Fluid/mineral interaction in UHP garnet peridotite

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/229391723 Fluid/mineral interaction in UHP garnet peridotite ARTICLE in LITHOS · JANUARY 2009 Impact Factor: 4.48 · DOI: 10.1016/j.lithos.2008.07.006 CITATIONS READS 48 66 3 AUTHORS: Nadia Malaspina Joerg Hermann 38 PUBLICATIONS 389 CITATIONS 183 PUBLICATIONS 6,631 CITATIONS Università degli Studi di Milano-Bicocca SEE PROFILE Universität Bern SEE PROFILE Marco Scambelluri Università degli Studi di Genova 121 PUBLICATIONS 2,450 CITATIONS SEE PROFILE Available from: Nadia Malaspina Retrieved on: 03 February 2016 This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright Author's personal copy Lithos 107 (2009) 38–52 Contents lists available at ScienceDirect Lithos j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / l i t h o s Fluid/mineral interaction in UHP garnet peridotite Nadia Malaspina a,⁎, Jörg Hermann b, Marco Scambelluri c a b c Dipartimento di Scienze della Terra, Università degli Studi di Milano, via Botticelli 23, 20133, Milano, Italy Research School of Earth Sciences, Australian National University, Mills Road, 0200, Canberra ACT, Australia Dipartimento per lo Studio del Territorio e delle sue Risorse, Università degli Studi di Genova, corso Europa 26, 16132, Genova, Italy a r t i c l e i n f o Article history: Received 8 November 2007 Accepted 6 July 2008 Available online 18 July 2008 Keywords: Fluid/peridotite interaction Ultrahigh pressure Polyphase inclusions Phlogopite Partitioning a b s t r a c t We present two case studies of metasomatised garnet peridotite from the Sulu (Zhimafang) and of garnet orthopyroxenite from the Dabie Shan (Maowu) ultrahigh-pressure terranes (Eastern China). The mantlederived peridotite from Zhimafang shows two ultrahigh-pressure (UHP) mineral assemblages. The older one is made of porphyroclastic garnet rich in inclusions (Grt1), coarse exsolved clinopyroxene (Cpx1) and coarse phlogopite flakes (Phl1). The younger paragenesis consists of fine-grained olivine + clinopyroxene (Cpx2) + orthopyroxene ± magnesite ± Phl2 equilibrated with neoblastic garnet (Grt2). The inclusions inside porphyroclastic Grt1 are polyphase secondary inclusions related to microfractures cutting the garnet core. They display irregular shapes and contain microcrystals of calcic-amphibole, chlorite, phlogopite and rare talc, associated with pyrite and/or spinel. The low Al2O3 content (b 0.2 wt.%) in orthopyroxene coexisting with garnets and clinopyroxenes indicates equilibration at P = 4.0–6.0 GPa and T = 700–1000 °C. The trace element composition of Cpx1 and Phl1 combined with the petrologic and isotopic data of Yang and Jahn [Yang, J.J., Jahn, B.M., 2000. Deep subduction of mantle-derived garnet peridotites from the Su-Lu UHP metamorphic terrane in China. Journal of Metamorphic Geology 18,167–180.] suggests that the Zhimafang garnet peridotite experienced metasomatism by a melt with alkaline character at high-temperature conditions (T ∼ 1000 °C and P N 5.0 GPa). The microtextural identification of pseudosecondary inclusions in the porphyroclastic garnet core and their geochemical characterisation indicate that an incompatible element- and silicate-rich fluid subsequently metasomatised the garnet peridotite and equilibrated with the newly formed Cpx2 probably during Triassic UHP metamorphism. Ultramafic metasomatic layers at Maowu Ultramafic Complex (Dabie Shan) consist of layered websterite and orthopyroxenite which preserve an old olivine+ orthopyroxene (Opx1) + garnet (Grt1) ± Ti-clinohumite paragenesis, overgrown by poikilitic Opx2. Grt2 is associated with Opx2 + phlogopite along the foliation, and fine-grained idiomorphic clinopyroxene also occurs. Grt2 cores contain disseminated primary polyphase inclusions. The textural and geochemical analyses of the primary polyphase inclusions indicate that they derive from a homogeneous fluid characterised by high LILE concentrations with spikes in Cs, Ba, Pb and high U/Th. These inclusions are interpreted as remnants of the LILE- and LREE-enriched residual fluid produced when a crustderived Si-rich metasomatic agent reacted with a previous harzburgite to form garnet orthopyroxenite. The insitu trace element analyses of the major phases garnet, clinopyroxene and phlogopite that formed at the same time as the polyphase inclusions at Maowu, permit the determination of empirical mineral/fluid partitioning at pressures relevant for element recycling in subduction zones. Our estimated DCpx/fluid suggests that all LILE are highly incompatible, Th and U are moderately incompatible, Pb is close to unity and Sr is moderately compatible. Phlogopite preferentially incorporates Rb and K with respect to Ba and Cs, and Th with respect to U. The similarity between the residual Maowu fluid with the secondary inclusions in the UHP wedge-type garnet peridotite from Sulu, indicates that the fluids produced from reactions at the slab–mantle interface may be effective metasomatic agents in the mantle wedge. Such reactions may produce phlogopite, which plays an important role in controlling the LILE characteristics of the slab-derived fluid in subduction zones. © 2008 Elsevier B.V. All rights reserved. 1. Introduction Arc volcanism at subduction zones is triggered by partial melting of a metasomatised mantle wedge that interacted with slab-derived agents produced by the dehydration and/or partial melting of subducted crust ⁎ Corresponding author. Tel.: +39 02 50315613; fax: +39 02 50315597. E-mail address: Nadia.Malaspina@unimi.it (N. Malaspina). 0024-4937/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.lithos.2008.07.006 (Plank and Langmuir, 1993; Elliot et al., 1997; Schmidt and Poli, 1998; Johnson and Plank, 1999). Evidence for metasomatism of these portions of mantle wedge is provided by comparing the composition of Island Arc Basalts (IAB), derived from them, with unmetasomatised Mid Ocean Ridge Basalts (MORB) derived from depleted mantle sources. The composition of IAB is enriched in incompatible elements preferentially residing in crustal reservoirs, and is characterised by the enrichment in LILE and LREE relative to HFSE. This signature has been explained by Author's personal copy N. Malaspina et al. / Lithos 107 (2009) 38–52 melting of a MORB-type mantle that was enriched in LILE and LREE after its equilibration with metasomatic agents produced by the slab (McCulloch and Gamble, 1991; Kelemen et al., 1993; Hawkesworth et al., 1993; Brenan et al., 1994). Experimental works together with studies of natural rocks demonstrated that a water-rich fluid produced by the dehydration of the subducting lithosphere is responsible for leaching elements from the slab (Scambelluri and Philippot, 2001; Ulmer, 2001; Poli and Schmidt, 2002). Once liberated, the enriched fluid phase can metasomatise the encountered mantle rocks. There is general consensus that reactive flow is one of the main mechanisms of material transport in many terrestrial environments (e.g. Kelemen et al., 1992). In subduction zones similar processes have been inferred from experimental works on fluid/melt/peridotite partitioning at P ≥ 2 GPa (e.g. Brenan et al., 1995; Ayers et al., 1997). During reactive flow in the mantle wedge slab-derived solute-rich aqueous fluids may exchange trace elements with the surrounding mantle rocks (Ayers, 1998). However, although recent petrologic and geochemical studies focussed on fluid– melt/rock interactions in subduction zones (Poli and Schmidt, 1998; Scambelluri and Philippot, 2001; Rampone and Morten, 2001; Spandler et al., 2003; Tenthorey and Hermann, 2004; Hermann et al., 2006; Tumiati et al., 2007), still few are analyses of the slab-derived metasomatic fluids. Information on deep metasomatism of the mantle wedge can be gained by the study of ultrahigh-pressure (UHP) terranes that contain felsic rocks and metasomatised peridotites (Brueckner, 1998; Liou et al., 2004; Scambelluri et al., 2006, 2008). Such associations represent natural laboratories to study the element exchange between crustal and mantle rocks at pressures corresponding to the sub-arc depth of the subducted slab (Hermann et al., 2006). A recent work on UHP garnet orthopyroxenites from the Maowu Ultramafic Complex (Dabie Shan, China) by Malaspina et al. (2006) proposes a model where a Si-saturated slab-derived fluid phase reacts with mantle peridotites to produce orthopyroxene- and garnet-rich layers and a residual aqueous fluid enriched in incompatible elements. Mantlederived peridotites are widespread in the southern Sulu UHP belt (China), where blocks or lenses of garnet peridotites are hosted by gneiss. These rocks are therefore good samples to investigate the fate and the interaction of crustally-derived subduction fluids with peridotites. In this paper we will investigate two case studies of metasomatised peridotite from the Sulu and of orthopyroxenite from the Dabie Shan UHP terrane (Eastern China). We will present (i) the petrologic and geochemical study of the major phases of metasomatised mantle peridotite; (ii) the microtextural and compositional analyses of polyphase inclusions in porphyroclastic garnets; and (iii) the partitioning between the major peridotite phases and an incompatible element-enriched slab-derived fluid represented by the polyphase inclusions. Major aims are: (i) to characterise the geochemical imprint of slab-derived fluid phases responsible for metasomatism of mantle peridotite at UHP conditions; and (ii) to investigate the interaction between metasomatic fluids and mantle wedge at sub-arc depths. 39 These units are overlain by Jurassic clastic sequences and by a Cretaceous volcano–sedimentary cover. The UHP belt mainly consists of amphibolite-facies orthogneiss, paragneiss, amphibolite, marble and quartzite. Coesite-bearing eclogites are widespread as lenses or interlayers in gneiss, marble and peridotite (Yang and Smith, 1989; Hirajima et al., 1990; Enami and Zhang, 1993; Zhang et al., 1995a). Coesite inclusions have been also discovered in zircon from the country-rock gneisses (Ye et al., 2000; Liu et al., 2002). Metre to kilometre-sized ultramafic bodies containing serpentinite, peridotite or pyroxenite occur throughout the UHP belt and contain records of UHP metamorphism (Yang et al., 1993). Zhang et al. (2000) identified two types of garnet peridotites in the Dabie–Sulu UHP terranes: (i) “type A” mantle-derived peridotites, considered as slices of mantle wedge sampled during the subduction or exhumation of deeply subducted crust (Brueckner, 1998; Brueckner and Medaris, 2000); and (ii) “type B” crustal peridotites, portions of mafic–ultramafic complexes intruded into the continental crust prior to subduction (Evans and Trommsdorff, 1978; Trommsdorff et al., 1998). In Donghai county located at the southeastern end of the Sulu terrane (Fig. 1), lenses of garnet peridotites and pyroxenites have been classified as type-A (mantle wedge) based on their textural and compositional features (Wang and Liou, 1991; Zhang et al., 1995b; Zhang et al., 2007; Yang et al., 2007). At Zhimafang (inset of Fig. 1) the peridotites are hosted by granitic gneiss and minor jadeitite, quartzite and marble. Large volumes of gneissic rocks in this belt experienced pervasive UHP metamorphism (Wang and Liou,1991; Okay and Sengor, 1992; Liu et al., 2002, 2004). Extensive serpentinisation and carbonation was recognised along the margins of the ultramafic bodies, where a talc + magnesite assemblage has been described (Yang and Jahn, 2000). To date the outcrop is covered by water and sediments, and sampling is not possible. Many of the peridotite samples in the Donghai area are obtained from old drill-cores (4–8 cm in diameter) which were collected at Zhimafang. The sample selected for this study (RPC684) comes from one of these drill-cores and is representative of fresh mantle-derived garnet peridotite. In the UHP unit of the Dabie Shan, type-B garnet peridotites occur in the Maowu Ultramafic Complex (Fig. 1; Zhang et al., 1995b, 1998; Liou and Zhang, 1998; Medaris, 2000; Jahn et al., 2003). This body consists of metre- to centimetre-thick layers of garnet orthopyroxenite and clinopyroxenite, websterite, and harzburgite, associated with eclogite and coesite-bearing omphacitite. Locally, decimetrethick phlogopite-rich layers are associated with the orthopyroxenite. 