Journal of Petrology Advance Access originally published online on October 20, 2005
Journal of Petrology 2006 47(2):355-383; doi:10.1093/petrology/egi078
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Roles of Melting and Metasomatism in the Formation of the Lithospheric Mantle beneath the Leizhou Peninsula, South China
1 STATE KEY LABORATORY FOR MINERAL DEPOSITS RESEARCH, DEPARTMENT OF EARTH SCIENCES, NANJING UNIVERSITY, NANJING 210093, P.R. CHINA
2 GEMOC ARC NATIONAL KEY CENTRE, DEPARTMENT OF EARTH AND PLANETARY SCIENCES, MACQUARIE UNIVERSITY, SYDNEY, N.S.W. 2109, AUSTRALIA
3 CSIRO EXPLORATION AND MINING, NORTH RYDE, N.S.W. 2113, AUSTRALIA
RECEIVED JUNE 16, 2004; ACCEPTED SEPTEMBER 20, 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
This study characterizes the nature of fluid interaction and melting processes in the lithospheric mantle beneath the Yingfengling and Tianyang volcanoes, Leizhou Peninsula, South China, using in situ trace-element analyses of clinopyroxene, amphibole and garnet from a suite of mantle-derived xenoliths. Clinopyroxenes from discrete spinel lherzolites exhibit large compositional variations ranging from extremely light rare earth element (LREE)-depleted to LREE-enriched. Trace-element modelling for depleted samples indicates that the Leizhou lherzolites are the residues of a mantle peridotite source after extraction of
1–11% melt generated by incremental melting in the spinel lherzolite field with the degree of melting increasing upwards from about 60 km to 30 km. This process is consistent with gradational melting at different depths in an upwelling asthenospheric column that subsequently cooled to form the current lithospheric mantle in this region. The calculated melt production rate of this column could generate mafic crust 5–6 km thick, which would account for most of the present-day lower crust. The formation of the lithospheric column is inferred to be related to Mesozoic lithosphere thinning. Al-augite pyroxenites occur in composite xenoliths; these are geochemically similar to HIMU-type ocean island basalt. These pyroxenites postdate the lithospheric column formation and belong to two episodes of magmatism. Early magmatism (forming metapyroxenites) is inferred to have occurred during the opening of the South China Sea Basin (32–15 Ma), whereas the most recent magmatic episode (producing pyroxenites with igneous microstructures) occurred shortly before the eruption of the host magmas (6–0·3 Ma). Trace-element traverses from the contacts of the Al-augite pyroxenite with the spinel peridotite wall-rock in composite xenoliths record gradients in the strength and nature of metasomatic effects away from the contact, showing that equilibrium was not attained. Significant enrichment in highly incompatible elements close to the contacts, with only slight enrichment in Sr, LREE and Nb away from the contact, is inferred to reflect the different diffusion rates of specific trace elements. The observed geochemical gradients in metasomatic zones show that Sr, La, Ce and Nb have the highest diffusion rates, other REE are intermediate, and Zr, Hf and Ti have the lowest diffusion rates. Lower diffusion rates observed for Nb, Zr, Hf and Ti compared with REE may cause high field strength element (HFSE) negative anomalies in metasomatized peridotites. Therefore, metasomatized lherzolites with HFSE negative anomalies do not necessarily require a carbonatitic metasomatizing agent. KEY WORDS: China; lithosphere; mantle xenoliths; clinopyroxene trace elements; mantle partial melting; mantle metasomatism; trace-element diffusion rates
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The lithospheric mantle commonly experiences a long and complex geological history and exhibits extensive horizontal and vertical heterogeneities. Mantle peridotite and pyroxenite xenoliths provide the best window to understand the nature of the mantle and its depletion and fluid movement processes. It has been documented that melt extraction and metasomatism are two of the most important processes influencing the composition of lithospheric mantle (e.g. Zangana et al., 1999
; Xu et al., 2000; Fodor et al., 2002). Using the trace-element composition of depleted peridotites, the partial melting style and degree of melting occurring in mantle peridotites have been modelled (e.g. Johnson et al., 1990; Niu, 1997; Norman, 1998; Xu et al., 2000). The composition of metasomatized peridotites has been used to reveal the nature of enrichment processes and the geochemical signature of metasomatic agents (Menzies et al., 1987; O'Reilly & Griffin, 1988; Rudnick et al., 1993; Baker et al., 1998; Gorring & Kay, 2000; Wang & Gasparik, 2001). In fact, many lithospheric mantle regions record several episodes of both partial melting and metasomatic overprints (Zangana et al., 1999; Xu et al., 2000). In addition, highly depleted peridotites record generally stronger metasomatism in their clinopyroxene (e.g. Frey & Green, 1974; O'Reilly & Griffin, 1988; Grégoire et al., 2000). Metasomatism can change, to varying degrees, the trace-element composition of peridotites even for the weakly incompatible elements or compatible elements, such as Sc, Y and Yb (Norman, 1998; Grégoire et al., 2000). In such cases, modelling partial melting processes using these elements without consideration of metasomatic effects may lead to false conclusions.
In recent studies of mantle metasomatism, the most common approach to defining the nature of metasomatic agents has been to use the partition coefficient between minerals and melt and mineral compositions. However, trace elements may have different diffusion properties depending on specific physico-chemical conditions (Zanetti et al., 1996). In weak metasomatism some elements with low diffusivity may not achieve equilibrium between the peridotite wall-rocks and metasomatic agents. On the other hand, the difference in partition coefficients for mineral/basaltic melt (Dmin/bas) and mineral/carbonatite melt (Dmin/carb) is relevant only for high field strength elements (HFSE) (Hart & Dunn, 1993; Klemme et al., 1995; Vannucci et al., 1998). Therefore, it is very important to seek other evidence to characterize the metasomatic agent.
