Available online at

Transcription

1 Available online at Geochimica et Cosmochimica Acta 75 (2011) Noble gas isotopic systematics of Fe Ti V oxide ore-related mafic ultramafic layered intrusions in the Panxi area, China: The role of recycled oceanic crust in their petrogenesis Tong Hou a, Zhaochong Zhang a,, Xianren Ye b, John Encarnacion c, Marc K. Reichow d a State Key Laboratory of Geological Process and Mineral Resources, China University of Geosciences, Beijing , China b Key Laboratory of Petroleum Resources Research, Institute of Geology and Geophysics, Chinese Academy of Sciences, Lanzhou , China c Department of Earth and Atmospheric Sciences, Saint Louis University, 3642 Lindell Avenue, St. Louis, MO 63108, USA d Department of Geology, University of Leicester, Leicester LE1 7RH, UK Received 15 April 2011; accepted in revised form 1 September 2011; available online 8 September 2011 Abstract Olivine and clinopyroxene grains have been separated from four large Fe Ti V oxide ore-bearing intrusions (Panzhihua, Hongge, Baima and Taihe) in the Panxi area, Emeishan large igneous province, Southwest China, for He and Ar isotope studies. The samples examined revealed extremely low 3 He/ 4 He ratios ( Ra with the mean value 0.78 Ra) for gases extracted by stepwise heating. This feature, combined with low 40 Ar/ 36 Ar ratios can be interpreted as due to addition of subduction-related fluids and melts that had been stored in the lithospheric mantle for long periods. Considering the regional geologic history, such addition can be attributed to the paleo subduction that occurred along the western margin of the Yangtze Block during the Neoproterozoic. The subducted oceanic crust beneath the Panxi area underwent eclogite-facies metamorphism and subsequent exhumation. The infiltration of subduction-related melts and fluids into the lithospheric mantle led to enriched isotopic signatures from that of the slightly depleted asthenopheric mantle which has been suggested by the Sr, Nd and Pb isotopic data of the Emeishan basalts and picrites. In addition, considerable amounts of eclogitic melts produced by partial melting of eclogite-facies oceanic crust extensively contaminated the lithospheric mantle. During the late Permian, partial melting of an upwelling mantle plume that contained an eclogite or pyroxenite component generated the parental Fe-rich magma that supplied the ore-bearing intrusions. The combination of these factors may have been the crucial reason that many world-class Fe Ti V oxides deposits are clustered in the Panxi area. Ó 2011 Elsevier Ltd. All rights reserved. 1. INTRODUCTION It is believed that the Emeishan large igneous province (ELIP) is genetically related to a major 260 Ma plume event (e.g., Thompson et al., 2001; Xu et al., 2004; Ali et al., 2010), which has a genetic connection with several world-class Fe Ti V oxide deposits, as well as several Corresponding author. Tel.: ; fax: address: zczhang@cugb.edu.cn (Z. Zhang). Cu Ni (PGE) sulfide deposits (e.g., Zhang et al., 2009 and references therein). These particular features distinguish the Emeishan LIP from other LIPs that formed near the end of the Permian in widely separated locations such as the Siberian Traps (Sharma, 1997; Dobretsov, 2005) and Panjal Traps of northwestern India (e.g., Bhat et al., 1981). An important question we address in this paper is: what factors led to the formation of so many large Fe Ti V oxide deposits in the ELIP? What role did the geologic setting, nature of mantle source(s), partial melting processes, and magma chamber processes play in the origin of this magmatic and ore province? /$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi: /j.gca

2 6728 T. Hou et al. / Geochimica et Cosmochimica Acta 75 (2011) Most previous petrologic and geochemical studies focused on magma chamber processes (e.g., Zhong et al., 2002, 2004, 2011; Zhou et al., 2005; Wang et al., 2008; Pang et al., 2010) and the depth of melting (e.g., Zhou et al., 2008). The metallogenesis of the Fe Ti V oxide deposits is generally thought to be closely related to the nature of their sources (e.g., Zhang et al., 2009). However, the aspect of the source composition has so far received little attention. For the ore-bearing mafic ultramafic intrusions in the ELIP, Zhou et al. (2008) proposed that both Cu Ni (PGE) and Fe Ti V oxide ore-bearing intrusions are derived from a heterogeneous mantle plume, but the former was generated by higher degrees of melting of shallower fertile mantle, whereas the latter was sourced from a deeper refractory garnet bearing peridotite-zone mantle. These conclusions were challenged by Zhang et al. (2009), based on isotopic analyses of three mafic ultramafic intrusions, which host both types of mineralization. These authors demonstrated that the Sr, Nd and Pb isotopic compositions of most samples from the three intrusions overlap with those of the Emeishan basalts and OIB, suggesting that they have been derived from a slightly enriched asthenospheric melt presumably generated by the same mantle plume that formed the Emeishan flood basalts. Zhang et al. (2009) also suggested that the ascending plume-derived magma was contaminated by a Fe- and Ti-rich lithospheric mantle and that this was a crucial factor in the generation of the Fe Ti V oxide ore-bearing intrusions. However, some important issues, such as the evidence for the presence of such a Fe- and Ti-rich lithospheric mantle and the details of the ore generation process remain poorly understood (e.g., Hergt et al., 1991; Hawkesworth et al., 2000). On the basis of trace element and Pb Sr Nd isotopic investigations, Song et al. (2004) and Xiao et al. (2004) proposed that the primitive magmas of the ELIP were produced by partial melting of a rising mantle plume that reacted with the lithospheric mantle, which had been previously modified and enriched by pelagic sediments during Neoproterozoic subduction. Nevertheless, in the case of the Fe Ti V oxide ore-bearing intrusions, the involvement of lithospheric mantle and the subducted oceanic crust in their source has not been evaluated. Moreover, the role and extent of each component, especially the old recycled component, in the source cannot be unambiguously assessed from a combination of trace elements, Sr, Nd and Pb isotopes alone, because different geochemical sources may produce similar geochemical signatures (e.g., Marty et al., 1996). For example, it is difficult to distinguish the effects of crustal assimilation from assimilation or melting of enriched subcontinental lithospheric