Introduction

While the details of the mechanism of Archean continental crust formation are still a matter of debate, it is clear that it involves partial melting of the upper mantle, resulting in depletion of FeO, CaO, Al2O3 and alkalis, and enrichment of MgO in the melt residue, which becomes the subcontinental lithospheric mantle (SCLM) (e.g., Jordan 1975, 1988; Boyd et al. 1985; Pollack 1986; Rudnick and Fountain 1995; King 2005). As a consequence of this depletion process, the residual SCLM has a lower density (about 3.0 g/cm3) than the asthenosphere (about 3.35 g/cm3), which helps to stabilize ancient cratonic lithosphere and explains why it behaves as a rigid plate in the scheme of global plate tectonics. This further implies that the structure of Archean cratons is such that Archean continental crust is underlain by Archean age lithospheric mantle, composed of highly refractory spinel-, garnet- to diamond-facies peridotite. This model for the structure of Archean cratons appears to be valid in southern Africa, as evidenced by mineral, geochemical and Re–Os isotopic studies on mantle peridotitic xenoliths from kimberlites of the Kaapvaal Craton (Richardson et al. 1984, 1985; Pearson et al. 1995; Pearson 1999).

While the lithospheric mantle beneath the Kaapvaal and Siberian cratons has been stable since their formation (Boyd 1989; Boyd et al. 1985, 1997; Pearson et al. 1995; King 2005), this is not the case for the North China Craton (Fig. 1). Here the thick and cold cratonic lithospheric mantle remained intact through the Paleozoic, but had become a thinner and hotter lithospheric mantle with oceanic affinity by the Cenozoic, as shown by data from mantle xenoliths entrained in the Paleozoic kimberlites and Cenozoic basalts (e.g., Fan and Menzies 1992; Griffin et al. 1992, 1998; Menzies et al. 1993; Menzies and Xu 1998; Zheng and Lu 1999; Fan et al. 2000; Xu 2001; Zheng et al. 2001; Rudnick et al. 2004). This means that a dramatic physico-chemical change occurred during the intervening time interval (e.g., Xu et al. 1998; Zheng et al. 1998; O’Reilly et al. 2001; Gao et al. 2002; Zhang et al. 2002, 2003, 2004; Wilde et al. 2003; Wu et al. 2003; Deng et al. 2004; Zhang 2005; Ying et al. 2006). The North China Cration thus offers an excellent opportunity to address the cause and mechanism of the lithospheric thinning, and the process for the compositional transformation or refertilization of the lithospheric mantle.

Fig. 1
figure 1

Simplified tectonic units of China and localities of diamondiferous and barren kimberlite fields (modified from Zhang et al. 1994). The North China Craton consists of two Archean nuclei, the Jiluliao nucleus in the east (the Eastern Block, Zhao et al. (2000)) and the Ordos nucleus in the west (the Western Block). The solid dash line extending roughly north-south direction shows the Daxing’anling-Taihangshan Gravity Lineamnet. Also shown are the Paleozoic kimberlite fields, Mesozoic basalts and Cenozoic basalts in eastern China mentioned in the text

Full size image

Mechanisms invoked for lithospheric thinning in eastern Asia include (1) delamination (Wu et al. 2003; Gao et al. 2004), (2) extension due to a mantle plume (Deng et al. 1994, 2004), and (3) collision between the Indian and Eurasian Plates along with subduction of the Paleo-Pacific Plate (Menzies et al. 1993). However, these explanations are not widely accepted, because (1) eclogite xenoliths are rare in the Paleozoic kimberlites and Mesozoic basalts; (2) Mesozoic plume-related magmatism has not been identified on the North China Craton; and (3) the collision between the Indian and Eurasian Plates occurred later than the Mesozoic (at ∼45 Ma). Based on the co-existence of high-Mg# and low-Mg# mantle xenoliths in the Cenozoic basalts from the Hebi region (Fig. 1) in Henan Province, Zheng et al. (2001) suggested that old refractory lithospheric mantle was replaced by young fertile compositions. Menzies and Xu (1998) and Xu (2001) noted the close relationship between the timing of lithospheric thinning and the collision of the North China Craton and the Yangtze Block. They argued that the collision changed the stress-state and stability of the lithospheric mantle, leading to thermal and mechanical erosion at the base of the lithosphere by the upwelling of hot asthenosphere. The widespread occurrence of isotopically enriched mafic rocks and carbonatites of Mesozoic age on the southeastern North China Craton (Fig. 1) demonstrate that subducted materials of the Yangtze lithosphere greatly contributed to generating enriched lithospheric mantle in this region during this time (Qiu et al. 2002; Zhang et al. 2002; Zhang and Sun 2002; Ying et al. 2004; Xu et al. 2004; Fan et al. 2004; Wang et al. 2005). The subduction also triggered alkaline magmatism, as reflected by syenites and monzonites occurring in a belt along the southern portion of the North China Craton (Zhang et al. 2005), generally parallel to the Qinling-Dabie orogenic belt. These observations led to the suggestion that subduction of the Yangtze lithosphere and subsequent collision with the North China Craton may have been the driving force for the rapid transformation of old refractory lithospheric mantle to fertile lithospheric mantle through melt-peridotite interaction during the Mesozoic, especially in the southern portion of the North China Craton (Zhang et al. 2002, 2007; Zhang and Sun 2002, Fan et al. 2004; Wang et al. 2005; Zhang 2005; Zheng et al. 2006a, b). However, these interpretations have not been widely accepted, and the key issue of the mechanism and dynamics of lithospheric thinning still remains controversial.

If the lithospheric mantle had an Archean age prior to the thinning, the Re–Os model ages (c.f. Shirey and Walker 1998) of the present-day lithospheric mantle can rigorously test the abovementioned models for lithospheric thinning. Delamination implies that the present-day lithospheric mantle should have young Re–Os model ages, since it has been newly accreted since the Paleozoic. If the thermo-mechanical erosion model is correct, the remaining lithospheric mantle should have Archean ages. As a third alternative, the compositional transformation model implies coexistence of both old and young ages. Until now, reliable Re–Os model age data for mantle peridotites of the North China Craton are rare, especially for peridotitic xenoliths from Paleozoic kimberlites (Gao et al. 2002; Wu et al. 2003, 2006). Pioneering work on two peridotitic xenoliths from the Fuxian diamondiferous kimberlites (Gao et al. 2002) indicates that the Archean lithospheric mantle existed beneath this region (Fig. 1) of the North China Craton. However, in the Mengyin kimberlites of the central Eastern Block of the North China Craton (Fig. 1) a peridotitic xenolith gives a much younger Mesoproterozoic “Re depletion age” (T RD determined by the time that the present-day 187Os/188Os of a sample matched the primitive mantle; c.f. Shirey and Walker 1998) of ∼1.46 Ga (Gao et al. 2002), while a chromite separate yielded an Archean T RD of 2.69 Ga (Wu et al. 2006). However, the complex origin of the chromite separate raises question about the implication of this “age”. Analyses of chromite separates from Fuxian and Tieling kimberlites (Fig. 1) give much younger Paleoproterozoic T RD of 1.56–1.77 Ga, at the same time giving old Paleoproterozoic to Archean “melt depletion ages” (T MA, determined by using the measured Re/Os ratio of the sample to calculate when the 187Os/188Os matched primitive upper mantle; c.f. Shirey and Walker 1998; this is completely analogous to the Sm-Nd model age) of 2.00–3.72 Ga (Wu et al. 2006). The Tieling kimberlites also contain highly serpentinized peridotite xenoliths that give T RD ages older than the chromites (1.9–2.3 Ga). However, the age of Tieling kimberlite emplacement is unknown, thus ages of the lithospheric mantle before the thinning are still not well constrained.

