Re–Os and S systematics of spinel peridotite xenoliths from east central China: Evidence for contrasting effects of melt percolation

Earth and Planetary Science Letters 239 (2005) 286 – 308 www.elsevier.com/locate/epsl

Re–Os and S systematics of spinel peridotite xenoliths from east central China: Evidence for contrasting effects of melt percolation Laurie Reisberg a,*, Xiachen Zhi b, Jean-Pierre Lorand c, Christiane Wagner d, Zicheng Peng b, Catherine Zimmermann a a

Centre de Recherches Pe´trographiques et Ge´ochimiques (CRPG/CNRS), BP 20, 54501 Vandoeuvre-les-Nancy Cedex, France School of Earth and Space Sciences, University of Science and Technology of China, Hefei, Anhui Province 230026, P.R. China CNRS UMR 7160 bMine´ralogie, Pe´trologieQ and Muse´um National d’Histoire Naturelle, USM 201, De´partement bHistoire de la TerreQ, CP 52, Paris, 61 Rue Buffon, 75005, France d CNRS UMR 7160 bMine´ralogie, Pe´trologieQ and Laboratoire Petrologie, mode´lisation des mate´riaux et processus (PMMP), Universite´ Paris 6, Boite 110, 4, Place Jussieu, 75252 Paris Cedex 05, France b

c

Received 31 March 2005; received in revised form 29 August 2005; accepted 3 September 2005 Available online 19 October 2005 Editor: K. Farley

Abstract Os isotopic compositions and Re and Os concentrations were determined for Cenozoic basalt-borne ultramafic xenoliths from the Subei Basin of east central China. Re–Os analyses were coupled with whole rock major and trace element and S abundance determinations, and with characterization of rock textures, modal phase proportions and sulfide petrography. The two main sampling areas, Lianshan and Panshishan, separated by only 6 km, have similar textures and major and moderately incompatible lithophile trace element compositions. The Os isotopic ratios of these two areas plot on the same trends relating 187 Os/188Os to indices of melt extraction such as whole rock Yb content. These Os isotopic systematics suggest that both areas were affected by an early Proterozoic (~1.8 Ga) melt extraction event. Thus the two areas apparently shared the same long term lithospheric history. Nevertheless, the sulfide abundances and whole rock S, Os and Re concentrations are strikingly lower in Lianshan than in Panshishan, and the two localities have different incompatible lithophile trace element signatures. These differences resulted from contrasting melt percolation styles between the two areas. Lianshan was affected by extensive percolation of sulfur undersaturated melts that removed Re, Os and S, while Panshishan experienced interaction with low degree or highly evolved melts that added Re, Cu and S, but had no affect on Os abundances. The lack of correlation between 187 Os/188Os and 187Re/188Os, compared with the good correlation between 187Os/188Os and Yb, indicates that the perturbation of the Re and Os concentrations was fairly recent, and perhaps related to Mesozoic or Cenozoic lithospheric thinning in eastern China. The Lianshan Os concentrations are typical of those of off-cratonic mantle xenoliths, while the Panshishan Os concentrations are closer to those of orogenic peridotites. This suggests that the low Os concentrations, and by extension, the low concentrations of all of the highly siderophile elements (HSE) typically observed in ultramafic mantle xenoliths, may result from recent melt percolation processes, probably directly or indirectly related to the magmatism that brought the xenoliths to the surface. Thus ultramafic xenoliths may not provide reliable estimates of the HSE contents of the upper mantle, and variations in HSE abundances between xenolith localities should not be used to define global scale processes. Os and other HSE abundances may prove to be sensitive indicators of melt percolation, and may provide information about the degree of sulfur saturation of the melts. Despite the

* Corresponding author. Fax: +33 3 83 51 17 98. E-mail address: [email protected] (L. Reisberg). 0012-821X/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2005.09.010

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

287

loss of Os during recent percolation, the Os isotopic ratios of most samples are nearly unmodified by this process, confirming the utility of this system for dating ancient melt extraction events. D 2005 Elsevier B.V. All rights reserved. Keywords: peridotites; osmium; Os isotopes; sulfur; chalcophile elements; mantle xenoliths

1. Introduction During Earth accretion, the highly siderophile elements (Ru, Rh, Pd, Re, Os, Ir, Pt, Au) were almost completely segregated into the core. The small quantities of these elements present in the silicate earth are thought to be derived from meteoritic material added after core formation [1]. Recent analytical improvements have permitted the HSE contents of mantle-derived rocks to be determined with greater precision, allowing variations in HSE abundances between localities to be established with confidence. It has been shown that the HSE abundances of peridotite xenoliths carried by alkali basalts are more variable and usually lower than those of peridotites exposed in orogenic massifs. These regional differences potentially reflect spatial variations in the late meteoritic veneer delivered to the earth and moon [2], or in core–mantle interaction and mantle differentiation. However, recent studies [3– 5] show that HSE abundances can be greatly altered by small scale processes such as melt percolation. The effects of these local processes must be understood before global scale processes can be considered. Osmium is particularly well suited for addressing this problem. Os behaves compatibly during partial melting, so Os abundances in mantle peridotites are much less affected than incompatible HSE (Re, Au, Pd, Pt) abundances by melt extraction. Thus partial melting leaves peridotite residues with low and variable Re / Os ratios. As 187Re is the radioactive parent of 187 Os, this fractionation leads with time to Os isotopic variations that correlate with the degree of ancient melt extraction. Because peridotites have high Os concentrations, their Os isotopic compositions are relatively resistant to modification during metasomatism, and reflect the long-term lithospheric history of these rocks. On the other hand, variations in Os abundances may result from superimposed recent processes [6,7]. Thus the Re–Os system provides an excellent tool for determining whether the low and highly variable HSE contents of peridotite xenoliths are of ancient or recent origin. We present results of an Os isotopic study of two ultramafic xenolith suites from eastern central China. Re–Os analyses were coupled with whole rock major and trace element analyses and characterization of the

rock textures. CO2 contents and FeO / Fe2O3 ratios were determined to explore possible relationships between extent of alteration and Os and Re concentrations. As sulfides are normally the major hosts of Os and other HSE in mantle rocks, the sulfides were examined and sulfur concentrations were determined. 2. Geologic and tectonic setting The xenoliths were collected from Neogene alkali basalts of the Subei basin, located east of the Tanlu fault zone in east central China (Fig. 1a). Seismic refraction data indicate that this Eocene fault-depression basin is underlain by about 30 km of continental crust [8,9]. This region is marked by the Triassic collision of the Yangtze (South China) block with the dominantly Archean North China block (NCB). While the Yangtze block is mainly Proterozoic, evidence for Archean crust has been found in its western part [10], and 2500 Ma zircons have been found in East Anhui province [11]. West of the Tanlu fault, the collision zone is defined by the ultrahigh pressure (UHP) Dabieshan metamorphic belt, but the location of the suture east of the fault is debated. The similarity between the Sulu and the Dabieshan UHP belts may imply (e.g. [12]) that the suture lies beneath or slightly north of the Sulu terrain [13]. In this case the mantle lithosphere beneath the Subei basin would belong to the Yangtze block. However, Li [14] proposed that the upper crust of the Yangtze block was detached and thrust to the north, while the lower crust and mantle lithosphere were subducted, with the suture located near the lattitude of Nanjing. If this model, which has some support from isotopic data [15], is correct, the uppermost mantle lithosphere beneath much of the Subei basin is actually part of the NCB, despite the association of the upper crust with the Yangtze block. The samples studied here come from just north of the proposed subsurface suture. Thus while a Yangtze lithospheric provenance seems more likely, a NCB provenance cannot be excluded. Geophysical and geochemical data argue that large thicknesses of mantle lithosphere have been removed from beneath eastern China (reviews in [16–18]). The timing and lateral and vertical extent of, as well as the mechanisms and tectonic driving forces responsible

288

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

Fig. 1. Map of eastern China, showing the main tectonic features. The studied xenoliths were taken from alkali basalts of the Subei basin, located east of the Tanlu Fault, between ~31 and 35 8N. While the upper crust in this region is associated with the Yangtze block, Li [14,87] suggests that the underlying lithosphere may be affiliated with the North China block. Inset shows alkali basalt outcrops in the northern part of Jiangsu Province, including the xenolith sampling localities of Lianshan, Panshishan, and Fangshan (modified after Zhi [88]).

for lithospheric removal are intensely debated. While the presence of deep garnet megacrysts in kimberlites from the eastern NCB defines a maximum Ordovician age for lithospheric thinning [17], further age constraints are limited. An early Cretaceous age has been suggested, based on the extensive magmatism that affected eastern China at that time [19]. If thinning post-dated the Jurassic NCB–Yangtze collision, it probably affected both the NCB and the Yangtze block, though most direct evidence for this process comes from the NCB. While geophysical data (summarized in [17]) indicate that the current lithospheric thickness beneath eastern China is 60 to 120 km, the lithosphere may have thickened by conductive cooling since the removal process. The vertical extent of lithospheric removal depends on the physical mechanism involved. Delamination would probably remove the entire mantle lithosphere [19], and perhaps even

part of the lower crust. In contrast, upwelling of asthenosphere induced by extension may remove only part of the mantle lithosphere. Tomographic imaging [20] suggests that break-off and dispersal of the ancient lithosphere may result from upwelling along weak zones in the Archean lithospheric roots [21]. Major structural features, notably the Tanlu fault, may serve as starting points for asthenospheric uplift [22,23]. As an alternative to lithospheric replacement by upwelling asthenosphere, Niu [24] argues for transformation of ancient lithosphere into mantle with the physical characteristics of asthenosphere by hydration with fluids derived from the underlying subducted slab. In addition to the uncertainty regarding the mechanisms of thinning, the large scale tectonic forces driving this process are unclear. Triassic collision between the North and South China plates [15,25], Jurassic to Tertiary subduction of the Kula and Pacific