2. Geological setting The Dabie Shan area (central-eastern China) is a 2000 km long Triassic orogenic belt formed after the collision between the SinoKorean and the Yangtze cratons (Fig. 1). The Sulu belt occupies the southeastern side of the Shandong Peninsula and is considered as a segment of the UHP Dabie terrane displaced about 500 km by the left lateral Tan-Lu Fault after the Mesozoic (Fig. 1). The Dabie–Sulu terrane consists of three fault-bounded metamorphic units (Zhang et al., 1996). From North to South, they are: 1) the northern Dabie hightemperature (HT) amphibolite–granulite unit, 2) the central Dabie– Sulu ultrahigh pressure (UHP) metamorphic unit, and 3) the southern high-pressure (HP) unit. The Sulu UHP belt is in turn divided into a HP and UHP belt, both intruded by post-orogenic Mesozoic granites. Fig. 1. Simplified map of the Dabie–Sulu UHP terrane (east-central China). Inset shows the distribution and locations of garnet peridotites in the Donghai area (black spots), with indication of the site of the drill-core at Zhimafang (modified after Zhang et al., 2000). Author's personal copy 40 N. Malaspina et al. / Lithos 107 (2009) 38–52 The mafic/ultramafic body is in fault contact with the country rocks, which mainly consist of garnet-bearing orthogneiss. Previous structural analyses and petrological studies suggested that the Maowu Ultramafic Complex and the host gneiss experienced coeval UHP metamorphism during the collision between the Sino-Korean and Yangtze cratons in the Triassic (Xue et al., 1996; Liou et al., 1996; Liou and Zhang, 1998; Ayers et al., 2002). Samples of garnet websterite (RPC171) and garnet orthopyroxenite (MWF2A and MWI2B) have been collected at Maowu, near the village of Shima. The outcrop mainly consists of garnet orthopyroxenite and is hosted by garnet-bearing gneisses where rare coesite inclusions are preserved in zircon. The orthopyroxenite is locally associated with phlogopite-rich layers. The orthopyroxenite contains an association of orthopyroxene + garnet ± clinopyroxene which formed at the expense of a previous ultramafic olivine-bearing paragenesis. Garnets contain core clusters of primary polyphase inclusions that are interpreted to derive from a solute-rich aqueous fluid enriched in LILE and LREE (Malaspina et al., 2006). Textural and geochemical data demonstrate that the Maowu Ultramafic Complex consists of metasomatic layers produced after the reaction of peridotites with a hydrous granitic melt sourced by the associated crustal rocks at UHP conditions (4.0–6.0 GPa, 700–750 °C). 3. Petrography 3.1. Sulu peridotite The Sulu garnet peridotite (RPC684) shows a porphyroclastic texture where coarse garnet (Grt1), clinopyroxene (Cpx1) and phlogopite (Phl1) occur in a matrix composed of olivine+ clinopyroxene (Cpx2) + orthopyroxene+ garnet (Grt2) ± phlogopite (Phl2) ± magnesite (Fig. 2). Late stage serpentine, chlorite and amphibole locally replace the rims of garnet, olivine and pyroxenes respectively. The coarse-grained Grt1 porphyroclast (4–10 mm in size) is zoned, with a deep-violet core containing abundant microinclusions and a clear rim (Fig. 2A). Finergrained garnet (Grt2) is inclusions-free and coexists with olivine, clinopyroxene and orthopyroxene sharing straight boundaries and triple junctions (Fig. 2B). Also clinopyroxene occurs both as coarse grain (Cpx1) and as fine-grained green euhedral crystal (Cpx2), in equilibrium with less abundant orthopyroxene. The cores of Cpx1 (Fig. 2C) contain magnetite + ilmenite exsolutions. Differently, the fine-grained Cpx2 (Fig. 2B) is clear and associated with fine-grained orthopyroxene, phlogopite and euhedral magnesite. Coarse olivine shows equilibrium textures with orthopyroxene and both types of clinopyroxene (Fig. 2B,C). Several olivine grains have oriented magnetite needles and tablets in their cores. Fig. 2. Photomicrographs (transmitted light) of representative minerals of the garnet peridotite from Zhimafang drill-core (Sulu). A: plane polarised light (PPL) image of coarsegrained porphyroclastic garnet (Grt1) and phlogopite (Phl1); B: equilibrium textures among euhedral olivine (Ol), fine-grained inclusion-free garnet (Grt2), orthopyroxene (Opx) and clear fine-grained clinopyroxene (Cpx2) (PPL); C: coarse-grained relict clinopyroxene (Cpx1) showing exsolutions parallel to the long crystallographic orientation (PPL); D: finegrained phlogopite (Phl2) in equilibrium with olivine and orthopyroxene (PPL); E: crossed polarised light (XPL) microtextural image of polyphase inclusions in porphyroclastic core domains; F: back scattered electron (BSE) image of polyphase solid inclusion in porphyroclastic garnet core (abbreviations: Tlc = talc, Chl = chlorite, Sp = spinel). Author's personal copy N. Malaspina et al. / Lithos 107 (2009) 38–52 Table Table 11 Mineral (Sulu) Mineral parageneses parageneses iningarnet garnetperidotite peridotitefrom fromZhimafang Zhimafangdrill-core drill-core (Sulu) Orthopyroxene occurs as fine-to-medium sized euhedral crystals and shares straight boundaries with both Cpx1 and Cpx2 (Fig. 2 B,C). Like clinopyroxene, phlogopite occurs in two textural modes. Coarse flakes of phlogopite (Phl1) occur around garnet porphyroclasts and contain calcite precipitates along oriented fractures (Fig. 2A). Phl2 is also present as fine-grained interstitial crystals among the matrix mineral phases (Fig. 2D) Coarse Phl1 is locally replaced by colourless Mg-chlorite. Euhedral to subhedral magnesite is only found in the matrix association and crystallises in the junctions between olivine and pyroxenes. Palisade dolomite and/or calcite replace the magnesite rims. The garnet porphyroclasts contain abundant polyphase microinclusions (Fig. 2A,E,F). They occur only in the core domains and are filled by hydrous minerals plus an opaque phase. The solid inclusions have irregular shapes and form clusters associated with microfractures that are in turn sealed by the garnet clear rim (Fig. 2A). Their occurrence in microcracks formed during the garnet rim growth refers them to as pseudosecondary inclusions (Roedder, 1984). The petrographic and electron microscope analyses indicate that these inclusions consist of chlorite (40%), calcic-amphibole (30%), mica (20%), pyrite and/or spinel (5–10%) and rare talc (5–10%) (Fig. 2F). The minerals in the polyphase inclusions occur in constant modal proportions. The time relations among the various mineral associations are difficult to reconstruct. Table 1 summarises the two main parageneses recognised from the petrographic observations. The older association is given by inclusion-rich Grt1 cores, clinopyroxene cores containing exsolution lamellae (Cpx1) and phlogopite flakes around garnet (Phl1) (Fig. 2A,C). The second paragenesis is represented by the clear Cpx2 equilibrated with Grt2, Phl2 and magnesite (Fig. 2B,D). Both mineral associations show equilibrium textures with olivine and orthopyroxene. However, petrographic evidence for recrystallisation of new olivine and orthopyroxene generations are not recognisable. The solid secondary inclusions occurring in Grt1 are related to microfractures cutting the porphyroclastic garnet core (Fig. 2A) and do not occur in any matrix mineral. This suggests that the inclusions formed before or during the second crystallisation stage. A retrograde equilibration occurs at low temperature and relatively high pressure, in the chlorite + serpentine stability field. 41 clinohumite and Ti-chondrodite. These minerals are overgrown by poikilitic orthopyroxene (Opx2). Olivine occurs as corroded grains and isolated blebs inside the coarse Opx2. Opx2 also includes older fine-grained Opx1, which is associated with the relict olivine and with fine-grained unzoned garnet (Grt1). Ti-clinohumite and Ti-chondrodite display the same textures as olivine, predating the Opx2 stage formation (Hermann et al., 2007). The garnet-rich layers are composed by millimetre to centimetre zoned garnet (Grt2) in textural equilibrium with Opx2. Grt2 displays dusty cores, disseminated by primary polyphase microinclusions (Fig. 3B), and clear rims. In RPC171 fine-grained and idiomorphic green clinopyroxene is also present along the garnet-rich layers. In the orthopyroxenite sample MWF2A, Grt2 is associated with phlogopite along the garnet+orthopyroxene foliation (Fig. 3A). Phlogopite is coarsegrained and shows equilibrium texture also with Opx2. Phlogopite flakes display exsolution-type microstructures given by talc intergrowths and are only locally replaced by retrograde chlorite. Polyphase inclusions within the core of porphyroblastic Grt2 have negative-crystal shapes and constant volume proportions of the infilling phases indicating their primary origin. Petrography and electron microprobe analyses reveal that the inclusion infillings consist of spinel (∼ 10–20 vol.%) and hydrous phases (∼80–90 vol.%) including amphibole + chlorite ± talc ± mica ± apatite. Sulfide was identified in rare cases. 4. Analytical techniques Major elements of rock-forming minerals were analysed by wavelength dispersive spectrometry using a Cameca SX 100 electron microprobe at the Research School of Earth Sciences (RSES) of The Australian National University (ANU), and a Jeol 8200 Super Probe at the Dipartimento di Scienze della Terra, University of Milano. Acceleration voltage was set to 15 kV, beam current was 15 nA and natural minerals were used as standard. Phlogopites were measured at 5 nA with defocussed beam to prevent K devolatilisation during the analyses. Mineral analyses were always assisted by detailed back scattered 3.2. Maowu websterite and orthopyroxenite The selected websterite and orthopyroxenite samples from the Maowu Ultramafic Complex (RPC171, MWF2A and MWI2B) display an assemblage dominated by orthopyroxene, garnet and minor clinopyroxene. Locally, thick phlogopite-rich layers are associated with the orthopyroxenites (see Fig. 1 from Malaspina et al., 2006). The samples are strongly foliated, with centimetre-thick layers of coarse oriented orthopyroxene, and layers where orthopyroxene coexists with coarse and zoned garnet (Fig. 3A). All samples show equilibrium texture among minerals with 120° triple junctions, but an older assemblage is still preserved in samples MWF2A and MWI2B. The relict mineral association consists of olivine, orthopyroxene (Opx1), garnet (Grt1), and locally Ti- Fig. 3. A: Garnet orthopyroxenite sample from the Maowu Ultramafic Complex where porphyroblastic garnet (Grt2), orthopyroxene (Opx2) and phlogopite (Phl) share macroscopic equilibrium texture (sample MWF2A); B: primary polyphase inclusion in porphyroclastic garnet core (Grt2) from Maowu garnet websterite RPC171; Chl = chlorite, Am = amphibole, Py = pyrite. Author's personal copy 42 Table 2 Representative major element compositions (wt.