This study focuses on mantle peridotite xenoliths and xenocrysts from the early Paleozoic Mengyin and Fuxian diamondiferous kimberlites from the North China Craton, and mantle xenoliths from Paleozoic-age barren kimberlites and Mesozoic basalts from Dahongshan, Ningyuan, and Fangcheng on the South China and North China Cratons (Fig. 1). We report major and trace element chemistry, including PGEs, Re–Os isotope systematics, and Sr-Nd isotopes on samples representing all the localities. The purpose is to characterize the chemical and isotopic compositions of the lithospheric mantle beneath eastern China, to constrain the age of the lithospheric mantle that existed during the Paleozoic and Mesozoic, and to address the mechanism for lithospheric thinning.

Geological background

Eastern China comprises several tectonic units, from south to north, the South China Craton, the Qilian-Qinling-Dabie-Sulu Orogenic Belt, the North China Craton and the eastern Central Asia Orogenic Belt (Fig. 1). These units amalgamated during the Late Paleozoic–Early Mesozoic. A brief description of the tectonic evolution is given below.

South China Craton

The South China Craton comprises the Yangtze Block and the Cathaysia Block (Fig. 1). The Yangtze Block is one of the three old continents in China. Its basement formed dominantly in the Proterozoic (1.85–0.85 Ga) and is covered by Phanerozoic sedimentary rocks (Zhang et al. 1994). However, recent work on zircon U-Pb ages and Hf isotopes of gneiss xenoliths in lamproite diatremes from Proterozoic terrains indicates a widespread presence of Archean basement with zircon ages of 2.5–2.6 and 2.8–2.9 Ga and Hf model ages of 2.6–3.5 Ga (Zheng et al. 2006c). There are three kimberlite and/or lamproite fields: the Dahongshan barren kimberlite-lamproite field in Hubei Province, the Ningxiang diamond-bearing lamproite field in Hunan Province, and the Zhengyuan diamond-bearing lamproite field in Guizhou Province (Fig. 1). The Dahongshan kimberlite-lamproite field comprises ca. 100 small kimberlite or lamproite vents or dikes. These dikes intruded Sinian-Ordovician carbonates and shales (Liu and Zhao 1991). Ten irregular kimberlite vents outcrop near Pengjiabang village where the xenolith of this study was collected (Zhang et al. 2001). The kimberlite breccias are composed of porphyritic phlogopite and olivine macrocrysts and aphanitic matrix as well as wallrock fragments. The aphanitic kimberlite in this area gave a K-Ar whole-rock age of 326 Ma (Liu et al. 1993).

The Cathaysia Block (Fig. 1) comprises the Western and Eastern Cathaysia, whose boundary broadly corresponds to the boundary between Mesozoic S-type granites to the northwest and I-type granites to the southeast (Li et al. 2003). The oldest crustal component of western Cathaysia is represented by 2.5 Ga inherited zircons in Mesozoic granites (Li et al. 1989). The Cathaysia Block underwent a series of reworking events in the Neoproterozoic (Li et al. 1997, 1999, 2000; Chen and Jahn 1998; Chen et al. 1991). However, there are no reports of Paleozoic magmatism in this block.

In the Daoxian–Ningyuan region of Hunan Province (Fig. 1), Mesozoic basalts erupted at 151–131 Ma along the Ningyuan-Jianghua fault (Wang 1991). Some xenoliths in this study are from the Ningyuan basalt field, where, in the Anyuan area of Jiangxi Province, there exists more than 20 Eocene lamprophyre pipes. The pipes explosively intruded into Late Jurassic rhyolites (Zheng et al. 2004) and contain abundant peridotite xenoliths and megacrysts of augite, amphibole and phlogopite. Neogene xenolith-bearing basalts are widespread in the eastern Cathaysia, but rare in the western Cathaysia.

North China Craton

The North China Craton (Fig. 1) comprises Eastern and Western Blocks and the intervening Trans-North China Orogen (Zhao et al. 2000, 2001, 2005; Li et al. 2001; Gao et al. 2002; Wang et al. 2004). The basement of the Eastern Block consists primarily of Archean tonalitic, trondhjemitic and granodioritic (TTG) gneisses and granitoids, with rafts of Archean granitic gneisses and supracrustal rocks (Zhao et al. 2000, 2001). The Western Block is composed of Late Archean to Paleoproterozoic metasedimentary belts (Li et al. 2001; Zhao et al. 2000, 2005), which mainly consists of granulite-facies TTG gneisses and charnockites. The Trans-North China Orogen (Zhao et al. 2001; Gao et al. 2002), roughly north-south trending across the North China Craton, consists of a series of 2.5–2.7 Ga greenschist-amphibolite to granulite facies terrains (Zhao et al. 2000, 2001, 2005; Kröner et al. 1988). This Orogen was formed by the collision between the Eastern and Western Blocks at ∼1.8 Ga (Zhao et al. 2000, 2001), representing the final cratonization event.

The North China Craton experienced widespread tectono-thermal reactivation since the Phanerozoic, manifested by emplacement of Early Paleozoic kimberlites, extensive Late Mesozoic basaltic rocks and granites, and Cenozoic alkali basalts (e.g., Fan et al. 2000; Zhang et al. 2002, 2004; Wu et al. 2005). Ordovician diamondiferous kimberlites occur mainly in Mengyin County, Shandong Province, and Fuxian, Tieling and Huairen Counties, Liaoning Province (Fig. 1) (Dobbs et al. 1994; Chi and Lu 1996). Kimberlite-borne xenocrystic minerals and xenoliths having refractory compositions indicate cool geotherms (36–40 mW/m2) and thick lithosphere (∼200 km) at the time of emplacement (Menzies et al. 1993; Griffin et al. 1998). These features are characteristics of lithospheric mantle beneath Archean cratons worldwide and indicate that an Archean lithospheric keel existed beneath the Eastern Block at least until the Paleozoic (Zheng et al. 2006b). In contrast, mantle xenoliths from Cenozoic basalts record shallower and hotter lithospheric mantle, predominantly composed of fertile spinel peridotites (Fan et al. 2000; Rudnick et al. 2004), consistent with an average present-day surface heat flow of 60 mW/m2 (Hu et al. 2000) and thin lithosphere of 80–60 km beneath the Eastern Block (Yuan 1996; Griffin et al. 1998; Chen et al. 2006). This suggests that a considerable amount of Archean lithosphere has been lost or transformed to relatively young lithosphere.

Xenolith localities and petrology

The xenoliths investigated in this study are from five localities (Fig. 1). On the South China Craton they are (1) the Early Carboniferous Dahongshan kimberlite–lamproite field on the northern margin of the Yangtze Block (Liu and Zhao 1991; Zhang et al. 2001); (2) the Late Jurassic–Early Cretaceous Ningyuan-Diaoxian basaltic field on the Cathaysian Block. On the North China Cration they are (3) the Mid-Ordovician Fuxian kimberlite field (Chi and Lu 1996; Griffin et al. 1998; Zhang and Yang 2007); (4) the Mid-Ordovician Mengyin kimberlite field (Dobbs et al. 1994; Lu et al. 1998); and (5) the Early Cretaceous Fangcheng basaltic field (Zhang et al. 2002; Zhang and Sun 2002).

