts from four areas in the Vedlozero-Segozero greenstone belt and two areas in the Tikshozero greenstone belt, Russian Karelia, to constrain their Ni-Cu sulfide mineralization potential using chalcophile element geochemistry. The platinum-group element (PGE) concentrations in these Munro-type komatiitic rocks are at a level of other similar S-undersaturated komatiites and komatiitic basalts globally, with Pt and Pd concentrations in the range of 5–20 ppb and Pd/Ir varying from < 10 (komatiites) to > 15 (komatiitic basalts and basalts). Generally, the metals of the iridium-group elements (IPGEs; Ir, Ru, Os) show a compatible behavior, decreasing in abundance with decreasing MgO, whereas the metals of the palladium group (PPGEs; Pt, Pd, Rh) exhibit an incompatible behavior. The poor correlation between Ir and MgO suggests that olivine fractional crystallization is not the main control on the behavior of IPGE. In contrast, Ir, Ru, and Os show positive correlations with Cr, which is consistent with the compatible behavior of IPGEs in chromite or during the co-precipitation of chromite and platinum-group minerals (PGMs). Palladium, Cu, and Au have been variably mobile during alteration and metamorphism whereas Pt appears to have been less mobile. Some samples from the Khizovaara area show low (Pt/Ti)N ratios, low Ni, and high La/Sm and La/Nb, suggesting localized sulfide saturation in response to crustal contamination. However, the potential of the Russian Karelian greenstone belts for Ni-Cu sulfide mineralization is considered relatively low because most samples show neither enrichment nor depletion of chalcophile elements, and the paucity of dynamic lava channel environments, as indicated by the scarcity of olivine-rich adcumulates. In addition, there appears to be a lack of exposed sulfidic sedimentary rocks in the region.


Introduction
Komatiites are highly magnesian (> 18 wt% MgO) volcanic rocks occurring predominantly in Archean and Paleoproterozoic greenstone belts (e.g., Arndt et al. 2008). Many komatiite-bearing greenstone belts host Ni-Cu sulfide deposits (Mudd and Jowitt 2014), and in some cases, they are economically important, notably in the Eastern Goldfields superterrane of the Yilgarn craton and the Abitibi greenstone belt of the Superior craton (Barnes and Fiorentini 2012 and references therein). Komatiites of the Karelian craton in the Fennoscandian Shield have been studied in a considerable detail, both in Russia (e.g., Puchtel et al. 1998Puchtel et al. , 1999Puchtel et al. , 2007Svetov 2005;Svetov et al. 2001Svetov et al. , 2010Svetov and Smolkin 2003) and Finland (e.g., Jahn et al. 1980;Hanski 1980;Papunen et al. 2009;Hölttä et al. 2012;Maier et al. 2013;Konnunaho et al. 2015), but so far, no significant komatiite-hosted Ni-Cu-(PGE) deposits have been discovered, though some small, disseminated sulfide deposits are known to exist in northern and eastern Finland (Konnunaho et al. 2015). A possible exception of significant mineralization is the Paleoproterozoic Sakatti Cu-Ni-PGE deposit, which may be related to magmas of komatiitic affinity (Brownscombe et al. 2015;Makkonen et al. 2017).
Komatiites in eastern Finland, including those in the Tipasjärvi-Kuhmo-Suomussalmi greenstone complex (TKS) and the Ilomantsi and Tulppio greenstone belts, have been dated at ca. 2.75-2.82 Ga  and are thought to have been formed in various tectonic settings (Papunen et al. 2009;Hölttä et al. 2012;Maier et al. 2013). Maier et al. (2013) studied the PGE systematics of komatiitic rocks in the greenstone belts in eastern Finland. Based on their relatively evolved composition, the lack of dynamic lava channel environments, and the paucity of PGE-enriched samples in most lavas, it was concluded that, compared with Archean komatiite belts globally, they may have a relatively low Ni sulfide mineralization potential. In Russian Karelia, Puchtel et al. (1998Puchtel et al. ( , 1999Puchtel et al. ( , 2004 determined PGE concentrations of the 2.8 Ga Kostomuksha komatiites, and Puchtel et al. (2007) reported PGE data for 2.9 Ga komatiites in the Volotsk suite. The PGE levels of both these belts seem to be similar to those of the Finnish belts. In the present study, we have determined PGE contents of komatiites in two Archean greenstone belts in Russian Karelia, Vedlozero-Segozero, and Tikshozero, in order to evaluate their prospectivity for Ni-Cu sulfide ore deposits.