The suite of xenoliths in this study comprises discrete lherzolites and composite xenoliths (lherzolite with pyroxenite) from the Yingfengling and Tianyang volcanoes of the Leizhou Peninsula, South China, and includes enriched as well as depleted peridotites, and particularly important composite xenoliths. This suite of xenoliths provides an excellent natural compositional range of Cr-diopside mantle wall-rocks that come from a large depth range and show many contact relationships with inferred mantle melt veins (Al-augite rock types), thus allowing detailed exploration of metasomatic processes.
In a previous paper (Yu et al., 2003a), we classified this suite of xenoliths and other discrete metapyroxenites and megacrystic aggregates according to their petrographic and mineral chemical features, calculated their equilibrium temperature and/or pressure conditions, constructed the paleogeotherm based on data from the garnet websterites, demonstrated a high heat flow (from both the geotherm and geophysical data), and confirmed the lithospheric rock-type section for the Leizhou region and the thickness of crust (30 km) and lithosphere (100 km) by integration of the xenolith information with seismic data. In this study, detailed trace-element data for clinopyroxenes, amphiboles and garnets from the range of xenolith types have been obtained by in situ laser-ablation inductively coupled plasma mass spectrometry (LA-ICPMS). These clinopyroxene compositions are then used to discuss partial melting processes recorded in the depleted lherzolites, to characterize the nature of the metasomatic processes, and especially to evaluate the diffusion rate of different trace elements and the resultant effect on the geochemical signatures and patterns of metasomatized lherzolites. In addition, we provide a geodynamic framework for the evolution of the lithospheric mantle in this region.
|
GEOLOGICAL BACKGROUND |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The Leizhou Peninsula and the nearby Hainan Island are located at the southernmost margin of South China continent and the northern edge of the South China Sea Basin (SCSB) (Fig. 1). There are extensive Cenozoic basalt eruptions spreading over this area and surrounding regions, such as in Vietnam and Thailand. These intracontinental volcanic rocks surrounding the SCSB, and those from seamounts within the SCSB, are commonly younger than the SCSB sea-floor extension, i.e. they are post-spreading volcanic rocks, and consist mainly of quartz tholeiite and olivine tholeiite (dominant in early eruptive series), with minor alkali basalts (increasing in later eruptive series: Flower et al., 1992
; Hoang & Flower, 1998). These latter basalts are characterized by depleted Sr–Nd isotopic, Dupal-like Pb isotopic compositions and enriched ocean island basalt (OIB)-type incompatible-element distributions, suggesting the interaction of asthenosphere with subcontinental lithosphere (Zhu & Wang, 1989; Flower et al., 1992; Tu et al., 1992). Isotopic data for these post-spreading basalts indicate the mantle source heterogeneity, reflecting mixing of EM2 and normal mid-ocean ridge basalt (N-MORB) mantle reservoirs. However, enriched trace-element composition and depleted Sr–Nd isotopic compositions suggest that partial melting of this mantle has probably taken place shortly after mantle metasomatism.
|
The SCSB was opened from late Oligocene to middle Miocene (32 to 15·5 Ma, Taylor & Hayes, 1983
; Briais et al., 1993), almost simultaneously with the tectonic extrusion of Indochina along the Red River fault as a result of the northward collision of India with Eurasia (Harrison et al., 1996; Wang et al., 1998). Therefore, some workers interpret the SCSB opening as the result of the southeastward extrusion of Indochina (Tapponnier et al., 1986; Briais et al., 1993), although others believe that the SCSB opening was simply initiated by the extension independent of the tectonic extrusion (Chung et al., 1997; Wang et al., 1998).
Seismic data for South China and the SCSB exhibit significant lithospheric structural changes and obvious thinning of crustal thickness from the northern continental margin to the centre of the SCSB (from 28 km to 8 km; Lin et al., 1988; Yao et al., 1994; Yan et al., 2001). The lower-crust thickness also decreases from the northern margin to the basin centre (Yan et al., 2001), suggesting that the lower crust formed prior to the basin pull-apart. The lithosphere of South China is thin; the old lithosphere appears to have been removed and replaced by new upwelling mantle, which took probably place during the Mesozoic (Xu et al., 2000). Studies on the granulite xenoliths from South China indicate that lower crustal mafic granulites were dominantly formed by Mesozoic basalt underplating (e.g. Yu et al., 2003c). Geological and geophysical data show only a small amount of magma intruded or erupted during the South China Sea opening (Yan et al., 2001), whereas extensive volcanism took place after the cessation of the SCSB floor spreading.
|
XENOLITH PETROGRAPHY |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
The studied xenolith samples were collected from Yingfengling and Tianyang basaltic volcanoes, 15 km apart, in the centre of the Leizhou Peninsula (Fig. 1). These two post-spreading Cenozoic volcanic vents are composed of several layers of lava and breccia, and interlayered sediments. The lavas are dated from 6·1 Ma to 0·3 Ma (from base to top: Zhu & Wang, 1989
; Xu et al., 1999). The xenoliths from the Yingfengling volcano represent a wide range of rock types, 2–10 cm across, and angular or subangular in shape, whereas those from the Tianyang volcano are peridotites generally 20 cm across with rounded shapes. The xenolith suite includes anhydrous Cr-diopside spinel lherzolites, spinel websterites, layered Cr-diopside xenoliths, and composite Al-augite–Cr-diopside spinel xenoliths (Table 1). Discrete Cr-diopside sp-lherzolites have large modal variations with 44·2–70·8% olivine, 15·1–37·2% orthopyroxene, 6·2–28·7% Cr-diopside and 0·7–3·6% spinel. Sp-websterites contain mainly orthopyroxene (38·5–50·5%) and Cr-diopside (45·0–57·8%). Layered Cr-diopside xenoliths consist of sp-lherzolite and sp-websterite. Al-augite composite xenoliths comprise sp-lherzolite domains in contact with Al-augite pyroxenite or very coarse-grained mafic assemblages referred to here as megacrystic aggregates that have been described in detail by Yu & O'Reilly (2001) and Yu et al. (2003a, 2003b) and listed in Table 1. The megacrystic aggregate in composite xenolith Lz-17 consists of Al-augite, Mg-, Ca-rich almandine, oligoclase and ilmenite, with rare small grains of biotite (0·04 mm long) occurring as inclusions in ilmenite. The megacrystic aggregate in sample Lz-29 consists of medium-grained Al-augite, Mg-, Ca-rich almandine, ilmenite and apatite (Yu & O'Reilly, 2001; Yu et al., 2003b).