South China Craton

The xenolith from the Dahongshan kimberlite field (Fig. 1) is a fresh garnet lherzolite, composed of Ol (60%), Opx (15%), Cpx (15%), and Gt (10%). It has a granuloblastic texture, with some features of deformation and recrystallization, such as kink bands in olivine and orthopyroxene and triple junctions between olivine and pyroxene. Most of the constituent minerals are 0.1–3 mm across. Olivines often possess small fractures and orthopyroxenes are colorless with well-developed cleavages. A few orthopyroxenes show poikilitic textures and enclose olivine and clinopyroxene. Clinopyroxene is light emerald diopside, small in grain size, and homogeneously scattered. Garnets are relatively large (dominantly 1–5 mm) and red-purple. Some garnets have kelyphitic rims composed of tiny pyroxenes and spinels. This xenolith appears to represent the residue after low degree partial melting of the upper mantle, which was refertilized by later stage mantle metasomatism, as indicated by high proportions of garnet and enrichment of incompatible trace elements (Zhang et al. 2001).

The Ningyuan–Daoxian basaltic field (Fig. 1) contains abundant spinel facies peridotite and gabbro xenoliths and rare granulitic xenoliths (Wang et al. 1997). The peridotites in this study were collected from the Ningyuan basalts. All of these spinel-facies xenoliths are angular, with an average size of 3 cm across, and have relatively low olivine (50–60%) and high clinopyroxene (10–15%) contents. Most of the peridotites have a porphyroclastic microstructure (Zheng et al. 2004), indicating strong deformation of the sub-continental lithospheric mantle along the Ningyuan–Jianghua fault. The olivines and pyroxenes contain abundant fluid inclusions, providing direct evidence for the presence of significant volume of a fluid phase. Four porphyroclastic spinel lherzolites were selected for analyses (Table 1).

Table 1 Sample description of mantle xenoliths entrained in Paleozoic kimberlites and Mesozoic basalts on the North China Craton and South China Block
Full size table

North China Craton

Fuxian kimberlites (Fig. 1) erupted through the Archean (2.9–2.5 Ga) Anshan terrain (Chi and Lu 1996). The peridotite xenoliths examined here, all from Pipe 50, have coarse granular textures and are strongly serpentinized and carbonated, as reflected by high loss on ignition values (15.3–25.6 wt%) and high CaO contents (7–18 wt%). The freshest sample (L50–2003) is a garnet bearing peridotite containing kelyphitic pyrope (as high as 5%) and olivine relicts with Fo as high as 92 and Mg# up to 90.6. The more altered peridotites contain small amounts of visible sulfides (Table 1). These samples have relatively low Mg# (85–89) compared to typical coarse-granular peridotites from other Archean cratons (Mg# = 92–93) (Boyd 1989; Gao et al. 2002). Although the absence of primary mineral phases makes it impossible to determine P-T conditions, the presence of garnet in some xenoliths suggests minimum derivation from depths greater than 75 km based on typical cratonic geotherms (Menzies and Xu 1998).

Fuxian sample F50–9229 is a composite spinel lherzolite xenolith interlayered with garnet granulite composed of garnet (30%), pyroxene (15%), plagioclase (50%) and Ba-bearing feldspar (5%) (Zhang 2006). This composite xenolith is thus derived from the crust-mantle transition zone. L42–13 is a serpentinized olivine megacryst collected from Pipe 42. Chromite separate (F50–01) is from kimberlites of Pipe 50 and is studied for comparison.

Mengyin diamondiferous kimberlites (Fig. 1) erupted through the Archean to Paleoproterozoic (2.7–2.3 Ga) Taishan terrain (Chi and Lu 1996). The garnet-bearing peridotite xenoliths have porphyroclastic textures, while a spinel peridotite xenolith has a coarse-granular texture. All were collected from Pipe Shengli 1, which erupted at 460–465 Ma based on U-Pb dating of perovskite (Dobbs et al. 1994) and Ar-Ar dating of a phlogopite megacryst (Zhang and Yang 2007). Like the Fuxian xenoliths, these peridotites are strongly serpentinized, and no primary minerals are preserved. They have high Mg# (89.4–91.8, Table 2), but are lower than the peridotite (92.2) reported in Gao et al. (2002). There are no obvious differences in Mg# between the porphyroclastic and coarse-granular peridotites, distinctive from the porphyroclastic and coarse-granular peridotites from other cratons (Smith and Boyd 1987).

Table 2 Major oxide, trace element concentration, PGE and Au abundances of mantle xenoliths entrained in the Paleozoic kimberlites and Mesozoic basalts on the North China Craton and South China Craton
Full size table

An amphibole-bearing pyroxenite (Sv1b-16) from Mengyin is a macrocrystic xenolith with a cumulate texture. It is dominantly composed of coarse-grained (0.5–1.5 cm) clinopyroxene (90% augite) with minor amphibole and plagioclase, and represents a cumulate of ancient basaltic magma underplated in the crust-mantle boundary.

The Fangcheng basalts (Fig. 1) erupted at ∼125 Ma (Zhang et al. 2002) and contain spinel-facies ultramafic xenoliths, occurring as olivine clinopyroxenites and websterites, and range in size from 2–8 cm in diameter. They are mainly composed of clinopyroxene and orthopyroxene with minor olivine, spinel and plagioclase. Clinopyroxenes and sometimes orthopyroxenes in these pyroxenite xenoliths occur as large crystals (2–6 mm) and show cumulate textures (Zhang et al. 2007). Olivine, spinel, and occasionally feldspar occur as small interstitial grains among clinopyroxenes. Three pyroxenite xenoliths were selected for this study (Table 1).

Analytical methods

Major and trace element analyses

Whole rocks were powdered with an agate mill. Major oxides (Table 2) were obtained with a Phillips PW 2400 sequential X-ray fluorescence spectrometer at the University of Hong Kong (HKU). Fused glass disks were used for XRF major oxide determinations. The analytical precision is better than ±2%. Trace elements (Table 2) were determined with a VG Elemental Plasma-Quad 3 ICP-MS at HKU after normal sample digestion using ultra-pure acid (HF-HNO3). Detailed sample preparation procedures are available in Zhang et al. (2001). A blank solution was prepared and the total procedural blanks were <50 ng for all the trace elements reported in this paper. Two standards (BEN, BHVO-1) were prepared using the same procedure to monitor the analytical reproducibility. Discrepancies between triplicate analyses are less than ±5% for all elements.

Platinum group elements (PGE) and Au

Platinum group elements and Au (Table 2) were also measured with the ICP-MS at HKU after a pre-concentration by Ni-sulfide fire assay and the Te co-precipitation (Sun et al. 1993; Zhou 1994). Mass 108 (108Pd) and mass 105 (40Ar65Cu and 105Pd) were measured in order to evaluate the contribution of 40Ar65Cu, so that 63Cu40Ar interference with 103Rh could be evaluated. Standard materials WPR-1 and TDB-1 were analyzed and analytical uncertainties were better than ±15% for Ru and Rh, and ±10% for Ir, Pd and Pt (Zhou et al. 2000).

Sr-Nd isotopic determination

For Sr-Nd isotopic analyses, sample powders were spiked with mixed isotope tracers (87Rb-84Sr and 149Sm-150Nd), dissolved in Teflon capsules with HF+HNO3 acid, and separated by conventional cation-exchange techniques. Isotopic measurements were performed on the VG-354 mass-spectrometer at the Institute of Geology and Geophysics, Chinese Academy of Sciences. The mass fractionation corrections for Sr and Nd isotopic ratios were based on 86Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219, respectively. The detailed sample preparation and analytical procedures are reported in Zhang et al. (2001). Over the period of analytical work, repeat analyses yielded an 87Sr/86Sr ratio of 0.710230 ±40 for the NBS-987 Sr standard and a 143Nd/144Nd ratio of 0.511845 ± 12 for the La Jolla standard. Total blanks for Rb, Sr, Sm and Nd were less than 100 pg. The results are given in Table 3.