Geology of the Vedlozero-Segozero greenstone belt
The Mesoarchean Vedlozero-Segozero greenstone belt is located at the western margin of the Vodlozero terrane (Arestova et al. 2012) in the Karelia craton, Russia (Figs. 1 and 2). It extends for~300 km in a N-S direction from Lake Vedlozero to Lake Segozero and is~50-60 km wide (Fig. 2). The underlying rocks mostly belong to the tonalite-trondhjemitegranodiorite (TTG) series rocks and have ages of 3.13-3.15 Ga (Arestova et al. 2012(Arestova et al. , 2015. The greenstone belt is unconformably overlain by Paleoproterozoic Sumian, Sariolian, and Jatulian supracrustal rocks composed of polymictic and quartz conglomerates, quartzites, quartzitic sandstones, and basaltic and basaltic andesite lava flows (Melezhik et al. 2012). All the komatiitic rocks in the Vedlozero-Segozero greenstone belt have undergone seafloor alteration and regional metamorphism, with the latter varying from greenschist to epidote-amphibolite facies, with the pressure ranging from 2 to 4 kbar and the temperature remaining below 540°C (Svetov 2005). During these processes, almost all the primary minerals have been replaced by secondary assemblages of actinolitic hornblende, anthophyllite, tremolite, serpentine, chlorite, talc, carbonate, epidote, magnetite, and quartz, but the rocks often contain relics of primary magmatic textures (e.g., ophitic, cumulative, and spinifex).
In the Vedloz ero-Sego zero green stone belt, Mesoarchean volcano-sedimentary rocks occur in several areas, including Hautavaara, Koikary, Palaselga, Semch, and Sovdozero, likely representing fragments of an originally larger belt (Svetov et al. 2001;Svetov 2005) (Fig.  2). In plane view, these areas form sub-parallel zones separated by gneissose granites (Arestova et al. 2015). We focused our study on four areas (see Fig. 2), which are described in more detail below.

Hautavaara area
The Hautavaara area is located in the southwestern part of the Vedlozero-Segozero greenstone belt (Fig. 2). It extends in a N-S direction for approximately 30 km and has a maximum width of 10-12 km. Komatiitic to basaltic rocks occur as a sequence varying from 1600 to 1800 m in its reconstructed thickness and consisting of high-Mg volcanic lava flows, interbedded tuffs, tuffites, and volcano-sedimentary and sedimentary rocks (Fig. 3) (Svetov et al. 2001;Slabunov et al. 2006).
The lowermost member of the sequence with a thickness of 400 m consists of komatiite lava flows and basalts with interbeds of tuff sandstone, tuffite, chert, and graphitic shale (see Online Resource 1). The middle member reaches a thickness of 700 m and is composed of massive and pillowed basalts (Online Resource 1) with rare interlayers of tuffitic material. The upper member consists of 350-700 m of interbedded pillow lavas and massive basalts, with less than 10 m of tuff and greywacke. Detrital zircon grains from the footwall terrigenous graywackes have yielded an age of 2947 ± 13 Ma (SHRIMP-II; Svetov et al. 2006;Fig. 3), defining a maximum age for the komatiitic rocks. Zircon from interlayered felsic volcano-sedimentary rocks in the Hautavaara komatiitebasalt sequence has yielded an age of 2917.2 ± 8.7 Ma (Svetov et al. 2010;Fig. 3), further constraining the age of the volcanic-sedimentary sequence. The komatiites and basalts from this area are characterized by positive initial ε Nd values that vary between + 0.4 and + 2.4 (Svetov 2005) (Online Resource 2).

