Journal of Petrology Advance Access originally published online on June 20, 2006
Journal of Petrology 2006 47(10):1943-1971; doi:10.1093/petrology/egl032
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Exhumation History of a Garnet Pyroxenite-bearing Mantle Section from a Continent–Ocean Transition (Northern Apennine Ophiolites, Italy)
1 DIPARTIMENTO DI SCIENZE DELLA TERRA, UNIVERSITÀ DI PARMA PARCO AREA DELLE SCIENZE 157A, 43100 PARMA, ITALY
2 DIPARTIMENTO DI SCIENZE DELLA TERRA, UNIVERSITÀ DI PAVIA I-27100 PAVIA, ITALY
3 IGG-CNR ISTITUTO DI GEOSCIENZE E GEORISORSE PAVIA, I-27100 PAVIA, ITALY
4 DEPARTMENT OF GEOLOGY, ROYAL HOLLOWAY, UNIVERSITY OF LONDON TW20 0EX EGHAM, UK
RECEIVED JUNE 2, 2005; ACCEPTED MAY 17, 2006
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONGarnet clinopyroxenite and garnet websterite layers occur locally within mantle peridotite bodies from the External Liguride Jurassic ophiolites (Northern Apennines, Italy). These ophiolites were derived from an ocean–continent transition similar to the present-day western Iberian margin. The garnet clinopyroxenites are mafic rocks with a primary mineral assemblage of pyrope-rich garnet + sodic Al-augite (Na2O
2·5 wt %, Al2O3 12·5 wt %), with accessory graphite, Fe–Ni sulphides and rutile. Decompression caused Na-rich plagioclase (An50–45) exsolution in clinopyroxene porphyroclasts and extensive development of symplectites composed of secondary orthopyroxene + plagioclase (An85–72) + Al-spinel ± clinopyroxene ± ilmenite at the interface between garnet and primary clinopyroxene. Further decompression is recorded by the development of an olivine + plagioclase-bearing assemblage, locally under syn-kinematic conditions, at the expense of two-pyroxenes + Al-spinel. Mg-rich garnet has been also found in the websterite layers, which are commonly characterized by the occurrence of symplectites made of orthopyroxene + Al-spinel ± clinopyroxene. The enclosing peridotites are Ti-amphibole-bearing lherzolites with a fertile geochemical signature and a widespread plagioclase-facies mylonitic foliation, which preserve in places a spinel tectonite fabric. Lu–Hf and Sm–Nd mineral isochrons (220 ± 13 Ma and 186.0 ± 1·8 Ma, respectively) have been obtained from a garnet clinopyroxenite layer and interpreted as cooling ages. Geothermobarometric estimates for the high-pressure equilibration have yielded T 1100°C and P 2·8 GPa. The early decompression was associated with moderate cooling, corresponding to T 950°, and development of a spinel tectonite fabric in the lherzolites. Further decompression associated with plagioclase–olivine growth in both peridotites and pyroxenites was nearly isothermal. The shallow evolution occurred under a brittle regime and led to the superposition of hornblende to serpentine veining stages. The garnet pyroxenite-bearing mantle from the External Liguride ophiolites represents a rare tectonic sampling of deep levels of subcontinental lithosphere exhumed in an oceanic setting. The exhumation was probably accomplished through a two-step process that started during Late Palaeozoic continental extension. The low-pressure portion of the exhumation path, probably including also the plagioclase mylonitic shear zones, was related to the Mesozoic (Triassic to Jurassic) rifting that led to continental break-up. In Jurassic times, the studied mantle sequence became involved in an extensional detachment process that resulted in sea-floor denudation. KEY WORDS: garnet pyroxenite; ophiolite; non-volcanic margin; mantle exhumation; Sm–Nd and Lu–Hf geochronology
|
INTRODUCTION |
|---|
The architecture and tectonic evolution of non-volcanic continental margins have been extensively investigated in the last two decades in the western Iberian margin (e.g. Boillot et al., 1989
; Beslier et al., 1996; Whitmarsh et al., 2001). These studies have shown the existence of a characteristic zone of exhumed continental mantle, between continent and newly formed oceanic crust. The record of Ligurian Tethys ophiolites [here taking ‘ophiolite’ as a descriptive term to denote an association of peridotites with mid-ocean ridge (MOR)-type gabbros, basalt lavas and pelagic sediments] from the Alpine belt has provided fundamental clues about the nature of this continent–ocean transition (e.g. Bertotti et al., 1993; Manatschal & Bernoulli 1999; Desmurs et al., 2001; Manatschal et al., 2001; Müntener & Hermann, 2001).