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

plates [17,26], or a combination of both [23] are commonly cited possibilities. Hemisphere scale processes, such as major mantle upwelling related to Gondwana breakup [27] and early Cretaceous superplume activity leading to increased subduction rates [19], have also been cited. Finally, seismic tomographic images [28], showing that eastern China is underlain by a horizontal subducted slab at the depth of the transition zone extending inward to the Daxin’anlingTaihang gravity lineament, provide a strong constraint on any successful model. East central China experienced widespread Cenozoic basaltic volcanism, probably directly or indirectly related to the lithospheric thinning. This volcanism was particularly intense in the Subei basin, where it included minor tholeiitic eruptions in the early Paleogene, and more extensive, xenolith-bearing alkali basalt activity in the Neogene, encompassing the sampling areas (Panshishan, Lianshan and Fangshan) of this study. This volcanism has been dated at about 9 Ma for Fangshan [29,30], but exact dates are not available for Panshishan and Lianshan. All three areas are located in northwestern Jiangsu Province (inset, Fig. 1). Panshishan and Lianshan are separated by about 6 km, while Fangshan is about 20 km southwest of Lianshan. Eighteen samples from Lianshan, eight from Panshishan, and one from Fangshan, were investigated. 3. Mantle xenoliths The xenoliths of both Lianshan and Panshishan are found embedded in alkali basalt lava flows. Estimated equilibration temperatures suggest derivation from the shallow lithosphere [31], mostly less than 40 km depth assuming a geotherm similar to that estimated for the Nushan locality [32]. Most of the xenoliths are rounded and small to moderate in size (typically 5–10 cm in diameter), though larger ones have been found. Nearly all are lherzolites or, rarely, harzburgites. Websterites and pyroxenites are extremely rare. Samples from Panshishan are quite fresh, while those from Lianshan appear slightly altered in hand specimen. This difference is apparent in the olivine color, which varies from light green in the freshest samples, towards yellow, brown or even red in the most altered. 3.1. Rock types and modal compositions Modal compositions (Table 1) were determined by counting 2000/3000 points per thin section. The sam-

289

Table 1 Modal analyses of Subei Basin mantle xenoliths (in vol.%) Mode (vol.%) Sample

Texture

Ol

Opx

Cpx

Sp

Cpx/Opx

pr po/eq pr pr pr

51.1 62.6 55.3 48.1 74.8

30.6 29.2 25.4 36.5 14.3

14.6 6.8 14.5 11.8 9.0

3.7 1.4 4.8 3.6 1.9

0.48 0.23 0.57 0.32 0.63

pr pr/po pr pr pr pr/po

53.3 59.0 58.1 50.2 80.8 53.0

27.7 27.9 25.1 31.4 14.1 27.1

14.5 10.7 15.9 17.2 4.5 15.7

4.5 2.4 0.9 1.2 0.6 4.2

0.52 0.38 0.63 0.55 0.32 0.55

Panshishan LHPSS-1 LHPSS-3 LHPSS-4 LHPSS-8 LHPSS-9 LHPSS-11 LHPSS-12

pr po/ta pr/po pr/po pr pr pr

51.3 54.6 56.7 57.7 49.8 57.1 55.4

28.5 25.7 27.2 22.5 34.2 28.8 28.2

15.4 13.7 14.1 16.5 13.5 11.8 13.9

4.8 6.0 2.0 3.3 2.5 2.3 2.5

0.54 0.53 0.52 0.73 0.39 0.41 0.49

Fangshan LHFS

pr

64.0

25.9

8.0

2.1

0.31

Lianshan LHLS-1 LHLS-2 LHLS-3 LHLS-4 LHLS-5 LHLS-6 LHLS-7 LHLS-8 LHLS-9 LHLS-10 LHLS-12 LHLS-18

pr = protogranular; po = porphyroclastic; eq = equigranular; ta = tabular.

ples are almost all spinel lherzolites (Fig. 2), as are most peridotites of eastern China [16,33,34]. Only one harzburgite xenolith was found, in Lianshan (LHLS-12). No primary hydrous phases were observed. Lianshan lherzolites have compositions ranging from highly fertile, with 15–17 vol.% clinopyroxene (cpx), to more depleted (~7 vol.% cpx, Table 1). Most have cpx / opx ratios of 0.5–0.6, but three are cpx-poor (cpx/opx = 0.2–0.3). Panshishan lherzolites have a more restricted range of fertile composition (11.8–16.5 vol.% cpx), and higher cpx / opx ratios (up to 0.7). The Fangshan sample (LHFS) contains 8 vol.% cpx. 3.2. Textures The xenoliths are coarse-grained and mostly protogranular or protogranular/ porphyroclastic (Table 1) [35,36]. Incipient melting along grain boundaries is observed in some xenoliths from both localities. 3.2.1. Lianshan xenoliths The protogranular textures typical of most Lianshan lherzolites show no preferential orientation and the crystals are not elongated. Olivine (ol) and orthopyr-

290

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

Fig. 2. A) Volume % cpx in Lianshan and Panshishan xenoliths. The open rectangle plotted with the Lianshan samples represents the Fangshan sample. B) Modal compositions (ol–opx–cpx) of peridotite xenoliths.

oxene (opx) are large (4–5 mm), while clinopyroxene (cpx) and spinel (sp) are smaller (1–3 mm). Cpx and sp are always in direct contact with the large opx grains and sp commonly forms vermicular crystals inside opx or between opx and cpx. Occasional cpx exsolution lamellae are seen in opx. The harzburgite (LHLS-12) also has a protogranular texture. Two samples (LHLS-8 and LHLS-18) are transitional between protogranular and porphyroclastic. These show a weak orientation with kink-banded ol porphyroclasts, a few neoblasts (0.5 mm) and disseminated sp grains either as holly-leaf-shaped crystals or ovoid inclusions inside ol and opx. Sample LHLS-2 is the only xenolith displaying a preferential orientation, with parallel elongated tabular ol crystals (1  3 mm). Holly-leaf sp grains and some 1208 triple junctions are present. This texture is transitional between porphyroclastic and tabular equigranular. In thin section, silicate phases show only limited evidence of alteration. In some samples, the olivines are slightly serpentinized. The most extreme case is harzburgite LHLS-12, which contains olivines with abundant serpentine-filled cracks.

Petrographic evidence for alteration is absent, with the exception of sample LHPSS-8, in which olivine fractures are filled with iddingsite. Mercier and Nicolas [36] suggested that mantle xenolith textures reflect increasing deformation from protogranular through porphyroclastic to equigranular types. Under this scenario, the xenoliths in the localities studied here suggest a mostly undeformed subcontinental lithospheric mantle. However, under appropriate temperature and stress conditions, porphyroclastic textured rocks eventually recrystallize to produce coarse granular or secondary porphyroclastic textures. Such recrystallization is facilitated by the presence of intergranular melt fractions [37]. Thus the interpretation of the apparently undeformed textures in our xenoliths is ambiguous. 4. Results Analytical procedures are described in the Appendix. Major and trace element data are listed in online supplemental information. 4.1. Whole rock major element systematics

3.2.2. Panshishan xenoliths The textures, mostly protogranular and protogranular-porphyroclastic (Table 1), resemble those of the Lianshan samples, but the average grain size is larger (~6 mm for ol and opx). One sample (LHPSS-3) shows tabular olivine crystals (2–3 mm) and weakly oriented vermicular spinel grains.

Most Lianshan samples are fertile spinel lherzolites, with high modal cpx (Table 1, Fig. 2) and high CaO and Al2O3 contents, reflecting only slight depletion in basaltic components; however, a few cpx-poor (b 9% modal cpx) spinel lherzolites and one harzburgite were also found. Most Panshishan samples are quite fertile,

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

and unlike in Lianshan, none have Al203b 2 wt.%. Several samples from both localities have CaO and Al2O3 contents comparable to primitive mantle estimates [38], as noted for peridotite xenoliths from elsewhere in southeastern China [34]. Most Lianshan samples, and nearly all Panshishan samples, lie within a small range (88.5–90.0) of Mg# [100  Mg/ (Mg + Fe)], coherent with their fertile character. Sample LHFS (Fangshan) has lower CaO and Al2O3 contents, similar to those of the depleted Lianshan samples. CaO / Al2O3 ratios of the fertile Lianshan lherzolites are close to or slightly greater than 1, while many of the refractory samples have much higher values (up to 2.16 for harzburgite LHLS-12). These high values suggest addition of cpx, or possibly apatite, though this phase was not observed. All but one of the Panshishan samples have CaO / Al2O3 ratios slightly less than 1, which are lower than those of Lianshan samples of similar fertility. The sole exception (LHPSS-11) has a value of 1.25. This sample does not have a higher cpx modal abundance, but that may reflect modal inhomogeneity at the thin section scale. Well defined negative correlations (not shown), including samples from all three areas, exist between whole rock concentrations of MgO and magmaphile elements such as Ti, Al, and Ca. Such trends, typical of other spinel peridotites from southeastern China [34] and worldwide locations, are generally interpreted to result from melt depletion. Samples LHLS-5, LHLS16, LHFS and to a lesser extent LHPSS-9, have unusually high Fe2O3 contents, considering their refractory character. Olivine Mg# is plotted against olivine modal proportions in Fig. 3. The Panshishan rocks plot near estimates for the primitive mantle composition (e.g. pyrolite) and form a more restricted area than do those from Lianshan. The peridotites from this study plot mainly within the field of other eastern Chinese peridotites, which include both on- and off-craton locations [16,34]. This field lies along the boceanic trendQ for residues of progressive low pressure basalt extraction [39], and mostly in the bPhanerozoicQ field defined by Griffin et al. [40]. Though the distinction between the Phanerozoic and Proterozoic fields is debated [41], the highly refractory nature of Archean lithosphere is widely acknowledged. The peridotites from eastern China Archean cratonic terranes are thus distinct from most other on-cratonic peridotites and more akin to off-cratonic peridotites ([16] and references therein).

291

Fig. 3. Olivine Mg# plotted against olivine modal proportions. Large symbols as in Fig. 2. Small solid triangles and black squares correspond respectively to on-craton and off-craton spinel peridotites of eastern China [16]. All of the Panshishan samples, and almost all of the Lianshan samples plot near the oceanic trend (arrows) defined by Boyd [39] and in the Phanerozoic field defined by Griffin et al. [40]. The exceptions are the Fe-rich samples LHLS-5, LHLS-6 and LHFS, which plot to the low Mg# side of this field.