Koikary area
The Koikary area occurs in the central part of the Vedlozero-Segozero greenstone belt (Fig. 2). The oldest rocks are exposed in the core of the central anticline, occurring as a lava sequence composed of komatiites and komatiitic basalts, which are overlain by a felsic volcanic-sedimentary formation ( Fig. 3). The lava flows alternate with tuffs and tuffites of komatiitic and basaltic composition. The sequence is 500-550 m in thickness and is succeeded by an approximately 500-m-thick basaltic unit (Fig. 3). The Koikary volcanic series is dominated by lavas, with pyroclastic and volcano-sedimentary rocks making up approximately 7-10% of the total volume.
The Koikary komatiites consist of massive lava flows, pillow lavas (Online Resource 1), and occasionally brecciated lava flows containing spinifex-textured lenses. The komatiite-   basalt sequence is crosscut by a dacite dike with a U-Pb zircon age of 2935 ± 20 Ma (Bibikova and Krylov 1983) while the overlying felsic volcanic rocks have yielded a U-Pb zircon age of 2860 ± 15 Ma (Samsonov et al. 1996). Svetov (2005) reported initial ε Nd values from + 1.3 to + 1.8 for komatiites and basalts in this area (Online Resource 2).

Palaselga area
The Palaselgа area is located approximately 45 km northeast of Koikary and has a strike length of 14-16 km (NW 340°) and a width of 1.5-3.0 km (Fig. 2). Basement rocks (granitegneisses) with an age of 3141 ± 9.7 Ma are overlain by the Palaselgа mafic-ultramafic volcanic sequence, which is dominated by komatiites and komatiitic basalts (Arestova et al. 2012;Fig. 3). This sequence contains three distinct units (Fig. 3), with the 460-to 600-m-thick lowest member containing massive and pillowed lava flows ranging from 10 to 40 m in thickness. The overlying middle member has a thickness of 400-450 m and consists of a series of interbedded komatiite lava flows (Fig. 4c,d) and komatiitic tuff layers varying from 0.5 to 6 m in thickness. The volume of the pyroclastic rocks at this level does not exceed 4-5%. The upper member is composed of a~800-m-thick series of interbedded pillowed and massive basalts, among which there are separate komatiitic lava flows (Fig. 3). Mafic dikes found within basement tonalities in the Palaselga area are geochemical analogs of komatiitic basalts in the greenstone belt. Arestova et al.

Sovdozero area
The Sovdozero area is located in the northwestern part of the Vedlozero-Segozero greenstone belt (Fig. 2), extending for 20 km in a north-south direction. The Mesoarchean volcano-sedimentary rocks dip to the west at an angle of 65-75°. The komatiite-basaltic sequence is composed of a 150-mthick lower basaltic unit overlain by a 600-m-thick komatiitic unit, which is capped by 200 m of basalts in the upper part of the sequence (Fig. 3).
The lower basaltic sequence is formed by alternating flows of massive and pillowed basalts (Online Resource 1). Pyroclastic and volcano-sedimentary rocks constitute no more than 1-2% of the sequence. Sub-volcanic rocks are represented by sills and dikes of metagabbro-diabase and porphyritic diabase. The lower basaltic unit is cut by numerous dacite dikes and microcline granite veins (Fig. 3). The basaltic rocks are intensely foliated near their contact with the granites (Svetov 2005).
The komatiitic unit consists of a series of lava flows with intercalated beds of tuffaceous material. The volcanic rocks are represented by massive and brecciated komatiitic pillow lavas and lapilli tuffs. The lavas occur as massive, polygonally jointed, brecciated, and spinifex-textured varieties. Autobreccias are developed in the lower and upper parts of massive lava flows, and cumulate zones occur as lenses in autobrecciated komatiitic flows.
The upper basaltic unit exhibits well-preserved lava and tuff textures, and the amount of pyroclastic and volcanosedimentary material increases to 25-30%. The sedimentary

Geology of the Tikshozero greenstone belt
The Tikshozero greenstone belt extends for around 300 km along the boundary between the Karelian granite-greenstone terrain and the Belomorian mobile belt (Fig. 1). It comprises several sub-belts, which are separated from each other by granitoids. All these sub-belts show broadly similar lithologies and structural styles, have undergone amphibolite facie regional metamorphism, with secondary mineral assemblage of serpentine, amphibole, talc, carbonate, and chlorite, and have been interpreted as fragments of an originally single Neoarchean greenstone belt (Kozhevnikov 1992(Kozhevnikov , 2000.