% oxide) and recalculated structural formulae of minerals from Sulu garnet peridotite (RPC684) and Maowu garnet websterite (RPC171) Sample Sulu garnet peridotite Mineral Grt1 core Grt2 core Grt2 rim Cpx1 core Cpx1 rim Cpx2 Opx core Opx rim Ol core Ol rim Phl1 Phl2 Grt1 core Grt1 rim Grt2 core Grt2 rim Cpx Opx1 Opx2 core Opx2 rim Ol⁎ Phl⁎ Phl⁎ 42.24 0.01 23.32 2.27 9.29 19.33 0.42 n.a. 4.93 n.a. 0.02 101.83 2.97 0.00 1.93 0.13 0.55 0.00 2.03 0.03 n.a. 0.37 n.a. 0.00 8.00 41.92 0.01 22.55 2.64 10.06 18.34 0.48 n.a. 4.95 n.a. 0.02 100.98 2.99 0.00 1.89 0.15 0.60 0.00 1.95 0.03 n.a. 0.38 n.a. 0.00 7.99 41.91 0.00 23.49 1.98 10.31 18.72 0.55 n.a. 4.55 n.a. 0.00 101.53 2.97 0.00 1.96 0.11 0.60 0.01 1.97 0.03 n.a. 0.34 n.a. 0.00 8.00 42.29 0.02 23.03 2.16 10.69 18.43 0.66 n.a. 4.65 n.a. 0.01 101.94 2.99 0.00 1.92 0.12 0.63 0.00 1.94 0.04 n.a. 0.35 n.a. 0.00 7.99 54.98 0.03 1.95 1.82 2.05 15.76 0.00 0.04 21.60 2.34 0.01 100.58 1.98 0.00 0.08 0.05 0.06 0.00 0.85 0.00 0.00 0.84 0.16 0.00 4.03 54.53 0.15 1.21 1.61 2.51 16.02 0.19 0.29 21.58 1.90 bdl 99.99 1.99 0.00 0.05 0.05 0.08 0.00 0.87 0.01 0.01 0.84 0.13 0.00 4.03 55.28 0.16 1.01 2.01 2.42 15.90 0.15 0.53 21.34 1.67 bdl 100.47 2.00 0.00 0.04 0.06 0.06 0.01 0.86 0.00 0.02 0.83 0.12 0.00 4.00 58.01 0.00 0.23 0.03 5.04 36.15 0.12 bdl 0.08 bdl 0.01 99.67 1.99 0.00 0.01 0.00 0.14a 57.30 bdl 0.13 0.08 5.29 36.87 0.07 0.08 0.09 0.01 bdl 99.93 1.96 bdl 0.01 0.00 0.15a 40.60 bdl 0.01 0.01 7.95 50.11 0.04 0.41 bdl bdl bdl 99.15 1.00 bdl 0.00 0.00 0.16a 40.15 bdl bdl bdl 7.97 49.45 0.05 0.41 bdl 0.01 bdl 98.04 1.00 bdl bdl bdl 0.17a 41.00 0.36 15.63 0.73 2.97 25.06 0.09 n.a. 0.12 0.31 8.93 95.73 2.87 0.02 1.29 0.04 0.17a 42.38 0.10 14.75 0.46 2.57 24.92 0.02 n.a. 0.00 0.25 8.90 94.35 2.97 0.01 1.22 0.03 0.15a 44.09 0.22 13.69 n.a. 2.78 25.83 0.02 n.a. 0.02 0.87 7.53 95.04 3.05 0.01 1.11 n.a. 0.16a 43.98 0.21 14.57 n.a. 2.82 24.01 0.01 n.a. 0.02 0.68 7.68 93.97 3.07 0.01 1.20 n.a. 0.16a 2.61 0.01 n.a. 0.01 0.04 0.80 7.87 2.61 0.00 n.a. 0.00 0.03 0.80 7.81 54.55 0.14 0.00 0.56 1.42 18.53 0.01 0.15 24.32 bdl bdl 99.68 1.98 0.00 0.00 0.02 0.03 0.01 1.00 0.00 0.00 0.95 bdl bdl 4.00 40.58 bdl bdl 0.01 7.49 51.35 0.05 0.18 bdl 0.01 0.02 99.69 0.99 bdl bdl 0.00 0.15a 1.83 0.00 0.01 bdl 0.00 bdl 3.00 41.62 0.03 23.11 1.53 10.47 17.71 0.54 n.a. 5.70 n.a. n.a. 100.71 2.98 0.00 1.95 0.09 0.63 0.00 1.89 0.03 n.a. 0.44 n.a. n.a. 8.00 57.98 bdl 0.08 0.01 6.26 35.92 0.05 0.19 0.08 bdl 0.01 100.57 1.98 bdl 0.00 0.00 0.18a 1.83 0.00 0.01 bdl bdl bdl 3.00 41.90 0.09 23.31 1.55 10.46 18.11 0.52 n.a. 5.25 n.a. 0.00 101.18 2.98 0.00 1.95 0.09 0.62 0.00 1.92 0.03 n.a. 0.40 n.a. 0.00 8.00 57.92 bdl 0.08 0.02 5.99 36.09 0.06 0.20 0.08 bdl bdl 100.43 1.98 bdl 0.00 0.00 0.17a 1.88 0.00 0.00 0.00 0.00 bdl 4.00 41.75 0.09 23.18 1.60 10.59 18.01 0.51 n.a. 5.20 n.a. 0.01 100.94 2.98 0.00 1.95 0.09 0.63 0.00 1.91 0.03 n.a. 0.40 n.a. 0.00 8.00 57.63 0.02 0.04 0.06 4.35 37.90 0.11 0.19 0.05 0.00 0.00 100.35 1.97 0.00 0.00 0.00 0.12a 1.85 0.00 bdl 0.00 bdl 0.00 4.00 41.63 0.06 23.86 0.77 10.75 17.69 0.53 n.a. 5.43 n.a. 0.00 100.72 2.97 0.00 2.01 0.04 0.64 0.00 1.88 0.03 n.a. 0.42 n.a. 0.00 8.00 1.92 0.00 0.01 0.00 0.00 0.00 4.02 1.84 0.00 0.01 0.00 bdl bdl 4.00 1.83 0.00 0.01 0.00 bdl 0.00 4.00 1.86 0.00 0.00 0.00 0.00 0.00 3.01 2.66 0.00 n.a. 0.00 0.12 0.66 7.77 2.49 0.00 n.a. 0.00 0.09 0.68 7.71 0.79 0.76 0.77 0.75 0.93 0.92 0.93 0.93 0.93 0.92 0.91 0.94 0.95 0.75 0.75 0.76 0.75 0.97 0.94 0.92 0.91 0.93 0.94 0.94 Analysed minerals from the other Maowu samples are indicated with the asterisk symbol (olivine from MWI2B and phlogopite from MWF2A). Garnet is normalised on the basis of 12 oxygens. Clino- and orthopyroxenes are normalised on the basis of 6 oxygens. Olivine is normalised on the basis of 4 oxygens. Phlogopite is normalised on the basis of 22 oxygens. Fe3+ for garnet and clinopyroxene is calculated assuming stoichiometry. Abbreviations: Grt = garnet, Cpx = clinopyroxene, Opx = orthopyroxene, Ol = olivine, Phl = phlogopite, bdl = below detection limit, n.a. = not analysed. a Total iron as Fe2+. N. Malaspina et al. / Lithos 107 (2009) 38–52 SiO2 TiO2 Al2O3 Cr2O3 FeOtot MgO MnO NiO CaO Na2O K2O Total Si Ti Al Cr Fe2+ Fe3+ Mg Mn Ni Ca Na K Cation sum Mg# Maowu garnet websterite Grt1 rim Author's personal copy 43 N. Malaspina et al. / Lithos 107 (2009) 38–52 electron (BSE) images to control the microtextural site. Mineral trace elements were acquired by Laser ablation inductively-coupled plasma mass spectrometer (LA ICP-MS) at the RSES (ANU). The LA ICP-MS technique employs an ArF (193 nm) EXCIMER laser coupled to an Agilent 7500 quadripole ICP-MS. Spot sizes of 112,142 and 187 μm were used and the counting time was 20 s for the background and between 40 and 45 s for sample analyses. All analyses were acquired in timeresolved mode, which allows to individuate microinclusions in minerals. Only clean signals have been integrated. 43Ca and 29Si were employed as the internal standard isotopes, based on CaO and SiO2 concentrations previously measured by electron microprobe. NIST-612 glass was used as the external standard, assuming the composition given by Pearce et al. (1997). A BCR-2G glass was used as secondary standard. Reproducibility about the mean values of 8 analyses were between 0.5 and 4% relative (1σ) for the majority of elements. The average trace element content of BCR-2G were typically within 2–6% of the certified values for this standard (Wilson, 1997). The mineral phases of polyphase inclusions in garnet were identified using a Philips SEM 515 electron microscope at the University of Genova. The trace element compositions of inclusions were acquired by LA ICP-MS at the RSES (ANU) at similar running conditions as those described above, with a spot size of 112 μm. The SiO2 concentration of bulk solid inclusions was estimated by calculating a weighted mean of the SiO2 content of the infilling mineral phases. Since the resulting SiO2 is comparable with the concentrations in the host garnet, 29Si was used as internal standard based on SiO2 concentrations measured in garnet (see Malaspina et al., 2006 for details). The single inclusion or cluster of inclusions occupies from 5 to 45 vol.% of the laser spot, whereas the host garnet comprises the remainder of the analysis. The measured trace element concentrations therefore do not correspond to the absolute concentrations of inclusions. Such values have to be interpreted as the sum of garnet + inclusion concentrations. 5. Mineral compositions The major and trace element mineral compositions of the Sulu garnet peridotite and Maowu garnet websterite are reported in Tables 2 and 3. 5.1. Sulu peridotite The rock-forming minerals of the Sulu garnet peridotite have homogeneous compositions, with the exception of porphyroclastic garnet and clinopyroxene which show a zonation. Grt1 (∼68–70 mol% pyrope, ∼19–20 mol% almandine and ∼ 13 mol% grossular) displays a slight depletion in MgO and enrichment in FeOtot and Cr2O3 from core to rim, while CaO is quite homogeneous. A slight depletion in Al2O3 and enrichment in MnO are complementary to the previously described variations (Table 2). Grt2 (∼ 66–68 mol% pyrope, ∼ 21–22 mol% almandine and ∼ 12 mol% grossular) is characterised by a major element composition very similar to that of Grt1 rim (Table 2). Both coarse and fine-grained clinopyroxenes are diopsidic in composition. The coarser Cpx1 is slightly zoned, with rim composition richer in FeOtot, MgO, CaO and depleted in Al2O3, Cr2O3 and Na2O contents. Cpx2 is homogeneous and characterised by lower MgO and CaO with respect to Cpx1. Also enstatitic orthopyroxene does not show any compositional zonation with low Al2O3 (b0.20 wt.%) and CaO (b0.1 wt.%). Olivine has high forsterite content (∼ 92 mol%) and NiO (0.41 wt.%), in agreement with the mantle origin of the peridotite. It does not show important core-to-rim major element variations, except for MgO Table 3 Representative trace element compositions (ppm) of minerals from Sulu garnet peridotite (RPC684) and Maowu garnet websterite (RPC171) Sample Sulu garnet peridotite Maowu garnet websterite Mineral Grt1 core Grt1 rim Cpx1 core Cpx1 rim Cpx2 Opx core Opx rim Ol core Ol rim Phl1 Phl2 Grt1 Grt2 core Grt2 rim Cpx Opx1 Opx2 core Opx2 rim Ol⁎ Phl⁎ Phl⁎ Li Be P K Sc Ti Cr Ni Rb Sr Y Zr Nb Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Er Yb Lu Hf Ta Pb Th U 0.59 bdl 171 7.8 110 149 13876 5.94 bdl 0.36 14.1 3.43 0.16 bdl 2.98 0.06 0.25 0.09 0.91 0.38 0.14 0.64 0.18 1.80 1.97 2.67 0.44 0.06 bdl 0.32 bdl 0.03 0.28 bdl 142 bdl 116 174 14558 6.39 bdl 0.15 15.0 3.78 0.21 bdl 0.11 0.03 0.23 0.11 1.07 0.44 0.17 0.74 0.19 2.00 2.07 2.75 0.46 0.07 bdl 0.30 bdl 0.03 3.09 0.51 29.7 3912 27.7 186 10010 326 13.5 296 1.03 4.3 0.92 0.54 274 10.1 33.1 4.69 15.2 1.28 0.23 0.56 0.06 0.29 0.09 0.04 bdl 0.24 0.09 11.3 1.66 0.14 2.75 0.44 25.4 7.08 25.8 121 8259 236 0.02 506 0.84 3.31 0.08 0.01 0.51 10.6 35.2 5.04 16.6 1.49 0.27 0.68 0.07 0.28 0.06 0.02 bdl 0.22 0.01 15.9 0.08 0.02 3.37 0.50 15.1 4.17 34.1 122 9273 223 bdl 581 1.69 6.11 0.11 bdl 1.26 11.1 36.9 5.21 17.2 1.68 0.33 0.87 0.10 0.45 0.13 0.07 0.01 0.28 0.01 16.2 0.07 0.02 0.16 0.08 21.4 bdl 1.5 33.9 257 569 bdl 0.02 0.01 0.03 0.03 bdl 0.01 bdl 0.02 bdl 0.01 0.01 bdl bdl bdl bdl bdl bdl bdl bdl 0.01 0.02 bdl bdl 0.18 0.10 21.3 bdl 1.61 44.0 297 560 bdl 0.02 0.01 0.27 0.03 bdl bdl bdl 0.02 bdl 0.01 0.00 bdl bdl bdl bdl bdl bdl bdl 0.01 0.01 0.01 bdl bdl 1.82 bdl 57.2 bdl 1.10 64.4 25.9 2847 bdl 0.03 bdl 0.03 2.12 bdl 0.02 0.01 0.02 bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl 0.14 0.01 bdl 0.01 1.66 bdl 40.3 3.36 1.14 58.2 50.67 2872 bdl 8.48 bdl 0.04 2.24 0.01 15.5 0.05 0.06 bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl bdl 0.15 2.00 0.01 0.02 11.0 0.11 4.76 – 3.00 634 2771 1445 312 61.7 0.49 0.10 7.01 13.4 6759 0.02 0.01 bdl bdl bdl 0.07 0.12 bdl bdl bdl bdl bdl 0.03 0.47 2.28 32.0 1.74 12.4 0.17 5.57 – 4.35 742 3350 1487 299 85.3 0.80 0.58 6.50 13.8 8133 0.88 3.45 0.48 1.62 0.18 0.07 0.13 0.02 0.11 0.07 0.09 0.02 0.03 0.43 3.89 38.1 5.32 2.81 0.30 223 65.4 43.2 18.4 n.a. 165 0.18 1.16 17.6 3.92 0.09 0.05 10.6 0.03 0.06 0.01 0.17 0.39 0.26 1.71 0.45 3.33 1.72 1.49 0.21 0.06 bdl 0.16 0.01 0.01 0.22 bdl 77.8 bdl 69.7 113 1178 5.03 bdl 0 42.2 5.15 0.01 bdl 0.06 0.01 0.05 0.03 0.52 1.20 0.78 4.39 0.95 7.23 4.96 5.07 0.76 0.08 bdl 0.1 0.02 0.02 0.33 bdl 70.4 bdl 73.6 73.0 1883 6.25 bdl 0.12 22.7 4.59 0.02 bdl 0.57 0.05 0.08 0.03 0.54 1.06 0.61 3.23 0.59 4.23 2.54 2.67 0.41 0.06 bdl 0.05 0.10 0.04 1.08 1.40 10.0 53.6 4.52 33.5 655 730 0.08 263 1.11 0.14 0.00 0.02 0.60 2.33 5.67 1.37 6.56 2.22 0.57 1.53 0.13 0.43 0.08 0.04 0.00 0.01 bdl 19.0 0.12 0.09 0.33 0.40 10.6 bdl 1.88 6.5 62.3 1368 bdl 0 0.02 bdl bdl bdl 0.03 bdl 0.01 bdl 0.01 0.01 bdl 0.01 bdl 0.01 bdl bdl bdl bdl bdl 0.0 bdl bdl 0.24 0.55 8.82 2.19 1.10 21.9 108 2740 bdl 0.13 0.05 0.01 bdl 0.01 0.22 0.02 0.01 0.01 0.04 0.02 bdl 0.02 bdl 0.02 0.01 bdl bdl bdl bdl 0.18 bdl bdl 0.24 0.55 9.39 bdl 1.24 29.5 112 2829 bdl 0.01 0.08 0.01 bdl bdl bdl bdl 0.01 bdl 0.02 0.02 bdl 0.02 bdl 0.02 0.01 0.01 bdl bdl bdl 0.01 bdl bdl 2.21 bdl 160 5.12 2.38 2.33 n.a. n.a. bdl 0.23 0.38 0.03 0.14 bdl 2.39 0.11 0.03 0.01 0.02 bdl bdl bdl bdl 0.02 0.02 0.02 bdl bdl bdl 0.16 bdl bdl 58.12 1.39 n.a. – 3.20 155 n.a. n.a. 117 136 0.25 0.13 0.04 29.9 1411 bdl bdl bdl bdl bdl 0.03 0.39 n.a. bdl bdl bdl bdl bdl 0.05 6.50 141 61.5 45.64 1.42 n.a. – 2.91 160 n.a. n.a. 119 144 0.23 0.11 0.04 31.8 1379 0.02 0.02 bdl bdl bdl 0.02 0.36 n.a. bdl bdl bdl bdl bdl 0.04 6.60 155 66.2 Symbols and abbreviations same as in Table 2. Analysed minerals from the other Maowu samples are indicated with the asterisk symbol (olivine from MWI2B and phlogopite from MWF2A). Author's personal copy 44 N. Malaspina et al. / Lithos 107 (2009) 38–52 In Fig. 4B and C the PM normalised patterns of the most incompatible trace elements are portrayed for the main mineral phases of the Sulu garnet peridotite. Both clinopyroxene generations share the same REE, Li, Zr, Hf and Sr compositions (Table 3). However, Cpx1 and Cpx2 display different patterns for many LILE (Fig. 4B). Cpx1 is enriched in Cs, Rb, Ba, K, Pb, with absolute concentrations up to 60 × PM and positive anomalies in Ba and Pb. It is also characterised by high Nb and Ta concentrations and displays UN/N b 1. A strong zonation features the trace element composition of Cpx1. The LILE concentration decreases from core to rim, together with Th and U, reaching the compositional pattern of Cpx2 (Fig. 4B). Cpx2 shows a depleted pattern in LILE but a pronounced positive Pb anomaly. The U–Th ratio is less steep with respect to Cpx1, with UN ≥ ThN and shows less enriched Nb and Ta concentrations. Similarly to Cpx1, the trace element signature of the coarse Phl1 (black triangles in Fig. 4C) is characterised by high LILE and HFSE concentrations. The normalised pattern is enriched in Cs, Rb, Ba, Th and U. Interestingly, Phl1 has more than one order of magnitude higher Th and U but significantly lower Pb contents than clinopyroxene. The low UN/ThN, and the negative anomaly in Pb, complementary to the spikes in Cpx1, suggests that these minerals share equilibrium partitioning during their formation. Also Phl2 is enriched in LILE and does not show significant differences in trace element concentrations with respect to Phl1 (Fig. 4C, Table 3). Therefore, phlogopites seem to record chemical variations only in some major elements (Table 2). As expected from a mantle peridotite paragenesis, the other anhydrous minerals, garnet, olivine and orthopyroxene, are strongly depleted in incompatible trace elements. Garnet displays a negative anomaly in Sr, while olivine and orthopyroxene show relatively high concentrations in Nb, Ta and Li (Fig. 4C). One interesting aspect of the dataset is the partitioning in P and Li among the major phases. For P, garnet has the highest concentration followed by olivine, clinopyroxene, orthopyroxene and phlogopite, whereas for Li, phlogopite has the highest concentration followed by clinopyroxene, olivine and orthopyroxene. Because olivine is by far the most abundant mineral in the garnet peridotite, and because it has relatively high Li and P contents, it is an important host for these two elements. 