4.2. Trace element results Moderately incompatible trace elements such as Y and the heavy rare earth elements (HREE) define strong negative correlations when plotted against MgO (not shown), consistent with the expected results of partial melt extraction. In contrast, no correlation exists between MgO and highly incompatible elements such as La and Sr. This behavior has been observed repeatedly in spinel peridotite xenoliths from throughout the world and is ascribed to metasomatic enrichment of rocks previously depleted by melt extraction (e.g., [42]). This enrichment is also evident in the chondrite normalized whole rock REE patterns of our samples (Fig. 4). None of the cpx-poor lherzolites or harzburgites has the LREE depleted pattern expected for residues of melt extraction. The two Lianshan samples (LHLS-5 and LHLS-16) and the Fangshan sample (LHFS) with anomalously high Fe contents have almost identical, LREE enriched patterns. The more fertile Lianshan lherzolites do have moderately LREE depleted patterns, but there is no relationship between the degree of fertility and the extent of LREE depletion. For example, no correlation exists between Ce/Yb and Yb (Fig. 5A), implying that the incompatible element systematics of even the fertile samples have been modified. This possibility is supported by the strong correlation be-

292

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

Fig. 4. Chondrite normalized whole rock rare earth element patterns of Panshishan lherzolites (upper), Lianshan lherzolites (middle) and Lianshan and Fangshan harzburgites and cpx-poor lherzolites (lower).

tween La/Sm and CaO/Al2O3 (Fig. 5B) among the Lianshan samples, suggestive of clinopyroxene addition from a LREE enriched melt. A clear difference is observed between the REE spectra of lherzolites from

Lianshan and Panshishan, with most Panshishan samples displaying mild enrichment in the lightest REE (Fig. 4). Unlike in Lianshan, no relationship exists between La/Sm and CaO/Al2O3 (Fig. 5B).

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

293

Fig. 5. A) Whole rock Ce/Yb vs. Yb content. B) Whole rock La/Sm vs. CaO/Al2O3.

4.3. Sulfide petrography and sulfur contents One striking difference between the two localities is that sulfides are abundant in the Panshishan samples, but are nearly absent in those from Lianshan. Lianshan samples contain no visible sulfides in thin section or only trace amounts of highly altered vein sulfides impossible to firmly identify. Panshishan samples are rich in small sulfide inclusions associated with secondary fluid inclusion trails (especially abundant in spongy clinopyroxene), in vein sulfides and in tiny droplets adhering to silicate grain boundaries. In addition to

intergranular vein sulfides, rounded sulfide blebs, up to 200 Am in diameter, may occur at triple junctions of the silicate matrix. Sulfide textures are similar to those described for intergranular sulfides in other mantle xenolith suites, i.e. finely intergrown pentlandite and pyrrhotite like phases veined or rimmed by Cu-rich sulfide [43,44]. Primary inclusions in silicates are very scarce, being represented by three partly altered Mss-like inclusions, a few tens of micrometers across, in olivine. This difference in modal sulfide abundances between the two areas is reflected in the whole rock sulfur contents. Lianshan samples have S contents (18–56

294

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

ppm) similar to those of the vast majority of basalthosted spinel peridotite xenoliths ([45–47]; [48] and ref. therein). Panshishan samples have higher S contents (86–294 ppm) similar to those of orogenic lherzolites [49] or of rare fertile spinel lherzolite xenoliths from Kilbourne Hole [50] or the southern Massif Central ([48] Group I). This difference cannot reflect great-

Fig. 6. Whole rock S (A), Os (B), Re (C) concentrations,

187

er extents of partial melting in Lianshan, because a marked contrast exists between the two localities even at a given Yb content (Fig. 6A). The two areas also display different correlation patterns between S and fertility indices (Al2O3; Yb) or between S and other chalcophile elements such as Cu (Fig. 7C). In Lianshan, a correlation exists between Yb and S (Fig. 6A) and a

Re / 188Os ratios (D), and Cu concentrations (E) plotted as a function of Yb content.

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

weaker correlation exists between Cu and S, although this is partly obscured by large variations of Cu/S ratios (0.2–1) in lherzolitic samples. In Panshishan, the relationship between S and Yb contents is much steeper, with one sample (LHPSS-11) deviating from the trend (Fig. 6A), and a good correlation is observed between S and Cu (Fig. 7C). For similar Cu contents, Panshishan lherzolites show less variable and much lower Cu/S ratios (0.11–0.19) than Lianshan xenoliths; similar ratios characterize orogenic lherzolites and Group I Massif Central lherzolite xenoliths [48,51]. Samples from both areas lie along a single, well defined trend relating Cu and Yb (Fig. 6E). 4.4. Re–Os results Lianshan Os concentrations range from 0.1 to 3.8 ppb, with a median value of 1.8 ppb (Table 2). No correlation exists between Os concentration and whole rock fertility, expressed by Yb content (Fig. 6B). Lianshan Os concentrations are typical of those of off-cratonic xenoliths (Fig. 8). In contrast, Panshishan Os concentrations range from 2.7 to 3.7 ppb, except for

295

one possibly spurious analysis of sample LHPSS-4 with 6.1 ppb, and have a median value of 3.2 ppb. A shallow anti-correlation is observed between Os and Yb concentrations (Fig. 6B). The Panshishan Os concentrations fall towards the high end of the non-cratonic xenolith range, and are comparable to those found in orogenic lherzolites (Fig. 8). This distinction is also seen in Re concentrations. Lianshan Re concentrations range from 0.02 to 0.14 ppb, with a median value of 0.063 ppb, while Panshishan Re concentrations range from 0.23 to 0.33 ppb, with a median value of 0.25 ppb. This difference cannot be wholly ascribed to the greater average fertility of the Panshishan samples, as Re concentrations are higher in Panshishan even at a fixed Yb or Al2O3 content (Fig. 6C). The Re / Os ratios of the Panshishan samples are also generally higher than those of Lianshan of similar fertility (Fig. 6D). Rough correlations between Re and Yb exist in both areas, though the slope of the correlation is steeper in Panshishan than in Lianshan (Fig. 6C). In Lianshan, neither Re nor Os correlates with S, the concentrations of both elements varying considerably at similar S contents (Fig. 7). In Panshishan, Os concentrations remain roughly constant

Fig. 7. Whole rock Os vs. S (A), Re vs. S (B), S vs. Cu (C) and Re vs. Cu (D) variation diagrams.

296

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

Table 2 Os isotopic ratios and Os, Re and S concentrations Sample

Al2O3

Yb (ppm)

S (ppm)

Lianshan LHLS-1

3.15

0.246

31

LHLS-2 LHLS-3 LHLS-4 LHLS-5 LHLS-6

1.17 3.32 2.62 1.43 3.08

0.031 0.245 0.218 0.119 0.255

20 50 49 31 38

LHLS-7 LHLS-8

3.14 2.99

0.296 0.259

33 46

LHLS-9 LHLS-10 LHLS-11

2.92 3.63 1.66

0.235 0.329 0.107

29 40 21

LHLS-12 LHLS-13

0.82 1.49

0.030 0.093

18 18

LHLS-14

2.54

0.217

16

LHLS-15 LHLS-16

2.97 1.22

0.241 0.092

53 25

LHLS-17 LHLS-18

2.99 3.55

0.252 0.370

51 56

Panshishan LHPSS-1 LHPSS-2 LHPSS-3 LHPSS-4

4.25 2.88 3.23 3.21

0.350 0.223 0.251 0.297

179 123 108 131

LHPSS-8 LHPSS-9 LHPSS-11 LHPSS-12

3.03 2.23 3.34 3.28

0.273 0.164 0.311 0.323

112 86 294

Fangshan LHFS

1.52

0.120

187

Os/188Os

Os (ppb)

188

0.12501 F 20 0.13373 F 15c 0.12302 F 13 0.12734 F 34c 0.12607 F 19 0.12389 F 17 0.11925 F 25 0.12527 F 21 0.12552 F 29b 0.12674 F 33 0.12507 F 20 0.12481 F15 0.12542 F 21 0.12615 F 20 0.13453 F 14c 0.12299 F 25 0.11715 F 07c 0.13029 F 08c 0.12520 F 29 0.12427 F 33 c 0.12226 F 22 0.12523 F 22b 0.11859 F 77 c 0.11775 F 16

1.460 3.135 1.083 3.784 2.140 1.951 2.449 1.607 1.475 1.692 1.853 1.959 1.827 1.519 0.111 1.231 1.005 0.143 0.181 1.942 1.980 1.415 2.072 2.161

1.02E-12 2.19E-12 7.57E-13 2.64E-12 1.49E-12 1.36E-12 1.71E-12 1.12E-12 9.99E-13 1.18E-12 1.29E-12 1.37E-12 1.28E-12 1.06E-12 7.77E-14 8.60E-13 7.03E-13 9.95E-14 1.26E-13 1.36E-12 1.38E-12 9.89E-13 1.45E-12 1.51E-12

0.12685 F 24

2.047

1.43E-12

0.12775 F 20 0.12480 F 23 0.12589 F 21 0.12544 F 30 0.12565 F 20b 0.12539 F 15b 0.12385 F 15 0.12764 F 20 0.12615 F 20

2.819 3.722 2.973 6.070 3.237 3.046 3.713 3.334 2.910

1.97E-12 2.60E-12 2.08E-12 4.26E-12 2.33E-12 2.13E-12 2.59E-12 2.33E-12 2.03E-12

0.12087 F 18

1.716

1.20E-12

Os M/g

Re (ppb)

187

Re/188Os

0.099 0.016a 0.151

0.441 0.021 0.338

0.026

0.050

0.092 0.110 0.080 0.060

0.301 0.313 0.207 0.147

0.074

0.233

0.067

0.260

0.038

1.009

0.038 0.027

0.093 0.090

0.024 0.110 0.080

0.053

0.259 0.272

0.442 0.350

0.241 0.252 0.194a 0.342 0.314

0.347 0.398 0.272 0.492 0.519

0.188

Repeat analyses were performed on dissolutions of separate powder splits. Uncertainties represent 2r-m in-run statistics. Samples in italics indicate cases where mass 233/mass 236 N 0.0005. a) Re and Os obtained on different powder splits. b) Samples digested without desilification in HF–HBr. c) Samples analyzed in integrative mode.

