Heterogeneous Os isotope compositions in the Kalatongke sulfide deposit, NW China: the role of crustal contamination

Re–Os isotope compositions of mantle-derived magmas are highly sensitive to crustal contamination because the crust and mantle have very different Os isotope compositions. Crustal contamination may trigger S saturation and thus the formation of magmatic Ni–Cu–(PGE) sulfide deposits. The ∼287-Ma Kalatongke norite intrusion of NW China are hosted in carboniferous tuffaceous rocks and contain both disseminated and massive sulfide mineralization. The Re–Os isotope compositions in the intrusion are highly variable. Norite and massive sulfide ores have γOs values ranging from +59 to +160 and a Re–Os isochron age of 239 ± 51 Ma, whereas disseminated sulfide ores have γOs values from +117 to +198 and a Re–Os isochron age of 349 ± 34 Ma. The variability of Os isotope compositions can be explained as the emplacement of two distinct magma pulses. Massive sulfide ores and barren norite in the intrusion formed from the same magma pulse, whereas the disseminated sulfide ores with more radiogenic Os isotopes formed from another magma pulse which underwent different degrees of crustal contamination. Re–Os isotopes may not be suitable for dating sulfide-bearing intrusions that underwent variable degrees of crustal contamination to form magmatic sulfide deposits.


Introduction
The precise determination of the formation age of magmatic sulfide mineralization is important for understanding the origin of mineral deposits and their host intrusions. The Re-Os isotopic system has been used to date magmatic Ni-Cu sulfide deposits because of different partition coefficients of Re and Os between silicate magmas and sulfide melts and between monosulfide solid solution (MSS) and sulfide melts (Brenan 2008). For example, Walker et al. (1994) obtained Re-Os ages of 245.7 ± 0.6 and 247.0 ± 3.8 Ma, respectively, for the Noril'sk I Talnakh and Kharaelakh intrusions in Siberia. These two ages are similar to zircon U-Pb ages of 248±4 (Campbell et al. 1992) and 251.2±0.3 Ma (Kamo et al. 1996). Sulfides of komatiitehosted magmatic sulfide deposits in Kambalda have a Re-Os age of 2,706±36 Ma, which is the same within error of zircon U-Pb ages of 2,703±4 and 2,709±6 Ma (Foster et al. 1996 and reference therein).
However, in other cases, the Re-Os isotope isochron ages of magmatic sulfide Ni-Cu deposits are different from zircon U-Pb ages of the host intrusions. Examples include the Voisey's Bay sulfide deposit in Canada, where the Re-Os ages (988-1,302 Ma; Lambert et al. 2000) are younger than the zircon U-Pb ages (1,332.7±1.0 Ma; Amelin et al. 1999).
The Jinchuan deposit in NW China has zircon and baddeleyite U-Pb ages of ∼830 Ma ), but Re-Os ages range from 833 to 1,117 Ma 2008).
Because mantle and crust have different Os isotopic compositions, Re-Os isotopic systematics of mantlederived magma are highly sensitive to crustal contamination (Walker et al. 1991;Shirey and Walker 1998). Many magmatic sulfide deposits have been interpreted to have involved crustal contamination (Naldrett 2004;Keays and Lightfoot 2010). The Os isotopic composition of sulfide-bearing mafic intrusions can be modified by contributions of crust with markedly different and potentially variable Re/Os ratios. In this paper, we present Re-Os isotope data for the Kalatongke Cu-Ni sulfide deposit in Xinjiang, NW China. Using this deposit as an example, we discuss the causes of the decoupling of Re-Os and zircon U-Pb ages and examine the implications of variable Os isotope compositions for the genesis of sulfide-bearing mafic intrusions.

Geological background
The Central Asian Orogenic Belt (CAOB), the largest Phanerozoic accretionary belt in the world, extends from the Urals in the west to the Pacific Ocean in the east and is bounded by the Siberia Craton to the north and the Tarim and North China Cratons to the south (Sengör et al. 1993) (Fig. 1a). Extensive Permian magmatism formed voluminous basalts and mafic-ultramafic intrusions in the Tarim Basin and surrounding regions of the CAOB. These rocks, which are thought to have been derived from a mantle plume, form the Tarim large igneous province (Zhou et al. 2004(Zhou et al. , 2009Pirajno et al. 2008).