|
Amphibole occurs as interstitial grains or as rims on spinel in lherzolite and is mainly confined to the composite xenoliths indicating a metasomatic origin. Discrete sp-lherzolite Lz-21 also contains amphibole (0·4%) and its low Mg-number indicates modal metasomatism.
Detailed major-element compositions of minerals from these xenoliths have been reported by Yu et al. (2003a). Most discrete lherzolites have similar major element compositions with high Mg-number of olivine (89·5–90·4), orthopyroxene (90·0–90·8), and clinopyroxene (89·7–91·9) (Table 1). Samples Lz-7 and Lz-21 have lower Mg-number, and Lz-7 shows large compositional variation on a centimeter scale with, for example, Mg-number in clinopyroxene ranging from 84·7 to 89·8. Sp-websterites exhibit similar mineral compositions to Cr-diopside sp-lherzolite, except for lower Cr-number of spinel. The minerals in the layered sp-lherzolite and sp-websterite domains of Cr-diopside composite xenoliths have identical mineral compositions indistinguishable from those of the discrete lherzolites and websterites. Minerals from Al-augite composite xenoliths have very low Mg-number and marked compositional variations, especially in lherzolite–megacryst aggregate composites (Table 1).
Temperatures of these xenoliths calculated by the Brey & Köhler (1990) two-pyroxene thermometer range from 870°C to 1130°C, spanning a depth range from 63 to 30 km when applied to the xenolith-derived geotherm for this region (Yu et al., 2003a). Sp-websterites, layered Cr-diopside xenoliths and spinel lherzolites with high-Mg clinopyroxene (Table 1) have the lowest temperatures (870–950°C) and are inferred to come from shallower levels (28–37 km). Al-augite composites generally have higher temperatures and thus occur deeper in the section. Identical calculated temperatures for metapyroxenite and lherzolite rock types in some composite xenoliths indicate thermal equilibrium (Table 1).
|
ANALYTICAL TECHNIQUES |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Olivine, orthopyroxene, diopside and spinel separates were hand-picked from coarsely crushed Cr-diopside sp-lherzolite xenoliths. These grains were mounted and polished for analysis by electron microprobe and LA-ICPMS. Polished sections 100 µm thick were also made for layered and composite xenoliths. Concentrations of 31 trace elements (Sc, Ti, V, Co, Ni, Zn, Ga, Rb, Sr, Y, Zr, Nb, Ba, REE, Hf, Pb, Th and U) for clinopyroxene, amphibole and garnet (Tables 2 and 3) were determined in situ by LA-ICPMS with a Perkin Elmer Elan 6000 instrument in the GEMOC National Key Centre laboratories at Macquarie University, Australia. Analytical procedure and precision are the same as those described in detail by Grégoire et al. (2000)
and Xu et al. (2000). Almost all clinopyroxenes have Rb abundances below the detection limit (0·1 ppm), so are not listed in Table 2.
|
|
At least five points were analysed for clinopyroxene grains in each discrete Cr-diopside sp-lherzolite and sp-websterite sample, and more than 13 points for grains in composite xenoliths (up to 41 points for Lz-10) to cover grains across the whole section or for a coverage of 5–7 cm2 for mounted grains in order to assess homogeneity. Analyses show that clinopyroxene compositions in most discrete Cr-diopside lherzolites, sp-websterites and layered Cr-diopside xenoliths are homogeneous, except for those from Cr-diopside lherzolites Lz-7 and Lz-21. Clinopyroxenes from these two lherzolites have large trace-element abundance variations, with Mg-number varying on a centimeter scale, similar to the clinopyroxene Mg-number variation for composite xenoliths. This indicates that these samples were close to pyroxenite veins and accordingly metasomatized, and thus they are described together with the composite xenoliths below.
|
Trace-element compositions of clinopyroxenes |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Transition metals
Clinopyroxenes from discrete Leizhou Cr-diopside lherzolites have relatively uniform Sc (55·2–69·2 ppm), V (237–265 ppm), Co (18·3–23·5 ppm) and Ni contents (282–373 ppm) (Table 2). Because of their similarity in these and other trace elements (see below), discrete Cr-diopside lherzolites, Cr-diopside sp-websterites and layered Cr-diopside xenoliths are considered together (Fig. 2).
|
Sc contents of clinopyroxenes from Al-augite composite xenoliths are significantly lower than those from Cr-diopside suite xenoliths, and decrease from lherzolite to pyroxenite domains with decreasing Mg-number (Table 2, Fig. 2a). V abundances are similar to those of the Cr-diopside suite xenoliths (Fig. 2b). Clinopyroxenes in Al-augite composite lherzolites have higher Co contents than those in Cr-diopside suite xenoliths; pyroxenite domains have the highest Co values, especially in megacrystic aggregate–lherzolite composites (Lz-17 and Lz-29; Fig. 2c). Clinopyroxenes of most Al-augite composites display similar Ni contents to those from Cr-diopside suite xenoliths, with a slight decrease in Ni from lherzolite to pyroxenite. In megacrystic rock types Lz-17 and Lz-29, the Ni content of clinopyroxene is very low (Fig. 2d).