mantle (SCLM) which had been modified by recycled materials, using incompatible element and Pb Sr Nd isotopic data. In addition, these ore-bearing intrusions have experienced post-magmatic low-temperature alteration, which may have changed their Pb and Sr isotopic signatures (Beswick, 1982). The relative proportions of radiogenic and primordial noble gas isotopes are dependent on the extent and timing of episodes of gas loss from the mantle (Kellogg and Wasserburg, 1990). Much of the utility of noble gases is based on the wide variations in their isotopic compositions. This is related to their overall depletion, which has made these elements sensitive to isotopic modification from addition of radiogenic isotopes derived from more abundant parent isotopes (Porcelli et al., 2002). The wide applicability of noble gas systematics is due to the range of such processes stated above. They provide a powerful tool to better understand the evolution of the Earth s interior, its chemical structure, and its volatile fluxes (e.g., Ozima and Podosek, 2002). However, recent research has largely concentrated on gases in submarine basalts (e.g., Moreira and Kurz, 2001; Raquin et al., 2008), mantle-derived volcanic rocks (e.g., Nuccio et al., 2008; Yamamoto et al., 2009; Starkey et al., 2009), or xenoliths of presumed upper-mantle origin (e.g., Yamamoto et al., 2004; Gautheron et al., 2005; Czuppon et al., 2009; Sumino et al., 2010; Hopp and Ionov, 2011). Over the past decades, limited results on the nonradiogenic gases in ordinary plutonic igneous rocks from continental crust have been reported (Rayleigh, 1939; Damon and Kulp, 1958; Butler et al., 1963; Kuroda and Sherrill, 1978; Smith, 1984). It is noteworthy that the gas extracted from these plutonic rocks by heating technique represent bulk sample analyses of multi grain aliquotes possibly holding different numbers, densities, and sizes of inclusions, thus the noble gas abundances cannot be easily related to any magmatic processes. For example, using the one-step heating, gases extracted from a bulk sample inevitably contain some adsorbed gases and those from the post-contemporary inclusions. These gases are regarded to have no obviously relations to the magmatic processes. Therefore, little importance is usually placed on noble gas data derived from bulk analysis in recent years (e.g., Nuccio et al., 2008). To assess the influences of recycled oceanic crust components on Fe- and Ti-rich lithospheric mantle, we present the first He and Ar contents and isotopic compositions by stepwise heating of olivine and clinopyroxene grains separated from gabbro and olivine-gabbro samples collected from the four renowned world-class Fe Ti V oxide deposits (Panzhihua, Hongge, Baima and Taihe) in the Panxi area. Our results provide important constraints on the petrogenesis of ore-bearing intrusions and related Fe Ti V oxide mineralization Regional geology 2. GEOLOGICAL BACKGROUND Southwestern China comprises the western margin of the Yangtze Block to the east and the easternmost part of the Tibetan Plateau to the west. The Yangtze Block consists of Mesoproterozoic granitic gneisses and metasedimentary rocks, which have been intruded by the Kangdian (800 Ma) granites (Zhou et al., 2002b). The Neoproterozoic granites are overlain by a series of marine and terrestrial rocks from the late Neoproterozoic (600 Ma) to the Late Permian (Yan et al., 2003).The Permian lithologies include carbonate-rich rocks and the Emeishan continental flood basalts. The Triassic comprise both continental and marine sedimentary rocks, whereas Jurassic to Cretaceous

3 Noble gas systematics of Panxi layered intrusions 6729 are entirely terrestrial. Neoproterozoic arc plutonic metamorphic assemblages occur along the western and northern margin of the Yangtze Block, which are believed to have been related to subduction of Rodinian oceanic lithosphere toward the Yangtze Block during a period from 860 to 760 Ma (Zhou et al., 2002a). A late Paleozoic to early Mesozoic (ca Ma) rifting event has also been recognized (Cong, 1988). The region was further deformed during the Paleogene India Asia collision (Yin and Harrison, 2000) Emeishen large igneous province (ELIP) and associated Fe Ti oxide deposits The ELIP is dominated by flood basalts (the Emeishan flood basalts) ranging in thickness from a few hundred meters to a maximum of 5 km and covering an area of at least km 2. The flood basalts were likely emplaced at or close to sea level (Thompson et al., 2001; Ukstins Peate and Bryan, 2009). In contrast to the Siberian flood basalts, which formed at relatively high northern latitude, emplacement of the Emeishan flood basalts occurred near the Equator (Enkin et al., 1992). Overall, the province appears to be slightly older than the geochronologically well-constrained 251 Ma Siberian Traps (e.g., Reichow et al., 2009) with reported 40 Ar/ 39 Ar ages of 254 ± 5 Ma (Boven et al., 2002), 255 Ma, and Ma (Lo et al., 2002) for lava flows and two late-stage intrusions. However, more recent SHRIMP U Pb dating for mafic intrusions, dykes and volcanic rocks have indicated emplacement between 257 and 263 Ma (Guo et al., 2004; Zhong and Zhu, 2006; He et al., 2007), which suggests a link between the ELIP and the end-guadalupian mass extinction (Middle Permian). The Panxi region lies in the central-western part of the ELIP where the flood basalts are variably deformed, uplifted and eroded due to strong tectonic activity in the Cenozoic. Magmatic Fe Ti V oxide deposits are documented in several layered intrusions in this region whose exposure is controlled by major N S trending faults. The ore deposits and host layered intrusions share the same names including Panzhihua, Hongge, Baima, Taihe and Xinjie. Most of the intrusions have been dated by the U Pb zircon method providing ages of 260 Ma (Fig. 2). Giant Fe Ti V oxide deposits occur in several relatively large layered intrusions which are spatially associated with contemporaneous flood basalts and copious granitoids (Zhang et al., 1999; Ma et al., 2001), including the Panzhihua, Hongge, Baima and Taihe intrusions (Fig. 1; Zhong et al., 2002, 2003, 2004, 2005; Zhou et al., 2005, 2008). The total oxide ore reserves of four large Fe Ti V oxide deposits are estimated to be 7544 million tonnes with an average ore grade of 36 wt.% Fe total, 0.28 wt.% V 2 O 5 and 12.6 wt.