with S content. On the other hand, good correlations exist between Re, S, and Cu in this locality (Fig. 7) suggesting that all three elements had similar geochemical behavior or reflecting the strong control of modal sulfide abundances. Regression lines in the Re vs. Cu and Re vs. S diagrams intercept the y axis at non-zero values. Among the Lianshan samples, 187Os / 188Os ratios range from 0.118 to 0.135. Panshishan 187Os/188Os ratios display a more restricted range, from 0.124 to

0.128, while the Fangshan sample has a value of 0.121. In Lianshan, the two samples with the lowest Re / Os ratios have the least radiogenic values, but otherwise no correlation is observed (Fig. 9). In contrast, in Panshishan, 187Os/188Os is correlated with 187Re/188Os. Despite the lack of correlation with Re/Os in Lianshan, a positive linear trend, including most of the samples from all three areas, relates 187Os/188Os to indices of melt extraction such as Yb or Al2O3 content (Fig. 10). The extension of this trend overlaps with the primitive

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

297

Fig. 8. Histograms of whole rock Os concentrations of ultramafic rocks from worldwide occurrences of off-cratonic ultramafic xenoliths (upper), Lianshan and Panshishan xenoliths (middle), and orogenic peridotites (lower). Data sources include: [3,7,25,52–54,58,72,89–94].

upper mantle (PUM) Os isotopic composition proposed by Meisel et al. [52,53], providing support for the validity of this estimate. A few samples have Os isotopic ratios plotting above the main trend. These include not only cpx-poor samples, as has been observed elsewhere (e.g. [54]), but also a fertile sample (LHLS-1), which is more unusual. Samples that fall on the trend have reproducible Os isotopic ratios and concentrations, within 1% and 10%, respectively. (The

only exception is sample LHPSS-4, which has a highly reproducible Os isotopic ratio that falls on the trend, but an Os concentration that varies by a factor of 2. The unusually high Os concentration (6.1 ppb) of one LHPSS-4 aliquot may reflect an analytical artifact, such as a spiking error.) In contrast, both the concentrations and the ratios of samples plotting above the trend are irreproducible, implying that they have internally heterogeneous Os compositions. Direct evi-

298

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

Fig. 9. Whole rock

187

Os/188Os plotted against

187

Re/188Os. Primitive Upper Mantle (PUM) values from Meisel et al. [52].

dence for internal Os isotopic heterogeneity in other xenolith suites has been provided by recent in situ sulfide studies [55–57]. In sample LHLS-1, which was

analyzed three times, Os concentration is positively related to 187Os/188Os. The opposite is true for samples LHLS-11 and LHLS-13.

Fig. 10. Whole rock 187Os/188Os plotted against Yb content. Repeat analyses of samples with irreproducible Os isotopic ratios are indicated by solid (LHLS-1), dashed (LHLS-11), and dotted (LHLS-13) lines. Isotopic analyses of separate powder splits of samples plotting along the main trend are reproducible within 1%.

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

5. Discussion 5.1. Relationship between Lianshan and Panshishan Lianshan and Panshishan are separated by only 6 km, yet have markedly different Re, Os and S concentrations. The low concentrations of these elements in the Lianshan samples are typical of non-cratonic mantle xenoliths from around the world, while those of Panshishan are similar to the higher concentrations found in orogenic lherzolites and abyssal peridotites (Fig. 8). This might suggest that the Lianshan samples represent the subcontinental lithosphere, while the Panshishan samples are derived from the asthenosphere. However this interpretation is not satisfying since ultimately, the subcontinental lithosphere must have been derived from the convecting upper mantle, and thus must have started with sulfur and HSE contents similar to those of the asthenosphere. The question then is, when and how did xenoliths from the non-cratonic subcontinental lithosphere lose most of their HSE? With the exception of a few internally heterogeneous samples, the Lianshan peridotites display a strong correlation between 187Os/188Os and Yb content, coupled with a near absence of correlation between 187Os/188Os and 187Re/188Os. This indicates that Os isotopic ratios were essentially unaffected by the loss of Re and Os and the ensuing modification of the Re / Os ratios. This loss of Re and Os must have been recent; otherwise the correlation linking 187Os/188Os and Yb would have been destroyed by radiogenic ingrowth. Conversely, the few xenolith localities with Re and Os concentrations comparable to those of orogenic peridotites are also among the rare xenolith suites displaying correlations between 187Os/188Os and 187Re/188Os. Examples include the Panshishan samples of this study, and xenoliths from Mt. Gambier, Australia [58], and Hannuoba, China [25]. If Lianshan is typical of subcontinental xenolith localities, Re, Os and sulfur loss in these suites must have occurred during, after, or shortly (ie, within ~100 Ma) before eruption. Thus most xenoliths cannot be used to define the HSE and sulfur abundances of the subcontinental lithosphere, because these abundances have been modified by processes directly (e.g., volatile loss during eruption) or indirectly (e.g., melt percolation linked to tectonic thinning) related to xenolith emplacement. 5.2. Possible petrogenetic processes Since the low Re and Os concentrations of Lianshan result from recent Re and Os loss, the contrasting

299

chalcophile element abundances of the two localities do not imply a fundamental difference in their mantle sources. Thus in the following discussion we assume that both Lianshan and Panshishan represent old lithospheric mantle, even if the two areas have experienced different processes recently. This assumption is based on the geographic proximity of the two localities, and the fact that both areas plot on the same trends in Cu vs. Yb (Fig. 6E), 187Os/188Os vs. Yb (Fig. 10) and major element variation diagrams. Several petrogenetic processes may explain the contrast in Os, Re and S concentrations between Lianshan and Panshishan and the Os isotopic systematics of these areas. These include: 1) syn or post-eruptive processes, such as devolatilization on eruption or sulfide weathering, entraining elemental loss during or after emplacement; 2) recent percolation of melts and/ or fluids (brecentQ is used to imply events sufficiently young so that any resulting modification of the Re / Os ratio did not notably affect the 187Os / 188Os ratio by radiogenic ingrowth, i.e., within the last ~100 Ma); 3) ancient melt extraction. All three types of processes affected both Lianshan and Panshishan, but the extent and the specific mechanisms involved differed between the two areas, leading to contrasting chalcophile element systematics. 5.2.1. Syn and post-eruptive processes In hand specimen, Lianshan samples appear more altered than Panshishan samples. This is confirmed by the higher bulk-rock Fe2O3 / FeO ratios of the Lianshan samples and the petrographic evidence of highly altered vein sulfides, coupled with a weak tendency for decreasing sulfur concentrations at increasing Fe2O3/FeO (Fig. 11A). The very high bulk-rock Cu / S ratios of the Lianshan samples (0.2–1), despite the absence of Cu-rich sulfides, argues for sulfur loss during groundwater circulation. Peridotite xenoliths containing weathered sulfides typically have Cu / S ratios higher than those of unaltered peridotites (typically 0.05–0.15 ; [48] and references therein) because Cu is less mobile than S in groundwaters. Like S, Os and Re may have been mobile at that stage, as concentrations of these elements also tend to decrease at increasing Fe2O3/FeO (Fig. 11B, C). Chalcophile element loss may also occur during eruption [59], by Mss decomposition and devolatilization when xenoliths come into contact with air [47,60]. Volatility in oxygenated environments (fO2 N magnetite-hematite buffer) increases from Cu to Re, Os and S [61] and thus could partially account for the depletion of Re, Os and S relative to Cu.

300

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

Fig. 11. Whole rock S (A), Os (B) and Re (C) concentrations plotted as a function of Fe2O3 / FeO ratio, a potential indicator of extent of alteration. Though the Lianshan samples are, in general, slightly more oxidized, for a fixed value of Fe2O3/FeO the S, Os and Re contents of the Panshishan samples are higher.

Nevertheless, surface processes were probably not the main cause of the difference in Re, Os and S concentrations between the two localities. For a given extent of alteration (same Fe2O3 / FeO ratio or CO2 content), the Panshishan xenoliths have higher concentrations of all of these elements (Fig. 11). Also, sulfide

modal abundances or their altered equivalents are dramatically lower in Lianshan xenoliths, implying that in this locality much of the sulfur was lost before emplacement. Thus an additional factor is needed to explain the contrasting chalcophile element contents of the two areas.

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

5.2.2. Recent melt percolation The loss of chalcophile elements, and concomitant disturbance of Re / Os ratios, evident in Lianshan does not result from mechanical removal of sulfides, as positive correlations are not observed between S and Re or Os. Also, as Os is hosted almost exclusively by sulfides in most peridotites, while a significant fraction of Re is located in silicates [62,63], mechanical sulfide removal should increase the whole rock Re / Os ratio [64]. Fig. 6D suggests, however, that Lianshan Re / Os ratios have been lowered. These observations are consistent with S, Re and Os removal during percolation of sulfur undersaturated basaltic melt at high melt / rock ratios, as has been suggested for other ultramafic suites [6,7,65–67]. As Os is much more chalcophile than Re, Os should be more strongly retained by the few sulfides remaining after percolation, while Re should partition into and be removed with the passing melts. Thus melt percolation should lower the Re / Os ratios of the peridotites, as observed. Melt percolation apparently did not lower the Cu contents of the Lianshan samples (Fig. 6E), despite the fact that Cu, like Re, is moderately chalcophile. This is probably because Cu is less incompatible than Re in silicates [68], so as melts percolated and absorbed the intergranular sulfides, most of the Cu was retained by the silicates. Volatile bearing phases (e.g., phlogopite, amphibole and apatite), frequently cited as evidence of melt percolation, are absent from our samples. This may be because, as suggested by the lack of sulfur saturation, the melts that percolated through the Lianshan peridotites were not very evolved and thus were undersaturated in volatile bearing phases. Nevertheless, whole rock rare earth spectra provide supporting evidence for extensive melt percolation beneath Lianshan. The fertile (Al2O3 N 2%) lherzolites all have LREE depleted to flat REE patterns, but no correlation exists between Yb content and Ce / Yb ratio (Fig. 5A). This suggests that their REE spectra reflect interaction with passing melts rather than pure melt extraction, as proposed for other ultramafic suites with similar features [37,69]. Most of the harzburgites and non-fertile lherzolites have LREE enriched or U-shaped REE spectra, coupled with low Yb contents and CaO/Al2O3 NN 1, indicative of cpx addition. A correlation exists between CaO/Al2O3 and La/Sm among all of the Lianshan samples (Fig. 5B), suggesting that they have experienced varying degrees of cpx addition from a LREE enriched melt. The refractory samples were most affected by this process, both because their low pre-percolation CaO and LREE contents were easily modified by cpx addition, and because they experienced the highest degrees of percolation.