Irinozero area
The Irinozero area is located approximately 100 km NW of the Khizovaara area (Figs. 1 and 4b). The lava sequence is overthrust onto a sequence of arc-derived felsic volcanic rocks and siliciclastic turbidites (Shchipansky et al. 2004). The bestpreserved remnant of the sequence occurs along the northern shore of Lake Irinozero (Fig. 4b). It begins with a gabbroic unit with a strong schistosity, which is overlain by a mafic dike unit. A volcanic unit composed of pillowed and massive lava overlies the mafic dike unit. The gabbros, mafic dikes, and

Sampling and analytical methods
In order to determine platinum-group element concentrations in mafic and ultramafic volcanic rocks from the studied greenstone belts, we collected 38 samples from the Koikary, Palaselga, Sovdozero, and Hautavaara areas in the Vedlozero-Segozero greenstone belt (Fig. 3)  Additional trace elements were determined by ICP-MS at Cardiff University and IG KRC RAS. For ICP-MS trace element analyses, samples were dissolved in a mixture of hydrofluoric and perchloric acids in a Teflon vessel, evaporated, and re-dissolved in nitric acid. To ensure total decomposition of the sample, the solution was filtered and the filter was ashed and fused with 0.2 g of lithium metaborate and 0.02 g sodium metaborate, followed by dissolution of the fused bead in nitric acid and in combination with the filtrate. This method digests most refractory minerals and provides acceptable precision for all elements. The error is less than 2% for elements at concentrations above 0.5 wt%, 3% at concentrations above 30 ppm, and 5% at concentrations below 30 ppm. Three reference materials, JB1a, NIM-N, and NIM-P, were used to monitor the reproducibility of both major and trace elements (McDonald and Viljoen 2006). Platinum-group elements and Au were determined by ICP-MS after Ni-sulfide fire assay and tellurium co-precipitation in the analytical laboratory of the University of Quebec at Chicoutimi (Laboratoire d'Analyses Géochimiques de l'UQAC) (for analytical details, see Savard et al. 2010). The weight of the analyzed samples varied from 3 to 6 g. For Pd and Au analyses, the precision was considerably lower than that for Ir, Ru, Rh, and Pt due to relatively high Pd and Au contents in the blank (0.17 ± 0.26 ppb for Pd, 0.4 ± 0.3 ppb for Au). Two reference material standards, OKUM (komatiite) and WMS-1a (Wellgreen massive sulfide), were used to monitor the reproducibility (Savard et al. 2010) (Table 1).

Major and trace elements
Average major and trace element compositions of komatiites and komatiite basalts of different areas from the Vedlozero-Segozero and Tikshozero greenstone belts are listed in Table 2 (original full data are available in Online Resource 3). For the purpose of plotting the data in diagrams, all major element compositions were recalculated to 100% on a volatile-free basis. Our samples have normalized MgO contents between 7.0 and 37.1 wt%, representing basalts, komatiitic basalts, and komatiites. The komatiitic samples from different areas show average Al 2 O 3 /TiO 2 ratios of 21.1 (Irinozero), 21.1 (Khizovaara), 18.7 (Koikary), 21.0 (Hautavaara), 21.3 (Sovdozero), and 26.3 (Palaselga). In Fig. 5, most of them plot in the field of Al-undepleted (or Munro-type) komatiites, the most common variety of Mesoarchean and Neoarchean komatiites globally (Nesbitt et al. 1979).
The samples from both greenstone belts have a negative correlation between MgO and SiO 2 , TiO 2 , Al 2 O 3 , and CaO ( Fig. 9a-d), forming continuous trends from komatiites to komatiitic basalts and basalts. This suggests that these rocks form a differentiation series related to olivine accumulation and fractionation. Samples from both greenstone belts also show a well-defined positive correlation on the Ni vs. MgO diagram (Fig. 6e), consistent with olivine-controlled trends.
Notably, a few samples from the Khizovaara area and one sample from the Irinozero area have slightly higher Ni contents plotting above the main trend (Fig. 6e).
In sulfide-undersaturated silicate melts, Cu abundances should increase during fractional crystallization of olivine and other silicate minerals, as Cu is incompatible in these minerals. However, the Cu contents of our samples show a significant scatter when plotted against MgO (Fig. 6f). Most of the samples have a Cu content of less than 200 ppm, with three exceptions that have higher Cu contents between 284 and 491 ppm. The scatter is likely due to secondary mobility caused by alteration and/or metasomatism during regional metamorphism.
On the Cr vs. MgO plot (Fig. 6g), our komatiitic samples with MgO greater than 25 wt% display two different trends: a negative correlation indicating mixing of olivine with chromite-undersaturated komatiitic liquid and a positive linear trend reflecting co-accumulation of olivine and chromite.
The Zr contents in the samples from the Segozero-Veldozero and Tikshozero greenstone belts plot at the lower end of a global database (Fig. 6h), indicating a relatively lower degree of crustal contamination compared with other greenstone belts globally. Chondrite-normalized rare earth element (REE) patterns and primitive mantle-normalized trace element patterns of the komatiites and komatiitic basalts from the Vedlozero-Segozero and Tikshozero greenstone belts are shown in Fig.  7. In these diagrams, the samples from the studied areas are divided into two groups, with one representing komatiites (MgO > 18 wt%) and the other komatiitic basalts (MgO 10-18 wt%) and basalts (MgO 5-10 wt%). Most samples from the Vedlozero-Segozero greenstone belt have flat or slightly fractionated patterns with enriched LREE contents and weak negative Nb-Ta anomalies (Fig. 8).
Samples from the Khizovaara area can be divided into two sub-groups. One group is slightly depleted in LREE (Fig. 7), while the other shows a relative LREE enrichment and significant negative Nb-Ta anomalies (Fig. 7). Samples from the Irinozero area show similar REE patterns to those of the LREE-depleted groups of the Khizovaara area, with most samples having weak Nb-Ta anomalies (Fig. 7).
In Fig. 8, primitive mantle-normalized (Nb/La) N ratio is plotted against (Nb/Th) N ratio. Samples from both the Vedlozero-Segozero and Tikshozero greenstone belts display positive correlations, with most samples showing lower (Nb/ La) N and (Nb/Th) N ratios than primitive mantle (PM, McDonough and Sun 1995), towards average upper continental crust composition (Rudnick and Gao 2003).