In the plots of Mg-number vs transition metals, Sc and Mg-number correlate positively, and Ni and Mg-number negatively in the Cr-diopside suite xenoliths (Fig. 2). However, in each Al-augite composite xenolith, Mg-number in clinopyroxene correlates positively with both Sc and Ni, and negatively with Co (Fig. 2). It is worth noting that the Ni correlation in Cr-diopside suite xenoliths is opposite to that for Al-augite composite xenoliths (Fig. 2d). Ni in clinopyroxenes from Cr-diopside suite xenoliths is positively correlated with the temperature (Fig. 3a), which implies that Ni distribution in clinopyroxene of Cr-diopside suite xenoliths is probably controlled by temperature. Ni variation in Al-augite composite xenoliths is evidently the result of interaction of mafic melt and wall-rock Cr-diopside lherzolite.
|
Incompatible trace elements
Clinopyroxenes of Cr-diopside suite xenoliths have uniform and low Ga contents (2·8–4·8 ppm) (Table 2, Fig. 4a), and those from composite xenoliths have higher Ga contents that increase significantly toward the pyroxenite portion with decreasing Mg-number (Fig. 4a). In contrast, Sr displays large variations in clinopyroxenes of different Cr-diopside suite xenoliths, but very limited variation within composite xenoliths despite the decrease in Mg-number from lherzolite to pyroxenite (Fig. 4b).
|
Ti in clinopyroxenes has the same pattern of variation as Ga, with low abundances and limited variation in the Cr-diopside suite xenoliths and high abundances and large variations in the composite xenoliths (Fig. 4c). Nb, La and Ce are similar to Sr, with large variations in Cr-diopside suite xenoliths and slight variation in each composite xenolith (Fig. 4d and e). Clinopyroxenes in discrete Cr-diopside suite xenoliths have lower abundances of Sr, Nb, Zr, Ti and light REE (LREE) but higher Y and heavy REE (HREE) than those in both the Cr-diopside lherzolite and Al-augite pyroxenite domains of composite xenoliths (Fig. 4b–g). All REE in clinopyroxenes from Cr-diopside suite xenoliths decrease as Mg-number increases, whereas LREE and HREE in clinopyroxenes from Al-augite composite xenoliths exhibit different variation trends (Fig. 4 and inset in Fig. 4b and g).
Clinopyroxenes of the Cr-diopside suite xenoliths display different REE patterns with large (La/Yb)n variations (0·05–1·98), including LREE-depleted patterns (Fig. 5a) and LREE- to middle REE (MREE)-depleted patterns with LREE inflections (Fig. 5a and b). However, discrete Cr-diopside lherzolites Lz-2 and Lz-9 show remarkable LREE–MREE enrichment and relatively low HREE [(La/Yb)n >1] with a convex-upward shape pivot at Nd (Fig. 5a), although they have the same high Mg-number as the other discrete lherzolites. This enriched REE pattern is similar to that of some Al-augite composites (Fig. 5c–h), indicating that lherzolite Lz-2 and Lz-9 have been strongly metasomatized. All clinopyroxenes show a negative correlation of (La/Yb)n ratios with Mg-number (Fig. 4h) and some show a positive correlation with temperature (Fig. 3d).
|
Clinopyroxene REE patterns for Cr-diopside lherzolite parts of Al-augite composite xenoliths are distinct from those for discrete depleted Cr-diopside suite xenoliths, and more similar to metasomatized samples Lz-2 and Lz-9 (Fig. 5c–h). Clinopyroxene commonly displays large differences in REE patterns (Fig. 5c–f) from lherzolite to pyroxenite domains in composite xenoliths. Clinopyroxenes in garnet-bearing pyroxenite of composites Lz-10, Lz-17, Lz-29 and Lz-37 have very low HREE abundances, probably as a result of HREE partitioning preferentially into garnet. However, those from garnet-free pyroxenites also have high LREE and low HREE relative to wall-rock lherzolite clinopyroxenes (e.g. composite xenolith Lz-20 and Lz-1), and clinopyroxenes from Cr-diopside domains in composite xenoliths have lower HREE contents than those from Cr-diopside suite lherzolites. This indicates that parental magmas of the pyroxenites are more enriched in LREE and more depleted in HREE than an assumed melt in equilibrium with wall-rock lherzolite.
Most clinopyroxenes from Cr-diopside suite xenoliths have normalized trace-element patterns (Fig. 6) that show positive slopes from Th through Sr to MREE with significant negative anomalies in Nb, Zr and Ti (Fig. 6a and b), similar to those in fertile lherzolites from eastern Australia (O'Reilly & Griffin, 1988) and abyssal peridotites (Johnson et al., 1990), but different from those in sp-lherzolites from Nushan and Niutoushan [more northerly localities in southeastern China (Xu et al., 2000)]. Nushan sp-lherzolite clinopyroxenes (Group B) have lower HREE contents and higher concentrations of highly incompatible elements with weak Nb anomalies, similar to lherzolites Lz-2 and Lz-9 of this study (Fig. 6a). Niutoushan sp-lherzolite clinopyroxenes exhibit universally high incompatible element concentrations with stronger Nb negative anomalies (Xu et al., 2000).
|
Incompatible-element abundance distributions of clinopyroxenes from Cr-diopside lherzolite domains of Al-augite composite xenoliths (including Lz-7 and Lz-21) are different from those of most Cr-diopside suite xenoliths (Fig. 6c–f). Composite xenolith clinopyroxenes often exhibit incompatible-element abundance distribution changes from the lherzolite to pyroxenite domains, especially in Sr, Zr and Ti anomalies. The strength of Zr and Ti anomalies decreases from lherzolite to pyroxenite domains and the Sr anomaly changes from positive to negative (e.g. Ti in composite xenolith Lz-10, Lz-17 and Lz-20 and Sr in sample Lz-10 and Lz-17; Fig. 6c–e).