Table 3 Sr-Nd isotopic ratios of mantle xenoliths from the Paleozoic kimberlites and Mesozoic basalts on the North China Craton and South China Craton
Full size table

Re–Os isotopic measurements

Sample preparation and sparging procedures for Os isotope measurements were based on Ravizza and Pyle (1997), Hassler et al. (2000), and Schoenberg and Kramers (2000). Re and Os concentrations and Os isotopes were measured on a VG Axiom MC-ICPMS at the Lamont-Doherty Earth Observatory by static multi-collection.

For Re analyses, 1 g of sample powder and 185Re spike solution was weighed into a 60 ml Teflon screw cap beaker. Concentrated HF+HNO3 (24 ml) was added to dissolve silicates. The beaker was sealed and left on the hotplate at 120°C for 2 days. After drying, 2 ml of concentrated HNO3 was added expel HF, and dried down. Concentrated HCl (10 ml) was added and the sample remained on a hotplate at 120°C overnight. The clear solution was dried down carefully to avoid complete dryness and 2 ml of concentrated HNO3 was added to form soluble nitrates. The solution was dried down carefully and dissolved in 10 ml 0.5 N HNO3, then centrifuged to separate any solid residue. About 1 ml of AG1-X8 resin was slurried into HDPE columns for Re separation. The column was washed four times with 2 ml 4 N HNO3. After conditioning the resin twice with 1 ml 0.5 N HNO3 and once with 5 ml 0.5 N HNO3, the sample was loaded and subsequently rinsed three times with each 5 ml 0.5 N HNO3. Rhenium was eluted three times with 5 ml 4 N HNO3 and the solution was collected in a clean 20 ml Teflon screw cap vial. Prior to analysis, the solution was dried down carefully at 90°C and dissolved again in 40 μl 4 N HNO3. This solution was then transferred to a clean centrifuge vial and diluted to 1 ml with ultraclean water. Instrumental mass fractionation was monitored by measurement of a Re standard solution.

For Os isotopes, 5 g of sample powder was weighed into a clean ceramic crucible capped with a ceramic lid and spiked with 190Os. The crucible was placed on a warm (60°C) oven to dry the spike solution. The possibility of volatilization of the Os spike at this stage could result in some high estimates of Os concentrations. However, this would be manifested in model ages that are too old, which we do not see in the results. The dry powder and spike were thoroughly mixed with 20 g borax, 2 g nickel powder and 1 g of sulfur powder. This mixture was fused for an hour in a muffle furnace at 11,00°C. After cooling, the crucible was crushed, and NiS beads were collected and dissolved in 6 N HCl on a hotplate at 200°C, in an Erlenmeyer flask covered with a Teflon lid. The green-colored solution was filtered through 0.45-μm cellulose filter paper, and the insoluble residue with the Os was collected. The filter paper containing the Os concentrate was digested with 5 ml of 16 N nitric acid in an Os sparging apparatus connected to the MC-ICPMS. The oxidized OsO4 vapor was directly sparged into the torch by an Ar gas stream where it is ionized to Os+ and measured. Total procedural blanks did not exceed 2 pg for Re and 3 pg for Os. Blank corrections were made to all samples using an average Re blank of 1 pg and Os blank of 2 pg. The measured isotope ratios were corrected for mass fractionation using 189Os/188Os = 1.2212. Replicate analyses by sparging of an in-house standard solution from Spex averaged 187Os/188Os = 0.10638 ± 0.00077 (2σ external reproducibility, n = 22). This is half the error reported by Hassler et al. (2000) for sparge analyses using a Finnegan Element single collector instrument. The accuracy of the measurements is verified by comparison of analyses on sample F50–9270 (Table 4), performed in this study and by Gao et al. (2002), which gave identical 187Os/188Os ratios (0.11015 ± 11 and 0.11011 ± 11, respectively). However, replicate Re and Os concentrations vary beyond analytical uncertainty (2%), indicating heterogeneity of the whole rock or the so-called “nugget effect”, since the concentrations were measured on powder aliquots. In addition, imperfect dissolution of Re-rich phases and some volatilization of the Os spike during the digestion step could contribute to imprecision of the concentrations.

Table 4 Re–Os isotopic compositions of mantle xenoliths from the Paleozoic kimberlites and Mesozoic basalts on the North China Craton and South China Craton
Full size table

Analytical results

Dahongshan garnet lherzolite xenolith

The Dahongshan garnet lherzolite xenolith has relatively high Al2O3 (2.7%) and CaO (2.8%), higher than average global kimberlite-borne garnet lherzolites from old cratonic lithospheric mantle (∼1.3 and 0.8%, respectively; McDonough 1990). It shows a V-shaped REE pattern (Figs. 2, 3), that is, with low MREEs compared to LREE and HREE. PGE and Os concentrations are very close to primitive mantle abundances (Barnes et al. 1988) but shows enrichment of Pd and Au (Fig. 4). The Pd/Ir ratio (2.55) is higher than that for garnet lherzolite from Matsoku kimberlites (typically <1.0; Mitchell and Keays 1981), which is again indicative of a less refractory composition.

Fig. 2
figure 2

Chondrite-normalized trace element patterns for the xenoliths from Paleozoic kimberlites and Mesozoic basalts, eastern China. Chondrite values are from Anders and Grevesse (1989)

Full size image
Fig. 3
figure 3

Primitive mantle-normalized trace element spidergrams for the xenoliths from Paleozoic kimberlites and Mesozoic basalts, eastern China. Primitive mantle values are from McDonough and Sun (1995). Ti concentration is calculated from the TiO2 content

Full size image
Fig. 4
figure 4

Primitive mantle-normalized Os, PGE and Au diagrams for the xenoliths and xenocrysts from Paleozoic kimberlites and Mesozoic basalts, eastern China. Primitive mantle values are from Barnes et al. (1988): Os 4.2 ppb, Ir 4.4 ppb, Ru 5.6 ppb, Rh 1.6 ppb, Pt 8.3 ppb, Pd 4.4 ppb, and Au 1.4 ppb. Solid line and dash line represent the spinel peridotite and garnet peridotite, respectively

Full size image

Sr and Nd isotopic ratios are slightly more “enriched” than bulk silicate Earth (Table 3 and Fig. 5). Zhang et al. (2001) show that regression of whole rock and cpx, gt, opx mineral separates yields a line in the Rb-Sr isotopic diagram corresponding to 519 Ma. This sample has relatively high Re (0.3 ppb) and Os (4 ppb) (Table 4) close to the primitive mantle values (Fig. 6) and 187Os/188Os slightly lower than primitive upper mantle (PUM = 0.1296). It gives a young Re depletion age (T RD = 1.04 Ga), reflecting its high 187Os/188Os (∼0.1240), and an unrealistic melt depletion age (T MA = 4.7 Ga, older than the age of the Earth), reflecting the high 187Re/188Os (∼0.36) (Table 4).