301

Despite the evidence for extensive chalcophile element loss, there is still a weak correlation between S and Yb content in the Lianshan xenoliths. This probably reflects the existence of a more strongly defined correlation prior to melt percolation. More fertile samples would have started with higher S contents, and thus might have retained more sulfur during percolation. In contrast, less fertile samples probably had low S contents before percolation, so comparable sulfur loss would have decreased the S contents to some background level, controlled by very rare sulfide inclusions, sulfur incorporated in the silicate structure and/or minor late-stage sulfide veins. Near total sulfide loss may explain why several of the less fertile Lianshan samples have very low and internally heterogeneous Os concentrations. As these refractory samples experienced particularly extensive melt percolation, their Os isotopic compositions may also have been modified. The radiogenic character of some samples (up to 0.1345, one aliquot of LHLS-11) could indicate a contribution from subducted slab derived fluids, which may carry radiogenic Os [70]. Xu et al. [71] argued that melt percolation related to Mesozoic and Cenozoic subduction transformed lherzolites to harzburgites beneath Huinan, NE China. This process cannot explain the refractory character of those Lianshan samples with very low 187 Os / 188Os ratios (harzburgite LHLS-12 and cpxpoor lherzolites LHLS-5 and LHLS-16), indicating depletion hundreds of millions of years ago rather than during recent subduction. Sample LHLS-2, on the other hand, has an Os isotopic composition similar to those of the fertile lherzolites, as well as an unusual spoonshaped REE pattern (Fig. 4) and a porphyroclastic/ equigranular texture different from the protogranular textures of the other Lianshan peridotites. It also has a very low cpx / opx ratio (0.23) and high opx content (29.2%), characteristic of reactive harzburgites. Thus this sample may have acquired its depleted character during recent melt percolation. Unlike the Lianshan samples, Panshishan peridotites have S, Os and Re contents comparable to those of abyssal peridotites and orogenic lherzolites (Fig. 8). Thus, in Panshishan, these elements were not lost during recent melt percolation. S, Cu and Re may even have been added to some of the samples. This is most evident for sample LHPSS-11, which has unusually high concentrations of these elements (Fig. 6), and a high CaO / Al2O3 ratio (Fig. 5B). Rough correlations exist between S, Cu and Re, though these are largely defined by sample LHPSS-11. When Re is plotted against S or Cu concentration (Fig. 7) the y-intercept value (~150 ppt) is significantly greater than zero.

302

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

While these correlations could represent mixtures between a S, Re and Cu bearing melt and a nearly sulfidefree, Re rich residue, this seems unlikely, as sulfur-poor depleted peridotites almost never contain 150 ppt of Re. More probably, prior to melt infiltration, a correlation already existed between Re and S. The percolating melt would then have added S, Cu and especially Re to the peridotites, translating and adding scatter to the preexisting trends. The Panshishan Os data support this suggestion. Os concentrations show only limited variation (assuming the aliquot of LHPSS-4 with 6.1 ppb Os is spurious), and a shallow negative correlation with Yb consistent with partial melt extraction. There is also no relationship between S and Os concentrations, indicating that the passing S-saturated melts did not add Os. On the other hand, a strong relationship exists between 187 Os/188Os and Yb content. This suggests that prior to recent melt percolation, Re/Os and Yb were correlated, which with time created the relationship between 187 Os/188Os and Yb. The liquids that infiltrated the Panshishan peridotites were S-saturated and rich in Re and Cu, implying that they were either low degree partial melts, or highly evolved fluids produced by differentiation of higher degree melts. Such liquids could impart the mildly LREE enriched patterns of most Panshishan lherzolites. Fractional crystallization can induce sulfur saturation, changing a sulfide dissolving liquid into a sulfide precipitating liquid. Thus the liquids that percolated through Lianshan and Panshishan may have been genetically related, despite their contrasting effects. Unlike in Lianshan, no relationship is observed in Panshishan between CaO/Al2O3 and La/Sm (Fig. 5B), implying that, with the exception of sample LHPSS-11, the passing melts or fluids did not add cpx. To sum up, both the Lianshan and the Panshishan peridotites show evidence for recent melt or fluid percolation, but the effects of this percolation were strikingly different. In Lianshan, pervasive infiltration of Sundersaturated basaltic melts led to major depletion of the S, Re, and Os contents. In Panshishan, percolating S-saturated fluids added S, Cu and Re to at least some samples. In both cases, these effects were superimposed on preexisting trends relating S, Cu, and Re concentrations, 187Os / 188Os ratios and indices of bulk rock fertility. Since the survival of the 187Os/188Os vs. Yb correlation (Fig. 10) implies that the modification of the Re / Os ratios was recent, this melt percolation was probably related to the magmatism that brought these samples to the surface. However, this relationship may have been indirect. In particular, partial melting may have occurred during Mesozoic or Cenozoic subduc-

tion-induced asthenospheric upwelling, producing magmas that infiltrated the lower lithosphere. The Lianshan mantle may have been infiltrated by high degree, sulfur undersaturated melts, while the Panshishan mantle was affected by lower degree melts and/or by highly evolved melts, but the source in both regions was probably the upwelling asthenosphere. The lithospheric thinning caused by this upwelling may in turn have facilitated Neogene volcanism. 5.2.3. Ancient magmatic processes Most samples from Lianshan and Panshishan, as well as the single sample from Fangshan, plot on a well-defined trend relating 187Os/188Os to Yb content (Fig. 10). This correlation does not result from recent melt percolation since, as shown above, Os was either lost (Lianshan) or not added (Panshishan) during this process. Instead, the trend probably reflects radiogenic ingrowth of 187Os since an ancient melt extraction episode that left variably depleted peridotites, with Re / Os ratio correlated with indices of melt extraction such as Yb concentration. The timing and tectonic significance of this event are discussed below. The details of this ancient melting event are difficult to decipher because of the overprinting by more recent processes. The more depleted samples probably formed mostly by simple melt extraction, while the more fertile samples could represent either the residues of simple melt extraction or a mixture of such residues with concomitantly percolating melts. In the latter case, the Re / Yb ratios of the ancient percolating melts must have been similar to that of the primitive upper mantle, since the 187Os/188Os vs. Yb correlation trends towards the PUM value (Fig. 10). Regardless of the details of melt extraction, this process produced residues with Re / Os ratios that varied systematically with the degree of fertility of the peridotite. 5.3. Tectonic significance of the Os isotopic data In a recent Os isotopic study of xenoliths from northeastern China, Wu et al. [26], suggested that the variable Re-depletion ages of the refractory samples indicated that Phanerozoic and Proterozoic mantle were complexly intermixed in the underlying subcontinental lithosphere. In the Subei basin, most of the samples plot along a well-defined trend relating 187 Os/188Os to Yb content. Those samples plotting above this trend have irreproducible Os compositions indicative of internal heterogeneity similar to that documented by in situ Os studies of xenolith sulfides from other localities [55–57]. The whole rock Os composi-

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

tions of such samples cannot be used to define depletion ages. Thus our chronological and tectonic interpretations are based only on the samples that plot along the main 187Os/188Os vs. Yb trend, which have reproducible and systematic behavior. This 187Os/188Os vs. Yb trend probably results from radiogenic ingrowth of 187Os subsequent to an ancient melt extraction event, whose age can be estimated by comparing the y-intercept of the correlation (~0.116) with the mantle evolution curve [72]. Assuming an evolution curve with the Re / Os ratio of PUM [52], this yields a model age of 1.6–2.0 Ga. This is consistent with the Re-depletion age [73] of ~1.5–1.8 Ga of the most depleted sample, which represents a minimum age for melt extraction. However, the tectonic interpretation of this event is ambiguous. While it is generally assumed that melt extraction occurred during creation of the subcontinental lithosphere, there is growing evidence [74–78] that Os isotopic heterogeneities caused by ancient melting can be preserved for over a billion years in the convecting mantle. If such ancient depleted material were added to the subcontinental mantle, the apparent model ages would be unrelated to the time of lithospheric formation. While bearing in mind this inherent uncertainty, in the following discussion we make the usual assumption that the model age represents the time of formation of the subcontinental lithosphere from a primitive or nearly primitive mantle. Under this assumption, these results suggest an early to mid Proterozoic age for the lithospheric mantle beneath the Subei basin. As lithosphere of this age exists beneath both the NCB and Yangtze blocks, the ~1.8 Ga model age neither confirms nor excludes the crustal detachment model of Li [14] for the suture zone in the Subei basin region (see Section 2). This Os model age is indistinguishable from the Re/Os age obtained for peridotites from Hannuoba [25], but this is probably coincidental, given the large distance between the two regions. The early Proterozoic model age implies that our samples, derived from depths close to the Moho, represent ancient lithosphere too shallow to be removed by the lithospheric thinning that has occurred since the Ordovician [17,18,26,71]. Thus these results argue against delamination, as this process is likely to remove the entire mantle lithosphere [19], and instead favor asthenospheric upwelling. While the lithosphere sampled here was not removed, the evidence discussed above indicates that it suffered extensive melt percolation, which most likely occurred during lithospheric thinning. This extensive percolation suggests that our samples were located only slightly above the base of the lithosphere at the time of thinning. No hydrous