% TiO 2 (Zhong et al., 2002, 2003; Ma et al., 2003; Zhou et al., 2005) Geology of the ore-bearing intrusions Detailed descriptions of the geology of these ore-bearing intrusions have been presented by several workers, i.e. Panzhihua (e.g., Zhang et al., 2009; Pang et al., 2010), Hongge (e.g., Zhong et al., 2002), Baima (e.g., Zhou et al., 2008; Pang et al., 2010) and Taihe (e.g., Pang et al., 2010) and a brief summary presented here. These four largest deposits i.e., Panzhihua, Hongge, Baima and Taihe, share similar characteristics, e.g. the major Fe Ti V layers (the ore bodies) occur as iron beds associated with the layered gabbros and are generally concentrated in the lower parts of the intrusion (Fig. 3; e.g., Zhou et al., 2005). Additionally, throughout the intrusion, the frequency of Fe Ti V oxide layers decreases upwards. Mineral compositions also show regular variations with for example Fo (Fo 82 Fo 63 ) and An (An 68 An 40 ) contents of olivine and plagioclase, respectively, decreasing upwards in Panzhihua. The compositions of clinopyroxenes are with i.e., En Fs Wo less variable. Moreover, these intrusions are undeformed, unmetamorphosed, and generally less affected by alteration. Local alteration consists of clinopyroxene and plagioclase being replaced by tremolite and albite, respectively. 3. SAMPLES AND EXPERIMENTAL PROCEDURE Samples were initially checked for weathering and traces of alteration were removed before the rocks were reduced to chips. The freshest chips were selected for analysis using a binocular microscope. Rock samples were crushed to a grain size below 2 mm using a jaw crusher, and the clinopyroxene and olivine grains separated by standard gravimetric and magnetic methods. Crystals for noble gas analysis were hand selected under a binocular microscope. These selected minerals were crushed into small grains with a diameter of about 0.5 mm and hand-picked again for the experiment. Noble gases were released from fresh olivine and clinopyroxenes by stepwise heating. Olivines and clinopyroxenes are retentive for noble gases; in-situ additions of radiogenic 4 He and 40 Ar are quite small because of very low contents of U, Th, and K in such minerals (e.g., Krause et al., 2007). Although U and Th in mineral inclusions or neighboring minerals can be a source of a-particles, the limited range of a-particles (tens of microns) means that the removal of the rims of mineral grains, e.g. by acid etching or physical abrasion, can reduce this nucleogenic component (Baxter, 2010). Stepwise heating is thought to be a powerful and widely used technique in noble gas analysis, for it has been quite successful in discriminating noble gases trapped in fluid or melt inclusions. The effect of cosmogenic nuclides due to surface exposure has a negligible effect on our samples because most samples were collected from locations that were only recently exposed by mining. Before stepwise heating, mineral grains were washed ultrasonically for 30 min with distilled water, and then ethanol and acetone, and dried at 80 C overnight. More than 500 mg of such prepared sample was wrapped in thin Alfoil. The sample bags were loaded into the sample turnplate connected with a stainless steel ultra-high vacuum extraction line. Up to ten samples could be mounted on the line at the same time. The sample chamber and the gas preparing system were baked out about 150 C for more than 24 h and the samples were also preheated during the baking period to remove atmospheric noble gases adsorbed in the

4 6730 T. Hou et al. / Geochimica et Cosmochimica Acta 75 (2011) Fig. 1. Geological map of the Panxi area and associated mineralized layered intrusions (modified from Zhong et al., 2011). Insert illustrates distribution of major terannes in China and study area (after Chung and Jahn, 1995). Abbreviations are as follows: IC = Indochina; NCB = North China block; QD = Qaidam; YZB = Yangtze block; SG = Songpan-Ganze accretionary complex. Fig. 2. Crystallization age of the ore-bearing intrusions in this study. The ages for Panzhihua and Baima are analyzed by sensitive high-resolution ion microprobe (Zhou et al., 2005), the age for Hongge are analyzed by isotope dilution thermal ionization mass spectrometry (Zhong and Zhu, 2006), the age for Taihe are analyzed by laser ablation inductively coupled plasma mass spectrometry (Zhong et al., 2009). sample grains. The tantalum and molybdenum crucibles used for melting samples were heated up 1700 C for more than 24 h in order to reduce the heat blank. A sample would fall into the molybdenum crucible when the sample turnplate was turned, and the gases contained in the sample were extracted. The gas released at each individual temperature was purified through a 800 C titanium furnace and a ZrAl getter held at room temperature. He, Ne, Ar, Kr and Xe were absorbed and separated by activated carbon at a constant temperature in a mixture of liquid nitrogen and ice water. The contents and isotopic compositions of noble gases were measured by a MM5400 mass spectrometer installed in the Laboratory of Gas Geochemistry (Lanzhou), Institute of Geology and Geophysics, Chinese Academy of Sciences. The gases were pumped away between samples to eliminate interference to the following sample. The accuracy of our data was checked by repeatedly measuring an air standard collected from the top of Gaolan Mountain at Lanzhou City, China. A total of 18 runs of air 3 He/ 4 He ratios were reproducible to within 3 He/ 4 He ratios of ± and 40 Ar/ 36 Ar ratios of ± 1.2. Hotblanks were run using the same procedure as the samples and the hot-blank levels of He and Ar at 1600 C (cm 3 STP) are 4 He = and 40 Ar = , respectively. All results were calibrated to hot-blank. All noble gases isotopic compositions in hot-blank were approximately air value. In order to confirm the experimental results, we have repeatedly verified our data. A relative error of less than 2.5% in our analytical techniques can be regarded to be very satisfactory Helium 4. RESULTS Noble gas isotopic results are summarized in Table 1. The measured 3 He/ 4 He ratios are plotted versus temperatures in Fig. 4. Except for the high 3 He/ 4 He ratio obtained in the 600 C fraction of clinopyroxene in HG02 (4.5 R A ; R A is the atmospheric ratio of ) during step-