Platinum-group elements
The concentrations of PGEs and Au of the analyzed komatiite and komatiite basalt samples are presented in Table 2. The PGE contents of most of the komatiite samples are in the global range of Al-undepleted komatiites, with up to 15.1 ppb Pt, 20.9 ppb Pd, 8.99 ppb Ir, 6.35 ppb Ru, 6.71 ppb Os, and 27.7 ppb Au. Three samples from the Vedlozero-Segozero belt show a higher Pd concentration than the others, but their Pt contents are in the range of the other samples.
Iridium and Ru show generally positive correlations with MgO with a few exceptions (Fig. 9). Platinum, Pd, and Au contents generally increase with falling MgO contents, but there is a sharp decrease when the MgO content falls below 10 wt%. Rhodium shows a slight increase with decreasing MgO. Due to the different compatibility of IPGE (iridium group, Ir, Ru, Os) and PPGE (palladium group, Pt, Pd, Rh), the Pd/Ir ratio broadly increases with decreasing MgO. On the primitive mantle-normalized chalcophile element diagrams, basalts and komatiitic basalts display more fractionated trends than komatiites and olivine cumulates (Fig. 10).
On the Pd vs. Pt diagram, most samples plot on a linear trend, excluding three samples from the Sovdozero area and one sample from the Khizovaara area, which plot above this trend (Fig. 11a). On the Rh vs. Pd plot, these four samples also stand out from the main linear trend (Fig. 11b). However, all the samples form a linear trend on the Rh vs. Pt diagram (Fig. 11c). This is evidence for Pd having been slightly mobile during post-magmatic processes.

Crystal fractionation
Platinum-group elements have high partition coefficients with regard to sulfide and silicate melts (e.g., Stone et al. 1990;Fleet et al. 1991Fleet et al. , 1999 and have therefore been widely used to evaluate whether or not mafic-ultramafic magmas have equilibrated with sulfides (e.g., Maier et al. 2003;Barnes and Fiorentini 2008;Fiorentini et al. 2010). However, the behavior of IPGE during magma fractionation is not fully understood. It has been proposed in many past studies that IPGEs are compatible in chromite (Puchtel and Humayun 2001;Stone and Crocket 2003;Brenan et al. 2003). In contrast, based on whole-rock geochemistry, Barnes and Fiorentini (2008) found no correlation between Cr and Ir and suggest that the IPGEs are not compatible in chromite but are hosted by alloys. Righter et al. (2004) obtained high partition coefficients of IPGE with regard to chromite, with the values varying from 5 to 22,000, which are higher than the partition coefficients of~20 determined experimentally (Capobianco and Drake 1990). These experimental data suggest that chromite may be an important phase controlling IPGE. In our data set, most of the samples show positive correlations between Ir, Ru, and Cr (Fig. 12), which could potentially be explained by the compatibility of IPGE in chromite. This is consistent with in situ LA-ICP-MS analyses that indicate, in komatiites, that the IPGE may be partly incorporated into the chromite lattice (Pagé et al. 2012;Locmelis et al. 2013Locmelis et al. , 2018   Std. Avg.