5.2. Maowu websterite and orthopyroxenite Fig. 4. A: REE Primitive Mantle (PM) normalised concentrations of representative rockforming minerals (Sulu garnet peridotite); B: trace elements PM normalised concentrations of the two clinopyroxene generations; C: trace elements PM normalised concentrations of garnet, olivine, orthopyroxene and phlogopites. Normalising values from McDonough and Sun (1995). which slightly decreases towards the rim (from 50.11 to 49.45 wt.%). Phlogopite flakes around garnet (Phl1) and fine-grained crystals (Phl2) display some differences in SiO2 and Al2O3 (Table 2). Phl1 shows higher Al2O3 (15.63 wt.%) and lower SiO2 (41.00 wt.%) with respect to Phl2 (14.75 wt.% Al2O3 and 42.38 wt.% SiO2 corresponding to a difference of 0.1 Si p.f.u.). K (0.8 p.f.u.) is identical in both phlogopites (Table 2). The Primitive Mantle (PM) normalised trace element concentrations of representative minerals composing the Sulu garnet peridotite are portrayed in Fig. 4. Garnet shows enrichment in HREE and is relatively homogeneous (Fig. 4A). Although the LREE are depleted with respect to the HREE, they display an abnormal pattern with nearly no fractionation between Nd, Sm and Eu. Both clinopyroxene generations (Cpx1 and Cpx2) are characterised by LREE-enriched and HREE-depleted patterns, with a strong enrichment from La to Nd. This indicates equilibrium partitioning with garnet and suggests that both clinopyroxene generations grew in the garnet-facies stability field. Cpx1 and Cpx2 have almost identical REE compositions, except for the HREE which are slightly higher in Cpx2 (Fig. 4A). Phlogopites have very low REE concentrations, showing only detectable LREE, Eu and Gd. In Table 2 the compositions of representative minerals forming the Maowu garnet websterite and orthopyroxenites are also reported. Microtextural analyses reveal two generations of garnet and orthopyroxenes. Both Grt1 and Grt2 are pyrope-rich (70–73 mol%). Grt1 shows a slight zonation in MgO and Cr2O3 which increase towards the rim, approaching the composition of Grt2 (Table 2). Grt2 is compositionally homogeneous, lacking a correlation with the observed optical zoning. With respect to the older Grt1, it has higher pyrope content. Also, the two enstatitic orthopyroxene generations show different compositions. Opx1 is characterised by lower SiO2 and FeOtot than Opx2 (Table 2). In contrast, MgO concentration decreases from Opx1 (1.92 p.f. u.) to Opx2 (1.83–1.86 p.f.u.). Both orthopyroxenes have low CaO and Al2O3 contents (b0.1 wt.%). Diopsidic clinopyroxene is characterised by low Al2O3 contents and is rich in MgO (∼17.5–18.5 wt.%) and CaO (∼23–25 wt.%). Olivine from sample MWI2B has high Mg number (91– 93) and NiO (0.84 wt.%), in agreement with the ultramafic composition of the protolith. Finally, the coarse phlogopite in equilibrium with Grt2 and Opx2 from sample MWF2A is characterised by high Si (up to 3.05– 3.07 p.f.u.) and relatively low K (0.66–0.6 p.f.u.). The trace element concentrations of rock-forming minerals of the Maowu garnet websterite RPC171 and of phlogopites from the garnet orthopyroxenite MWF2A are reported in Table 3 and have been normalised to the PM (Fig. 5). Grt1 and Grt2 share HREE-enriched patterns, with absolute concentrations up to 10–20 × PM. Opx1 and Opx2 have similar contents in MREE and both are HREE-depleted, with absolute concentrations at or below the detection limits (Fig. 5A, Table 3). This indicates that both generations of orthopyroxene formed in the presence Author's personal copy N. Malaspina et al. / Lithos 107 (2009) 38–52 45 of garnet. On the other hand, the two orthopyroxenes of RPC171 show differences in LREE concentrations. Opx1 is LREE-depleted, whereas Opx2 is relatively LREE-enriched (Fig. 5A). Clinopyroxene shows equilibrium partitioning with garnet with LREE-enriched and HREEdepleted patterns (Fig. 5A). Clinopyroxene is depleted in LILE and HFSE, whereas it shows positive anomalies in Pb and Sr, and UN N ThN. This pattern mirrors the trace element signature of phlogopite, which is characterised by enrichment in LILE, negative anomaly in Pb, surprisingly high Th–U (UN/ThN ≥ 1) and low LREE. This indicates that clinopyroxene, phlogopite (and talc associated with phlogopite) formed at equilibrium conditions in the garnet stability field. 5.3. Polyphase inclusions Fig. 5. A: REE PM normalised concentrations of representative minerals forming the Maowu garnet websterite and orthopyroxenite samples; B: trace elements PM normalised mineral concentrations of clinopyroxene and the two orthopyroxene generations (from RPC171); C: trace elements PM normalised mineral concentrations of garnets and phlogopite + talc intergrowths (from MWF2A). Normalising values from McDonough and Sun (1995). Numerous polyphase inclusions occupy the core domains of both porphyroclastic Grt1 of the Sulu peridotite, and porphyroblastic Grt2 of the Maowu orthopyroxenite. The petrographic description indicates that the inclusions occurring in garnet of the Sulu peridotite are pseudosecondary inclusions related to microfractures cutting the core domains of the crystal. These inclusions are texturally different with respect to the primary inclusions occurring in the garnet of Maowu websterite (c.f. Figs. 2F and 3B). However, the compositions and the relative proportions of the mineral infillings of both types of inclusions show important similarities. All inclusions in the Sulu peridotite and Maowu websterite are composed of 10–20 vol.% of opaques and 80– 90 vol.% of hydrous phases. Due to the very fine inclusions size, the major element concentrations of infilling minerals have been mostly qualitatively determined by electron microscopy. The opaque phase mainly consists of Al-spinel, or occasionally Fe–Ni-rich sulphides. The remaining 80–90 vol.% of the inclusions consists of hydrous minerals. K-rich calcic-amphibole is the most abundant phase and is always associated with chlorite, minor talc, Ba-rich mica and rare apatite. To characterise the trace element composition of the polyphase microinclusions, Laser Ablation analyses were performed on the bulk inclusions following the procedure of Heinrich et al. (2003) and Malaspina et al. (2006). In Fig. 6A and B the PM normalised patterns of the most incompatible trace elements of the polyphase inclusions analysed in the Sulu (RPC 684) and Maowu samples (RPC171) (grey areas) are compared with the average pattern of inclusion-free domains of the host garnets (dark solid lines). The white area in the background representing the trace element patterns of primary inclusions from previously studied Maowu garnet orthopyroxenites (Malaspina et al., 2006) is also reported in both figures for comparison. Overall, the trace element patterns of the Sulu and Maowu polyphase inclusions presented here show compositional similarities. The relatively large spread of the inclusion compositions is a consequence of the different volume proportion of the inclusions with respect to the host garnet affected by the laser spot Fig. 6. A: PM normalised trace element compositions of polyphase inclusions in porphyroclastic garnet from Sulu garnet peridotite (RPC684); B: PM normalised trace element compositions of polyphase inclusions in porphyroblastic garnet and from Maowu garnet websterite (RPC171). Both patterns are compared with: (i) an average pattern of inclusionfree domains in the host garnet (black solid line) and (ii) the trace element signature of polyphase inclusions in Maowu garnet orthopyroxenite (white area in the background) from Malaspina et al. (2006). Normalising values after McDonough and Sun (1995). Author's personal copy 46 Table 4 Trace element concentrations (ppm) of representative polyphase inclusions and inclusion-free garnet domains in Sulu garnet peridotite RPC684 (pseudosecondary inclusions) and Maowu garnet websterite RC171 (primary inclusions) Sample Garnet peridotite (RPC684) Garnet websterite (RPC171) Polyphase inclusions 20 ± 10 vol.% 0.38 bdl 117 bdl 134 197 102 13842 8.37 bdl 0.14 17.18 3.92 0.16 bdl 0.16 0.01 0.24 0.11 1.12 0.45 0.16 0.72 0.21 2.15 2.50 3.55 0.62 0.07 0.01 0.07 bdl 0.03 1.02 bdl 148 22.4 123 155 94.2 13001 10.2 0.10 1.77 15.48 3.73 0.20 0.02 7.75 0.09 0.31 0.10 0.98 0.44 0.16 0.72 0.20 2.02 2.10 2.94 0.49 0.08 0.01 0.62 0.04 0.04 30 ± 15 vol.% 1.82 bdl 123 16.5 116 158 97.5 15140 5.58 0.05 3.31 14.54 3.54 0.14 bdl 4.63 0.21 0.46 0.11 0.98 0.40 0.15 0.67 0.17 1.87 2.11 2.94 0.53 0.07 0.01 0.60 0.04 0.08 1.80 bdl 119 15.7 127 164 94.9 14838 11.7 0.06 3.71 16.88 3.72 0.13 0.02 9.68 0.10 0.34 0.10 1.03 0.43 0.16 0.72 0.20 2.10 2.48 3.70 0.64 0.08 0.01 2.40 0.04 0.05 0.86 bdl 157 15.3 113 139 93.9 16136 6.41 0.08 4.44 13.86 3.11 0.11 0.02 13.16 0.10 0.30 0.09 0.80 0.37 0.14 0.64 0.18 1.79 1.87 2.59 0.44 0.06 bdl 3.65 0.01 0.04 1.21 bdl 160 122 121 141 91.8 12612 18.4 0.44 5.44 15.9 3.81 0.17 0.10 19.4 0.17 0.44 0.11 0.94 0.44 0.15 0.74 0.20 1.98 2.15 2.94 0.50 0.07 0.01 2.36 0.06 0.07 1.88 bdl 126 45.3 121 167 99.0 13762 8.36 0.12 3.06 15.2 3.70 0.25 0.01 57.0 0.12 0.41 0.12 1.04 0.47 0.15 0.70 0.19 1.90 2.13 3.04 0.53 0.08 0.01 0.59 0.03 0.08 2.88 bdl 141 118 108 99.2 83.2 16973 4.59 0.48 15.9 14.9 3.59 0.22 0.09 18.6 0.31 0.71 0.13 0.94 0.41 0.15 0.68 0.19 1.92 2.00 2.73 0.46 0.07 0.01 2.47 0.14 0.11 0.49 0.02 153 225 119 147 94.4 14724 62.3 0.55 12.9 14.9 3.34 0.16 0.10 27.4 0.07 0.23 0.09 0.86 0.41 0.15 0.68 0.18 1.94 2.04 2.76 0.47 0.07 0.01 0.33 0.13 0.04 0.32 bdl 115 bdl 135 193 100 13314 8.46 bdl 0.13 17.5 4.03 0.15 bdl 0.02 0.0 0.24 0.11 1.14 0.48 0.16 0.74 0.20 2.21 2.53 3.57 0.62 0.08 0.01 0.03 0.01 0.03 Avg estimate from inclusion/host proportion Polyphase inclusions 10 ± 5 vol.% 20 ± 10 vol.% 30 ± 15 vol.% 10 ± 5 vol.% 20 ± 10 vol.% – – 1050 ± 116 58 – – – – – 0.34 4.1 ± 2.7 – – 1.6 ± 0.05 bdl 6.9 ± 5.2 0.34 ± 0.22 2.9 ± 0.5 1.2 ± 0.1 11 ± 0.3 – – – – – – – – – 0.09 3.4 ± 2.7 0.07 ± 0.01 0.43 ± 0.11 – – 684 ± 81 87 ± 12 – – – – – 0.36 ± 0.08 16.5 ± 3.8 – – 0.74 ± 0.14 0.07 ± 0.04 44 ± 13 0.63 ± 0.22 1.8 ± 0.3 0.50 ± 0.03 4.7 ± 0.4 – – – – – – – – – 0.03 ± 0.02 9.1 ± 6 0.16 ± 0.06 0.27 ± 0.07 – – 480 ± 34 380 ± 160 – – – – – 1.1 ± 0.6 28 ± 16 – – 1.8 ± 1.7 0.22 ± 0.12 102 ± 35 0.57 ± 0.21 1.6 ± 0.5 0.38 ± 0.04 3.2 ± 0.2 – – – – – – – – – 0.44 ± 0.64 5.9 ± 3.5 0.33 ± 0.14 0.27 ± 0.08 0.57 0.05 77 4.0 69.9 9426 463.7 3698 71.6 0.10 0.59 43.5 7.68 3.76 0.23 0.4 0.06 0.10 0.03 0.59 1.28 0.83 4.56 1.03 7.71 4.95 5.01 0.77 0.17 0.35 0.21 0.07 0.20 0.99 0.04 79.8 195 67.2 115 92.1 2581 61.6 0.92 4.51 44.2 4.78 0.02 0.29 47.4 0.24 0.08 0.08 0.82 1.45 0.92 5.05 1.11 8.12 4.92 4.75 0.71 0.09 bdl 1.76 0.04 0.10 Host Grt 30 ± 15 vol.% 1.81 0.06 217 106 67.4 164 126 1359 26.5 0.35 8.62 42.1 5.00 0.01 0.10 6.63 0.32 0.48 0.09 0.91 1.58 0.94 4.94 1.02 7.47 4.93 4.95 0.76 0.09 bdl 0.85 0.02 0.04 1.96 0.34 83.3 392 67.7 130 90.2 2159 26.7 0.95 28.0 39.1 5.39 0.04 0.18 88.4 0.18 0.44 0.11 1.03 1.74 1.08 5.42 1.08 7.30 4.45 4.57 0.70 0.09 bdl 1.78 0.03 0.05 4.08 0.75 83.9 291 63.0 119 94.3 2083 18.6 1.19 18.3 46.1 5.09 0.02 0.21 40.2 0.09 0.18 0.06 0.76 1.48 0.99 5.58 1.24 8.66 5.02 4.90 0.73 0.09 bdl 2.40 0.02 0.05 1.05 0.01 69.9 188 65 229 82.6 2113 13.5 0.71 6.89 29.2 4.07 0.08 0.11 242 0.08 0.12 0.03 0.43 0.88 0.58 3.11 0.66 5.05 3.53 3.99 0.64 0.06 bdl 3.58 0.03 0.06 2.05 0.13 76.8 232 73.7 166 96.4 2099 18 1.02 40.9 40.0 5.72 0.09 0.27 53.8 0.39 0.46 0.09 0.79 1.35 0.87 4.86 1.04 7.48 4.63 4.77 0.74 0.09 0.00 12.2 0.07 0.08 5.79 0.83 147 1100 63.6 127 94.1 2502 53.6 3.33 14.7 43.8 5.44 0.04 0.59 116 0.22 0.27 0.12 0.97 1.48 0.91 4.93 1.09 8.06 4.85 4.60 0.67 0.12 bdl 19.2 0.01 0.06 0.17 bdl 81.0 bdl 89.6 243 106 6700 10.1 bdl 0.18 31.7 5.62 0.10 bdl 0.10 0.01 0.05 0.02 0.56 1.33 0.82 4.49 0.90 6.19 3.55 3.41 0.52 0.10 0.01 0.06 0.03 0.03 Avg estimate from inclusion/host proportion 10 ± 5 vol.% 20 ± 10 vol.% 30 ± 15 vol.% – – 852 ± 81 78 ± 38 – – – – – 0.87 ± 0.18 6.4 ± 0.4 – – 19 ± 18 1.20 7.9 ± 4.2 0.57 ± 0.01 2.2 ± 1.2 0.81 ± 0.49 8.8 ± 3 – – – – – – – – – 1.8 ± 1.7 4.1 ± 2 0.43 ± 0.30 1.3 ± 0.7 – – 846 ± 298 653 ± 214 – – – – – 2.7 ± 1.3 39 ± 11 – – 0.07 ± 0.02 0.80 ± 0.43 111 ± 84 1.4 ± 0.1 1.2 ± 0.8 0.46 ± 0.03 4.4 ± 0.2 – – – – – – – – – bdl 8.4 ± 2.8 0.13 ± 0.05 0.3 ± 0.1 – – 296 ± 56 1450 ± 740 – – – – – 4.9 ± 2 78 ± 45 – – 0.14 ± 0.08 1.0 ± 0.4 301 ± 169 3 ± 3.4 4.7 ± 5.3 0.61 ± 0.48 3.8 ± 1.6 – – – – – – – – – bdl 19 ± 19 0.52 ± 0.59 0.34 ± 0.21 On the basis of the inclusion/garnet volume proportion (vol.