5 Noble gas systematics of Panxi layered intrusions 6731 Fig. 3. Stratigraphy of the Panzhihua, Hongge, Baima and Taihe intrusions, simplified from Li et al. (1981) and Pang et al. (2010). Note the vertical exaggeration to illustrate the details of the ore-bearing part of the intrusions. Abbreviations are as follows: FW = footwall or roof of intrusion, HW = hanging wall or floor of intrusion, UZ = upper zone, MZ= middle zone, LZ = lower zone, and MGZ = marginal zone. heating gas extraction, the 3 He/ 4 He ratios cover a narrow range from 2.5 R A down to <1 R A, and do not vary among different localities. This may indicate that these minerals have trapped the same fluid and that post-emplacement processes have not significantly affected the 3 He/ 4 He ratio. The observed values are much lower than those of MORB- and plume-type mantle (Kurz, 1991; Hilton et al., 1993), xenoliths from Eastern Australia (Matsumoto et al., 1998, 2000; Czuppon et al., 2009, 2010), European sub-continental lithospheric mantle (SCLM; Dunai and Baur, 1995; Gautheron et al., 2005), Japanese island arc (Nagao and Takahashi, 1993; Ikeda et al., 2001), and back arc basalts (Honda et al., 1993; Bach and Niedermann, 1998; Ikeda et al., 1998; Sano et al., 1998). However, they overlap with values of some mantle xenoliths which are thought to be influenced by ancient recycled materials, such as eastern China (Tang et al., 2007) and far eastern Russia (Yamamoto et al., 2004) (Fig. 6.) Argon The measured 40 Ar/ 36 Ar ratios show large variations ranging from 297.1, which is close to the atmospheric value of (Nier, 1950), to Low temperature extractions yield low values whereas the high temperature fractions were generally characterized by higher isotopic ratios, especially the values obtained at 1100 C, which are always the highest in our samples. The lack of correlation in 40 Ar/ 36 Ar vs. 40 Ar (Fig. 5b) suggests that the isotopic ratios have not been dramatically affected by K decay in the samples after emplacement. 5. DISCUSSION 5.1. Genesis of extremely low 3 He/ 4 He ratios in mafic ultramafic intrusions This study shows that mafic minerals, i.e. olivines and clinopyroxenes, in mafic ultramafic Fe Ti V oxide orebearing intrusions in Panxi area display extremely low 3 He/ 4 He ratios (the lowest is 0.1 R A ) for gases extracted by stepwise heating. Such low 3 He/ 4 He ratios are rarely seen in rocks or minerals separated from LIPs. Previous work (e.g., Marty et al., 1996) has revealed that only some low-ti basalt samples in Oligocene continental flood basalts from Ethiopia show similar extremely low 3 He/ 4 He ratios (0.035 R A ), which have been explained by accumulation of radiogenic 4 He in their mantle source with concomitant crustal contamination. Since helium has high diffusivities in the mantle (e.g., Hart, 1984; Lux, 1987; Trull and Kurz, 1993), an extensive mantle heterogeneity would be required, in order to maintain the He isotope anomaly for significant periods of time. For instance, in order to maintain a source body with 6.8 R A in a mantle with 30 R A for a billion years,

6 6732 T. Hou et al. / Geochimica et Cosmochimica Acta 75 (2011) Table 1 Noble gas abundance and isotopic compositions of olivine and clinopyroxene from ore-bearing intrusions in Pan-Xi area, analyzed by stepwise heating method. Sample Mineral Weight (g) Temperature 4 He(10 7 ) r 40 Ar (10 7 ) r 3 He/ 4 He(R/R A ) r 40 Ar/ 36 Ar r PZH01 Ol C C C Total PZH02 Cpx C C C Total PZH03 Ol C C C Total HG01 Cpx C C C Total HG02 Cpx C C C Total HG03 Cpx C C C Total HG04 Cpx C C C Total HG05 Cpx C C C Total HG03 Ol C C C Total BM01 Ol C C C Total BM02 Ol C C C Total BM01 Cpx C C C Total BM02 Cpx C n.d. n.d C C Total TH01 Cpx C C C Total All tabulated data were corrected for blanks. Unit of concentrations are cm 3 STP g 1. Errors are 1 s.d. Ol and Cpx are short for olivine and clinopyroxene, respectively. n.d. is not detectable. R A : atmospheric 3 He/ 4 He =

7 Noble gas systematics of Panxi layered intrusions 6733 Fig He/ 4 He ratios versus stepwise heating temperatures. To avoid overlapping of symbols the data are partly offset from temperatures. The uncertainties are given as 1r. the thickness of the source body is required to be on the order of 1 km (Hanyu and Kaneoka, 1998). To explain the extremely low 3 He/ 4 He ratios observed in our separated olivine and clinopyroxene minerals, we raise several possibilities and discuss each of them below Atmospheric contamination The concentration of He in the atmosphere is 5 ppm and He is generally thought to be not recycled back into the mantle (e.g., Ozima and Podosek, 2002). The 3 He/ 4 He ratios much lower than the atmospheric value cannot be readily explained by a simple binary mixing with atmospheric component. Thus, contamination of atmospheric He directly in the magma reservoir of these intrusions was unlikely Crustal contamination The magmas forming these intrusions ascended through and were emplaced into continental crust. It is therefore important to assess the extent to which crustal contamination may have influenced these intrusions before we can determine the noble gas isotopic signature of their mantle Fig He/ 4 He versus 4 He ratios of olivine and pyroxene separates from Fe Ti V oxide-bearing intrusions in the Panxi area. For comparison, arrays of published data are shown for xenolith data from Japan island arc (Nagao and Takahashi, 1993; Ikeda et al., 2001), European subcontinental lithospheric mantle (SCLM; Dunai and Baur, 1995; Gautheron et al., 2005), Eastern Australia (Matsumoto et al., 1998, 2000; Czuppon et al., 2009, 2010), Eastern China (Tang et al., 2007), and Far Eastern Russia (Yamamoto et al., 2004). The compiled backarc basin basalts 3 He/ 4 He data are from Honda et al. (1993), Bach and Niedermann (1998), Ikeda et al. (1998) and Sano et al. (1998). MORB data are from Kurz (1991) and Hilton et al. (1993). source. Assuming the most favorable condition for contamination of zero helium content in magma, any 4 He contribution from crustal fluids would diffuse through the magma by tens of centimeters over a time scale of years (based on He diffusivity in basalt melts; Lux, 1987; Heber et al., 2007; Tolstikhin et al., 2010). The diffusion of crustal He into the melt would be less efficient if the concentration of magmatic He is higher than zero and if we recognize that degassing processes cause a He flux from magma to country rocks. Based on Sr, Nd, Pb, and O isotope data, Zhang et al. (2009) argued that the magma that supplied the Panzhihua intrusion were derived from the mantle with little or no crustal contamination making a crustal contribution to the He in these samples less likely. Thus, we infer that the Fig He/ 4 He ratios versus 4 He. (a) and 40 Ar/ 36 Ar ratios versus 40 Ar (b) of olivine and pyroxene separates from Fe Ti V oxide -bearing intrusions in Panxi area. Effects on isotope ratios and 4 He abundance by addition of cosmogenic and radiogenic components are shown by arrows. The uncertainties are 1r. Symbols as in Fig. 2.