303

phases were observed, arguing against the suggestion that lithosphere was transformed into asthenosphere by hydration [24]. In summary, in the Subei basin region, asthenospheric upwelling led to incomplete removal of the lithosphere, coupled with profound modification of the remaining lithosphere by melt percolation. Given the location of our sampling areas, either south of (e.g. [13]) or at the latitude of [14], the suture zone, it is difficult to imagine how the Yangtze–NCB collision could have provoked this upwelling. Thus in this region, subduction of the Pacific plate seems to be a more plausible tectonic driving force. 6. Summary and conclusions The ultramafic xenoliths carried by alkali basalts of the Subei Basin provide evidence for two mantle processes. The first of these was a major melt extraction episode, which affected the mantle now present beneath all of the localities studied. The effects of this episode are evident in the major and the compatible and moderately incompatible trace element systematics. Both Lianshan and Panshishan xenoliths plot in indistinguishable fields on variation diagrams involving these elements, with the exception of a few highly refractory samples found in Lianshan but not in Panshishan. Samples from both areas, as well as the single Fangshan sample, define correlations between Os isotopes and melt extraction indices such as Yb or Al2O3 content. These correlations suggest melt extraction at about 1.8 Ga, which presumably represents the time of formation of the subcontinental lithosphere. The highly incompatible elements, and the S, Re and Os abundances, provide evidence for a second process, melt percolation, superimposed on the results of the earlier melt extraction episode. This process did not erase the correlations between 187Os/188Os and melt extraction indices, indicating that the Os concentrations of the percolating magmas were low. The percolation event was fairly recent, since the modified Re / Os ratios did not have time to significantly perturb the 187Os / 88Os ratios, and may have been linked to Mesozoic to Cenozoic subduction responsible for lithospheric thinning in eastern China. The effects of magma and/or fluid percolation were strikingly different in Lianshan and in Panshishan. In Lianshan, pervasive percolation of sulfur undersaturated melts led to a marked decrease in Re, Os and S abundances, and to a correlation between CaO/Al2O3 and La/Sm suggestive of varying degrees of clinopyroxene addition from a LREE enriched melt. In Panshishan, passage of

304

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

highly evolved, sulfur-saturated melts or fluids led to enrichment in the most incompatible REE and to addition of S, Cu and Re to at least some of the samples. Os abundances in Panshishan were unaffected by this percolation event, and retain a correlation with Yb contents consistent with the results of partial melt extraction. Panshishan and Lianshan are separated by only 6 km and until recently probably shared a similar mantle history, recorded in their similar Os isotopic systematics. Nevertheless, their Os and Re concentrations are very different. This observation adds to the growing evidence [3–5,47,79] indicating that platinum group elements in mantle rocks are often affected by melt percolation and other secondary processes that obscure their original abundances. Thus the concentrations of these elements in mantle peridotites cannot be used to determine primordial regional variations in the HSE contents of the upper mantle. This is particularly true for mantle xenoliths. As shown in Fig. 8, the Lianshan samples have Os contents typical of those of offcratonic mantle xenoliths. As the Lianshan xenoliths have suffered recent loss of Os, Re and S, it seems likely that most other ultramafic xenoliths have experienced similar processes. Thus, most xenoliths provide unreliable estimates of the sulfur [45] and chalcophile element abundances of the upper mantle. Mechanisms linked to the processes that allow these xenoliths to arrive at the surface, such as melt percolation related to lithospheric thinning and syn- and post-emplacement volatile loss, greatly perturb the chalcophile element budgets. As Os and Re abundances are highly sensitive to melt percolation, they may provide evidence that this process occurred, even when this is not obvious from the lithophile element systematics. Moreover, they provide information about the level of sulfur saturation of the passing melts. Finally, despite the loss of Os during melt percolation, Os isotopic compositions appear to be mostly unchanged by this process, and can be used with care to estimate the age of ancient melt extraction. Acknowledgements This research was supported in part by China National Natural Sciences Foundation grants (Nos. 40173009; 49873005) to X.C.Zhi, and by CNRS/ ASC Project no. 16292. We thank C. Parmentier for assistance in the laboratory. We thank Igor Puchtel and an anonymous reviewer for their thorough and constructive comments. This is CRPG contribution number 1766.

Appendix A. Analytical techniques A.1. Whole rock major and trace element analyses Whole rock major element compositions were determined by conventional XRF on fused glass discs at the Institute of Geophysical and Geochemical Exploration, Chinese Academy of Geology Sciences, using China national standard GB/T14506.28-93. Whole rock trace element compositions were determined at the Service d’Analyse des Roches et des Mine´raux (SARM) of the CRPG. Samples were digested by fusion using a LiBO2 flux, then dissolved in HNO3. Analyses were performed by ICPMS. A.2. Sulfur analyses Sulfide petrography was examined in detail in reflected light. Bulk-rock S concentrations were determined by iodometry of the SO2 produced by combustion at 950 8C of 500 mg to 1 g powder aliquots. Accuracy and detection limits of this technique were reported by Lorand [45], Lorand et al. [48] and Gros et al. [80]. A.3. Re-Os analyses About 2 g of sample powder were spiked with Re and 190Os tracer solutions. In most cases, the samples were then desilicified in a reducing (HF–HBr) solution. This solution was dried down and redissolved in concentrated HCl, and then again dried down to drive off excess HF. The residue was digested in a Carius tube (CT) [81] at 230 oC using a 2 : 1 mixture of HCl and HNO3. After digestion, about 30% of the sample solution was removed for Re analysis. Concentrated HNO3, Br2 and CrVI were added to the remaining liquid and Os was extracted into the Br2 [82]. Several samples were digested without desilicification in a 2 : 1 HNO3:HCl mixture. In these cases CrVI was not added since the Carius tube solution was already sufficiently oxidizing, and the Re chemistry was performed on the residue of the Os extraction. Samples digested by these modified techniques are indicated in Table 2. Digestion of the same peridotites by both methods in this study (LHLS-6) and others [7,83] suggests that this difference in technique creates no significant bias in the observed Os and Re concentrations or Os isotopic ratios. (Sample LHPSS-4 was also digested by both techniques, but we suspect that the very high Os concentration obtained for the desilicified aliquot reflects an analytical error, since it 185

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

is so different from those of all of the other samples of this suite digested by the same technique). After purification of the separated Os by microdistillation [84], Os isotopic ratios were measured by negative thermal ionization [85,86] using a Finnigan MAT262 mass spectrometer at CRPG. Measurements were made by peak jumping on an electron multiplier. Signal intensities were obtained by ion counting in most cases, though for a few of the earlier measurements (indicated in Table 2) the multiplier was operated in integrative mode. For ion counting measurements, the 187Os / 188Os ratio of the standard was reproducible within 0.3% (2 r) over the course of this study. The standard reproducibility was slightly worse (0.5%) for the integrative measurements. Os concentrations were determined by isotope dilution (ID). The average Os blank was 1.3 F 0.7 pg. The solution aliquot saved for Re separation was dried down and redissolved in 0.4 N HNO3. Re was extracted using AG1 X8 anion exchange resin columns. Spiked Re isotopic ratios were determined using a Micromass Isoprobe ICPMS at CRPG. Re blanks were about 4 pg. Daily Re standard reproducibility varied from 0.1 to 0.4% (1 r). Appendix B. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j. epsl.2005.09.010. References [1] J.W. Morgan, Osmium isotope constraints on Earth’s late accretionary history, Nature 317 (1985) 703 – 705. [2] J.W. Morgan, R.J. Walker, A.D. Brandon, M.F. Horan, Siderophile elements in Earth’s upper mantle and lunar breccias: data synthesis suggests manifestations of the same late influx, Meteorit. Planet. Sci. 36 (2001) 1257 – 1275. [3] C.T. Lee, Platinum-group element geochemistry of peridotite xenoliths from the Sierra Nevada and the Basin and Range, California, Geochim. Cosmochim. Acta 66 (2002) 3987 – 4005. [4] A. Luguet, O. Alard, J.P. Lorand, N.J. Pearson, C. Ryan, S.Y. O’Reilly, Laser-ablation microprobe (LAM)-ICPMS unravels the highly siderophile element geochemistry of the oceanic mantle, Earth Planet. Sci. Lett. 189 (2001) 285 – 294. [5] M. Rehka¨mper, A.N. Halliday, J. Alt, J.G. Fitton, J. Zipfel, E. Takazawa, Non-chondritic platinum-group element ratios in oceanic mantle lithosphere: petrogenetic signature of melt percolation? Earth Planet. Sci. Lett. 172 (1999) 65 – 81. [6] A. Bu¨chl, G. Bru¨gmann, V.G. Batanova, C. Mu¨nker, A.W. Hofmann, Melt percolation monitored by Os isotopes and HSE abundances: a case study from the mantle section of the Troodos Ophiolite, Earth Planet. Sci. Lett. 204 (2002) 385 – 402.