%), the analyses have been divided in three main groups. Also reported is the average (Avg) of the best estimated incompatible elements fluid concentration (see text for explanation) from more than 10 analyses for each group and the relative standard errors. Abbreviations same as in Table 2. N. Malaspina et al. / Lithos 107 (2009) 38–52 Li Be P K Sc Ti V Cr Ni Rb Sr Y Zr Nb Cs Ba La Ce Pr Nd Sm Eu Gd Tb Dy Er Yb Lu Hf Ta Pb Th U 10 ± 5 vol.% Host Grt Author's personal copy N. Malaspina et al. / Lithos 107 (2009) 38–52 (Table 4). Compared with the host garnet, both the pseudosecondary inclusions from Sulu and the primary inclusions from Maowu display strong enrichments in LILE, with spikes in Cs, Ba, Pb, Sr and high UN/ThN ratios. The inclusions in garnet of the Sulu peridotite are characterised by lower absolute concentrations with Cs, Ba and Pb normalised values up to 10 × PM (Fig. 6A). On the other hand, Cs, Ba and Pb spikes in the Maowu garnet websterite inclusions patterns reach 100–200 × PM normalised values, with UN N /ThN (Fig. 6B). It is important to note that the LILE patterns of all polyphase inclusions analysed here (RPC684 and RPC171) resemble the ones of garnet orthopyroxenites from Maowu studied by Malaspina et al. (2006). 6. Discussion 6.1. Metamorphic evolution and multistage metasomatism of the Sulu garnet peridotite The petrographic observations indicate that the Sulu peridotite preserves two main parageneses (Table 1). The older one consists of coarse inclusion-rich Grt1, coarse exsolved Cpx1 and phlogopite flakes (Phl1) (Fig. 2A,C). The younger one is formed by clear Cpx2, magnesite and Phl2 in equilibrium with Grt2, (Fig. 2B,D). Phl2 and euhedral magnesite share equilibrium textures with the other rock-forming minerals suggesting that they both formed in the garnet-facies stability field (P N 2.0 GPa for a lherzolite composition). The low Al2O3 content (b0.2 wt.%) in orthopyroxene coexisting with the garnet rim and Cpx2 indicates equilibration in a pressure range of 4.0–6.0 GPa at 700–1000 °C (Brey and Köhler, 1990). The mineral assemblage associated with Grt2 and Grt1 rim, and low-Al orthopyroxene in the peridotite is therefore considered to represent a UHP recrystallisation stage. Grt1 shows a slight core-to-rim zonation given by the decrease in MgO relative to an increase of FeOtot and Cr2O3 (Table 2). A similar chemical variation has been described by Yang and Jahn (2000) who proposed a double stage UHP equilibration for the Zhimafang garnet peridotite. The first stage records HT conditions (T ≈ 1000 °C and P N 5.0 GPa) and was estimated based on the Mg-rich core of porphyroblastic garnet and orthopyroxene. The second UHP stage, occurring at lower temperature (T ≈ 760 °C and P ≈ 4.2 GPa), has been inferred from Al-in-orthopyroxene and orthopyroxene–garnet exchange thermometers. The second UHP recrystallisation stage coincides with our estimates for the UHP event, which has been attributed by Yang and Jahn (2000) to derive from metasomatism by a SiO2-rich melt affecting the peridotites at UHP conditions. The major and trace element compositions of single minerals performed in this study can give information on the timing of mantle metasomatism with respect to the formation of different parageneses. Phl1 shares straight boundaries and triple junctions with garnet, clinopyroxene, orthopyroxene and olivine (Fig. 2C,D). Additionally, the trace element patterns suggest geochemical equilibrium between Phl1, Cpx1 and porphyroclastic garnet (Fig. 4). These observations provide evidence that a first metasomatic process occurred in the garnet stability field prior to or during the formation of the first paragenesis (Table 1). The Cpx1 trace element composition is characterised by a strong enrichment in LILE, with absolute concentrations up to ∼10 × PM, by elevated LREE contents, by relatively high Nb and Ta concentrations (Table 1), and high U–Th (Fig. 4B). Also Phl1 is characterised by enrichment in LILE, particularly in Cs, Rb and Ba (∼1000 × PM), high Nb–Ta and high U, Th contents. The textural relations and the trace element signatures of Cpx1 and Phl1 therefore suggest that these two minerals crystallised from a metasomatic agent rich in LILE, LREE and HFSE at HP/UHP conditions. Several experimental works demonstrated that HFSE are highly mobile in silicate melts rather than aqueous fluids (e.g. Brenan et al., 1995; Stalder et al., 1998). Also, enrichment in Th with respect to U can be considered as indicative of melt-derived metasomatism (e.g. Johnson and Plank, 1999 and references therein). These observations indicate that the Sulu peridotite was percolated by a metasomatic melt rising 47 from a deeper part of the mantle. The high LILE and HFSE concentrations that characterise the Cpx1 core and Phl1 compositions indicate that the metasomatic silicate melt might have had an alkaline character. Similar observations have been reported for clinopyroxenes of Ti–Al–Fe-rich harzburgites/lherzolites, dunites, wehrlites and clinopyroxenites from Canary Islands (Neumann et al., 2004). Clinopyroxene in metasomatised xenoliths from this suite is enriched in Th, U, Nb, Ta and LREE relative to MREE. These rocks and their clinopyroxenes have been interpreted as showing affinity to mildly alkaline basaltic lavas (Neumann et al., 1999). A recent work on mantle-derived xenoliths from different geological settings indicates that peridotite minerals from intraplate domains record HFSE enrichment (Nb, Ti) compared to samples from suprasubduction zones (Coltorti et al., 2007). Our new data on the trace element composition of minerals formed in the first stage of metasomatism combined with the petrologic and isotopic data by Yang and Jahn (2000) suggests that the Zhimafang garnet peridotite experienced metasomatism by a melt with alkaline character at high-temperature conditions (T ≈ 1000 °C and P N 5.0 GPa) probably at ∼380 Ma. Hence, this stage of metasomatism might be unrelated to the Triassic HP/UHP metamorphic event. Based on several petrological and isotopic studies, it has been extensively demonstrated that the Sulu UHP garnet peridotites correspond to mantle-derived ultramafic bodies tectonically emplaced in the crust (Yang et al., 1993; Yang and Jahn, 2000; Zhang et al., 2000). According to Zhang et al. (2000), the Sulu orogenic garnet peridotites possibly represent different portions of a suprasubduction mantle wedge sampled by the subducting crust during the Triassic Dabie–Sulu collision. Alternatively, the geochronological and geochemical results by Yang and Jahn (2000) suggest that the Zhimafang peridotites represent an old mantle slice (∼380 Ma) emplaced in the crust before the Triassic subduction and then subducted at UHP together with the continental crust. Several Sulu mantle-derived peridotites record multistage recrystallisation and it has been suggested that some metasomatism also occurred during the Triassic HP/UHP metamorphic event (Zhang et al., 2000; Yang and Jahn, 2000; Zheng et al., 2005; Yang et al., 2007). This can be evaluated using the second metamorphic stage documented in sample RPC684, which is related to lower temperature (T ≈ 750–800 °C and P ∼ 4.0–4.5 GPa) UHP conditions. Minerals belonging to the second paragenesis do not record the same strong incompatible element enrichments as those of Cpx1 and Phl1. As shown in Fig. 4B Cpx2 LILE concentrations do not reach values over 1 × PM, ThN/UN is closed to unity and Nb–Ta concentrations are lower than Cpx1. In contrast, the similarity of Phl1 and Phl2 suggests that Phl2 recrystallised from Phl1 but maintained the initially high LILE signature. This indicates that Phl2 is not in trace element equilibrium with Cpx2. The occurrence of euhedral magnesite in equilibrium with Cpx2 indicates that a carbonate component was added during the second metasomatic event. Zhang et al. (2007) proposed a metasomatism caused by a mantle-derived carbonatitic melt for core samples of peridotites at Zhimafang. In these rocks magnesite coexists with olivine + clinopyroxene + phlogopite ± orthopyroxene ± garnet in the matrix. The oxygen isotopic determination in the Zhimafang peridotitic rocks indicates that carbonatitic melts were the metasomatic agent that produced magnesite (Zhang et al., 2007). However, as will be discussed in the following sections, the microtextural identification of pseudosecondary inclusions in the porphyroclastic garnet core of the studied sample and their geochemical characterisation suggest that an incompatible element-rich fluid likely metasomatised the garnet peridotite, equilibrating with the newly forming Cpx2. 6.2. Formation of polyphase inclusions and constraints on the UHP fluid phase composition There is increasing evidence that the solid polyphase inclusions preserved in UHP minerals derive from trapped fluids (Scambelluri and Philippot, 2001; Scambelluri et al., 2001, 2008; van Roermund et al., Author's personal copy 48 N. Malaspina et al. / Lithos 107 (2009) 38–52 2002; Ferrando et al., 2005; Korsakov and Hermann, 2006; Malaspina et al., 2006). Some of these works provide unequivocal textural evidence (e.g. negative-crystal shapes of inclusions and constant proportions of infilling phases) that the solid polyphase inclusions are microprecipitates from homogeneous fluid phases (aqueous or melt), rather than relics of former mineral assemblages overgrown by the host garnets. In some cases, the presence of UHP minerals like diamond in the inclusion assemblage clearly constrains deep conditions for the formation of the inclusion infillings (Stöckhert et al., 2001; Van Roermund et al., 2002; Korsakov and Hermann, 2006). In other cases, recognition that the solid polyphase inclusions contain minerals unstable at UHP, requires that the mineral assemblage formed during exhumation. In such cases, post-entrapment changes of the inclusions and their re-equilibration and chemical exchange with the host mineral are common. The primary and pseudosecondary polyphase inclusions hosted in the Maowu and Sulu UHP garnets contain a number of hydrous phases, such as calcic-K amphibole, chlorite, talc, Ba-bearing mica. The presence of amphibole in the polyphase inclusions indicates that the inclusion assemblage crystallised after entrapment at UHP (P ≤ 3–2.5 GPa), i.e. during rock exhumation. The occurrence of more than 80 vol.% of hydrous phases in the inclusions suggests that some water was retained in the inclusions even after decompression. However, the precipitation of solid minerals at lower pressure suggests a volume change of the inclusions with a possible fluid loss. The constant volume proportions of oxide (10–20 vol.%) and hydrous silicates (80–90 vol.