8 6734 T. Hou et al. / Geochimica et Cosmochimica Acta 75 (2011) addition of crustal He could hardly cause the observed extremely low 3 He/ 4 He ratios in the samples Preferential depletion of helium If significant He loss occurred after emplacement of these intrusions, mass-dependent fractionation would significantly lower the 3 He/ 4 He ratio in the minerals. To explain the observed low 3 He/ 4 He values by this process from an assumed initial value of a plume-type mantle such as the Hawaiian mantle source, an anomalously high initial He concentration is required; much higher than that of popping rocks ( cm 3 STP/g; Marty and Ozima, 1986; Burnard et al., 1997) which are the most gas-rich basaltic glass. There is also a potential for loss of helium from clinopyroxene after emplacement of the magma. This hypothesis is predominantly based on the common idea that pyroxene crystallization postdates olivine crystallization and continues to exchange helium with the magma, due to the lower effective closure temperature for He in pyroxene (Shaw et al., 2006). Indeed, olivine has a crystal structure that reduces any secondary exchange after crystallization, thus preserving the original magmatic signature (Craig and Lupton, 1976; Ozima and Podosek, 1983; Martelli et al., 2004). However, in the case of the intrusions in the Panxi area, both olivines and clinopyroxenes are early phases in the crystallization of the ore-bearing magma (e.g., Zhou et al., 2005). Moreover, both our olivine and clinopyroxene data have similar values (Table 1) and no systematic difference in noble gas concentrations and isotopic compositions in samples from different localities have been observed. Hence, it appears that both olivine and clinopyroxene trapped similar fluids and melts during the crystallization processes and it is difficult to explain the low 3 He/ 4 He ratios by significant helium loss from the fluid and melt inclusions unless similar amounts of He loss and fractionation happened for both the olivines and clinopyroxenes, which is unlikely Accumulation of radiogenic 4 He after emplacement Radiogenic and/or nucleogenic 4 He is mostly generated within the crystal lattice after emplacement of the intrusion. Such components are thought to be exclusively extracted at relatively higher temperature during the stepwise heating experiment. Considering the ages of these intrusions (260 Ma), it is plausible that the accumulation of nucleogenic 4 He can account for the low 3 He/ 4 He ratios. However, production of 4 He by U and Th decay within the crystal lattice is very low, due to the very low content of these radioactive nuclides inside the minerals (Zindler and Hart, 1986). If the gases extracted at high temperature were affected by nuclides generated in-situ within the crystal lattice, the 3 He/ 4 He ratios should show a negative correlation with 4 He concentrations. Given that such a correlation is not been observed (Fig. 5a), the low 3 He/ 4 He ratios in the samples are unlikely the result of addition of radiogenic nuclides generated in situ after emplacement Addition of radiogenic 4 He in the mantle Based on the above discussion, we conclude that the low 3 He/ 4 He ratio in the samples from the Panxi area must reflect an inherent feature of the mantle source for these orebearing intrusions. A reason other than those discussed above must be assumed to explain the extremely low 3 He/ 4 He ratios. Considering that variable contributions of lithospheric materials were involved in generating the parental magmas of these intrusions based on Sr Nd Pb isotopic analyses (Zhang et al., 2009), we prefer to attribute the low 3 He/ 4 He ratios to the addition of radiogenic 4 He predominantly to the lithospheric mantle. The addition of radiogenic 4 He could be due to the addition of mantle fluids enriched in 4 He, or due to in-situ growth of radiogenic 4 He after influx of a fluid with a high U/He ratio into the mantle, or a combination of both. Such additions have usually been interpreted as the addition of a component recycled in the mantle via subduction process (e.g., Yamamoto et al., 2004). Regarding the former possibility, some source is required to produce the mantle fluids with low 3 He/ 4 He ratios. For example, it is possible that an old subducted slab was stored in the upper mantle for a sufficiently long time to have accumulated enough radiogenic 4 He. In the latter case, for example, infiltration of U-bearing fluids into the Panxi lithospheric mantle could cause heterogeneous and low 3 He/ 4 He ratios. Neoproterozoic arc plutonic metamorphic assemblages occur along the western and northern margin of the Yangtze Block, which are believed to have been related to subduction of Rodinian oceanic lithosphere toward the Yangtze Block during the period from 760 to 860 Ma (Zhou et al., 2002a). Hence, the lithospheric mantle beneath Panxi area, which is located at the western margin of the Yangtze Block, was probably affected by fluids and/or melts derived from the ancient subducted slab. Infiltration of U-bearing fluids into the Panxi lithospheric mantle could cause heterogeneous and low 3 He/ 4 He ratios, because U is soluble in slab-derived fluids to considerable depth (e.g., Brenan et al., 1995; Kogiso et al., 1997) and tends to be relatively enriched in potential silicate melts due to its incompatibility during partial melting processes. This would allow U-bearing fluids and/or melts to infiltrate the mantle wedge from the subducted slab. Moreover, the infiltration of subduction-related melts and fluids into the lithospheric mantle led to enriched isotopic signatures from that of the slightly depleted asthenopheric mantle which has been suggested by the Sr, Nd and Pb isotopic data of the Emeishan basalts and picrites (Xiao et al., 2004; Zhang et al., 2006) Assessment of addition of subduction-related components on 40 Ar/ 36 Ar ratios In contrast to He, heavy noble gases like Ar, have isotopic compositions of mantle end-members such as MORBs and OIBs that are difficult to distinguish, mainly because of high abundance of relatively nonradiogenic Ar in the atmosphere ( 40 Ar/ 36 Ar = 295.5) coupled with highly radiogenic Ar in mantle samples ( 40 Ar/ 36 Ar P12,000 for OIBs (Poreda and Farley, 1992), and P28,000 in MORBs (Staudacher et al., 1989; Fisher, 1994). Although 40 Ar/ 36 Ar ratios in our samples reach up to 16,000, it is not possible to assert conclusively that the relatively high 40 Ar/ 36 Ar ratio compared to the atmospheric value are due to plume-de-