Fig. 5
figure 5

Sr and Nd isotopic diagram for the xenoliths from Paleozoic kimberlites and Mesozoic basalts, eastern China. εNd(t) and εSr(t) are the deviations from the Chondrite Uniform Reservoir values in parts per 104 at the time of eruption. Data of Fuxian and Mengyin kimberlites (Chi and Lu 1996; Tompkins et al. 1999; Zhang and Yang 2007) and global kimberlites and lamproites (Hawkesworth et al. 1990) are plotted for comparison

Full size image
Fig. 6
figure 6

Histograms of Os (a) and Re (b) concentrations for the xenoliths from Paleozoic kimberlites and Mesozoic basalts, eastern China. Data for Paleozoic kimberlite-borne xenoliths from Gao et al. (2002) are included. “P.M.” denotes primitive mantle values from McDonough and Sun (1995)

Full size image

Ningyuan spinel lherzolite xenoliths

The Ningyuan spinel lherzolite xenoliths show fertile compositions (Table 2; e.g., CaO = 3.5–4.0%, Al2O3 = 3.2∼3.6%), similar to PUM (∼3.3% and 4.1%, respectively; Hart and Zindler 1986), and Mg#s are also low (88–89). They show nearly flat REE patterns (Fig. 2) with a slightly LREE enrichment in sample NY-G2 and REE concentrations slightly lower than average primitive mantle (Fig. 3). They are enriched in Cs and depleted in Ba, but do not show HFSE anomalies (Fig. 3). Rb and Sr abundances are close to the value of the primitive mantle, as are PGE and Os concentrations (Barnes et al. 1988) although again Pd is enriched along with Au (Pd/Ir = 2–2.53, Table 2 and Fig. 4).

The Ningyuan xenoliths have high εNd(t) (11–13.5) and Sr isotope ratios slightly higher than bulk Earth (εSr(t) = 6.6–9.2) at the time of basalt eruption (131–151 Ma, Wang (1991); an intermediate value of 141 Ma is used for the correction). Os concentrations (Fig. 6) are close to primitive mantle abundances (2–3.8 ppb; Table 4), but there is a large range in Re abundances (0.12–0.98 ppb). All samples show nearly identical 187Os/188Os ratios (Fig. 7b), similar to the Dahongshan garnet lherzolite. Three xenoliths give T RD ages of ∼1 Ga like the Danhongshan sample (Table 4; Fig. 7a); one sample gives a younger T RD of ∼0.7 Ga. The xenoliths yielded variable T MA ages of 2.3–3.3 Ga. The T MA age of one sample is negative, due to its very high 187Re/188Os ratio (∼1.2 compared to ∼0.3 in the other samples, Table 4).

Fig. 7
figure 7

Histograms of Re-depletion (T RD) model ages in billions of years (a) and initial 187Os/188Os (b) for the xenoliths at the time of host magma eruption. PUM is primitive upper mantle composition of Meisel et al. (2001). Data for Paleozoic kimberlite-borne xenoliths from Gao et al. (2002) are included

Full size image

Fuxian peridotitic xenoliths and xenocrysts

Peridotites

Garnet and spinel peridotites from the Fuxian kimberlite field are strongly serpentinized and carbonatized, except for one relatively fresh sample (L50–2003; Table 2), which is a garnet-bearing peridotite containing kelyphitic pyropes and olivine relicts with Fo as high as 92. The highly serpentinized peridotites have very high REE concentrations and highly fractionated LREE-enriched patterns with HFSE depletions (Figs. 2, 3). The relatively fresh sample L50–2003 also shows LREE-enrichment and extremely high LREE/HREE (LaN/YbN = 474), despite lower REE abundances than the highly serpentinized samples (Fig. 2). All the peridotite xenoliths from Fuxian kimberlites display very similar and “sinusoidal” primitive mantle-normalized Os-PGE-Au patterns (Fig. 4), with a positive slope from Os to Ru, a negative slope from Ru to Pt, and a positive slope from Pt to Au. Sample L50–9281 has much higher PGE abundances than the others (Fig. 4), consistent with the presence of sulphides (Table 1).

Fuxian peridotitic xenoliths show εNd(t) slightly lower than the bulk Earth (-1.7 to -4.2) and extremely high initial Sr isotope ratios (εSr(t) = 122–223; Table 3 and Fig. 5), calculated to a kimberlite emplacement age of 465 Ma (Zhang and Yang 2007). The high Sr likely reflects alteration by a fluid with high 87Sr/86Sr. The observation that the peridotites have the same initial Nd isotopic ratios but much higher initial Sr isotopic ratios than the host kimberlites (Fig. 5) indicates that the alteration was not associated with the kimberlite emplacement but likely occurred in the lithospheric mantle.

The peridotite xenoliths have variable Os (1–6 ppb) and very low Re (0.01–0.12 ppb; Fig. 6), compatible with the Re–Os data of Fuxian peridotite xenoliths of Gao et al. (2002). The garnet lherzolites have slightly higher Re and lower Os than the spinel peridotites (Fig. 6). Nevertheless, all of them have similar 187Re/188Os (∼0.11) and 187Os/188Os ratios (0.106–0.116, γOs = -9.3 to -16.0; Fig. 7b). They have T RD ages of 2.0–3.2 Ga, similar to the Fuxian chromite separate (Fig. 8a). They also yield very similar T MA ages of 3.0–3.4 Ga (Table 4), slightly older than the T RD ages. On a Re–Os isochron diagram, even if sample LN-9830 is excluded, the peridotite xenoliths form a positive correlation yielding an errorchron of 3.1 ± 1.5 Ga (Fig. 8b). There are no apparent differences between the garnet and spinel peridotites in terms of their Os isotopic compositions and PGE abundances.

Fig. 8
figure 8

a187Re/188Os versus 187Os/188Os for the xenoliths from Paleozoic kimberlites and Mesozoic basalts with the Re–Os errorchrons for the xenoliths from Fuxian and Mengyin Paleozoic kimberlites. b A magnification of the rectangle area in a. PUM is primitive upper mantle composition of Meisel et al. (2001). Shaded field represents range of 187Os/188Os in abyssal peridotites (Brandon et al. 2001 and references therein). Data for Paleozoic kimberlite-borne xenoliths from Gao et al. (2002) are included. Data regressed and MSWD determined using Isoplot program of Ludwig (2003)

Full size image

Composite xenolith

A composite xenolith (F50–9229) is a spinel lherzolite interlayered with garnet granulite (c.f. Zhang 2006). It has low Mg# (77) and high Al2O3 (∼17%, vs. 0.7–2.2%) relative to Fuxian peridotite xenoliths (Table 2). It is LREE-enriched and shows a positive Eu anomaly (Fig. 2). It is extremely enriched in some LILE (e.g., Cs = 12.3 ppm, Ba = 717 ppm, Sr = 350 ppm). PGE abundances are extremely low, and IPGE (Os, Ir, Ru) are depleted relative to the PPGE (Rh, Pt, Pd) such that Pd/Ir = 13 (Fig. 4). It has slightly higher Nd and much lower Sr isotope ratios than Fuxian peridotites (Fig. 5). The presence of garnet granulite rather than eclogite likely reflects the conditions of the cratonic geotherm beneath the Fuxian region of the North China Craton during the Paleozoic.

Olivine megacryst

An elliptoid and egg-sized olivine xenocryst (L42–13) is highly serpentinized. The olivine has very low PGE contents and IPGE are depleted relative to PPGE (Pd/Ir = 7.15, Fig. 4). It has extremely low Re and Os concentrations (0.04 and 0.05 ppb, respectively, Table 4 and Fig. 6), but high 187Re/188Os compared with the peridotites (>3 vs. <0.21), as well as high 187Os/188Os (∼0.14 vs. ∼0.11).