Recent studies have proposed that extension leading to final continental break-up is accommodated through different fault systems (Whitmarsh et al., 2001; Manatschal, 2004) and showed that exhumation at the sea floor of the subcontinental mantle is a late process accomplished through relatively shallow downward concave faults. Petrological studies demonstrated the effects of the interaction between rising asthenospheric melts and lithospheric mantle rocks, and suggested that these processes played a significant role in modifying the lithospheric rheology, influencing the localization of continental rupture and the onset of sea-floor spreading (Müntener & Piccardo, 2003). However, geochronological constraints on the timing of mantle evolution during rifting are commonly lacking and little is known about the timing of high-temperature structures observed within the exhumed mantle rocks. The involvement of deep levels of subcontinental mantle during the rifting stages that precede the formation of an oceanic basin still remains an open issue.
The opening of the Ligurian Tethys in Middle Jurassic times led to the separation of the continental plates of Europe and Adria and gave rise to uplift and denudation of subcontinental lithospheric mantle, which is at present exposed in the Alpine chain. Some of these mantle bodies were affected by thermochemical erosion and refertilization processes related to asthenospheric upwelling during rifting and oceanization (Müntener & Piccardo, 2003), but rare occurrences of peridotites retaining their subcontinental origin are still preserved (e.g. Rampone et al., 1995; Müntener et al., 2004). New findings of garnet pyroxenite layers within ophiolitic mantle peridotites from the Northern Apennines provide an opportunity to throw light onto the evolution of subcontinental lithospheric mantle affected by extensional rift-related processes. This garnet pyroxenite-bearing mantle section is not contaminated by ascending asthenospheric melts and preserves relics of a deep-seated lithospheric history. Melting of garnet-bearing mafic rocks has been invoked to explain the ‘garnet signature’ of mid-ocean ridge basalt (MORB) (e.g. Hirschmann & Stolper, 1996) and of isotopically enriched MORB compositions (Salters & Dick, 2001). However, garnet pyroxenite layers are characteristic of mantle bodies exhumed in orogenic settings (e.g. Davies et al., 1993; Pearson et al., 1993). Garnet pyroxenites are absent in peridotite exposures on the present-day ocean floors and, to our knowledge, only two occurrences in ophiolitic mantle rocks have been reported (Peters, 1968; Müntener & Hermann, 1996; Blichert-Toft et al., 1999).
To unravel the exhumation history of the garnet pyroxenite-bearing mantle from the Northern Apennines, we combine field, petrological and geochronological investigations. In particular, Sm–Nd and Lu–Hf mineral isochrons for a garnet pyroxenite allow us to constrain the age of the mantle uplift. The determination of the pressure–temperature–time evolution of this mantle section suggests that only the low-pressure portion of the exhumation path, probably including also the development of plagioclase-facies shear zones, is related to rifting and continental break-up leading to the opening of the Ligurian Tethys. We argue that this has implications for the general understanding of the processes leading to the opening of oceanic basins along non-volcanic margins.