Std.
Avg.      Park et al. 2017), though IPGE is rather low in chromites of basalt (Ballhaus and Sylvester 2000;Godel et al. 2007;Godel and Barnes 2008). A few of our samples with moderate Cr contents but rather high Ir contents (up to 8 ppb) plot away from the main trend, indicating that IPGE may not only be controlled by chromite but also by IPGE-rich platinum-group minerals or alloys (e.g., laurite, Fig. 12; Barnes and Fiorentini 2008;Maier et al. 2015). The elevated Ir level could not be explained by the presence of sulfide, because these samples do not show unusual high Pt and Pd contents (about 2 ppb). The komatiites in eastern Finland show similar patterns, with most samples plotting on a positive correlation trend between IGE and Cr and a few samples plotting above the trend (Maier et al. 2013). Some authors have proposed that Ir is hosted in olivine (Brenan et al. 2005). However, based on empirical data, Puchtel et al. (2004) evaluated that the olivine/melt partition coefficients for IPGE in komatiitic systems fall in the range of 0.2 to 1.0, indicating a moderately incompatible nature of IPGE. On the Ir vs. MgO diagram (Fig. 9), our samples show a weak positive correlation. Considering that MgO tends to be mobile during alteration and metamorphism, the data do not provide clear indications for or against the olivine control for Ir. A similar pattern has also been observed in komatiites from the Yilgarn craton (Barnes and Fiorentini 2008) and Finnish greenstone belts (Maier et al. 2013).

Interaction between magma and crustal rocks
Komatiites host some of the world's most important Ni sulfide deposits, notably in the Yilgarn craton (Mudd and Jowitt 2014). Smaller Ni deposits also occur in komatiites of the Zimbabwe craton and the Abitibi greenstone belt in Canada ( Barnes and Fiorentini 2012 and references therein). Nickel sulfide mineralization has been found in different types of komatiites (Al-depleted and Al-undepleted komatiites) and in different tectonic settings (e.g., Arndt et al. 2008). Crustal contamination is an important factor in the formation of komatiite-hosted sulfide deposits, by increasing the S content of the magma through addition of external S and by decreasing the sulfur solubility in magma through addition of Si and other crustal components (Lightfoot and Hawkesworth 1997;Lesher and Keays 2002;Arndt 2008;Lesher and Barnes 2009;Fiorentini 2008, 2012;Barnes et al. 2011). In the Vedlozero-Segozero greenstone belt, most samples show flat LREE or slightly LREE-enriched patterns with weak negative Nb-Ta anomalies on primitive mantle normalized multi-element plots (Fig. 7). In the Tikshozero greenstone belt, a group of samples from the Khizovaara area shows a LREE-enriched chemical signature, which is accompanied by negative Nb-Ta anomalies (Fig. 7). These features could be explained by crustal contamination. This is supported by the observed positive correlation between (Nb/Th) N and (Nb/La) N and the relatively low values of these two ratios in most samples compared with the primitive mantle value, as continental crust has low (Nb/Th) N and (Nb/La) N ratios (Puchtel et al. 1998).
As komatiites are normally generally sulfur-undersaturated during eruption, contamination with S-bearing crustal rocks is considered an important, if not essential factor in the formation of komatiitic Ni-sulfide deposits (e.g., Keays and Lightfoot 2010;Barnes and Fiorentini 2012). In the Agnew-Wiluna greenstone belt of Western Australia, Ni-Cu sulfide-mineralized komatiites are spatially associated with abundant volcanic massive sulfide (VMS) deposits, which have potentially provided external sulfur to the komatiitic magma, thereby triggering sulfide melt saturation and the formation of Ni sulfide mineralization . The Alexo, Texmont, and Hart sulfide deposits in the Abitibi greenstone belt have also been regarded as results of sulfide immiscibility triggered by assimilation of sulfidic sedimentary rocks (Lahaye et al. 2001). In the Karelia craton, Konnunaho et al. (2013) presented multiple sulfur isotope evidence that the komatiites related to the Vaara Ni deposit in the Suomussalmi greenstone belt have assimilated external sulfur. Archean sulfide-bearing BIF have been reported to be present in the Ilomantsi and Kostomuksha greenstone belts (Sorjonen-Ward et al. and Luukkonen 2005;Kuleshevich and Gor'kovets 2008), which could potentially have provided external sulfur to the komatiites to form sulfide deposits. In the Sovdozero area of the Vedlozero-Segozero greenstone belt, BIF occurs above the komatiite rocks in the upper part of the volcanic sedimentary sequence and was thus formed later than the major komatiitic magmatism. This is consistent with the slightly older age of the komatiites in the Vedlozero-Segozero belt compared with that of the Ilomantsi and Kostomuksha belt komatiites. There are no reports of the occurrence of BIF in the