HFSE negative anomalies are very common in clinopyroxenes of mantle peridotites (Salters & Shimizu, 1988; Johnson et al., 1990; Xu, et al., 2000). Similar to the abyssal peridotites, clinopyroxenes of Cr-diopside suite xenoliths from the Leizhou Peninsula (except for Lz-2 and Lz-9) show increasingly negative Zr anomalies with decreasing Zr content of clinopyroxenes; that is, positive correlation of Zr/Zr* with (Zr)n (Fig. 7a). Clinopyroxenes in Al-augite composites and Lz-2 and Lz-9 all plot to the right of the Zr/Zr* vs (Zr)n correlation line for the Cr-diopside suite xenoliths, and show distinct differences in variation trend between lherzolite and pyroxenite. For example, the Zr content increases from lherzolite domain to pyroxenite in composite xenoliths Lz-10, Lz-17, Lz-20 and Lz-21, whereas the Zr/Zr* ratio decreases at first, and then increases (Fig. 7a).
|
Similarly, most clinopyroxenes of the Cr-diopside suite xenoliths show positive correlations between Nb/Nb* and Nb (Fig. 7b), but Lz-2, Lz-8, Lz-9 and Lz-12 deviate significantly from the correlation line with high Nb contents, possibly reflecting their metasomatic signature as indicated by their REE patterns (Fig. 5a). Clinopyroxenes of composite xenoliths have higher Nb and Nb/Nb*, and their Nb/Nb* values increase with Nb from lherzolite domain to pyroxenite domain (Fig. 7b). Ti/Ti* of clinopyroxenes in Cr-diopside suite xenoliths are low and show little variation (Fig. 7c), consistent with the observation of Johnson et al. (1990)
, whereas Ti/Ti* of clinopyroxenes from composite xenoliths are higher and increase with increasing Ti. Sr/Sr* of clinopyroxene may be the best indicator of metasomatic intensity. Clinopyroxenes from most Cr-diopside suite xenoliths (Fig. 7d) have high Sr/Sr* ratios, low Sr contents and a positive correlation of Sr/Sr* with Sr, whereas those from Al-augite composite xenoliths and metasomatized lherzolite (e.g. Lz-2 and Lz-9) exhibit negative correlations between Sr/Sr* and Sr as well as a marked negative Sr anomaly (Sr/Sr* < 1, Fig. 7d). This indicates that LREE to MREE (i.e. Ce and Nd) increase more than Sr during the type of metasomatism that affected these rocks.
|
TRACE ELEMENTS IN AMPHIBOLE AND GARNET FROM COMPOSITE XENOLITHS |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
All analyzed amphiboles (Table 3) have REE patterns similar to those of the enriched clinopyroxenes (Fig. 8a). However, amphiboles have distinctive incompatible trace-element abundance patterns, with obvious Ba, Nb, Sr and Ti positive anomalies and Th, U, Pb, Zr and HREE negative anomalies (Fig. 8b), similar to those for Vitim xenoliths (Litasov et al., 2000
) and Lherz composite peridotite samples (Zanetti et al., 1996), but different from those for Kerguelen xenoliths (Grégoire et al., 2000). Amphibole from megacrystic composite xenolith Lz-29 has higher LREE and MREE than other samples and exhibits slightly different incompatible trace-element patterns with no positive Ti anomaly (Fig. 8a and b), indicating that these amphiboles were generated from melts with different compositions, analogous to amphibole origins for western Victorian (southeastern Australia) xenoliths (Powell et al., 2004).
|
Garnet occurs only in the metapyroxenites of Al-augite composites Lz-10 and Lz-37 and in megacrystic aggregates of composite xenoliths Lz-17 and Lz-29. These latter garnets show much lower Mg-number (Table 1) and higher CaO content (7·07–7·67%) than garnets from metapyroxenites Lz-10 and Lz-37 (CaO 5·11–5·49%) (Yu et al., 2003a
). Compared with garnets in garnet peridotites from Nushan and Mingxi, southeastern China (Xu et al., 2000) and Vitim, Russia (Ionov et al., 1993), which have low LREE and high HREE patterns, Leizhou pyroxenite garnets have REE patterns with highest MREE and low LREE and HREE (Table 3; Fig. 8c) and their trace-element patterns have strong negative Sr and Ti anomalies (Fig. 8d). Garnets from megacrystic composites Lz-17 and Lz-29 show higher LREE and MREE and slightly lower HREE than those from metapyroxenite Lz-10 and Lz-37. Combined with their lower Mg-number, this suggests that the megacrystic aggregates have crystallized from a relatively more evolved magma than the metapyroxenites. |
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONGEOLOGICAL BACKGROUNDXENOLITH PETROGRAPHYANALYTICAL TECHNIQUESTrace-element compositions of...TRACE ELEMENTS IN AMPHIBOLE...DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Partial melting origin for Leizhou depleted peridotites
Clinopyroxene is a major carrier of most trace elements and commonly has a similar incompatible-element pattern to its host peridotite (but 1–2 orders higher), except for lower Ba and Nb, and slight negative Ti anomalies (Chalot-Prat & Boullier, 1997
; Zangana et al., 1999; Grégoire et al., 2000; Xu et al., 2003). Accordingly, it has been used to effectively model the melting process and metasomatic enrichment in mantle peridotites (e.g. Johnson et al, 1990; Norman, 1998; Grégoire et al., 2000; Xu et al., 2000; Wang & Gasparik, 2001; Neumann et al., 2004).