|
GEOLOGICAL AND PETROLOGICAL FRAMEWORK |
|---|
The Northern Apennine ophiolites are remnants of lithosphere from the Ligurian Tethys ocean, which opened in Middle Jurassic times (Lemoine et al., 1987
; Bill et al., 2001). They occur in two distinct ophiolite-bearing tectonic units, ascribed to different palaeogeographical domains (Fig. 1). The Internal Liguride (IL) ophiolites have lithostratigraphic characteristics that have been related to an intra-oceanic setting, i.e. as on-land analogues of oceanic lithosphere that originated at slow-spreading centres (Barrett & Spooner, 1977; Cortesogno et al., 1987; Tribuzio et al., 2000). In the External Liguride units (EL), ophiolites consist of fertile lherzolites, MOR-type basalts and rare gabbroic rocks occurring as large olistoliths within Cretaceous sedimentary mélanges, together with rare continental crust bodies locally displaying primary relationships with the ophiolites (Marroni et al., 1998). The gabbros have trace element and Nd–Sr isotope characteristics consistent with derivation from N-MORB magmas and their emplacement predated the oldest pelagic sediments of the Ligurian Tethys ophiolites (Tribuzio et al., 2004), thus testifying to the occurrence of a syn-rift MOR-type magmatic activity. The association of continental crust with the ophiolites was attributed to their formation in a fossil ocean–continent transition zone (Marroni et al., 1998), similar to that of modern non-volcanic continental margins. In particular, the External Liguride mantle rocks bear close petrological and geochemical similarities (Rampone & Piccardo, 2000; Tribuzio et al., 2004) to those from the western Iberia margin (e.g. Boillot et al., 1989; Whitmarsh et al., 2001).
|
Detailed petrological and geochemical studies were carried out on mantle peridotite slide-blocks located in the southern sector of the External Liguride units (namely Suvero, Mt. Nero and Mt. Ragola massifs, Fig. 1). These peridotites are mainly spinel lherzolites with protogranular to tectonite fabrics, commonly displaying extensive growth of plagioclase (Piccardo, 1976
; Beccaluva et al., 1984; Piccardo et al., 1990). These lherzolites contain disseminated Ti-rich amphibole and sporadic spinel pyroxenite layers. Temperature estimates for the spinel-facies equilibration (1000–1050°C) are compatible with continental geothermal gradients (Rampone et al., 1995). The relatively fertile nature of the southern External Liguride peridotites is shown by high amounts of clinopyroxene (10–15 vol. %), high whole-rock Al2O3 contents and flat to slightly light rare earth element (LREE)-depleted clinopyroxene REE patterns (Rampone et al., 1995). Nd–Sr–Os isotope investigations indicate a Proterozoic age for the accretion of the Suvero mantle body to the subcontinental lithosphere (see also Snow et al., 2000). In addition, recent studies have indicated that the southern External Liguride peridotites were affected by refertilization by asthenospheric melts during the Ligurian Tethys formation (Piccardo et al., 2004a, 2004b).
In the northern sector of the External Ligurides (Fig. 1), ophiolites are scarce and almost exclusively represented by the ultramafic rocks that form the basis of this study (Bernini et al., 1997; Marroni et al., 2002). These mantle rocks form several bodies embedded in a narrow, NW–SE-trending belt of sedimentary mélanges of Late Cretaceous age that crop out along the margin of the Northern Apennine chain close to the Po Plain. The enclosing sediments do not show any significant metamorphic recrystallization related to the Alpine orogenic cycle. The mélange also includes rare MOR-type basalts (Montanini & Tribuzio, 2003) and fragments of continental crust (mafic granulites, amphibolites, granitoids). The ultramafic bodies are of variable size, ranging from only some tens of metres to a few kilometres (Fig. 1).
|
FIELD RELATIONS |
|---|
The mantle peridotites commonly show a mylonitic fabric characterized by relatively large (up to 1 cm across) pyroxene porphyroclasts. A thickness of several hundred metres can be conservatively estimated for the studied mylonitic shear zones. The peridotites contain pyroxenite layers concordant with the foliation of the host peridotite, ranging in thickness from a few millimetres to
2 m. The garnet clinopyroxenite layers are, in the central part of the thickest layers, locally isotropic and coarse-grained. In such cases, the external portions of the garnet clinopyroxenite layers display a weak foliation defined by elongated clinopyroxene porphyroclasts. The isotropic rocks are composed of pinkish Al-augite and embayed red–orange garnets with greenish pyroxene–spinel coronas; clusters of tiny graphite flakes occur in places. Garnet-free layers with evident foliation are also present. These layers are fine-grained and dark grey, and show parallel millimetre-scale pyroxene-rich and olivine-rich bands, concordant with the main peridotite foliation.