Metal depletion or enrichment related to sulfide segregation
Traditionally, chalcophile element depletion in maficultramafic lavas has been used in the evaluation of the Ni prospectivity of greenstone belts (Lesher and Keays 2002;Lesher and Barnes 2009;Barnes and Fiorentini 2012), as it is a useful indicator of sulfide saturation. Fiorentini et al. (2010) presented a systematic comparison of the PGE abundances in sulfide-free samples from well-mineralized and barren komatiites and found that the well-mineralized komatiites show a large scatter in the PGE abundances with both enrichment and depletion, whereas the barren komatiites show neither PGE enrichment nor depletion. On the Pd vs. MgO diagram (Fig. 9), the samples of this study show a considerable scatter, with three samples from the Vedlozero-Segozero belt plotting above the rest of the data. These samples have high MgO contents (27-31 wt%) and likely represent komatiitic cumulates. It is unlikely that the  Hence, the high Pd concentration in these samples probably results from the mobility of Pd during alteration and/or metasomatism during metamorphism. This is consistent with the relatively poor correlation between Pt and Pd (Fig. 11), whereas Pt and Rh show a good correlation (Fig. 9). However, most of the samples follow a negative trend on the Pt and Pd vs.
MgO diagrams with only a few komatiites and basalts being depleted in Pt and Pd (Fig. 9). In Fig. 9, komatiite samples