Many clinopyroxenes of Leizhou Cr-diopside suite xenoliths are LREE-depleted (Fig. 5a and b), and their incompatible element contents and (La/Yb)n, Zr/Zr* and Sr/Sr* ratios decrease with increasing Mg-number (Figs 4 and 7), suggesting variable degrees of melt extraction. The good positive correlation of incompatible trace elements and (La/Yb)n ratios with temperature (Fig. 3c and d) is consistent with the inference that the Leizhou upper mantle represents an earlier residual mantle column generated by decompression melting at different depths of an adiabatic upwelling asthenosphere (rather than a metasomatic event and/or mafic magma intrusions), analogous to mid-ocean ridge melting (McKenzie & Bickle, 1988; Yang et al., 1998). Therefore, the trace-element contents of clinopyroxenes from the depleted lherzolites in this study may provide clues about the geodynamic history of the lithospheric mantle beneath the Leizhou Peninsula, South China.
So far three melting models, batch, fractional and incremental melting, have been used to fit the composition variation of the mantle peridotites (e.g. Johnson et al., 1990; Norman, 1998; Yang et al., 1998; Xu et al., 2000). Fractional melting depletes the residue in incompatible elements far more effectively than batch melting does and the two processes are easily distinguished in REE diagrams and plots of incompatible elements in the residues (Johnson et al., 1990; Norman, 1998). Incremental melting is intermediate to batch and fractional melting. In this study all three melting models have been tested to choose the best for explaining compositional variation observed in Leizhou lherzolites.
Melting style and extents were modelled using REE, Y, Zr, and Sr concentrations in clinopyroxenes in this study. The equations are those given by Johnson et al. (1990). The input parameters needed in these models are starting source composition, partition coefficients, and proportions of minerals in the source and those contributing to the liquid. The model starting composition was inferred to be a residue after removing 1–3% melt generated by batch melting from the primitive mantle (Hart & Zindler, 1986; Johnson et al., 1990; Hirschmann & Stolper, 1996; Yang et al., 1998) which these workers preferred, based on mass balance and 143Nd/144Nd and 177Hf/176Hf ratios. The present study accepts such depleted peridotite after 2% melt removal as the source composition. The mineral proportions of this slightly depleted source and those phases entering the melt are taken from Yang et al. (1998) and Johnson et al. (1990), respectively, and partition coefficients for different minerals are from published data (Table 4).
|
Y and Yb modelling
Y and Yb concentrations of clinopyroxenes were used by Norman (1998)
and Xu et al. (2000) to model the partial melting process, because of their low mobility relative to LREE. However, our data show that metasomatism can result in not only enrichment in LREE and large ion lithophile elements (LILE) but also at least relative depletion in HREE and Y in the Cr-diopside lherzolite xenoliths (Figs 4 and 5), especially in strongly metasomatized ones. The same observations have been reported in other metasomatized samples worldwide, e.g. Shadwell peridotites, southeastern Australia (Norman, 1998), Dariganga peridotites of Mongolia (Wiechert et al., 1997) and clinopyroxene inclusions in diamonds from Liaoning, North China (Wang & Gasparik, 2001). Thus modelling based on Y and Yb concentrations of metasomatized clinopyroxenes may lead to overestimation of the extent of partial melting. Therefore melt modelling was carried out only for LREE-depleted clinopyroxenes to avoid such ambiguity. Modelling results show that Y and Yb contents of clinopyroxenes from Leizhou lherzolites can be reproduced by both batch and fractional melting with reasonable degrees of melting (Fig. 9a). Only 1–7% fractional melting is necessary to generate Y and Yb contents observed in these lherzolite clinopyroxenes, and about 10% melt needs to be removed in batch melting model to form the most depleted lherzolite Lz-11. Evidently, Y and Yb concentrations are insensitive to melting styles for the lherzolites with low degrees of melting.
|
Zr and Sr modelling
Zr and Sr concentrations of clinopyroxenes were used by Yang et al. (1998)
to model melting processes, because they may more effectively distinguish batch melting from fractional melting models. However, Sr is affected easily by metasomatism, so possible metasomatic effects must be ruled out before modelling. The Zr–Sr plots of Cr-diopside suite xenoliths (Fig. 9b–d) show that depleted and slightly metasomatized clinopyroxenes result in a consistent correlation whereas more strongly metasomatized samples, such as Lz-2, Lz-9, Lz-12 and Lz-8, deviate significantly and are more enriched in Sr at given Zr levels.
Melting trends based upon three models (batch, fractional and incremental melting) are shown in Fig. 9b–d. The compositional variations for Zr and Sr in the Leizhou Cr-diopside wall-rock xenoliths can be successfully modelled using a Sr content of 80 ppm for the source (source 2 of Fig. 9b–d), rather than the 38·1 ppm used in the modelling of Yang et al. (1998) (source 1 of Fig. 9b–d).
Although all three models (batch, fractional and incremental) reproduce the natural trends in Sr and Zr of the clinopyroxenes, different melting degrees are required for each melting style. To reproduce Zr and Sr contents of the most depleted sample Lz-11, an unrealistic 50% batch melting would be required (Fig. 9b). In contrast, fractional melting and incremental melting can both produce residual peridotite with the Sr and Zr values of Lz-11 with reasonable degrees of melting of about 9% (Fig. 9c) and 12% (Fig. 9d), respectively. However, this Zr and Sr modelling method cannot further discriminate effectively between these two melting styles.
REE modelling
Total REE patterns in peridotite clinopyroxenes provide important constraints to test melting models for unmetasomatized mantle peridotites (Johnson et al., 1990; Yang et al., 1998). All three melt models generate REE patterns that are consistent with the observed REE abundances in lightly depleted Cr-diopside wall-rock xenoliths from the Leizhou Peninsula (Fig. 10). However, batch melting does not yield matching patterns for REE for the highly depleted clinopyroxenes (Fig. 10a), for example, as 20% batch melting generates a pattern with higher LREE and lower HREE than the natural samples (e.g. Lz-11).
|
Fractional melting produces a matching REE pattern for the slightly depleted lherzolite at 4% melting (Fig. 10b), but more than 5% fractional melting more rapidly depletes the LREE relative to HREE, inconsistent with the observed clinopyroxene patterns for the Leizhou highly depleted lherzolites (Fig. 10b). Incremental melting with 2% increments best fits the REE characteristics of all Leizhou lherzolite clinopyroxenes, and Fig. 10c illustrates that the range of Leizhou lherzolites may be generated by from about to (1 to 10%) incremental melting (at increments of 1–2%) of a depleted mantle source.