There are numerous Permian mafic intrusions, including the Kalatongke intrusion, in the NW-SE trending Perkin-Ertai terrane of the Paleozoic Altay orogenic belt in the CAOB (Windley et al. 2002). They are distributed along the Irtysh River, where they are hosted in sandstones, argillaceous slates, volcanic breccias, and tuffs of the Carboniferous Nanmingshui Formation.
There are 11 segments of the Kalatongke intrusion. The Y1 segment, the largest exposed body, is 695 m long and up to 270 m wide. The Y2 and Y3 segments are unexposed and occur to the southeast of the Y1 segment. There are six weakly mineralized segments to the east of the Y1 intrusion (Fig. 1b). All of these intrusions have similar lithologies and consist from the base upward of biotite-hornblende gabbro, olivine norite and troctolite, hornblende norite and biotite diorite. Sulfide ores are only hosted in the olivine norite and troctolite. The Y1, Y2, and Y3 segments contain significant magmatic sulfide ores with 420,000 tons Cu and 250,000 tons Ni at grades of 1.3% Cu and 0.8% Ni (Liu et al. 2005), whereas the other six segments (Y4-9) are weakly mineralized. More than 50 vol.% of the Y1 segment contains sulfide ores, but only the basal parts (∼10 vol.%) of the Y2 and Y3 segments are mineralized (Fig. 1c).
There are disseminated and massive Ni-Cu and massive Cu-rich sulfide ores in Kalatongke (Fig. 2). Disseminated Ni-Cu sulfide ores are hosted in olivine norite and troctolite and occur in all three segments, whereas massive Ni-Cu sulfide ores occur only in the Y1 and Y2 segments and their adjacent country rocks. In the Y1 segment, massive Ni-Cu sulfide ores are in sharp contact with disseminated sulfide ores, whereas massive Cu-rich sulfide ores occur as veins in the norite in the lower part of the intrusion. Massive Cu-rich sulfide ores also occur as veinlets in the massive Ni-Cu sulfide ores.

Analytical methods
Sample preparation and inductively coupled plasma mass spectrometry (ICP-MS) measurements were performed in the National Research Center of Geoanalysis (NRCG), Chinese Academy of Geological Sciences, Beijing. Re and Os concentrations were determined by isotope dilution. 185 Re and 190 Os spikes from Oak Ridge National Laboratory, USA were calibrated at the NRCG. Detailed procedures for sample preparation are available in Du et al. (2001Du et al. ( , 2004 and Yang et al. (2008). Re and Os concentrations were measured with a Thermo X7 Quadrupole ICP-MS. Because of the low concentration of Os, signal intensities of the Os isotopes were measured with a Pt screen. The signal intensities of both Re and Os in this mode are ∼200,000 cps/ppb. 190 Os intensity was used to correct for the interference of 187 Os on 187 Re, and 185 Re intensity was used to correct for Re on Os. The total procedural blanks were about 5 pg for Re and 1 to 0.2 pg for Os based on blank runs analyzed together with the samples. 187 Os/ 188 Os ratios of the blank are 0.25 based on routine measurements. The standard material, GBW04436 (JDC, a molybdenite from the Jinduicheng deposit, China), was used to monitor the accuracy of the measurements. Re and Os contents and Re-Os ages of the JDC molybdenite determined during the course of this work are 17,340± 40 ppb, 25.25±0.07 ppb, and 138.9±0.5 Ma (n017), respectively, which coincide within uncertainties with the reference values (Re017,390±320 ppb, Os025.5±0.6 ppb, and t0139.6±3.8 Ma; Du et al. 2004).

Results
Two tuff samples from the country rocks of the Nanmingshui Formation were taken from drill core 7501 that penetrated the Y3 segment. They contain 1.07-2.00 ppb Re, and 0.015-0.027 ppb common Os, with 187 Re/ 188 Os ratios of 339-361 and 187 Os/ 188 Os ratios of 2.122-2.339. When calculated to an age of 287 Ma (Han et al. 2004), γ Os values for the two samples are +295 and +384.