The most common layers are represented by boudinaged pinkish grey medium- to fine-grained websterites, of 2–10 cm thickness. Websterites may contain thin stretched lenses of olivine. Interlayering of centimetre-scale websterite and olivine-rich bands is also common.
Brittle deformation with multiple stages of veining, hydration and the formation of tectonic breccias is widespread and shows similar patterns in the different mantle bodies. Both the earliest veining events (F1 and F2) yielded orthogonal fracture systems subparallel and normal to the mylonitic foliation. The oldest veins (F1) are up to 2–3 mm thick and filled with dark green amphibole ± chlorite. F2 veins have variable thickness (millimetre- to centimetre-sized) and are filled with white fibrous or massive dark green serpentine (± opaque minerals). Anastomosing F2 veins are also locally present. In strongly foliated serpentinitized peridotites, there are locally millimetre- to centimetre-thick F2 veins of light green massive serpentine or densely spaced subparallel millimetre-sized veinlets of fibrous white serpentine, both subparallel to the mylonitic foliation.
A brittle event (F3) pervasively overprinted the former vein systems, yielding cohesive tectonic breccias (‘jigsaw breccia’). These cataclasites consist of millimetre- to metre-sized angular clasts in a fine-grained, grey–bluish, serpentine matrix with little displacement and no significant rotation of the clasts, as attested to by the coherence of the mylonitic fabric and pyroxenite layer orientation between adjacent clasts. Carbonates are absent in the early vein generations, but they may occur as tiny spherules in the matrix of the jigsaw breccia. Multiple calcite vein generations, commonly associated with Fe-sulphides and massive green serpentine and/or talc, may crosscut the jigsaw breccia and bound the clasts, yielding ophicalcite-like rocks (e.g. Lemoine et al., 1987; Treves & Harper, 1994).
|
PETROGRAPHY AND MICROSTRUCTURES |
|---|
Garnet clinopyroxenites
The garnet clinopyroxenites are commonly affected by a complex retrograde evolution. In particular, retrogression can be divided into two distinct stages characterized, respectively, by two-pyroxene + spinel + plagioclase and olivine + plagioclase assemblages.
The high-pressure protoliths were characterized by a coarse-grained isotropic texture and an anhydrous bimineralic assemblage of garnet + clinopyroxene (Cpx I). The oldest clinopyroxene generation (Cpx I) is preserved only as inclusions in garnet. Accessory graphite, Fe–Ni–Cu sulphides and rutile have been found locally. Graphite occurs as small flakes and stacks of flakes with grain size up to 2–3 mm. No diamond pseudomorphs have been observed. Raman spectra (Montanini et al., 2005) indicate a highly ordered structure similar to that of graphite crystallized at high temperature in peridotite and eclogite xenoliths (Pearson et al., 1993). Sulphides may occur both as tiny (20–50 µm) inclusions in garnet or, locally, as interstitial, lobate larger crystals (0·1–1·0 mm).
Complex pyroxene–spinel–plagioclase symplectites are present at the interface between garnet and Cpx I (Fig. 2a). Fine-grained (µm-scale) radial and fibrous symplectites composed of orthopyroxene (Opx I) + green spinel (Spl I) + plagioclase ± ilmenite (Fig. 3a) occur at the rims of the garnet relics. This inner symplectite is commonly rimmed by coarser symplectitic intergrowths of the same minerals and, locally, by an outer shell of plagioclase + clinopyroxene (Cpx IIa) with vermicular spinel inclusions.