MgO (wt%) MgO (wt%)
MgO (wt%) Fig. 9 Plots of platinum-group elements vs. MgO for komatiites in the Vedlozero-Segozero and Tikshozero greenstone belts. For comparison, analytical data from a global komatiite database are also shown, with the outline representing the field of komatiites from unmineralized greenstone terrane (Barnes and Fiorentini 2012) from both Russian Karelia and Finland plot within the global field of barren komatiites, being clearly different from wellmineralized komatiites. Consequently, these data indicate that the komatiites in Russian Karelia did not undergo extensive sulfide liquid saturation, which is not a good indication for the presence of Ni sulfide mineralization.  Fig. 11 Pt, Pd, and Rh contents in komatiites in the Vedlozero-Segozero and Tikshozero greenstone belts compared with analytical data from a global komatiite database (Barnes and Fiorentini 2012) Volcanology Several authors have emphasized the role of the eruption environment in the formation of Ni-Cu sulfide mineralization (e.g., Lesher and Keays 2002;Fiorentini et al. 2010Fiorentini et al. , 2012Fiorentini 2008, 2012). Two main types of komatiite-related Ni-Cu sulfide mineralization can be distinguished: (1) massive sulfide at the base of olivine-rich mesocumulate and orthocumulate zones and (2) disseminated sulfide in olivine-rich adcumulates. In both types, the Ni-Cu sulfide mineralization occurs in close association with high-Mg olivine-rich cumulate zones. In komatiitic lava flows, olivine-rich cumulates are interpreted to represent a channelized lava conduit, where relatively unevolved, hot, and turbulently flowing lava is particularly effective in eroding the substrate, assimilating external sulfur, and precipitating sulfides in flow dynamic traps, thereby facilitating ore formation (e.g., Arndt et al. 2008;Barnes and Fiorentini 2012). In addition, komatiite sequences with the thickest cumulate package are favored for the formation of Ni sulfide ores as they may reflect a particularly high magma influx, probably resulting from the presence of craton-scale, deep lithospheric structures (Barnes and Fiorentini 2012). The komatiite samples from the Vedlozero-Segozero and Tikshozero greenstone belts have similar MgO contents to those of the komatiites in eastern Finland, but they seem to be less magnesian than most mineralized komatiite fields elsewhere (e.g., Abitibi and Kalgoorlie) (Maier et al. 2013). The lower MgO content could result from a lower degree of mantle melting or a relatively advanced fractionation. Given the good overlap in the MgO contents between the komatiites in the Vedlozero-Segozero and Tikshozero greenstone belts, and those in eastern Finland, this interpretation may also apply to the komatiites in the current study, though olivine compositional data are not available due to the high degree of alteration.
In the Vedlozero-Segozero and Tikshozero greenstone belts, another potential explanation for the paucity of Ni-Cu mineralization is the lack of olivine-rich cumulate units and the dominance of sheet flows, including spinifex-textured komatiites, breccia-textured komatiites, and pillow basalts. Only small serpentinite bodies that may represent lava channels occur in the Sovdozero, Koikary, and Palaselga areas. This is a significant difference compared with the komatiites in the Kalgoorlie area, which contain abundant thick (up to km scale) olivine-rich adcumulates (Fiorentini et al. 2004;Barnes and Fiorentini 2012). The scarcity of olivine-rich cumulates suggests a low magma flux rate, representing relatively unfavorable conditions to form large and extensive channelized lava tubes, which can efficiently erode their substrate and form Ni-Cu sulfide mineralization. The age of the Vedlozero-Segozero greenstone belt is estimated to be ca. 2.92 Ga (Svetov et al. 2001), i.e., similar to the ages of 2916 ± 117 and 2892 ± 130 Ma obtained for the komatiites in the Sumozero-Kenozero greenstone belt using the Sm-Nd and Pb-Pb methods, respectively (Puchtel et al. 1999). On the other hand, these ages are slightly older than the age of ca. 2.8 Ga determined for the Kostomuksha greenstone belt (Puchtel et al. 1998(Puchtel et al. , 2005, the age of 2.82 Ga reported for the TKS greenstone belts in eastern Finland , and the age of 2.8 Ga for the Tikshozero greenstone belt. These multiple generations of komatiitic magmatism in the Fennoscandian Shield are clearly older than the global Neoarchean peak in mantle plume magmatism and Ni-Cu sulfide mineralization at around 2.7 Ga (Arndt et al. 2008 and references therein). For reasons presently not well understood, it appears that the slightly older magma influx has less potential to form major Ni sulfide ore deposits (cf. Weihed et al. 2005;Hanski 2015). However, as the available database is relatively limited, it is difficult to assess the Ni sulfide prospectivity of these belts accurately. Future exploration work should focus on the identification of relatively prospective lava channel-facies rocks (i.e., olivine mesocumulate and adcumulates) in Archean greenstone belts.

Conclusions
Komatiites in the two studied greenstone belts in Russian Karelia show considerable similarities to those in the neighboring Archean terrane in Finland, though some of the Russian komatiites, namely those in the Vedlozero-Segozero greenstone belt, seem to be around 100 Ma older than the Finnish komatiites. All studied Russian lavas are relatively evolved (mostly < 25 wt% MgO), and their PGE contents are within the range of other unmineralized komatiites globally. The palladium-group PGE behave incompatibly, whereas the IPGEs show positive correlations with Cr, from which we infer that they are dominantly controlled by chromite and possibly IPGE-rich phases. Considerable scatter in the Pd abundances and depletion in Cu and Au suggest that these elements were mobile during metamorphism and alteration. The absence of sulfide-rich sedimentary rocks in the two studied greenstone belts, the lack of PGE-enriched or depleted samples, the relatively differentiated magma composition compared with sulfide-mineralized komatiites globally, and the paucity of unevolved olivine adcumulates suggest a relatively low prospectivity for Ni sulfide mineralization, though localized crustal contamination and sulfide saturation seem to have occurred. However, as the available database is relatively limited, the Ni sulfide prospectivity of the studied belts remains incompletely understood, and future exploration work should focus on the identification of more prospective lava channelfacies rocks.