Chromite separate

Chromite separate F50–01 has very high PGE contents (for instance, Ir = 52.2 ppb, more than an order of magnitude higher than the next enriched Fuxian sample) and enrichment of IPGE relative to the PPGE (Pd/Ir = 0.11), but also low Os and high Au (Fig. 4). The Re and Os abundances and 187Os/188Os are generally consistent with those reported by Wu et al. (2006) for chromite separates from the same kimberlite field. The fact that chromites separated from the Fuxian kimberlites have generally higher Re and lower Os than typically found in chromites from ophiolites (Walker et al. 2002) indicates that these might be disaggregated from the lithospheric mantle peridotites, and thus represent the litospheric mantle. If true, then chromite Os isotopic data can be used to constrain the age of the lithospheric mantle. The chromite separate studied here yields a T RD age of 2.83 Ga and falls right in the middle of ages for the North China Craton (Table 4, Fig. 7a). If this chromite separate and the olivine xenocryst plotted in a Re–Os isochron diagram with the peridotites, all the Fuxian data form an errorchron of 0.53 ± 0.11 Ga with an initial 187Os/188Os ratio of 0.1086 (Fig. 8a). Although we do not put any emphasis on any implications of this age, we note that the initial Os isotopic ratio gives a PUM model age of 2.9 Ga, similar to the T RD ages given by the chromite separate as well as the peridotites.

Mengyin peridotitic and pyroxenite xenoliths

Peridotites

Garnet (SD-9804, SL-2001, Sv1b-12) and spinel (Sv1b-11) peridotites from the Mengyin kimberlite field are highly serpentinized, but not as strongly as those from the Fuxian kimberlite field, as shown by lower loss on ignition (13–14%). Unlike the Fuxian peridotites, those from Mengyin are not strongly carbonatized, as indicated by low CaO (<1.2%; Table 2). They have high REE abundances and highly fractionated LREE-enriched patterns (LaN/YbN = 150–369; Fig. 2). Also unlike the Fuxian xenoliths, they are slightly depleted in HFSE such as Ti, Zr, Hf, Ta (Fig. 3). The three garnet peridotitic xenoliths from the Mengyin kimberlites display similar U-shaped primitive mantle-normalized Os-PGE-Au patterns (Fig. 4), with a negative slope from Os to Pt and a positive slope from Pt to Au. The spinel peridotitic xenolith shows a sinusoidal pattern, similar to the Fuxian peridotites. The PGE concentrations are similar to those in Fuxian peridotites.

Like the Fuxian peridotites, the Mengyin peridotites show a small range of εNd(t) (−0.6 to 0.6), in this case close to bulk Earth values. The three garnet peridotites have higher Sr isotope ratios (εSr(t) = 50–60) than the spinel peridotite (εSr(t) = 30), calculated to a kimberlite emplacement age of 465 Ma (Table 3, Fig. 5).

The Mengyin peridotite xenoliths have variable Os (1.5–5.2 ppb) concentrations, similar to Fuxian peridotites (Table 4, Fig. 6). However, Mengyin peridotites have slightly higher Re (0.05–0.28 ppb). The three garnet peridotites have much lower Re (<0.06 ppb) than the spinel peridotite (>0.11 ppb, Table 4, Fig. 6), a distinction not seen in the Fuxian garnet and spinel peridotites. All the Mengyin peridotite xenoliths possess nearly identical initial 187Os/188Os (0.110–0.115, γOs = −10 to −12.4), and give similar T RD ages of 2.2–2.7 Ga, typical of North China Craton xenoliths, but with a slightly lower average value than the Fuxian peridotites. These initial 187Os/188Os values are lower than the Mengyin sample analyzed by Gao et al. (2002) (∼0.118, SD-9405, listed in Table 4), which also gives a younger T RD age (∼1.5 Ga) and a future T MA age. Peridotites from both the Mengyin and Fuxian kimberlites have similar T MA ages, averaging ∼3.0 Ga. The Mengyin peridotitic xenoliths form an errorchron on a Re–Os isochron diagram with an age of 1.66 ± 0.64 Ga (Fig. 8b). Though its meaning is unclear, the initial Os ratio gives a model age of 2.98 Ga, consistent with the T MA ages. Although the Mengyin garnet and spinel peridotites show differences in PGE and Re, Os contents, the Os isotopic compositions and model ages do not show apparent differences.

Pyroxenite

A macrocrystic pyroxenite xenolith (Sv1b-16) from the Mengyin kimberlite has low MgO (14 vs. >20%) and is more enriched in Al2O3, CaO and K2O (Table 2) than pyroxenite xenoliths from Mesozoic Fangcheng basalts, reflecting its mineral assemblage with some amphibole and plagioclase. It is LREE-enriched (LaN/YbN ∼47.6) and has very high REE but has low HFSE and Sr (Figs. 2, 3). While Nb and Ta are lower than the peridotites, Sr is higher.

The pyroxenite shows “EM1-like” Sr-Nd isotopes (Fig. 5), that is, a very low Nd isotope ratio (εNd(t) = −26) and at the same time low initial 87Sr/86Sr (εSr(t) ∼11.3). It has low Os (0.133 ppb) and extremely high 187Re/188Os (2.4). Nevertheless, it has a high initial 187Os/188Os (0.154, γOs = 22, Fig. 7b), thus is completely different from the Re–Os isotopes of the peridotites (Table 4). It has a future T RD age and a T MA age of 1.32 Ga. It should be noted that this T MA age overlaps the Re–Os errorchron age of the Mengyin peridotites, perhaps indicating that this pyroxenite is a sample of the melts that caused metasomatism of the lithospheric peridotites.

Fangcheng pyroxenite xenoliths

As noted above, the pyroxenite xenoliths from the Fangcheng basalts have higher MgO and lower CaO and Al2O3 then the Mengyin pyroxenite (Table 2). They also have similarly LREE-enriched patterns, and low HFSE and LILE (Table 2 and Figs. 2, 3).

The Fangcheng pyroxenites have enriched initial Sr-Nd isotope ratios, corrected to 125 Ma (εSr(t) = 76–81 and εNd(t) ∼ −12; Fig. 5). These Sr-Nd isotopic ratios are very similar to the host basalts (εSr(t) = 75 to 82; εNd(t) = −12.8 to −14.4; Zhang et al. 2002). These pyroxenites show extremely large variations in Os abundances (0.06–2.36 ppb), 187Re/188Os (0.08∼3.54) and initial 187Os/188Os (0.12–0.16) (Table 4; Figs. 6, 7, 8) though the Re abundances are relatively constant, leading to exceedingly large γOs (−6.4 to 24). The currently available Fangcheng Re–Os data are not as systematic as the other localities, and less easy to interpret, and therefore are not used in the following discussion.