Geodynamic considerations of melt modelling results
Incremental melt models have been shown to be the best matches for the geochemical parameters of the natural samples from the Leizhou lithospheric mantle and this can be reconciled with geodynamic processes relevant to its formation.
Johnson et al. (1990) suggested that the compositional variations observed in abyssal peridotites, inferred to be the residues of partial melting processes that produce MORB, reflect a process of melt extraction that involves the removal of small increments of melt (about 0·1% increment). Yang et al. (1998) considered that a 1% incremental batch melting model yields the best fit to the REE, Sr and Zr abundances in clinopyroxenes from Hawaiian lherzolites. Experimental studies and theoretical calculations also indicate that mantle melting beneath mid-ocean ridge environments favours a near fractional melting process rather than pure fractional melting and batch melting (McKenzie, 1984; Riley & Kohlstedt, 1991; Langmuir et al., 1992; Iwamori et al., 1995). Perfect fractional melting is dynamically more difficult for large mantle melt columns or domains because of the difficulty in segregating sufficient melt volumes, and the mantle wall-rock xenoliths in this study have been shown (by the large temperature range) to represent samples from a large vertical section (30 km). Upwelling asthenospheric mantle such as occurs during lithosphere extension or other convective upwelling would be characterized by this type of incremental melt column segregation process, similar to that described by McKenzie & Bickle (1988). A higher incremental melting fraction (1–2%) for Leizhou xenolithic lherzolites than that of abyssal peridotites (0·1%, Johnson et al., 1990) probably reflects the difference in the mantle upwelling rate. Smaller incremental step-melting may correspond to conditions of fast upwelling and strong extension (e.g. Pacific mid-oceanic ridge), whereas larger incremental step-melting was limited for conditions of slow extension and low rates of asthenosphere upwelling (e.g. rifting near a continental margin; Leizhou, this study).
Melt production rate (dF/dP) and mafic lower crustal thickness in the Leizhou lithosphere
Melt production rate is the ratio of the difference between the highest and lowest degrees of melting (dF) to the difference in their equilibrium pressures (dP) in a residual mantle column (Langmuir et al., 1992; Yang et al., 1998). The melt modelling calculations above indicate that the highest and lowest melt fractions are 10–12% and 1% respectively for the Leizhou lithospheric mantle observed at present, and thus (dF) is about 10%. Figure 3c and d shows that incompatible element concentration is positively correlated with temperature, suggesting that the melting degrees of lherzolites decreased with increasing temperature (proxy for depth). As there are no robust methods for geobarometry for spinel lherzolites, the pressures of Leizhou lherzolites are estimated referring calculated temperatures (Brey & Köhler, 1990, two-pyroxene thermometer) to the well-constrained xenolith-derived geotherm (Yu et al., 2003a). The temperature range of the sampled mantle wall-rock lherzolites (except for four highly metasomatized lherzolites) is from 870°C to 1090°C (Table 1), corresponding to pressures of 9 and 18 kbar respectively and a depth range of about 30–60 km. This results in an estimated linear mean melt production rate of about 1·1%/kbar for the Leizhou lithospheric mantle over its 30 km depth range.
This melt production rate is similar to the estimations (1·2–1·3%/kbar) for the oceanic mantle beneath mid-ocean ridges (Langmuir et al., 1992; Iwamori et al., 1995; Shen & Forsyth, 1995). It is therefore suggested that Leizhou lithosphere mantle was formed in a strong extensional setting, probably related to Mesozoic lithospheric thinning of South China (Xu et al., 2000), and represents an upwelling asthenospheric melting column that cooled to become new lithospheric mantle.
The extent of crustal growth as a result of intrusion of mafic melts near the crust–mantle boundary from such a melting event can be calculated to be about 5–6 km according to the melt production rate and the formula proposed by Forsyth (1993), and previously had been modelled for a similar tectonic environment in eastern Australia by Cull et al. (1991).
Seismic surveys show that the crustal thickness of the studied area is about 28 km and that the lower crust velocities are consistent with mafic granulites from about 20 km to 28 km (Lin et al., 1988). In addition, Yu et al. (2003a, c) used xenolith suites to demonstrate that the lower crust in this region is indeed composed of mafic granulites and cumulates. These observations and the calculation of the thickness of basaltic addition from the inferred degrees of partial melting suggest that the lower crust beneath the Leizhou Peninsula was probably generated by underplating during the formation of new lithospheric mantle by cooling of a column of upwelling asthenosphere that had undergone differential and gradational partial melting over its vertical extent.
Metasomatic modification of the Leizhou lithospheric mantle
Metasomatic effects
Almost all discrete lherzolites from the Leizhou Peninsula lack modal metasomatic phases, but show a wide range of modes and geochemical compositions, reflecting significant vertical mantle heterogeneities. The melting calculations detailed above explain trace-element variations of depleted lherzolites as due to different degrees of depletion. However, all the enriched lherzolites, including those in composite xenoliths, have geochemical signatures (Figs 4–7) that cannot be interpreted by melt depletion processes and reflect subsequent metasomatic events.
The similarity in clinopyroxene REE patterns in both Cr-diopside lherzolites and Al-augite pyroxenites juxtaposed in some composite xenoliths (e.g. Lz-1 and Lz-29; Fig. 5g and h) suggests that lherzolites were strongly metasomatized by mafic melts parental to the intruding Al-augite rock types (or that both rock types were metasomatized by a similar agent).