Nine samples from the Kalatongke sulfide-bearing segment were analyzed for Re-Os isotopic compositions (Table 1). A barren norite sample from drill core 7501 has 1.89 ppb Re, and 0.181 ppb common Os with a Fig. 1 a Location of the study area in the CAOB (modified from Sengör et al. 1993). b Simplified geological map of the Kalatongke mine area, North Xinjiang, China (modified from Wang and Zhao 1991). c Cross section of the Y1, Y2, and Y3 segments of Kalatongke (modified from Wang and Zhao 1991 Fig. 3). Han et al. (2007) and Zhang et al. (2008) reported Os isotope data for disseminated and massive sulfide ores, respectively. Massive sulfide ores from Zhang et al. (2008) and this study have a limited variation of γ Os values (Fig. 3a). Disseminated sulfide ores from Han et al. (2007) and this study plot between the norite and country rocks (Fig. 3a) and form a trend with a positive correlation between γ Os values and 187 Re/ 188 Os ratios, different from the massive sulfide ores with relatively consistent Os isotope compositions.

Os isotope compositions in the Kalatongke intrusion
The heterogeneous Os isotope composition in Kalatongke yielded highly variable Re-Os isochron ages. The four massive sulfide samples yielded a poor Re-Os isochron age of 239±51 Ma (MSWD012), and four disseminated sulfide ores yielded a Re-Os age of 349 ± 34 Ma (MSWD05.0). Previous studies reported Re-Os isochron ages of 282.5±4.8 Ma (MSWD01.5) for the Y1 segment, 290.2±6.9 Ma (MSWD01.02) for the Y2 segment , and 305±15 Ma (MSWD02.1) (Han et al. 2007) for massive and disseminated sulfide ores, which are within errors of our Re-Os ages. When all data are plotted in a 187 Re/ 188 Os-187 Os/ 188 Os diagram (Fig. 3b), the disseminated sulfides plot roughly on the 305-Ma isochron line determined by Han et al. (2007) for disseminated ores, whereas massive sulfide ores plot along the isochron line for norites (∼287 Ma; Han et al. (2004)). The barren norite analyzed in our study also plots on this 287-Ma line (Fig. 3b). Thus, the disseminated sulfide ores are at least 20 Ma older than the massive sulfide ores. The difference between Re-Os ages emphasizes the variability of the Os isotope composition in the sulfide-bearing mafic intrusions of Kalatongke.
Heterogeneous Os isotope compositions of mafic intrusions are known in the literature, but their origin is still a matter of debate. Heterogeneous Os isotope compositions of MORB or OIB are commonly attributed to heterogeneous mantle sources (Walker et al. 1999;Gannoun et al. 2004). Heterogeneous Os isotope composition in the Sudbury complex, Canada was explained as having been derived from different sources (Walker et al. 1991) or mixing between mantle-derived magmas and crustal rocks (Cohen et al. 2000). Walker et al. (1994) attributed heterogeneous Os isotope compositions of the sulfide ore-bearing intrusions in the Noril'sk region, Russia to interaction between mantlederived magmas and various lithospheric components. Variable Os isotope composition was also thought to be produced by different diffusion rates of Os or later disturbances (Cohen et al. 2000;Lambert et al. 2000;Yang et al. 2008). However, Re-Os isotope variations of massive sulfide ores from Kalatongke cannot be explained by later disturbances or different diffusion rates because they have Os isotope compositions similar to the barren norites, but different from the disseminated sulfide ores.