Discussion

Ancient SCLM during the Paleozoic beneath the eastern North China Craton

The Re–Os data for garnet and spinel peridotitic xenoliths from the Fuxian and Mengyin diamondiferous kimberlites document the presence of Archean lithospheric mantle beneath the eastern North China Craton during Ordovician time. Two Fuxian xenoliths in the pioneering work of Gao et al. (2002) are clearly Archean, with T RD and T MA ages greater than 2.5 Ga, suggesting that the SCLM beneath the Fuxian region was formed in the Archean. Their conclusion is further confirmed by this study on additional peridotitic xenoliths from the same kimberlite field, which dominantly give T RD ages of 2.7–3.2 Ga and the T MA ages of 2.9–3.4 Ga except for one sample (L50–2003 for the T RD age and LN-9830 for the T MA age; Table 4, Fig. 7a). A chromite separate from the Fuxian kimberlites also gives an Archean T RD age of ∼2.8 Ga (Table 4). The low Re in these xenoliths may be a primary feature and suggests that intensive serpentinization did little to increase the Re/Os ratio or alter 187Os/188Os of peridotitic xenoliths. The peridotite xenoliths from the Mengyin kimberlites analyzed in this study all yield effectively the same T RD ages of ∼2.6 Ga and T MA ages of 3.0–3.3 Ga (Table 4; Fig. 7a). This is consistent with an Archean T RD age of 2.69 Ga of a chromite separate from Mengyin kimberlites by Wu et al. (2006). However, a Mengyin peridotite analysed by Gao et al. (2002) has significantly higher Re/Os and 187Os/188Os and a younger T RD age of ∼1.5 Ga. This sample also has a negative T MA age, indicating that it experienced Re addition after the original melt depletion event. If we assume that melt depletion occurred in the Archean (based on our data), then the Proterozoic Re depletion age may indicate Re addition during the Proterozoic, perhaps at around 1.5 Ga. This is consistent with the Proterozoic melt depletion age of the pyroxenite xenolith from the Mengyin kimberlite (T MA age = 1.32 Ga, Table 4). Ages of ∼1.4 Ga can also be found in peridotites from the Dabie-Sulu orogen (Zheng et al. 2006a, d). The 1.6 Ga Re–Os errorchron for the Mengyin peridotites (Fig. 8) also supports the idea that something important happened to these rocks in the mid-Proterozoic. In addition, the SCLM beneath the northern margin of the eastern North China Craton might have slightly younger ages as the mantle peridotitic xenoliths from the Tieling kimberlites give the T RD ages of Paleoproterozoic (1.8–2.3 Ga) (Wu et al. 2006).

Overall, the Re–Os data demonstrate that SCLM beneath the central part of the eastern North China Craton had Archean and Paleoproterozoic ages during the Paleozoic. Moreover this is the case for both shallow and the deeper lithospheric mantle, as the ancient ages are observed in both garnet and spinel facies xenoliths. However, this Archean lithospheric mantle was later modified by melts. The best evidence for melt modification of the Archean lithosphere comes from the overlapping mid-Proterozoic Re–Os errorchron of the Mengyin peridotites and the 1.32 Ga T MA age of the Mengyin pyroxenite (Fig. 8).

Compositional stratification in the SCLM before lithospheric thinning?

Compositional distinctions of mantle peridotite xenoliths sampled by the Paleozoic kimberlites, (such as forsterite compositions of olivines) compared with those in the Cenozoic basalts, have raised the question of whether the SCLM beneath the eastern North China Craton was compositionally stratified before it was thinned. For example, one might argue that the highly refractory garnet peridotitie xenoliths sampled by the Paleozoic kimberlites (Fo > 92; Zheng et al. 2006b) were derived from deeper mantle that was completely lost during lithosphere thinning. That is, the relatively fertile spinel peridotite xenoliths from the widespread Cenozoic basalts of the North China Craton (generally Fo ≤ 92; Fan et al. 2000; Rudnick et al. 2004) are from shallower mantle that may be compositionally different from the deeper garnet-facies mantle. The few spinel-bearing peridotitic xenoliths entrained in Paleozoic kimberlites have high Fo contents in olivines similar to those from the garnet peridotitic xenoliths (Zheng et al. 2006b), suggesting that there were no obvious compositional differences between the garnet- and spinel-facies SCLM before the lithospheric thinning. This conclusion is supported by the PGEs and Re–Os isotopic systematics of the garnet and spinel peridotite xenoliths in this study.

Both the garnet and spinel peridotitic xenoliths from the Fuxian kimberlites display similar primitive mantle-normalized Os-PGE-Au patterns (Fig. 4). They have very similar initial 187Os/188Os (0.106–0.116, Table 4; Fig. 7b), although the garnet lherzolites have slightly higher Re and lower Os than the spinel peridotites (Table 4; Fig. 6). All these peridotite xenoliths give similar (Archean) T RD ages of 2.7–3.2 Ga and T MA ages of 3.0–3.4 Ga (Table 4; Fig. 7a). Thus, there are no apparent differences in Os isotopes or PGE concentrations between the garnet and spinel peridotites in the Fuxian locality.

From Mengyin, three garnet peridotitic xenoliths display similar U-shaped primitive mantle-normalized Os-PGE-Au patterns (Fig. 4), slightly different from the sinusoidal pattern for a spinel peridotite. The spinel peridotite shows lower εSr(t) than the garnet peridotites (Fig. 5). The slight differences may be produced by Proterozoic metasomatism in the garnet-facies lithospheric mantle. More significant is that all the garnet and spinel peridotites show effectively the same initial 187Os/188Os and the similar (Archean) T RD (∼2.7 Ga) and T MA (3.0–3.3 Ga) ages (Table 4; Fig. 7a).

We conclude that apparent compositional stratification of the SCLM before its thinning is not favored by the Fo numbers, PGEs and Re–Os isotopic data. This is also the case for the SCLM beneath the Archean Kaapvaal Craton (Pearson et al. 1995, 1999).

Mechanisms for lithosphere thinning and its transformation

Re–Os isotopic data of mantle xenoliths can be used to test models of lithospheric thinning. Our data indicate that the lithospheric mantle of the North China Craton had ages as old as Archean prior to kimberlite emplacement during the Paleozoic, and lacked compositional stratification in garnet and spinel facies mantle. The delamination model of Wu et al. (2003) and Gao et al. (2004) suggests that the present-day lithospheric mantle should have young Phanerozoic ages, since it must be newly accreted since the thinning. On the other hand, the thermo-mechanical erosion model of Xu (2001) implies an Archean age for the remaining lithospheric mantle. Compositional transformation through peridotite-melt reaction as suggested by Zhang (2005) implies a lithospheric mantle with ages from Archean to Phanerozoic, as this process could reset the ages of parts of the lithosphere.

Re–Os data have been published for Cenozoic basalt-hosted peridotitic xenoliths from different localities of the North China Craton (Fig. 1). In the central zone, peridotite xenoliths from Hannuoba give a Paleoproterozoic Re–Os isochron age of 1.91 ± 0.22 Ga (Gao et al. 2002), while the crustal age is Archean. Mantle xenoliths from Fanshi, about 200 km to the south give Archean T RD ages (Rudnick et al. 2006). These data are consistent with the suggestion by Gao et al. (2002) that the lithosphere beneath Hannuoba formed in the Archean, but was replaced in the Paleoproterozoic. In the eastern North China Craton, peridotite xenoliths from Qixia give predominantly Phanerozoic T RD ages (0.22–0.75 Ga), although the sample having the highest Fo in olivine and Cr# in spinel gives an older age (T RD = 1.32 Ga; Gao et al. 2002) (Fig. 9), and those from Kuandian show a large T RD age range from Paleoproterozoic to Tertiary (T RD = 0.3–2.18 Ga; Wu et al. 2006) (Fig. 9). Although 187Os/188Os ratios do not correlate with melt depletion or fertility indicators such as Al2O3 in the Kuandian xenoliths, the two peridotites with the highest Fo in olivines and Cr# in spinels give the oldest T RD ages (Wu et al. 2006), implying that older Re–Os ages may generally correlate with the Fo contents in olivines. On the northern and southern margins of the North China Craton at Longgang and Nushan, respectively, Re–Os ages in mantle xenoliths are Proterozoic to Phanerozoic (Wu et al. 2003; Zhi and Qin 2004) (Fig. 9).