305

[7] L. Reisberg, J.-P. Lorand, R.M. Bedini, Reliability of Os model ages in pervasively metasomatized continental mantle lithosphere: a case study of Sidamo spinel peridotite xenoliths (East African Rift, Ethiopia), Chem. Geol. 208 (2004) 119 – 140. [8] H.S. Chen, The essentials of geological-geophysical integrated interpretation of the line HQ-13 in the lower Yangtze Basin on the Yangtze Metaplatform, in: Q.X. Ou (Ed.), Petroleum Exploration in Southern China, Geological Publishing House, Beijing, 1988, pp. 239 – 250 (in Chinese with English abstract). [9] X.-L. Huang, Y.-G. Xu, D.-Y. Liu, Geochronology, petrology and geochemistry of the granulite xenoliths from Nushan, east China: implication for a heterogeneous lower crust beneath the Sino-Korean Craton, Geochim. Cosmochim. Acta 68 (2004) 127 – 149. [10] Y.M. Qiu, S. Gao, N.J. McNaughton, D.I. Groves, W.-L. Ling, First evidence of N 3.2 Ga continental crust in the Yangtze craton of south China and its implications for Archean crustal evolution and Phanerozoic tectonics, Geology 28 (2000) 11 – 14. [11] Y. Tu, X. Yang, Y.-F. Zheng, H. Li, U–Pb dating of zircon from gneiss at Nanhuang in East Anhui, Acta Pet. Sin. 17 (2001) 157 – 160 (in Chinese with English abstract). [12] M. Zhai, B. Cong, J. Guo, W. Liu, Y. Li, Q. Wang, Sm–Nd geochronology and petrography of garnet pyroxene granulites in the northern Sulu region of China and their geotectonic interpretation, Lithos 52 (2000) 23 – 33. [13] M. Faure, W. Lin, N. Le Breton, Where is the North China– South China block boundary in eastern China? Geology 29 (2001) 119 – 122. [14] Z.-X. Li, Collision between the North and South China blocks: a crustal-detachment model for suturing in the region east of the Tanlu fault, Geology 22 (1994) 739 – 742. [15] S.-L. Chung, Trace element and isotope characteristics of Cenozoic basalts around the Tanlu fault with implications for the eastern plate boundary between North and South China, J. Geol. 107 (1999) 301 – 312. [16] W.M. Fan, H.F. Zhang, J. Baker, K.E. Jarvis, P.R.D. Mason, M.A. Menzies, On and off the North China craton: where is the Archaean keel? J. Petrol. 41 (2000) 933 – 950. [17] W.L. Griffin, A. Zhang, S.Y. O’Reilly, C.G. Ryan, Phanerozoic evolution of the lithosphere beneath the Sino-Korean craton, in: M.F.J. Flower, S.-L. Chung, C.-H. Lo, T.-Y. Lee (Eds.), Mantle Dynamics and Plate Interactions in East Asia, vol. 27, AGU, Washington, D.C., 1998, pp. 107 – 126. [18] M.A. Menzies, Y. Xu, Geodynamics of the North China Craton, in: M.F.J. Flower, S.-L. Chung, C.-H. Lo, T.-Y. Lee (Eds.), Mantle Dynamics and Plate Interactions in East Asia, vol. 27, AGU, Washington, D.C., 1998. [19] F.-Y. Wu, J.-Q. Lin, S.A. Wilde, X.O. Zhang, J.-H. Yang, Nature and significance of the early cretaceous giant igneous event in eastern China, Earth Planet. Sci. Lett. 233 (2005) 103 – 119. [20] X. Yuan, Velocity structure of the Qinling lithosphere and mushroom cloud model, Sci. China, Ser. D: Earth Sci. 39 (1996) 235 – 244. [21] S.Y. O’Reilly, W.L. Griffin, Y.H. Poudjom Djomani, P. Morgan, Are lithospheres forever? Tracking changes in subcontinental lithospheric mantle through time, GSA Today 11 (2001) 4 – 10. [22] J. Zheng, S.Y. O’Reilly, W.L. Griffin, F. Lu, M. Zhang, Nature and evolution of Cenozoic lithospheric mantle beneath Shandong Peninsula, Sino-Korean craton, eastern China, Int. Geol. Rev. 40 (1998) 471 – 499. [23] Y.-G. Xu, S.-L. Chung, J. Ma, L. Shi, Contrasting Cenozoic lithospheric evolution and architecture in the western and eastern

306

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

[33]

[34]

[35]

[36]

[37]

[38] [39] [40]

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308 Sino-Korean craton: constraints from geochemistry of basalts and mantle xenoliths, J. Geol. 112 (2004) 593 – 605. Y. Niu, Generation and evolution of basaltic magmas: some basic concepts and a hypothesis for the origin of the Mesozoic–Cenozoic volcanism in eastern China, Geol. J. China Univ. 11 (2005) 9 – 46. S. Gao, R. Rudnick, R.W. Carlson, W.F. McDonough, Y.-S. Liu, Re–Os evidence for replacement of ancient mantle lithosphere beneath the North China craton, Earth Planet. Sci. Lett. 198 (2002) 307 – 322. F.Y. Wu, R.J. Walker, X.W. Ren, D.Y. Sun, X.H. Zhou, Osmium isotopic constraints on the age of lithospheric mantle beneath northeastern China, Chem. Geol. 196 (2003) 107 – 129. S.A. Wilde, X.-H. Zhou, A.A. Nemchin, M. Sun, Mesozoic crust–mantle interaction beneath the North China craton: a consequence of the dispersal of Gondwanaland and accretion of Asia, Geology 31 (2003) 817 – 820. H. Ka´rason, R. van der Hilst, Constraints on mantle convection from seismic tomography, Geophys. Monogr. 121 (2000) 277 – 288. D. Chen, Z. Peng, K–Ar ages and Pb, Sr isotopic characteristics of some Cenozoic volcanic rocks from Anhui and Jiangsu Provinces, China, Acta Pet. Sin. 2 (1988) 3 – 12 (in Chinese with English abstract). R. Liu, The K–Ar age and tectonic environment of Cenozoic volcanic rock in China, in: L.R. (Ed.), The Age and Geochemistry of Cenozoic Volcanic Rocks in China, Seismology Press, Beijing, 1992, pp. 1 – 43 (in Chinese with English abstract). L. Reisberg, X. Zhi, Z. Peng, C. Wagner, W. Sun, The mantle lithosphere beneath eastern China: trace element and Os isotopic results from spinel peridotite xenoliths from Jiangsu Province, EUGXI, Cambridge Publications, Strasbourg, 2001, p. 355. X. Xu, S. O’Reilly, W. Griffin, X. Zhou, X. Huang, The nature of the Cenozoic lithosphere at Nushuan, eastern China, in: M.F.J. Flower, S.-L. Chung, C.-H. Lo, T.-Y. Lee (Eds.), Mantle Dynamics and Plate Interactions in East Asia, vol. 27, AGU, Washington, D.C., 1998, pp. 167 – 195. R.L. Cao, S.-H. Zhu, Mantle xenoliths and alkali-rich host rocks in eastern China, in: P.H. Nixon (Ed.), Mantle Xenoliths, John Wiley and Sons, Chichester, 1987, pp. 167 – 180. X. Xu, S.Y. O’Reilly, W.L. Griffin, X. Zhou, Genesis of young lithospheric mantle in southeastern China: an LAM-ICPMS trace element study, J. Petrol. 41 (2000) 111 – 148. B. Harte, Rock nomenclature with particular relation to deformation and recrystallization textures in olivine-bearing xenoliths, J. Geol. 85 (1977) 29 – 288. J.-C. Mercier, A. Nicolas, Textures and fabrics of upper mantle peridotites as illustrated by xenoliths from basalts, J. Petrol. 16 (1975) 454 – 487. R.M. Bedini, J.-L. Bodinier, J.-M. Dautria, L. Morten, Evolution of LILE-enriched small melt fractions in the lithospheric mantle: a case study from the East African Rift, Earth Planet. Sci. Lett. 153 (1997) 67 – 83. W.F. McDonough, S.-s. Sun, The composition of the earth, Chem. Geol. 120 (1995) 223 – 253. F.R. Boyd, Compositional distinction between oceanic and cratonic lithosphere, Earth Planet. Sci. Lett. 96 (1989) 15 – 26. W.L. Griffin, S.Y. O’Reilly, C.G. Ryan, The composition and origin of sub-continental lithospheric mantle, in: Y. Fei, C.M. Bertka, B.O. Mysen (Eds.), Mantle Petrology: Field Observations and High Pressure Experimentation: A Tribute to Francis

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50] [51]

[52]

[53]

[54]

[55]

[56]

[57]

R. (Joe) Boyd, The Geochemical Society, Special Publication, vol. 6, 1999. R.L. Rudnick, S. Gao, W.-L. Ling, Y.-S. Liu, W.R. McDonough, Petrology and geochemistry of spinel peridotite xenoliths from Hannuoba and Qixia, North China craton, Lithos 77 (2004) 609 – 637. F.A. Frey, M. Prinz, Ultramafic inclusions from San Carlos, Arizona: petrologic and geochemical data bearing on their petrogenesis, Earth Planet. Sci. Lett. 38 (1978) 129 – 176. J.-P. Lor, F. Conque´re´, Contribution a` l’e´tude des paragene`ses sulfure´es dans les enclaves de basalte alcalin du Massif Central et du Languedoc (France), Bull. Mineral. 106 (1983) 585 – 606. C. Szabo, R.J. Bodnar, Chemistry and origin of mantle sulfides in spinel peridotite xenoliths from alkaline basaltic lavas, Nograd-Gomor volcanic field, northern Hungary and southern Slovakia, Geochim. Cosmochim. Acta 59 (1995) 3917 – 3928. J.P. Lorand, Are spinel lherzolite xenoliths representative of the abundance of sulfur in the upper mantle? Geochim. Cosmochim. Acta 54 (1990) 1487 – 1492. D.A. Ionov, J. Hoefs, K.H. Wedepohl, U. Wiechert, Content and isotopic composition of sulphur in ultramafic xenoliths from central Asia, Earth Planet. Sci. Lett. 111 (1992) 269 – 286. M.R. Handler, V. Bennett, Behaviour of platinum-group elements in the subcontinental mantle of eastern Australia during variable metasomatism and melt depletion, Geochim. Cosmochim. Acta 63 (1999) 3597 – 3618. J.-P. Lorand, O. Alard, A. Luguet, R.R. Keays, S/Se systematics of the subcontinental lithospheric mantle beneath the Massif Central, Geochim. Cosmochim. Acta 67 (2003) 4137 – 4153. J.-P. Lorand, Sulphide petrology and sulphur geochemistry of orogenic lherzolites: a comparative study of the Pyrenean bodies (France) and the Lanzo massif (Italy), J. Petrol. Special Volume 1991 – Orogenic Lherzolites and Mantle Processes, 1991, pp. 77 – 95. J.W. Morgan, Ultramafic xenoliths: clues to Earth’s late accretionary history, J. Geophys. Res. 91 (1986) 12,375 – 12,387. J.P. Lorand, Abundance and distribution of Cu-Fe-Ni sulfides, sulfur, copper and platinum-group elements in orogenic-type spinel lherzolite massifs of Arie`ge (northeastern Pyrenees, France), Earth Planet. Sci. Lett. 93 (1989) 50 – 64. T. Meisel, R.J. Walker, A.J. Irving, J.-P. Lorand, Osmium isotopic compositions of mantle xenoliths: a global perspective, Geochim. Cosmochim. Acta 65 (2001) 1311 – 1323. T. Meisel, R.J. Walker, J.W. Morgan, The osmium isotopic composition of the Earth’s primitive upper mantle, Nature 383 (1996) 517 – 520. A.H. Peslier, L. Reisberg, J. Ludden, D. Francis, Re–Os constraints on harzburgite and lherzolite formation in the lithospheric mantle: a study of Northern Canadian Cordillera xenoliths, Geochim. Cosmochim. Acta 64 (2000) 3061 – 3071. O. Alard, W.L. Griffin, N. Pearson, J.P. Lorand, S.Y. O’Reilly, New insights into the Re–Os systematics of sub-continental lithospheric mantle from in-situ analyses of sulfides, Earth Planet. Sci. Lett. 203 (2002) 651 – 663. W.L. Griffin, S. Graham, S.Y. O’Reilley, N.J. Pearson, Lithosphere evolution beneath the Kaapvaal Craton: Re–Os systematics of sulfides in mantle-derived peridotites, Chem. Geol. 208 (2004) 89 – 118. N.J. Pearson, O. Alard, W.L. Griffin, S.E. Jackson, S.Y. O’Reilly, In situ measurement of Re–Os isotopes in mantle sulfides by laser ablation multicollector-inductively coupled