|
|
Coarse clinopyroxene grains (up to 4–5 mm; Cpx IIb) display plagioclase exsolution lamellae and occur as porphyroclasts in medium-grained granoblastic domains composed of clinopyroxene (Cpx III) + orthopyroxene (Opx II) + minor plagioclase and green spinel (Spl II; Fig. 2b). Several garnet clinopyroxenite layers are almost completely retrogressed to foliated rocks characterized by this latter assemblage, which locally preserves pyroxene–spinel symplectitic domains after garnet and clinopyroxene IIb porphyroclasts showing breakdown to secondary clinopyroxene + plagioclase. In these rocks, small (1–2 mm) clinopyroxene and orthopyroxene grains (corresponding to Cpx III and Opx II) are rimmed by fine-grained neoblastic Cpx (IV) + Opx (III) + plagioclase + spinel (Spl III). The garnet textural domain is recognized by orthopyroxene including vermicular grains of spinel (locally with a thin rind of secondary plagioclase) and minor clinopyroxene (Fig. 3b).
An olivine + plagioclase (± Spl IV) assemblage occurs as fine-grained aggregates replacing either the pyroxene–plagioclase–spinel symplectites after garnet, or the garnet relics within the symplectites. The association of olivine + plagioclase is also found as coronas and/or symplectites (Fig. 2c) around pyroxenes at the contact with spinel. In a few layers, the olivine + plagioclase assemblage syn-kinematically overgrows pyroxene + spinel (II) + plagioclase domains, locally associated with trace amounts of Spl (IV) and brown amphibole.
The garnet clinopyroxenites do not record an extensive low-temperature hydration process. Veining and replacement of the pyroxene + spinel + plagioclase assemblage by pale green amphibole + chlorite related to the F1 stage is locally observed. Serpentine F2 veins and serpentine ± chlorite pseudomorphs after olivine occur in places.
Websterites
The websterites are medium-grained rocks consisting of clinopyroxene (40–60%), orthopyroxene (20–30%), spinel (5–10%), plagioclase (0–10%), olivine (0–5%) and brown amphibole (0–3%); Fe–Ni sulphides are common accessory minerals. The most common texture is protomylonitic, characterized by highly stretched, elongated orthopyroxene crystals with high aspect ratios (up to 15; Fig. 2f) and fine-grained neoblastic aggregates of orthopyroxene + clinopyroxene + spinel ± amphibole around pyroxene porphyroclasts. Evidence for a previous garnet-bearing assemblage comes from rounded domains composed of Opx + Cpx + Spl symplectites (Fig. 2d). In one case, we have found tiny garnet relics within these symplectites. Coarsening and progressive annealing of the symplectites after garnet led to the formation of pyroxene–spinel clusters and, locally, to aggregates with allotriomorphic texture (Fig. 2e and f).
The pyroxene + spinel assemblage is locally mantled by fine-grained olivine–plagioclase-rich mylonitic bands (Fig. 2e and f), similar to those occurring in the peridotite (see the next section). Strain localization associated with this deformation is commonly observed at the contact with peridotite. The clinopyroxene porphyroclasts frequently show plagioclase exsolution and rims. Green to greenish brown spinel occurs both as small rounded or vermicular grains within orthopyroxene and as larger allotriomorphic crystals, commonly rimmed by plagioclase. Olivine of probably primary origin occurs as irregularly shaped crystals or thin seams surrounded by pyroxenes similar to the ‘olivine flames’ of Burg et al. (1998). Secondary olivine forms: (1) radial crystals, associated with plagioclase + greenish brown spinel ± ilmenite ± brown amphibole, forming coronas between pyroxenes and green spinel, locally stretched along the main foliation; (2) granoblastic domains composed by the same phases; (3) fine-grained plagioclase-bearing (± brown amphibole) mylonitic bands.
Peridotites
The peridotites are clinopyroxene-rich (10–15 vol. %) spinel–plagioclase lherzolites with accessory brown amphibole. The oldest recognizable texture is preserved in small domains characterized by coarse-grained (up to 5–10 mm across) orthopyroxene and clinopyroxene grains surrounded by smaller polygonal or irregular-shaped neoblasts of olivine + pyroxene + spinel ± brown amphibole (Fig. 2g). These domains thus represent relics of a spinel-facies low-strain tectonite. Mutual pyroxene exsolution may occur in these porphyroclasts.