9 Noble gas systematics of Panxi layered intrusions 6735 rived components with even higher Ar ratios that had been simply mixed with atmospheric components which is thought to be added during subduction. In addition, old oceanic crust and continental crust also have high 40 Ar/ 36 Ar ratios as a result of growth of radiogenic nuclides during K decay. Hence, the origin of high 40 Ar/ 36 Ar ratios in our data is not clearly identified. Nevertheless, it seems possible that the low 40 Ar/ 36 Ar ratios analyzed in our samples are due to the addition of low 40 Ar/ 36 Ar ratio components such as air and/or deep-sea sediments which were contaminated by atmospheric components (Igarashi et al., 1987). As mentioned above, the Panxi upper mantle may have been metasomatized by subduction-related melt or fluid. Such an effect should be more strongly reflected in the 40 Ar/ 36 Ar ratio, if the fluid was surface-derived. However, there are four possibilities that can cause the addition of the atmospheric gases into the samples: (1) Atmospheric adsorption on the surface of the samples; (2) Incorporation of surface water with dissolved Ar in samples as hydrate or hydrous minerals formed by alteration after emplacement of magma; (3) Infiltration of ground water with dissolved Ar circulating within the crust into the samples; (4) Percolation of subducted atmospheric noble gases dissolved in sea water through the mantle as the solution dehydrated from the subducted slab. Ballentine and Barfod (2000) have argued that most atmospheric addition is only due to a very superficial contamination that might occur in the laboratory especially for heavier noble gases. However, for noble gas analysis, samples and analytical lines are always preheated in order to reduce adhered atmospheric gases. The amounts of 36 Ar in procedural blanks were in some cases a few orders of magnitude lower than those extracted from the samples in heating experiments. Furthermore, atmospheric contaminants borne by hydrous minerals generated by low temperature alteration or weathering are likely to be removed in relatively low temperature fractions (600 C) and by careful handpicking of mineral separates. Our samples were collected in freshly exposed sections uncovered by shot-firing in the open pit, and the influence of ground water is very limited. These lines of evidence suggest that possibilities (1), (2), (3) are unlikely. Accordingly, the low 40 Ar/ 36 Ar ratios we measured are likely to reflect the isotopic compositions of the mantle source, and require that the mantle source was contaminated by materials with lower 40 Ar/ 36 Ar ratios, such as fluids derived from subduction-related components containing atmospheric argon (Sarda et al., 1999, 2000; Matsumoto et al., 2001). crust/continental crust region (C) are shown on a 3 He/ 4 He vs. 40 Ar/ 36 Ar diagram (Kaneoka and Takaoka, 1985) (Fig. 7). Noble gas compositions in most samples from these intrusions in Panxi area plot within the field enclosed by the lines connecting P, A, and C. Some samples show lower 3 He/ 4 He ratio than the atmospheric value together with moderately higher 40 Ar/ 36 Ar ratios. Accordingly, the samples plotting within the P A C field can be plausibly explained by three-component mixing of noble gases among plume-derived magma, an atmospheric component derived from subducted deep-sea water and a C-type source strongly enriched in radiogenic 4 He and 40 Ar from a crustal material like oceanic sediments (Kaneoka and Takaoka, 1985). However, the extremely low 3 He/ 4 He ratios require clarifying the contribution from a putative plume. Based on radiogenic Sr-, Nd-, and Pb data (e.g., Zhang et al., 2009), and as mentioned above, the ore-bearing intrusions in the Panxi area are interpreted to be derived from the interaction between the Permian plume and lithospheric mantle. Therefore, we incline to attribute the noble gas isotopic signature to be inherited from both the plume and the lithospheric mantle. Although partial melting of a plume source contributed largely to the generation of magmas in ELIP, it is noteworthy that noble gas concentrations in plume-derived rocks, e.g. OIBs, are comparatively low (Ozima and Podosek, 2002) This in particular as they are at least 2-orders of magnitude lower compared to the metasomatic xenoliths from the enriched lithospheric mantle (e.g., Yamamoto et al., 2004). Thus, since the lithospheric mantle was metasomatized by subduction-related melt or fluid derived from the ancient subducted slab, a relatively high abundance of noble gas in the lithospheric mantle 5.3. Role of recycled components in the lithospheric mantle What has been discussed above suggests an important recycling of atmospheric argon back into the mantle and challenges the so-called noble gases subduction barrier which says that noble gases of the subducted oceanic crust (and sediments) are largely returned back to the atmosphere through arc volcanism (Staudacher and Allègre, 1988). The isotopic compositions for plume-type mantle such as the mantle source of the Hawaiian hotspot (P), MORB-type (M) mantle, atmosphere (A), and old oceanic Fig He/ 4 He versus 40 Ar/ 36 Ar diagram for samples analyzed in the present study. The isotopic compositions of the mantle plume (P), MORB (M), atmosphere (A), old oceanic crust field (C) and Samoan hotspot lavas (Somoa) (Data sources: Kaneoka and Takaoka (1985), Staudacher and Allègre (1988) and Farley et al. (1992)). The field enclosed by the lines connecting P A C represents the isotopic ratios produced by three-component mixing (Kaneoka and Takaoka, 1985; Nagao and Takahashi, 1993). The uncertainties are 1r. Symbols as in Fig. 4.