Os can easily diffuse between sulfide minerals in massive sulfide ores but not between the silicate minerals and sulfide minerals . In Kalatongke, the difference of Os isotope compositions between Cu-rich massive sulfide ores and Ni-Cu massive sulfide ores cannot be caused by Os diffusion. On the other hand, if the Os composition was Re/ 188 Os ratios vs 187 Os/ 188 Os ratios for different sulfide ores and silicate rocks from the Kalatongke mafic intrusion. R is the ratio of the mass of silicate magma to the mass of sulfide and is calculated using the following formula: C i 0C 0 ×D×(R i +1)/(R i +D), where C i is the final concentration of metal i in the sulfide melt, C 0 is the metal concentration in the original silicate melt, D is the sulfide/silicate melt partition coefficient (Campbell and Naldrett 1979). D Re 0500 and D Os 030,000 were used in this study homogeneous in the parental mafic magmas, massive and disseminated sulfide ores produced from these magmas should have achieved Os equilibrium instead of heterogeneous Os composition in different sulfide ores. Systematic variation of Re-Os isotopes in Kalatongke also cannot be explained by later disturbances or hydrothermal alteration. Later disturbances may yield younger Re-Os ages than zircon U-Pb ages (Cohen et al. 2000;Lambert et al. 2000), but Re-Os ages of the Kalatongke ores are either older or younger than the zircon U-Pb ages. Hydrothermal alteration can also be ruled out because the host intrusion and sulfide ores in Kalatongke have undergone little or no hydrothermal alteration and the samples selected for analyses are fresh.
Interaction between mantle-derived magmas and various lithospheric components, particularly crust, has been recognized in many sulfide-bearing mafic intrusions (e.g., Walker et al. 1994). Sulfide mineralization in mafic intrusions is generally thought to be caused by crustal contamination and fractional crystallization (Naldrett 2004), and it has been argued that contamination by S-rich crustal materials is required for the genesis of magmatic Cu-Ni sulfide deposits (e.g., Keays and Lightfoot 2010). Therefore, it is possible that Os isotope composition heterogeneities in sulfidebearing mafic magma systems may be related to crustal contamination and sulfide segregation under different R factors. γ Os values and Re/Os ratios of sulfides are similar to those of the contaminant at low R factor, whereas γ Os values and Re/Os ratios of sulfides approach those of the magma at high R factor (Lesher and Burnham 2001). However, modeling of the R factor suggests that sulfides in Kalatongke were not formed with a single R factor process (Fig. 3a). In sulfide-bearing mafic magma systems, most Re and Os are hosted in sulfides, which may have formed at different stages during the magma evolution and thus record different Os isotope compositions. In Kalatongke, both the massive sulfide ores and barren norite have similar Re/Os ratios and γ Os values and are thought to be have derived from the same pulse of sulfide-bearing magma. On the other hand, the disseminated sulfide ores have higher γ Os values and formed from a different pulse of magma.
Formation of the massive sulfide ore and barren norite Both Ni-Cu and Cu-rich massive sulfide ores have a limited variation of Os isotope compositions, similar to those of the barren norite, suggesting that large amounts of sulfide melts were segregated from silicate magmas that formed the norite and concentrated possibly at the bottom of the magma chamber at high temperature and low viscosity. Crystallization of MSS from massive sulfide melts may produce Nirich and Cu-rich sulfide ores (e.g., Ebel and Naldrett 1996), but Os isotope compositions should remain the same in a closed system. In Kalatongke, the massive sulfide ores have nearly homogeneous γ Os values but variable Re/Os ratios which may be due to either fractionation of sulfide melts or interaction of such melts with silicate magmas because Os is significantly more compatible than Re in sulfides than in silicate magmas (Brenan 2002(Brenan , 2008. Formation of the disseminated sulfide ore Os isotope compositions and Re/Os ratios of disseminated sulfide ores from Kalatongke correlate positively and form a trend towards the hosting country rocks (Fig. 3a). The disseminated sulfide ores contain more radiogenic Os than both the massive Ni-Cu and Cu-rich sulfide ores. Because the immediate country rocks in Kalatongke have the highest radiogenic Os compositions, we suggest that the disseminated sulfide ores were formed from highly contaminated melts. As temperature decreases, the crystal content and viscosity of silicate magmas increase, leading to more heterogeneous contamination. Variable Os isotope compositions for disseminated sulfides ores can be explained by variable abundances of crustal Os. If we assume that the samples with the least radiogenic Os have the minimum contamination, modeling suggests that the highest degree of contamination was more than 50% (Fig. 3a). Experimental data have confirmed that sulfur and Os from decomposing crustal sulfides can be transferred into silicate magmas as part of the vapor phase (Baker et al. 1998). Thus, Os in crustal sulfides can be easily added into the mafic magmas and modify the Os composition of the magma. Sulfide droplets cannot be separated and segregated as sulfide melts to form massive sulfide orebodies under high viscosity. These individual sulfide droplets in the silicate magmas would have been trapped to form disseminated sulfide ores and record the additional contamination history during magma evolution. Thus, disseminated sulfide ores would have different radiogenic Os compositions because of different degrees of crustal contamination.