However, Cr-diopside lherzolites in other composite xenoliths (Lz-10, Lz-17 and Lz-7) have clinopyroxene REE patterns distinct from those of the pyroxenites, with lower (La/Yb)n and higher (La/Pr)n (Figs 5c–f and 11). The extent of metasomatism (or enrichment) decreases distinctly with distance from the contact and the sequence of changing REE pattern is typical of chromatographic changes in percolation processes (Navon & Stolper, 1987; O'Reilly & Griffin, 1988; O'Reilly et al., 1991).
|
Clinopyroxenes from discrete lherzolites Lz-2 and Lz-9 show strong cryptic metasomatism (Figs 5a, 6a and 11) with enrichment in LREE, Sr and Nb, and relative depletion in Y and HREE. Moderately metasomatized Cr-diopside lherzolites (e.g. Lz-8 and Lz-12) have clinopyroxene REE patterns with (La/Yb)n
1 and (La/Pr)n >1 (Figs 5a and 11). In this metasomatic type, most of the highly incompatible elements are moderately enriched, and Y and HREE weakly depleted. Clinopyroxenes from weakly metasomatized lherzolites are slightly enriched in La, Ce, Sr and Nb compared with the depleted ones, but still have (La/Yb)n <1 with an upward inflection at La and Ce with (La/Pr)n >1 (e.g. Lz-11, xw97-1 and Lz-7).
Good discrimination of different metasomatic types can be shown in plots of (La/Yb)n vs (La/Pr)n (Fig. 11), similar to the plots used by Xu et al. (2000), which distinguish different types of REE patterns. Original depleted lherzolite clinopyroxenes plot in the lower-left field of this plot, and highly enriched clinopyroxenes from lherzolites and pyroxenites are in the upper-left field. Weakly metasomatized samples plot in the lower-right quadrant, and moderately metasomatized ones in the upper-right quadrant.
Nature of metasomatic agents
Differences in clinopyroxene compositions between enriched discrete lherzolite and composite lherzolite document distinct metasomatic agents and variations in the degree of metasomatism. Previous studies have demonstrated various metasomatic agents occurring in mantle peridotites, including carbonatitic melts (e.g. Baker et al., 1998; Yaxley et al., 1998; Gorring & Kay, 2000; Wang & Gasparik, 2001), siliceous melts (e.g. Menzies et al., 1987; Vannucci et al., 1998; Zangana et al., 1999; Grégoire et al., 2000), adakitic (Kepezhinskas et al., 1995, 1996; Schiano et al., 1995), melilitic or melanephelinitic melts (Chalot-Prat & Boullier, 1997) and fluids (CO2-, H2O-, halogen- or P-rich) (O'Reilly & Griffin, 1988; Baker et al., 1998; Gorring & Kay, 2000; Larsen et al., 2003). These metasomatic agents may give rise to distinct chemical variations for the mantle peridotites but may be difficult to distinguish using lherzolitic clinopyroxene compositions alone (Vannucci et al., 1998).
Composite xenoliths
The compositions of clinopyroxenes vary sequentially from Cr-diopside lherzolites to juxtaposed pyroxenites in the composite xenoliths (Figs 4–6), and thus it is inferred that the lherzolites were metasomatized by the siliceous melts or fluids released from them that formed the pyroxenites. These clinopyroxenes are relatively low in MgO, very low in HREE, Y, Sc and Ni, and enriched in LREE–MREE (La to Gd), Sr, Nb, Zr, Hf, Ti, Ga and Co (Figs 2 and 4) compared with clinopyroxenes in the depleted discrete Cr-diopside lherzolites. Because of the strong compositional gradients observed in the lherzolites in contact with pyroxenites, it is generally not possible to simply calculate the composition of the original metasomatic agent using partition coefficients and enriched lherzolitic clinopyroxenes. However, the composition of the pyroxenites most distant from the contact zone may directly reflect the chemical features of their parental magma, and this magma may represent the original metasomatic agent. In addition, amphibole compositions in pyroxenite and lherzolite near pyroxenite also may reflect chemical fingerprints of the metasomatic agent (Bodinier et al., 2004).
The parental magmas of pyroxenites in composite xenoliths have been calculated using the clinopyroxene compositions and the partition coefficients of Hart & Dunn (1993). The metasomatic melts in equilibrium with amphiboles were calculated using the partition coefficients of McKenzie & O'Nions (1991), Adam & Green (1994) and La Tourrette et al. (1995). The calculation results show that the melts deduced from clinopyroxene and amphibole have similar compositions (Fig. 12), consistent with the observation of Bodinier et al. (2004), suggesting the clinopyroxenes have re-equilibrated with the metasomatic melt. These calculated melts have very high incompatible element abundances with obvious Pb and slight Ti negative anomalies (Fig. 12). These trace-element signatures resemble those of basanites from Nushan, Xilong and Mingxi, southeastern China (Zou et al., 2000) with slightly higher contents of moderately incompatible elements (from Zr to Gd), but differ from those of basalts from the Leizhou Peninsula and Hainan Island (Flower et al., 1992; Liu et al., 1994) that have much lower contents of most incompatible elements. This indicates that these pyroxenites, especially the metapyroxenites, are not high-pressure products of the host basalt or the basalts widespread in the study area, implying that the metasomatism event occurred before host magmatism (6–0·3 Ma) and possibly took place during the SCSB opening (or earlier). Megacrystic aggregates Lz-17 and Lz-29 show igneous microstructures and different geochemical characteristics [much lower Mg-number and higher (La/Yb)n] than the metapyroxenites. Geochemical modelling for clinopyroxene and garnet megacrysts indicates that these megacrysts crystallized from a highly evolved magma consistent with derivation by 70–80% fractional crystallization of a quartz tholeiite similar to those in the Leizhou and Hainan area (Yu & O'Reilly, 2001; Yu et al., 2003b). This observation suggests that they were formed recently (probably related to early post-spreading magmatism) and represent a different episode of metasomatism in Leizhou mantle.
|