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

[58]

[59]

[60]

[61]

[62]

[63]

[64]

[65]

[66]

[67]

[68]

[69]

[70]

[71]

[72]

[73]

plasma mass spectrometry: analytical methods and preliminary results, Geochim. Cosmochim. Acta 66 (2002) 1037 – 1050. M.R. Handler, V.C. Bennett, T.M. Esat, The persistence of offcratonic lithospheric mantle: Os isotopic systematics of variably metasomatised southeast Australian xenoliths, Earth Planet. Sci. Lett. 151 (1997) 61 – 75. J.C. Lassiter, Rhenium volatility in subaerial lavas: constraints from subaerial and submarine portions of the HSDP-2 Mauna Kea drillcore, Earth Planet. Sci. Lett. 214 (2003) 311 – 325. D.G. Pearson, G.J. Irvine, D.A. Ionov, F.R. Boyd, G.E. Dreibus, Re–Os isotope systematics and platinum group element fractionation during mantle melt extraction: a study of massif and xenolith peridotite suites, Chem. Geol. 208 (2004) 29 – 59. S.A. Wood, Thermodynamic calculations of the volatility of the platinum group elements (PGE): the PGE content of fluids at magmatic temperatures, Geochim. Cosmochim. Acta 51 (1987) 3041 – 3050. K. Righter, E.H. Hauri, Compatibility of rhenium in garnet during mantle melting and magma genesis, Science 280 (1998) 1737 – 1740. K.W. Burton, P. Schiano, J.-L. Birck, C.J. Alle`gre, Osmium isotope disequilibrium between mantle minerals in a spinellherzolite, Earth Planet. Sci. Lett. 172 (1999) 311 – 322. G.A. Gaetani, T.L. Grove, Wetting of mantle olivine by sulfide melt: implications for Re / Os ratios in mantle peridotite and late-stage core formation, Earth Planet. Sci. Lett. 169 (1999) 147 – 163. J.-P. Lorand, O. Alard, Platinum-group element abundances in the upper mantle: new constraints from in situ and whole-rock analyses of Massif Central xenoliths (France), Geochim. Cosmochim. Acta 65 (2001) 2789 – 2806. J.-P. Lorand, L. Reisberg, R.M. Bedini, Platinum-group elements and melt percolation processes in Sidamo spinel peridotite xenoliths, Ethiopia, East African rift, Chem. Geol. 196 (2003) 57 – 75. J.-P. Lorand, G. Delpech, M. Gre´goire, B. Moine, S.Y. O’Reilley, J.-Y. Cottin, Platinum-group elements and the multistage metasomatic history of Kerguelen lithospheric mantle (South Indian Ocean), Chem. Geol. 208 (2004) 195 – 215. W. Sun, V.C. Bennett, S.M. Eggins, R.J. Arculus, M.R. Perfit, Rhenium systematics in submarine MORB and back-arc basin glasses: laser ablation ICP-MS results, Chem. Geol. 196 (2003) 259 – 281. D. Van der Wal, J.-L. Bodinier, Origin of the recrystallisation front in the Ronda peridotite by km-scale pervasive porous mantle flow, Contrib. Mineral. Petrol. 122 (1996) 387 – 406. A.D. Brandon, R.Z. Creaser, S.B. Shirey, R.W. Carlson, Osmium recycling in subduction zones, Science 272 (1996) 861 – 864. Y.-G. Xu, M.A. Menzies, M.F. Thirwall, X.-L. Huang, Y. Liu, X.-M. Chen, bReactiveQ harzburgites from Huinan, NE China: products of the lithosphere–asthenosphere interaction during lithospheric thinning? Geochim. Cosmochim. Acta 67 (2003) 487 – 505. L. Reisberg, J.-P. Lorand, Longevity of sub-continental mantle lithosphere from osmium isotope systematics in orogenic peridotite massifs, Nature 376 (1995) 159 – 162. R.J. Walker, R.W. Carlson, S.B. Shirey, F.R. Boyd, Os, Sr, Nd and Pb isotope systematics of southern African peridotite xenoliths: implications for the chemical evolution of subcontinental mantle, Geochim. Cosmochim. Acta 53 (1989) 1583 – 1595.

307

[74] A.C. Brandon, J.E. Snow, R.J. Walker, J.W. Morgan, T.D. Mock, 190 Pt-186Os and 187Re-187Os systematics of abyssal peridotites, Earth Planet. Sci. Lett. 177 (2000) 319 – 335. [75] S. Esperanc¸a, S.E. Sichel, M.F. Horan, R.J. Walker, T. Juteau, R. Hekinian, Some abyssal peridotites are old!, Ninth Annual V.M. Goldschmidt Conference, pp. Abstract #7389, LPI Contribution No. 971, Lunar and Planetary Institute, Houston (CD-ROM), Boston, 1999. [76] M. Griselin, J. Lassiter, Extreme unradiogenic Os isotopes in Hawaiian mantle xenoliths: implications for mantle convection, Goldschmidt 2002 CD, Cambridge, Davos, 2002. [77] I.J. Parkinson, C.J. Hawkesworth, A.S. Cohen, Ancient mantle in a modern arc: osmium isotopes in Izu-Bonin-Mariana forearc peridotites, Science 281 (1998) 2011 – 2013. [78] B.F. Schaefer, S. Turner, I. Parkinson, N. Rogers, C. Hawkesworth, Evidence for recycled Archaean oceanic mantle lithosphere in the Azores plume, Nature 420 (2002) 304 – 307. [79] J.T. Chesley, R.L. Rudnick, C.-T. Lee, Re–Os systematics of mantle xenoliths from the East African rift: age, structure, and history of the Tanzanian craton, Geochim. Cosmochim. Acta 63 (1999) 1203 – 1217. [80] M. Gros, J.-P. Lorand, A. Luguet, Analysis of Platinum Group Elements (PGE) and Gold in Geological Materials using NiS Fire Assay and Te-Coprecipitation: The NiS Dissolution Step Revisited, 2001. [81] S.B. Shirey, R.J. Walker, Carius tube digestion for low-blank rhenium–osmium analysis, Anal. Chem. 67 (1995) 2136 – 2141. [82] J.-L. Birck, M. Roy Barman, F. Capmas, Re–Os isotopic measurements at the femtomole level in natural samples, Geostand. Newsl. 21 (1997) 19 – 27. [83] T. Meisel, L. Reisberg, J. Moser, J. Carignan, F. Melcher, G. Bru¨gmann, Re–Os systematics of UB-N, a serpentinized peridotite reference material, Chem. Geol. 201 (2003) 161–179. [84] M. Roy Barman, Mesure du Rapport 187Os/186Os dans les Basaltes et les Pe´ridotites: Contribution a` la Syste´matique 187 Re-187Os dans le Manteau, Universite´ de Paris VII, 1993. [85] R.A. Creaser, D. Papanastassiou, G.J. Wasserburg, Negative thermal ion mass spectrometry of osmium, rhenium, and iridium, Geochim. Cosmochim. Acta 55 (1991) 397 – 401. [86] J. Volkening, T. Walczyk, K.G. Heumann, Osmium isotope determinations by negative thermal ionization mass spectrometry, Int. J. Mass Spectrom. Ion Phys. 105 (1991) 147 – 159. [87] Z.X. Li, Tectonic history of the major East Asian lithospheric blocks since the mid-Proterozoic – a synthesis, in: M.F.J. Flower, S.-L. Chung, C.-H. Lo, T.-Y. Lee (Eds.), Mantle Dynamics and Plate Interactions in East Asia, vol. 27, AGU, Washington, D.C., 1998. [88] X.C. Zhi, Trace element geochemistry of Tertiary continental alkali basalts from Liuhe-Yizheng Jiangsu Province, China, Acta Pet. Sin. 2 (1990) 30 – 42 (in Chinese, with English abstract). [89] O.M. Burnham, N.W. Rogers, D.G. Pearson, P.W.v. Calsteren, C.J. Hawkesworth, The petrogenesis of the eastern Pyrenean peridotites: an integrated study of their whole-rock geochemistry and Re–Os isotope composition, Geochim. Cosmochim. Acta 62 (1998) 2293 – 2310. [90] J.S. McBride, D.D. Lambert, A. Greig, I.A. Nicholls, Multistage evolution of Australian subcontinental mantle: Re–Os isotopic constraints from Victorian mantle xenoliths, Geology 24 (1996) 631 – 634. [91] A. Peslier, L. Reisberg, J. Ludden, D. Francis, Os isotopic systematics in mantle xenoliths; age constraints on the Canadian Cordillera lithosphere, Chem. Geol. 166 (2000) 85 – 101.

308

L. Reisberg et al. / Earth and Planetary Science Letters 239 (2005) 286–308

[92] L.C. Reisberg, C.-J. Alle`gre, J.-M. Luck, The Re–Os systematics of the Ronda Ultramafic Complex of southern Spain, Earth Planet. Sci. Lett. 105 (1991) 196 – 213. [93] A.E. Saal, E. Takazawa, F.A. Frey, N. Shimizu, S.R. Hart, Re– Os isotopes in the Horoman peridotite: evidence for refertilization? J. Petrol. 42 (2001) 25 – 37.

[94] G. Schmidt, J. Snow, Os isotopes in mantle xenoliths from the Eifel volcanic field and the Vogelsberg (Germany): age constraints on the lithospheric mantle, Contrib. Mineral. Petrol. 143 (2002) 694 – 705.