%) in all inclusions indicate that the loss of H2O if occurred, took place homogeneously. The equilibration process might have been accompanied by exchange between inclusions and garnet of those major elements which reside in garnet such as Mg, Al, Si, thus preventing their use to reconstruct major element concentrations and solubility in such fluids. On the other hand, this behaviour did not affect the incompatible trace elements, which are largely below detection in garnet (see Fig. 6A and B) and which, after detailed search, have not been found in microphases or haloes in garnet domains around the inclusions. One would expect that if there was a significant loss of LILE during post-entrapment processes this would have affected the most incompatible element Cs and would have lead to changing Cs/Rb. However, the trace element ratios for LILE are quite constant, independent of the size of the inclusions analysed suggesting that no significant and selective post-entrapment loss of LILE occurred. In the previous work on polyphase inclusions from Maowu garnet orthopyroxenites Malaspina et al. (2006) re-homogenised the primary solid inclusions in porphyroblastic garnet by piston cylinder experiments. The homogenisation experiments were run on inclusion-rich garnets by the addition of Al(OH)3, which produced about 8% of H2O. The authors demonstrated that the inclusions formed from a solute-rich aqueous fluid rather than a melt. Laser Ablation analysis performed on the resulting quench also indicated that the composition of the homogeneous solute-rich aqueous fluid was very similar to that of the starting polyphase inclusions. The identical trace element patterns of polyphase and re-homogenised inclusions, mostly for fluid-mobile elements such as the LILE, indicates that a possible H2O gain or loss does not affect the relative ratios of the trace element inclusions compositions. Also, element solubilities are likely significantly lower at lower pressures than at peak UHP conditions (Manning, 2004; Kessel et al., 2005; Manning, 2006) and partitioning with infilling phases such as amphibole will probably result in negligible loss of LILE. Finally, the host garnet in Maowu orthopyroxenites and websterites does not show major and trace element variations in correspondence of the inclusion sites. All these considerations point to a conservative behaviour of the fluidmobile trace elements. The Laser Ablation analyses of polyphase inclusions therefore can give a reliable information on the incompatible trace element signature of the fluid present at peak metamorphic conditions. While it is difficult to estimate absolute concentrations of the fluid phase (see below) the trace element patterns of the polyphase inclusions are consistent within a sample. The polyphase inclusions in the Maowu garnets have been studied in detail by Malaspina et al. (2006), who recognised that they represent trapped fractions of residual aqueous fluids likely produced by the reaction of a depleted garnet harzburgites with a hydrous granitic melt sourced from the adjacent gneisses. The reaction led to garnet orthopyroxenites plus a residual fluid locally trapped in solid polyphase inclusions inside garnet. The polyphase inclusions presented here for the Maowu sample RPC171 are texturally and compositionally similar to those of Malaspina et al. (2006). They have very high fluid-mobile trace elements concentrations, with spikes in Cs, Ba, Pb, Sr, and a high U–Th ratio (Table 4 and Fig. 6B) and their trace element patterns fully overlap the ones of Malaspina et al. (2006), represented by the white area in Fig. 6A. The polyphase inclusions found in the garnet peridotites from Sulu are secondary inside coarse garnets: they represent a fluid phase that infiltrated this mantle wedge after an earlier metasomatism by alkaline melt and equilibrated with the matrix at UHP conditions. These inclusions have mineral infillings and trace element compositions surprisingly similar to the ones of Maowu (Fig. 6A). The trace element patterns of the polyphase pseudosecondary inclusions of RPC684 show enrichments in LILE, with positive anomalies in Cs, Ba, Pb relative to Rb and K, and UN/ThN N 1. This similarity points to a genetic relation with the residual fluid from Maowu. The similarity between the residual Maowu fluid with the secondary inclusions in the UHP wedge-type garnet peridotite from Sulu, indicates that the fluids escaping from melt–peridotite reaction zones at the slab–mantle interface may be effective metasomatic agents in the mantle wedge. 6.3. Clinopyroxene/fluid and phlogopite/fluid partitioning It is useful to calculate qualitative partition coefficients (D =Cmineral/ C ) for the trace elements expected to partition in the fluid phase, i.e. the trace elements from Cs to Nd following the order of incompatibility showed in Fig. 6. In Table 4 the best estimates for the incompatible element concentrations calculated for the Sulu and Maowu inclusions are reported together with the uncertainties from the volume proportion between the inclusion site and Laser spot. In order to estimate the absolute trace element concentrations in the fluid (Cfluid) we considered the ratio between the volume percentage of the original inclusion and the host garnet within the Laser spot. It must be specified that the current volume observed is not necessarily the original volume of the inclusion during the fluid entrapment. Potential fluid loss can cause a decrepitation with consequent volume reduction whereas fluid/host interaction would apparently increase the inclusion volume. We therefore report a range of reasonable inclusion/host volume ratios in Table 4. Because the Laser Ablation analyses a volume of sample, we can estimate the absolute ppm concentrations of trace elements in inclusions by subtracting the volume percentage of the host garnet that mixes with the inclusion. We also calculated the standard error related to the different size of inclusions or cluster of inclusions occupying the Laser spot. The recalculated concentrations of fluid-mobile elements and the relative standard deviations are reported in Table 4. Because in the Sulu garnet peridotite Nb, Ta and Nd are difficult to separate from the contribution of the host garnet (Fig. 6A), these elements have not been considered in the determination of mineral/fluid partition coefficients. The clinopyroxene/fluid (DCpx/fluid) partition coefficients calculated using clinopyroxenes and polyphase inclusions of the Sulu peridotite and Maowu websterite are shown in Fig. 7A. Error bars refer to the standard deviation of the average trace element mineral concentration used to calculate the partition coefficients, combined with the standard error of the fluid concentration (Table 4). Overall, the clinopyroxene/ fluid trace element partition coefficients achieved for Maowu and Sulu may range over three orders of magnitude for the LILE. Since the Sulu samples contain at least two generations of clinopyroxene, we will adopt the Maowu sample RPC171 as reference guideline to evaluate the clinopyroxene/fluid partitioning at UHP, as this rock has only one fluid Author's personal copy N. Malaspina et al. / Lithos 107 (2009) 38–52 49 Fig. 7. Mineral/fluid partition coefficients for incompatible fluid-mobile elements in clinopyroxenes (A) and phlogopites (B) from Sulu garnet peridotite (RPC684) and Maowu websterites (RPC171 and MWF2A). Error bars indicate the standard deviation relative to the combination of the error in the fluid and the error on the average trace element mineral concentration (Table 3). The clinopyroxene/fluid partition coefficients (A) are compared with reference clinopyroxene/fluid partition coefficients calculated at 2.0 GPa and 900–1100 °C (Ayers, 1998, and references therein; Brenan et al., 1995). The phlogopite/fluid partition coefficients in Maowu sample (B) are compared with DPhl/fluid of phlogopites in garnet websterites from Western Norway (Scambelluri et al., 2008). generation of UHP clinopyroxene equilibrated with the primary polyphase inclusions in garnet. We thus consider this partitioning as the most robust. The DCpx/fluid indicates that LILE are highly incompatible while U and La are moderately incompatible. In contrast, D values for Pb, Sr and Ce are close to unity. These partitioning values can now be used to evaluate which clinopyroxene generation observed in the Sulu sample fits the clinopyroxene/fluid partitioning criteria achieved for RPC171. Cpx2 in sample RPC684 displays a very similar partitioning to the Maowu sample RPC171. On the other hand, the core of Cpx1 displays DCpx/fluid ≥ 1 for Cs, Rb, Ba, K, Th, U, and La, indicating that this clinopyroxene is not in equilibrium with the fluid. The rim of Cpx1 is transitional between the partitioning for Cpx1 core and Cpx2 and does not match the partitioning of the reference sample RPC171 as good as the partitioning obtained from Cpx2. These observations suggest that the fluid is most likely in equilibrium with Cpx2. The clinopyroxene partition coefficients reported in Fig. 7A have been compared with the experimentally determined partition coefficients between aqueous fluids and clinopyroxene in mantle-like systems (Brenan et al., 1995; Ayers, 1998 and references therein). Ayers et al. (1997) performed experiments on partitioning between peridotite rocks and H2O fluids at pressures comparable with the ones estimated for the metasomatism of the studied samples, but at higher temperatures (900– 1100 °C). The DCpx/fluid values calculated in this study are comparable with the reference experimental data, particularly clinopyroxene of Maowu websterite RPC171 and Cpx2 from the Sulu peridotite. This suggests that RPC171 clinopyroxene equilibrates with the residual fluid released after the harzburgite/Si-rich metasomatic agent reaction. Additionally, Cpx2 in the Sulu peridotite appears to be in equilibrium with the fluid associated with the microfractures of garnet cores. This observation indicates that the LILE-enriched fluid percolated a previously metasomatised mantle peridotite (Cpx1 and Phl1) and then reequilibrated with the outer rims of clinopyroxenes and with the newly growing Cpx2. Our estimated partitioning suggests that all LILE are highly incompatible whereas Th and U are moderately incompatible in clinopyroxene whereas Sr is moderately compatible. Phlogopite is another phase rich in incompatible elements that occurs in both the studied samples. However, we have shown that in the Sulu peridotite the fluid is only in equilibrium with Cpx2. Consequently we are not able to plot a phlogopite–fluid partitioning for this sample because Phl2 is not in trace element equilibrium with Cpx2. Phlogopite analysed in the phlogopite-rich layer of MWF2A from the Maowu Complex compared to the estimated fluid of RPC171 shows DPhl/fluid well LILE and DPhl/fluid close to 1. The apparent DPhl/fluid is above unity and DPhl/fluid Pb Sr Th,U well above unity and is related to the unexpectedly high Th and U contents of the phlogopite. In the time-resolved LA-ICP-MS analyses of these phlogopites, the Th and U signals are smooth and there is no indication of Th–U rich inclusions. It must be noted that the phlogopites of the Maowu sample come not exactly from the same sample as the polyphase inclusions, which might introduce a systematic error. Trace element analyses of metasomatic phlogopites formed at 7 GPa and 1000 °C in garnet websterites from Bardane (Western Norway) and DPhl/fluid for a number of incompatible trace elements at such UHP conditions have been recently published by Scambelluri et al. (2008). Compared with the Maowu phlogopite presented here, the Norwegian phlogopite has much lower Th and U (about 1 and 10 × PM, respectively). This suggests a substantial difference in the metasomatic agents responsible for phlogopite formation in the two settings. The Maowu phlogopite was produced in a Th- and U-rich environment. This agrees with our conclusion that the Maowu phlogopite rinds around garnet orthopyroxenites represent first reaction products between mantle rocks and the fluid phase sourced from the adjacent silica-saturated, incompatible element-rich orthogneisses. In contrast the Bardane phlogopites seem to form in an Th–U poor environment, that was not in immediate contact with felsic rocks. The different U and Th contents of the Bardane and Maowu phlogopites yield a quite different phlogopite/fluid partition coefficients for such elements (Fig. 7B). Moreover, in the Bardane UHP rocks DPhl/fluid was calculated adopting the experimental fluid released by Kdeficient mafic systems at 5 GPa–1000 °C (Kessel et al., 2005), rather than using the natural polyphase inclusions deriving from natural fluids present in the rock at UHP. As a consequence, comparison between the Dabie–Sulu and the Norwegian DPhl/fluid is also hampered by the uncertainties in the fluid phases employed. Nevertheless, it is worth to note that the DPhl/fluid for Cs, Rb, Ba, and Pb are in rather good agreement. This indicates that phlogopites might be important repositories of LILE with respect to fluid phases in a range of pressures and temperatures. 