Fig. 9
figure 9

Emplacement age (Ma) of host magma versus Re depletion age (Ga) of mantle xenoliths from Paleozoic kimberlites and Mesozoic and Cenozoic basalts from the North China and South China Cratons. Re–Os data of mantle xenoliths from the Cenozoic basalts on the North China Craton are collected from Zhi et al. (2001), Gao et al. (2002), Wu et al. (2003, 2006), Xia et al. (2004), Zhi and Qin (2004). All the T RD ages were re-calculated from the original data relative to the value of PUM: 187Os/188Os = 0.1296 and 187Re/188Os = 0.433 (Brandon et al. 2001)

Full size image

These studies show that the present-day lithospheric mantle beneath the North China Craton has a mixture of ages, from Archean or Paleoproterozoic to Phanerozoic (Fig. 9). Therefore, we can conclude that the Archean (and Proterozoic) mantle peridotitic xenoliths represent the remnants of thinned Archean lithospheric mantle, and the Phanerozoic ones represent the lithospheric mantle that is newly accreted after the thinning, or formed through interaction with the massive addition of melts during the Mesozoic (and Cenozoic). The latter case would imply that melt addition transformed highly refractory peridotites to fertile ones, and also introduced Re into peridotites, which would lower the Re–Os isotopic ages. Evidence for melt circulation in the lithospheric mantle and interaction with peridotites has been identified in xenocrysts and pyroxenite xenoliths from the Mesozoic Fangcheng basalts (Zhang et al. 2007) as well as in composite xenoliths from the Cenozoic Hannuoba basalts (Liu et al. 2005). The pyroxenite xenoliths in this study (Table 4) show variable 187Re/188Os and high 187Os/188Os ratios, and these characteristics could be the products of the melt circulation. The evidence suggests that melt infiltration plays an important role in the transformation of the lithospheric mantle.

Our conclusion that the Archean refractory lithospheric mantle had been partially refertilized to Phanerozoic fertile lithospheric mantle is also supported by the variable mineral compositions, such as Fo 93–83 values in olivines, found to be widespread in mantle peridotite xenoliths and xenocrysts from Mesozoic and Cenozoic basalts of the North China Cration (Song and Frey 1989; Fan et al. 2000; Zheng et al. 1998, 2001; Yan et al. 2003; Rudnick et al. 2004; Zhang 2005; Tang et al. 2004; Ying et al. 2006; Wu et al. 2006). The xenoliths with Archean T RD ages from the Mengyin and Fuxian kimberlites still have olivine relicts with Fo as high as 92–94. However, a compilation of published data (Zhang et al. 2006) shows that such high Fo in olivines is rare in mantle peridotitic xenoliths or xenocrysts from Mesozoic and Cenozoic basalts on the North China Craton, except for the Hebi (Zheng et al. 2001). This also implies that ancient refractory lithospheric mantle was transformed to young fertile mantle through multiple events of peridotite-melt interaction. This could be the result of decompressional melting during upwelling of asthenosphere in association with lithospheric thinning.

Neoproterozoic Re–Os isotopic resetting of the Yangtze Block

The South China Craton, like the North China Craton, has Archean crust, particularly in the Yangtze Block. Recent work on zircon U-Pb ages and Hf isotopes of gneiss xenoliths from early Carboniferous lamproite/kimberlite diatremes suggest the widespread presence of Archean basement with zircon ages of 2.5–2.6 and 2.8–2.9 Ga and Hf model ages of 2.6 to 3.5 Ga (Zheng et al. 2006c). This implies that the lithospheric mantle beneath the craton originally formed in the Archean. However, the only garnet peridotitic xenolith analyzed from the Yangtze Block (from Dahongshan, Fig. 1) gives a much younger Neoproterozoic T RD age of 1.04 Ga (Table 4). The xenolith has relatively high Re (0.3 ppb) and Os (4 ppb), close to primitive mantle values (Fig. 6). Its initial 187Os/188Os, slightly lower than PUM (Fig. 7b), falls in the range of global abyssal peridotites (Brandon et al. 2001). Interestingly, spinel peridotite xenoliths from the Mesozoic Ningyuan basalts have very similar Os (2–3.8 ppb) concentrations and initial 187Os/188Os ratios (Table 4) to the Dahongshan garnet lherzolite, despite a large range of Re concentrations (0.12–0.98 ppb) and 187Re/188Os (0.3–1.2). Three of the Ningyuan xenoliths also show very similar Neoproterozoic T RD ages of ∼1 Ga and variable T MA ages of 2.3–3.3 Ga (Table 4). This may indicate that the Archean lithospheric mantle beneath the South China Craton has undergone considerable Re–Os isotopic resetting since its formation, perhaps during the Neoproterozoic amalgamation of the Cathaysia Block in the south and the Qinling-Dabie belt in the north. It has been suggested that extensive Neoproterozoic magmatism along the southeastern margin of the Yangtze Block was associated with the Jiangnan island arc, and the adjacent ocean existed as early as ∼1.0 Ga (Zhou et al. 2002 and references therein). The mafic protolith for the Dabie eclogites, with positive ɛHf(t) values, give depleted mantle Hf model ages of ∼0.8 Ga (Zheng et al. 2006e). These Hf model ages are similar to ages for bimodal magmatism in the periphery of the Yangtze Block, and indicate new addition of juvenile continental crust in the region at that time (Li et al. 2003; Zheng et al. 2006c, e).

Lithospheric reworking is also indicated by the petrological and geochemical characteristics of the Dahongshan xenolith (Zhang et al. 2001). Its high Al2O3 and CaO contents, compared with global kimberlite-borne garnet lherzolites from Archean cratons (McDonough 1990), indicate a fertile mantle source. The V-shaped REE pattern (Fig. 2) is also considerably different from the general LREE-enriched patterns for garnet peridotite xenoliths of global Archean cratons (McDonough 1990). The PGE concentrations similar to primitive mantle (Fig. 4) are also indicative of less depletion. The enrichments in the PPGE and Re over the IPGE for the Dahongshan and Ningyuan peridotites provide additional evidence for the refertilization of the lithosphere through the mantle metasomatism (Fig. 4). Together, these observations suggest that lithospheric mantle refertilization resulted in considerable resetting of the PGE features and Re–Os isotopic systems.

Conclusions

We draw the following primary conclusions from this study.

  1. (1)

    Re–Os isotopes in garnet and spinel peridotite xenoliths hosted by the Mengyin and Fuxian diamondiferous kimberlites, emplaced during the early Paleozoic in the North China Craton, further confirm that SCLM of Archean age existed beneath the North China Craton during the Paleozoic.

  2. (2)

    Comparison of spinel and garnet facies peridotite xenoliths indicates an absence of compositional stratification in the Paleozoic lithospheric mantle.

  3. (3)

    The Re–Os isotopic features of mantle xenoliths from the Paleozoic kimberlites of the North China Craton are different from peridotitic and pyroxenite xenoliths from Cenozoic and Mesozoic basalts, which show younger Re–Os model ages. The present-day lithospheric mantle beneath this craton has both ancient and recent Re–Os ages.

  4. (4)

    In the South China Craton, there is a decoupling of ages between the crust and the underlying SCLM in the Yangtze Block, with mantle xenoliths emplaced during the Paleozoic and Mesozoic showing Neoproterozoic T RD ages, suggesting Neoproterozoic resetting of the SCLM beneath the Yangtze Block.

  5. (5)

    The large range of Re–Os isotopic ages in Phanerozoic xenoliths indicate that lithospheric transformation through melt-peridotite interaction could be a more important mechanism for compositional changes during the Phanerozoic than delamination or thermo-mechanical erosion.