10 6736 T. Hou et al. / Geochimica et Cosmochimica Acta 75 (2011) would be expected. Infiltration of subduction-related melts and/or fluids into the Panxi lithospheric mantle could cause heterogeneous and extremely low 3 He/ 4 He ratios and addition of low 40 Ar/ 36 Ar ratio components such as air and/or deep-sea sediments which were contaminated by atmospheric components in the lithospheric mantle as mentioned in Section Therefore, even a small addition of material from the lithospheric mantle could significantly influence the noble gas signature of the magma. In the case of the ELIP this could have lead to an extreme decrease in 3 He/ 4 He and 40 Ar/ 36 Ar ratios in the magma generated by the Permian plume-lithosphere interaction. Comprehensive noble gas studies from these intrusions in the Panxi area have attempted to relate He isotopic variations to different contributions of radiogenic He from subducted materials (e.g., Craig et al., 1975). However, there are still two alternative mechanisms for producing the low 3 He/ 4 He ratio in the mantle. The first is an incorporation of fluids and/or melts derived from the ocean floor sediments (e.g., Matsuda and Nagao, 1986), and the second is derived from the oceanic basaltic crust (e.g., Staudacher and Allègre, 1988). Correlations of 3 He/ 4 He with 87 Sr/ 86 Sr in oceanic rocks have been used to infer the large-scale structure of the mantle and interactions between sources (Kurz et al., 1982; Poreda et al., 1986). The high 3 He/ 4 He ( /27R/ R A ) and relatively low 87 Sr/ 86 Sr ( ) of hot spots such as Hawaii are explained by undegassed lower mantle (Kaneoka and Takaoka, 1985; Valbracht et al., 1997). Combined with published Sr isotopic data of these intrusions (Zhong et al., 2002; Zhou et al., 2005, 2008; Shellnutt and Zhou, 2007; Zhang et al., 2009; Shellnutt et al., 2009), the 3 He/ 4 He vs. 87 Sr/ 86 Sr diagram for the intrusions is shown in Fig. 8. The near-vertical fields for these intrusions from the Panxi area in He Sr isotope space point to an EMI-type mantle end-member. The low 3 He/ 4 He ratios of the samples indicate that the mantle source should have elevated time-integrated (U + Th)/ 3 He. This can be interpreted to reflect a recycled component of degassed oceanic sediment and crust with low He/(Th + U) and elevated 87 Sr/ 86 Sr (Kurz et al., 1982; Graham et al., 1990; Farley et al., 1992). If these recycled materials have been stored in the mantle for long periods, the Sr, Nd, and Pb isotopic compositions would have EMI signatures (e.g., Cohen and O Nions, 1982; Weaver et al., 1986). As stated earlier, during the Neoproterozoic period, a subduction zone was present on the western margin of Yangtze Block, including the Panxi area. Widely distributed adakitic plutons, such as Datian and Dajianshan felsic plutons in the Panxi area, were the products of melting of a subducted oceanic slab, in the presence of garnet as a residue in the source, i.e. in the form of eclogite (e.g., Zhou et al., 2002a, 2008; Zhao et al., 2008). Such melts are thought to have contaminated the lithospheric mantle beneath the Panxi area, perhaps remaining in the lithospheric mantle as a grain-boundary component or as melt inclusions. These melts may also have contained an atmospheric component inherited from the deep-sea water or sediments in the subducted slab. The fluids derived from the subducted slab are likely to have high U/ 3 He ratios (Farley, 1995) Fig. 8. Simplified plot of 3 He/ 4 He versus 87 Sr/ 86 Sr for samples from Fe Ti V oxide-bearing intrusions in the Panxi area and other terrestrial fields. Data sources: the 87 Sr/ 86 Sr isotopic compositions of Panxi samples are from Zhong et al. (2002), Zhou et al. (2005, 2008), Shellnutt and Zhou (2007), Zhang et al. (2009), Shellnutt et al. (2009), and Zhong et al. (2002). Compositional fields of Samoa, Iceland and Loihi Seamount are taken from Kurz et al. (1982), Farley et al. (1992) and Starkey et al. (2009). Effects to isotope ratios by addition of radiogenic components are indicated by arrows. due to the mobility of U in water and therefore likely to partition into aqueous fluids when the slab dehydrates (Brenan et al., 1995; Kogiso et al., 1997). The high U/ 3 He ratio in the fluids will consequently lead to a rapid decrease in the 3 He/ 4 He ratio over time. These inferences are supported by low d 18 O values ( &) of clinopyroxenes separated from Panzhihua intrusion (Zhang et al., 2009) that are lower than that of lithospheric and asthenosphric mantle (6&; e.g., Kyser et al., 1986; Mattey et al., 1994). This is consistent with the addition of subduction-related fluids and melts Implications for petrogenesis of the ore-bearing intrusions As discussed earlier, during the Neoproterozoic, subducted oceanic crust beneath the Panxi area underwent eclogite-facies metamorphism and subsequent exhumation. The infiltration of subduction-related melts and fluids into lithospheric mantle led to an addition of subducted atmospheric argon and pronounced decrease in the 3 He/ 4 He ratio over geological time. Considerable amounts of melts were produced by partial melting of eclogite-facies oceanic crust as reflected by the Neoproterozoic adakitic plutons in the Panxi area (e.g., Zhao et al., 2008). Such melts would have moved upwards through the lithospheric mantle and reacted with olivine in the surrounding peridotite to form orthopyoxene and garnet (Gibson, 2002). Re-melting of residual eclogite could also produce andesitic partial melts that react with the surrounding peridotite and enrich it in garnet and clinopyroxene (Yaxley, 2000), i.e. equivalent of eclogite. Melting of this relatively anhydrous, re-fertil-