6.4. The role of phlogopite in the trace element fractionation It is widely accepted that phlogopite is an important carrier of K and H2O in the Earth's upper mantle. Experimental works demonstrated that in ultramafic systems phlogopite has a large P–T stability, exceeding depths of 150 km (Wendtland and Eggler, 1980; Sudo and Tatsumi, 1990; Konzett and Ulmer, 1999). It has also been evidenced that phlogopite composition changes with different P–T conditions. Particularly, Author's personal copy 50 N. Malaspina et al. / Lithos 107 (2009) 38–52 6.5. Fluid–peridotite interaction in the metasomatised mantle wedge Fig. 8. PM normalised trace element patterns of clinopyroxene and phlogopite from Maowu websterite and orthopyroxenite (RPC171 and MWF2A), and the best estimate of the residual fluid composition, compared with country granitic gneiss (from Xia et al., 2008). synthesis experiments performed at HP/UHP conditions produced phlogopitic micas showing increasing Si content with increasing pressure (Wunder and Melzer, 2002; Fumagalli, 2003; Fumagalli and Stixrude, 2007). Phlogopites in KNCMASH and phlogopite lherzolite systems from the experimental work of Konzett and Ulmer (1999) have Si from 3.03 to 3.09 p.f.u. in runs at P = 6.0–6.5 GPa. Similarly, piston cylinder experiments carried out in the KCMASH system by Hermann (2002) yielded biotite containing Si exceeding the upper limit of pure phlogopite (3.0 p.f.u.) at P N 2.5 GPa. Our data on phlogopites from Maowu garnet orthopyroxenite MWF2A are in agreement with formation at UHP conditions. As shown in Table 2, compared with the Sulu peridotite RPC684, the Maowu phlogopites are characterised by much higher SiO2 contents and lower K2O. The recalculated structural formulae indicate Si= 3.05–3.07 p.f.u. and K ∼ 0.67 p.f.u. in the Maowu phlogopites, whereas the Sulu phlogopites have Si= 2.86–2.93 p.f.u. and K ∼ 0.8 p.f.u. The increase in Si content coupled with the systematic decrease in K has been explained by the occurrence of a talc component, in solid solution with phlogopite (Hermann, 2002; Wunder and Melzer, 2002, Fumagalli and Stixrude, 2007). Such interpretation is supported by the occurrence of talc intergrowths in our phlogopites from MWF2A sample. The geochemical analyses also demonstrate that talc associated with phlogopite is in equilibrium with the UHP mineral paragenesis (Fig. 5C). Formation of phlogopite layers was assumed to occur at the slab/ mantle wedge interface as the result of the percolation of K-enriched slab-derived fluid (Sudo and Tatsumi, 1990; Manning, 2004). Metasomatic phlogopite may hence play an important control on the LILE compositions of slab-derived agents responsible for metasomatism of the mantle wedge. Melzer and Wunder (2001) determined the distribution of K–Rb and Cs–K between phlogopite and K, Rb, Cs and Clbearing aqueous solutions at P = 2–4 GPa and T = 800 °C. According to their work all LILE are compatible in phlogopite, but in a slightly different way, leading to a fractionation of the LILE. Cs fractionates into the fluid, whereas Rb is preferentially partitioned into phlogopite together with K. The phlogopite/fluid partition coefficients of Fig. 7B and together with the ones achieved for UHP phlogopites (Scambelluri et al., 2008) confirm that this phase is a repository for Cs, Rb, Ba, K. Finally, a recent work by Hermann and Spandler (2008) suggests that the occurrence of phlogopite in a mantle peridotite equilibrating with a slab-derived fluid may affect K2O/H2O of the fluid reaching the locus of partial melting in the mantle wedge. This is well illustrated in the Maowu samples by the formation of the metasomatic garnet, orthopyroxene and phlogopite. While garnet and orthopyroxene efficiently extract the main components SiO2 and Al2O3 from the hydrous granitic fluid phase and thus passively enrich H2O, phlogopite will act as a sink of K2O and lower also other LILE in the residual fluid phase. Several authors identified the subducted sediments and the underlying mafic crust as main repositories for the LILE and incompatible elements in the subducted crust (Plank and Langmuir, 1993, 1998; Elliot, 2003). Both sediment and mafic crust have much higher incompatible element concentrations than typical mantle rocks and thus have the potential to impress a peculiar chemical signature if a fluid phase that equilibrated with the subducted sediments is added to the mantle wedge. It has been emphasised that even the ultramafic rocks may produce LILE-enriched fluids (Scambelluri et al., 1997). Fluid-enhanced partial melting of sediments at sub-arc depths occurs at T N 700 °C and produces hydrous granitic melts rich in incompatible elements (Hermann and Green, 2001; Hermann et al., 2006). Also mafic crust dehydration at P N 5 GPa can produce silica-saturated LILE–LREE-enriched supercritical fluids (Kessel et al., 2005). Fluid phases extracted from metasediments or from the mafic part of the subducted crust may potentially interact with the mantle wedge peridotites to form a garnet pyroxenite layer at the slab–mantle interface and a free LILE–LREE-enriched aqueous fluid (Malaspina et al., 2006). We have shown in Fig. 7A that LILE preferentially partition into the residual fluid rather than clinopyroxene. On the other hand, they are strongly compatible in phlogopite which also fractionates Rb and K from Cs and Ba (Fig. 7B). The composition of the residual fluid is therefore influenced only by phlogopite precipitation in terms of LILE fractionation (e.g. Melzer and Wunder, 2001). This can be qualitatively illustrated by Fig. 8. The PM normalised patterns of clinopyroxene, phlogopite and residual fluid from the Maowu pyroxenite samples have been plotted together with an average composition of country-rock gneisses from Maowu (Xia et al., 2008). The geochemical and isotopic study of these granitic gneisses indicates that they underwent variable degrees of dehydration and partial melting during subduction and exhumation of the continental crust (Xia et al., 2008). It is important to note that at the peak UHP conditions the composition of the fluid phase produced by the gneiss approaches that of a hydrous granitic melt rather than an aqueous fluid (Malaspina et al., 2006). The trace element pattern of the agent produced by the dehydration or melting of gneisses is therefore expected to be richer in LILE than the gneisses showed in Fig. 8. In contrast, the residual aqueous fluid, represented by the grey line, has lower LILE concentrations than the country-rock gneisses (black area). This indicates that newly formed phlogopite might have retained some LILE and thus acted as a LILE filter at the interface between subducted crust and overlying mantle wedge. Once the LILE-enriched aqueous fluid escapes the slab–mantle reaction front, it is able to migrate in the mantle wedge. Evidence of interaction of such a fluid with mantle wedge peridotites is given by the polyphase inclusions in the Sulu sample which are characterised by trace element signatures that are very similar to the residual fluid observed at the reaction front in the Maowu sample. We expect that such a fluid can reach the locus of partial melting in the mantle wedge without significant further modification of the LILE signatures. Acknowledgments We are grateful to R. Compagnoni for providing us some of the samples studied in this work, and to F. Rolfo and S. Xu for guiding the field trip in Dabie–Shan. The analytical assistance of A. Norris, C. Allen and M. Shelley for laser ablation and electron microprobe at the Research School of Earth Sciences are greatly acknowledged. Critical reviews by J.C. Ayers and an anonymous reviewer, and discussion with S. Poli, P. Fumagalli and S. Tumiati significantly improved the manuscript. N. Malaspina and M. Scambelluri acknowledge funding by the Italian MIUR-Cofin and by the Universities of Milano and Genova. J. Hermann acknowledges financial support by the Australian Research Council and the Swiss National Science Foundation. Author's personal copy N. Malaspina et al. / Lithos 107 (2009) 38–52 References Ayers, J., 1998. Trace element modeling of aqueous fluid; peridotite interaction in the mantle wedge of subduction zones. Contributions to Mineralogy and Petrology 132, 390–404. Ayers, J., Dittmer, S.K., Layne, G.D., 1997. Partitioning of elements between peridotite and H2O at 2.0–3.0 GPa and 900–110 °C, and application to models of subduction zone processes. Earth and Planetary Science Letters 150, 381–398. Ayers, J.C., Dunkle, S., Gao, S., Miller, C.F., 2002. Constraints on timing of peak and retrograde metamorphism in the Dabie Shan Ultrahigh-Pressure Metamorphic Belt, east-central China, using U–Th–Pb dating of zircon and monazite. Chemical Geology 186, 315–331. Brenan, J.M., Shaw, H.F., Phinney, D.L., Ryerson, F.J., 1994. Rutile-aqueous fluid partitioning of Nb, Ta, Hf, Zr, U and Th: implications for high field strength element depletions in island arc basalts. Earth and Planetary Science Letters 128, 327–339. Brenan, J.M., Shaw, H.F., Ryerson, F.J., Phinney, D.L., 1995. Mineral–aqueous fluid partitioning of trace elements at 900 °C and 2.0 GPa; constraints on the trace element chemistry of mantle and deep crustal fluids. Geochimica et Cosmochimica Acta 59, 3331–3350. Brey, G.P., Köhler, T., 1990. Geothermobarometry in four lherzolite II. New thermobarometers and practical assessment of existing thermobarometers. Journal of Geology 31, 1553-1378. Brueckner, H.K., 1998. A sinking intrusion model for the introduction of garnet-bearing peridotites into continent collision orogens. Geology, 26, 631–634. Brueckner, H.K., Medaris, L.G., 2000. A general model for the intrusion and evolution of “mantle” garnet peridotites in high-pressure and ultrahigh-pressure metamorphic terranes. Journal of Metamorphic Geology 18, 123–133. Coltorti, M., Bonadiman, C., Faccini, B., Grégoire, M., O'Reilly, S.Y., Powell, W., 2007. Amphiboles from suprasubduction and intraplate lithospheric mantle. Lithos 99, 68–84. Elliott, T., 2003. Tracers of the slab. In: Eiler, J. (Ed.), Inside the Subduction Factory. Geophysical Monograph, vol. 138, pp. 23–45. Elliott, T., Plank, T., Zindler, A., White, W., Bourdon, B., 1997. Element transport from slab to volcanic front at the Mariana arc. Journal of Geophysical Research 102, 14991–15019. Enami, M., Zhang, Q., 1993. High-pressure eclogite from northern Jiangsu and southern Shandong province, eastern China. Journal of Metamorphic Geology 11, 589–603. Evans, B.W., Trommsdorff, V., 1978. Petrogenesis of garnet lherzolite, Cima di Gagnone, Lepontine Alps. Earth and Planetary Science Letters 40, 415–424. Ferrando, S., Frezzotti, M.L., Dallai, L., Compagnoni, R., 2005. Multiphase solid inclusions in UHP rocks (Su-Lu, China): remnants of supercritical silicate-rich aqueous fluids released during continental subduction. Chemical Geology 223, 68–81. Fumagalli, P., 2003. Subsolidus phase relations in K-doped lherzolites and the composition of phlogopite to 6.0 GPa: an experimental study. Geophysical Research Abstracts 5, 12436. Fumagalli, P., Stixrude, L., 2007. The 10 Å phase at high pressure by first principles calculations and implications for the petrology of subduction zones. Earth and Planetary Science Letters 260, 212–226. Hawkesworth, C.J., Gallagher, K., Hergt, J.M., McDermott, F., 1993. Trace element fractionation processes in the generation of island arc basalts. Philosophical Transactions of the Royal Society A 342, 179–191. Heinrich, C.A., Pettke, T., Halter, W.E., Aigner-Torres, M., Audétat, A., Günther, D., Hattendorf, B., Bleiner, D., Guillong, M., Horn, I., 2003. Quantitative multi-element analysis of minerals, fluid and melt inclusions by laser-ablation inductively-coupled-plasma mass-spectrometry. Geochimica et Cosmochimica Acta 67, 3473–3497. Hermann, J., 2002. Experimental constraints on phase relations in subducted continental crust. Contributions to Mineralogy and Petrology 143, 219–235. Hermann, J., Gerald, J.F., Malaspina, N., Berry, A., Scambelluri, M., 2007. OH-bearing planar defects in olivine produced by the breakdown of Ti-rich humite minerals from Dabie Shan (China). Contributions to Mineralogy and Petrology 153, 417–428. Hermann, J., Green, D.H., 2001. Experimental constraints on high pressure melting in subducted crust. Earth and Planetary Science Letters 188, 149–168. Hermann, J., Spandler, C., 2008. Sediment melts at sub-arc depths: an experimental study. Journal of Petrology 49, 717–740. Hermann, J., Spandler, C., Hack, A., Korsakov, A., 2006. Aqueous fluids and hydrous melts in high- pressure and ultra-high pressure rocks: implications for element transfer in subduction zones. Lithos 92, 399–417. Hirajima, T., Ishiwatari, A., Cong, B., Zhang, R.Y., Banno, S., Nozaka, T., 1990. Coesite from Mengzhong eclogite at Donghai county, northeastern Jiangsu province, China. Mineralogical Magazine 54, 579–583. Jahn, B.M., Fan, Q., Yang, J.J., Henin, O., 2003. Petrogenesis of the Maowu pyroxenite– eclogite from the UHP metamorphic terrane of Dabieshan: chemical and isotopic constraints. Lithos 70, 243–267. Johnson, M.C., Plank, T., 1999. Dehydration and melting experiments constrain the fate of subducted sediments. Geochemistry Geophysics Geosystems 1, 00014. Kelemen, P.B., Dick, H.J.B., Quick, J.E., 1992. Formation of harzburgite by pervasive melt/ rock reaction in the upper mantle. Nature 358, 635–641. Kelemen, P.B., Shimizu, N., Dunn, T., 1993. Relative depletion of niobium in some arc magmas and the continental crust: partitioning of K, Nb, La and Ce during melt rock reaction in the upper mantle. Earth and Planetary Science Letters 120, 111–134. Kessel, R., Ulmer, P., Pettke, T., Schmidt, M.W., Thompson, A.B., 2005. The water–basalt system at 4 to 6 GPa: phase relations and second critical endpoint in a K-free eclogite at 700 to 1400 °C. Earth and Planetary Science Letters 237, 873–892. Konzett, J., Ulmer, P., 1999. The stability of hydrous potassic phases in lherzolite mantle; an experimental study to 9.5 GPa in simplified and natural bulk compositions. Journal of Petrology 40, 629–652. Korsakov, A.V., Hermann, J., 2006. Silicate and carbonate melt inclusions associated with diamonds in deeply subducted carbonate rocks. Earth and Planetary Science Letters 241, 104–118. 