Implications for Re-Os dating and sulfide formation Mafic intrusions that crystallized from PGE-rich magmas, such as the Noril'sk I and Talnakh-Kharaelahk intrusions, Siberia and the Kambalda deposit in Australia, have similar Re-Os and zircon U-Pb ages. In these deposits, the systems have high Os contents (>10 ppb) and low Re/Os ratios, such that their Re-Os compositions are not easily altered by crustal contamination.
The Kalatongke intrusion are depleted in PGE relative to Ni and Cu and are thought to have undergone sulfide removal (Song and Li 2009). Os isotope compositions of mafic magmas that are depleted in PGE, particularly Os, due to sulfide removal or fractional crystallization of olivine and/or chromite (Crocket 2002), can be easily modified by crustal contamination (Shirey and Walker 1998). If such a contamination results in heterogeneous Os isotope compositions in the mafic intrusions, Re-Os dating of these sulfides and intrusions cannot yield the true formation age.
Os isotopes of sulfide melts can reach equilibrium under high temperature because of their high diffusion rate (Brenan et al. 2000). Therefore, MSS grains commonly have Os isotope compositions similar to those of the co-existing Cu-rich sulfide melts. In Kalatongke, there are only small differences in γ Os between massive Ni-Cu and Cu-rich sulfide ores, yielding only small differences between the Re-Os isochron and zircon U-Pb ages (e.g., Zhang et al. 2008; this study).
The formation of disseminated sulfide ores typically involves assimilation and fractional crystallization (AFC). Assimilation of S-rich country rocks has been proposed to explain the origin of magmatic sulfide ore deposits (Lesher and Burnham 2001), and selective crustal contamination has been invoked to explain the behavior of different isotope systems during the interaction between mantle-derived magmas and crustal rocks (Watson 1982;Walker et al. 1994;Ripley et al. 1999Ripley et al. , 2001Lambert et al. 2000). In an AFC process, if the systems are variably contaminated, the Re-Os data will plot around the mixing lines between parental magmas and contaminants and form "pseudoisochrons" (DePaolo 1981). In sulfide-bearing mafic intrusions, a different degree of contamination can be recorded by the disseminated sulfide , such that the Re-Os isochron produced by the disseminated sulfide ores in Kalatongke is a pseudoisochron that does not have geological significance. That is why the disseminated sulfide ores have variable Re-Os ages and are not useful for precise geochronology.
Both massive and disseminated sulfide ores are generally considered to have formed at the same stage (Barnes and Lightfoot 2005), which is not supported by this study. Identification of the timing of sulfide segregation is difficult, but the Os isotope composition can be used to trace multiple pulses of sulfide-bearing magmas and/or crustal contamination recorded in sulfides and thus aid in the study of sulfide-forming processes in sulfide-bearing mafic intrusions.

Conclusions
Massive sulfide ores and barren norite in the Kalatongke intrusion have Re-Os isotope compositions different from those of the disseminated sulfide ores, reflecting variable amounts of crustal contamination. Heterogeneous Os isotope compositions may record the magma evolution, but are not suitable for dating Cu-Ni sulfide mineralized intrusions. Massive sulfide ores, particularly massive Ni-rich sulfide ores, may have homogeneous Os isotope compositions and Re-Os ages closer to the true age of mineralization. Os isotope compositions of both massive Cu-rich sulfide ores and disseminated sulfide ores may have undergone additional modification and are problematic for dating.