11 Noble gas systematics of Panxi layered intrusions 6737 ized peridotite produces Fe-rich melts that are enriched in incompatible trace elements (Yaxley and Green, 1998). Such a re-fertilized peridotite source has a lower solidus temperature than normal peridotite (Yaxley, 2000). During the late Permian, when the Emeishan plume ascended from the lower mantle (Zhang et al., 2008) it encountered the enriched lithospheric mantle containing the eclogitic material. Thus, the primary magmas were likely produced by partial melting of both, the plume and the eclogitic material. The magmas fed the ore-bearing intrusions, carrying an enriched noble gas signature such as low 3 He/ 4 He ratios, derived from a hybridized mantle source which contained normal garnet-zone lherzolitic components (e.g., Zhou et al., 2008; Zhang et al., 2009) and a mixture of re-fertilized peridotite and recycled material. It was assumed by several researchers that the parental magma of these Fe Ti V oxide ore-bearing intrusions are ferrobasalt (Zhou et al., 2005) or ferropicrite (Zhang et al., 2009) based on bulk major element data. Moreover, it has been inferred that the Fe-rich parental magma had been derived from an Fe-rich mantle source. However, whether such a Fe-rich mantle exists and how it was formed is still not clear (e.g., Pang et al., 2008). Tuff et al. (2005) demonstrated that ferropicritic magmas can be generated by melting of garnet pyroxenite under high pressure (5 GPa) and temperature (1550 C). This is consistent with previous experiments suggesting that the iron content of primary magmas increases with both increasing mean degree of partial melting and increasing mean temperature at a given pressure (Langmuir and Hanson, 1980). In the Panxi area, the Fe-rich magma could have been generated by this process, i.e. partial melting of a mixture of an upwelling mantle plume and the subcontinental lithospheric mantle that comprises the equivalent of eclogite and other recycled materials due to the ancient subduction-related enrichment. This type of melting involving an eclogite or pyroxenite component is also capable of explaining the significant volumes of melts seen in the ELIP and other LIPs (e.g., Cordery et al., 1997; Campbell, 1998; Takahashi et al., 1998; Yaxley, 2000; Leitch and Davies, 2001). However, it is also possible that the parental ferropicritic and ferrobasaltic magma was produced by partial melting of upwelling mantle plume that entrained garnet pyroxenite streaks derived from subducted mafic oceanic crust (Gibson et al., 2000). Considering that subduction occurred in the Neoproterozoic whereas the plume activity occurred during the Permian, it is unlikely that the upwelling plume entrained the streaks or blobs of eclogite/pyroxenite matrix. Therefore, the equivalent of eclogite and other recycled material should have been stored in the lithospheric mantle due to the ancient subduction-related enrichment. The discussion and evidence provided here that the He and Ar noble gas isotopic composition of the studied intrusions is best explained by involving subducted crustal components, which is consistent with the ideas above that the parental magma of these ore-bearing intrusions is ferropicritic and derived from possible eclogitic materials of former oceanic crust. The junction of these subductiongenerated mantle sources and the Emeishan plume can well explain why many world-class Fe Ti V oxide deposits are clustered in the Panxi area. 6. CONCLUSIONS Extremely low 3 He/ 4 He ratios are a widespread feature of the Fe Ti V oxide ore-bearing intrusions in Panxi area. This feature, combined with the variable 40 Ar/ 36 Ar ratios, is best explained by addition of subduction-related fluids and melts that have been stored in the lithospheric mantle for long periods. A paleo-subduction zone that formed in the western margin of Yangtze Block including the Panxi area during the Neoproterozoic period probably accounts for such a mantle enrichment process. Eclogite-derived melts from the deep-subducted oceanic slab extensively affected the lithospheric mantle. Partial melting of an upwelling mantle plume that involved an eclogite or pyroxenite component in the lithospheric mantle could have resulted in the parental Fe-rich magma that supplied the ore-bearing intrusions. The junction of these sources and processes may have been the crucial factor in generating many world-class Fe Ti V oxide deposits clustered in the Panxi area. ACKNOWLEDGEMENTS Constructive reviews and suggestions by Dr. Joyashish Thakurta, Dr. Franco Pirajno and an anonymous reviewer helped to improve the manuscript. The authors greatly acknowledge Dr. Tang, H-Y for helpful discussion and the management of the mining company of Panzhihua Iron & Steel Group for logistical support during fieldwork in the Panxi area. Sichuan Department of Land & Resources is thanked for their help. Financial support for this work was supported by Special Fund for Scientific Research in the Public Interest ( ), National Nature Science Foundation of China (Grant No ), 973 Project (Grant No. 2012CB and 2009CB421002), The Fundamental Research Funds for the Central Universities, 111 Project (B07011) and PCSIRT. REFERENCES Ali J. R., Fitton J. G. and Herzberg C. (2010) Emeishan large igneous province (SW China) and the mantle plume up-doming hypothesis. J. Geol. Soc. London 167, Bach W. and Niedermann S. (1998) Atmospheric noble gases in volcanic glasses from the southern Lau Basin: origin from the subducting slab?. Earth Planet Sci. Lett Ballentine C. J. and Barfod D. N. (2000) The origin of air-like noble gases in MORB and OIB. Earth Planet. Sci. Lett. 180, Baxter E. F. (2010) Diffusion of noble gases in minerals. Rev. Mineral Geochem. 72, Beswick A. E. (1982) Some geochemical aspects of alteration and genetic relations in komatiitic suites. In Komatiities (eds. N. T. Arndt and E. G. Nisbet). George Allen and Unwin, London, pp Bhat I., Zainuddin S. M. and Rais A. (1981) The Panjal Trap chemistry; witness to the birth of Tethys. Geol. Mag. 118, Boven A., Pasteels P., Punzalan L. E., Liu J., Luo X., Zhang W., Guo Z. and Hertogen J. (2002) 40 Ar/ 39 Ar geochronological