51 Liou, J.G., Tsujimori, T., Zhang, R.Y., Katayama, I., Maruyama, S., 2004. Global UHP metamorphism and continental subduction/collision: the Himalayan model. International Geology Review 46, 1–27. Liou, J.G., Zhang, R.Y., 1998. Petrogenesis of an ultrahigh-pressure garnet-bearing ultramafic body from Maowu, Dabie Mountains, east-central China. Island Arc 7, 115–134. Liou, J.G., Zhang, R.Y., Eide, E.A., Matuyama, S., Wang, X., Ernst, W.G., 1996. Metamorphism and tectonics of high-P belts in Dabie–Sulu Regions, eastern China. In: Yin, A., Harrison, T.M. (Eds.), The Tectonics Evolution of Asia. Cambridge Univ. Press, pp. 300–343. Liu, F., Xu, Z., Liou, J.G., Katayama, I., Masago, H., Maruyama, S., Yang, J., 2002. Ultrahigh-P mineral inclusions in zircons from gneissic core samples of the Chinese Continental Scientific Drilling Site in eastern China. European Journal of Mineralogy 14, 499–512. Liu, F.L., Xu, Z.Q., Liou, Z.G., Song, B., 2004. SHRIMP U–Pb ages of ultrahigh-pressure and retrograde metamorphism of gneisses, south-western Sulu terrane, eastern China. Journal of Metamorphic Geology 22, 315–326. Malaspina, N., Hermann, J., Scambelluri, M., Compagnoni, R., 2006. Polyphase inclusions in garnet–orthopyroxenite (Dabie Shan, China) as monitors for metasomatism and fluid-related trace element transfer in subduction zone peridotite. Earth and Planetary Science Letters 249, 173–187. Manning, C.E., 2004. The chemistry of subduction-zone fluids. Earth and Planetary Science Letters 223, 1–16. Manning, C.E., 2006. Mobilizing aluminum in crustal and mantle fluids. Journal of Geochemical Exploration 89, 251–253. McCulloch, M.T., Gamble, J.A., 1991. Geochemical and geodynamical constraints on subduction zone magmatism. Earth and Planetary Science Letters 102, 358–374. McDonough, W.F., Sun, S.S., 1995. Composition of the Earth. Chemical Geology 120, 223–253. Medaris, L.G., 2000. Garnet peridotites in Eurasian high-pressure and ultrahigh-pressure terranes: a diversity of origins and thermal histories. In: Ernst, W.G., Liou, J.G. (Eds.), Ultrahigh-Pressure Metamorphism and Geodynamics in Collision-type Orogenic Belts. International Book Series, vol. 4. Geological Society of America and Columbia (Md.). Bellwether Publ. Ltd., Boulder (Co.), pp. 57–73. Melzer, S., Wunder, B., 2001. K–Rb–Cs partitioning between phlogopite and fluid: experiments and consequences for the LILE signatures of island arc basalts. Lithos 59, 69–90. Neumann, E.R., Griffin, W.L., Pearson, N.J., O'Reilly, S., 2004. The evolution of the upper mantle beneath the Canary Islands: information from trace elements and Sr isotope ratios in minerals in mantle xenoliths. Journal of Petrology 45, 2573–2612. Neumann, E.R., Wulff-Pedersen, E., Simonsen, S.L., Pearson, N.J., Mitjavila, J., Martí, J., 1999. Evidence for fractional crystallization of periodically refilled magma chambers in Tenerife, Canary Islands. Journal of Petrology 40, 1089–1123. Okay, A.T., Sengor, A.M.C., 1992. Evidence for intra-continental thrust related exhumation of the ultra-high pressure rocks in China. Geology, 20, 411–414. Pearce, N.J.G., Perkins, W.T., Westgate, J.A., Gorton, M.P., Jackson, S.E., Neal, C.R., Chenery, S.P., 1997. A compilation of new and published major and trace element data for NIST SRM 610 and NIST SRM 612 glass reference materials. Geostandard News Letters 21, 115–144. Plank, T., Langmuir, C.H., 1993. Tracing trace elements from sediment input to volcanic output at subduction zones. Nature 362, 739–743. Plank, T., Langmuir, C.H., 1998. The chemical composition of subducting sediment: implications for the crust and mantle. Chemical Geology 145, 325–394. Poli, S., Schmidt, M.W., 1998. The high-pressure stability of zoisite and phase relationships of zoisite-bearing assemblages. Contributions to Mineralogy and Petrology 130, 162–175. Poli, S., Schmidt, M.W., 2002. Petrology of subducted slabs. Annual Review of Earth and Planetary Science Letters 30, 207–235. Rampone, E., Morten, L., 2001. Records of crustal metasomatism in the garnet peridotites of the Ulten Zone (Upper Austroalpine, Eastern Alps). Journal of Petrology 42, 207–219. Roedder, E., 1984. Fluid Inclusions. Reviews in Mineralogy, vol. 12. Mineralogical Society of America, Washington. Scambelluri, M., Bottazzi, P., Trommsdorff, V., Vannucci, R., Hermann, J., GòmezPugnaire, M.T., Lòpez-Sànchez Vizcàino, V., 2001. Incompatible element-rich fluids released by antigorite break-down in deeply subducted mantle. Earth and Planetary Science Letters 192, 457–470. Scambelluri, M., Hermann, J., Morten, L., Rampone, E., 2006. Melt versus fluid induced metasomatism in spinel to garnet wedge peridotites (Ulten Zone, Eastern Italian Alps): clues from trace elements and Li abundances. Contributions to Mineralogy and Petrology 151, 372–394. Scambelluri, M., Pettke, T., van Roermund, H.L.M., 2008. Majoritic garnets monitor deep subduction fluid flow and mantle dynamics. Geology 36, 59–62. Scambelluri, M., Philippot, P., 2001. Deep fluids in subduction zones. Lithos 55, 213–227. Scambelluri, M., Piccardo, G.B., Philippot, P., Robbiano, A., Negretti, L., 1997. High salinity fluid inclusions formed from recycled seawater in deeply subducted alpine serpentinite. Earth and Planetary Science Letters 148, 485–500. Schmidt, M.W., Poli, S., 1998. Experimentally based water budgets for dehydrating slabs and consequences for arc magma generation. Earth and Planetary Science Letters 163, 361–379. Spandler, C.J., Hermann, J., Arculus, R.J., Mavrogenes, J.A., 2003. Redistribution of trace elements during prograde metamorphism from lawsonite blueschist to eclogite facies; implications for deep subduction-zone processes. Contributions to Mineralogy and Petrology 146, 205–222. Stalder, R., Foley, S.F., Brey, G.P., Horn, I., 1998. Mineral–aqueous fluid partitioning of trace elements at 900–1200 °C and 3.0–5.7 GPa: new experimental data for garnet, clinopyroxene, and rutile, and implications for mantle metasomatism. Geochimica et Cosmochimica Acta 62, 1781–1801. Author's personal copy 52 N. Malaspina et al. / Lithos 107 (2009) 38–52 Stöckhert, B., Duyster, J., Trepman, C., Massonne, H.J., 2001. Microdiamond daughter crystals precipitated from supercritical COH + silicate fluids included in garnet, Erzgebirge, Germany. Geology 29, 391–394. Sudo, A., Tatsumi, Y., 1990. Phlogopite and K-amphibole in the upper mantle: implication for magma genesis in subduction zones. Geophysical Research Letters 17, 29–32. Tenthorey, E., Hermann, J., 2004. Composition of fluids during serpentinite breakdown in subduction zones: evidence for limited boron mobility. Geology 32, 865–868. Trommsdorff, V., Risold, A.C., Reusser, E., Ulmer, P., 1998. Subduction of hydrated mantle to conditions of garnet peridotite, Central Alps. In: Liou, J.G., Ernst, W.G. (Eds.), International Workshop on UHP Metamorphism and Exhumation, pp. 6–7. Tumiati, S., Godard, G., Martin, S., Klotzli, U., Monticelli, D., 2007. Fluid-controlled crustal metasomatism within a high-pressure subducted melange (Mt. Hochwart, Eastern Italian Alps). Lithos 94, 148–167. Ulmer, P., 2001. Partial melting in the mantle wedge—the role of H2O in the genesis of mantle-derived “arc-related” magmas. Physics in Earth Planetary Interiors 127, 215–232. Van Roermund, H.L.M., Carswell, D.A., Drury, M.R., Heijboer, T.C., 2002. Microdiamonds in a megacrystic garnet websterite pod from Bardane on the island of Fjørtoft, western Norway: evidence for diamond formation in mantle rocks during deep continental subduction. Geology 11, 959–962. Wang, X., Liou, J.G., 1991. Regional ultrahigh-pressure coesite-bearing eclogitic terrane in central China: evidence from country rocks gneiss, marble and metapelite. Geology 19, 933-930. Wendtland, R.F., Eggler, D.H., 1980. The origins of potassic magmas: 2. Stability of phlogopite in natural spinel lherzolite and in the system KalSiO4–MgO–SiO2–H2O– CO2 at high pressures and high temperatures. American Journal of Science 280, 421–458. Wilson, S.A., 1997. The collection, preparation, and testing of USGS reference material BCR-2, Columbia River, Basalt, U.S. Geological Survey Open-File Report 98. Wunder, B., Melzer, S., 2002. Interlayer vacancy characterization of synthetic phlogopitic micas by IR spectroscopy. European Journal of Mineralogy 14, 1129–1138. Xia, Q.X., Zheng, Y.F., Zhou, L.G., 2008. Dehydration and melting during continental collision: constraints from element and isotope geochemistry of low-T/UHP granitic gneiss in the Dabie orogen. Chemical Geology 247, 36–65. Xue, F., Rowley, D.B., Baker, J., 1996. Refolded syn-ultrahigh-pressure thrust sheets in the south Dabie complex, China: field evidence and tectonic implications. Geology 5, 455–458. Yang, J.J., Godard, G., Kienast, J.R., Lu, Y., Sun, J., 1993. Ultrahigh-pressure (60 kbar) magnesite-bearing garnet peridotites from northeastern Jiangsu, China. Journal of Geology 101, 541–554. Yang, J.J., Jahn, B.M., 2000. Deep subduction of mantle-derived garnet peridotites from the Su-Lu UHP metamorphic terrane in China. Journal of Metamorphic Geology 18, 167–180. Yang, J.J., Smith, D.C., 1989. Evidence for a former sanidine–coesite eclogite at Lanshantou, eastern China, and the recognition of the Chinese “Su-Lu Coesite– Eclogite Province”. Terra Nova, Abstract 1, 26. Yang, J.S., Zhang, R.Y., Li, T.F., Zhang, Z.M., Liou, J.G., 2007. Petrogenesis of the garnet peridotite and garnet-free peridotite of the Zhimafang ultramafic body in the Sulu ultrahigh-pressure metamorphic belt, eastern China. Journal of Metamorphic Geology 25, 187–206. Ye, K., Yao, Y., Katayama, I., Cong, B.L., Wang, Q.C., Maruyama, S., 2000. Large areal extent of ultrahigh-pressure metamorphism in the Sulu ultrahigh-pressure terrane of East China: new implications from coesite and omphacite inclusions in zircon of granitic gneiss. Lithos 52, 157–164. Zhang, R.Y., Hirajima, T., Banno, S., Cong, B., Liou, J.G., 1995a. Petrology of ultrahighpressure rocks from the southern Su-Lu region, eastern China. Journal of Metamorphic Geology 13, 659–675. Zhang, R.Y., Liou, J.G., Cong, B., 1995b. Talc-, magnesite- and Ti-clinohumite-bearing ultrahigh- pressure meta-mafic and ultramafic complex in the Dabie Mountains, China. Journal of Petrology 36, 1011–1037. Zhang, R.Y., Liou, J.G., Tsai, C.H., 1996. Petrogenesis of a high-temperature metamorphic terrane: a new tectonic interpretation for the Northern Dabieshan, central China. Journal of Metamorphic Geology 14, 271–284. Zhang, R.Y., Liou, J.G., Yang, J.S., Yui, T.F., 2000. Petrochemical constraints for dual origin of garnet peridotites from the Dabie–Sulu UHP terrane, eastern-central China. Journal of Metamorphic Geology 18, 149–166. Zhang, R.Y., Li, T., Rumble, D., Yui, T.F., Li, L., Yang, J.S., Pan, Y., Liou, J.G., 2007. Multiple metasomatism in Sulu ultrahigh-P garnet peridotite constrained by petrological and geochemical investigations. Journal of Metamorphic Geology 25, 149–164. Zhang, R.Y., Rumble III, D., Liou, J.G., Wang, Q.C., 1998. Low d18O, ultrahigh-P garnet-bearing mafic and ultramafic rocks from Dabie Shan, China. Chemical Geology 150, 161–170. Zheng, J.P., Zhang, R.Y., Griffin, W.L., Liou, J.G., O'Reilly, S.Y., 2005. Heterogeneous and metasomatized mantle recorded by trace elements in minerals of the Donghai garnet peridotites, Sulu UHP terrane, China. Chemical Geology 221, 243–259.