The spinel-facies assemblage is overprinted by plagioclase-facies recrystallization associated with the development of a mylonitic fabric. The mylonite microstructure is characterized by aligned porphyroclasts of pyroxene + brown spinel + brown amphibole and by an ultrafine-grained (20–50 µm) polyphase aggregate composed of olivine + pyroxenes + plagioclase (± spinel ± accessory brown amphibole) occurring as millimetre-sized bands, lenses and porphyroclast tails (Fig. 2h). Crystals in this polyphase matrix commonly display grain boundary alignment parallel to the mylonitic foliation.
Orthopyroxene porphyroclasts may be rounded, lozenge-shaped or elongate along the mylonitic foliation with high aspect ratio (up to 10). Clinopyroxene is usually more equant than orthopyroxene. Both pyroxenes commonly show evidence for intracrystalline deformation (i.e. undulose extinction, bending and kinking). Plagioclase occurs as (1) thin rims between spinel and pyroxene porphyroclasts, (2) tiny neoblastic grains in the mylonitic matrix and, rarely, (3) exsolution in clinopyroxene porphyroclasts.
The mylonitic foliation is locally crosscut by thin fractures filled with prismatic pale green amphibole, which is commonly associated with minor amounts of chlorite (F1). Amphibole and chlorite also grow adjacent to the fractures, as pseudomorphs after clinopyroxene and brown amphibole, and as coronas around spinel, respectively (Fig. 4a).
|
The association of amphibole + chlorite is followed by multiple serpentine veining episodes (F2), associated with serpentinization, to various extents, of the host rock. X-ray diffraction analyses have shown that lizardite is the dominant phase in the serpentinized peridotite. Vein serpentine may be fibrous with antiaxial filling or isotropic, rarely associated with magnetite. Olivine alteration is mainly accomplished through growth of ‘mesh texture’ serpentine (with frequent preservation of olivine cores) and, to a lesser extent, of ‘hourglass’ serpentine. Magnetite abundance in olivine pseudomorphs usually increases with increasing degree of serpentinization. Pyroxenes may be pseudomorphed by light green or bluish fibrous serpentine (‘bastite’). Relics of the former mantle assemblages, mainly represented by sparse brown spinel grains with opaque rims, are only rarely preserved in the serpentinized plagioclase mylonites. Two main types of subparallel veins follow the original plagioclase-bearing foliation: (1) thin ribbon lizardite veins (± magnetite seams) transected by (2) millimetre- to centimetre-thick massive lizardite veins with low birefringence and undulose extinction.
All the structures described above are pervasively disrupted by the F3 stage of brecciation. These breccias show a jigsaw-puzzle pattern at both meso- and micro-scales, and show evidence for polyphase cracking and veining. The breccia matrix (Fig. 4b) is made of fine-grained serpentinite derived from comminution of the larger clasts and by yellow–light brown serpentine in tiny fibrous–spherulitic aggregates. The serpentine matrix also contains clusters of micrometre-scale, curved needles of Ni-sulphide (millerite), and locally abundant calcite spherules, studded with dark fluid inclusions, some tenths of a micrometre across.
|
ANALYTICAL METHODS |
|---|
Major element mineral analyses were obtained on samples representative of the different lithologies and different stages of retrograde evolution. Mineral analyses were performed at the Dipartimento di Scienze della Terra of Università di Parma using a JEOL-6400 electron microprobe equipped with LINK-ISIS energy-dispersive microanalytic system. Operating conditions were accelerating voltage of 15 kV and probe current of 0·25 nA; both natural minerals and synthetic compounds were used as standards. Trace amounts of Na2O in garnet were analysed at CNR–Istituto di Geoscienze e Georisorse, Sezione di Firenze, using a JEOL JXA-8600 microprobe. Operating conditions were accelerating voltage of 15 kV, probe current of 100 nA and counting times of 20 s. These conditions provided a detection limit of
40 ppm for Na.
Sm–Nd and Lu–Hf dating techniques were applied to a garnet clinopyroxenite layer (sample AM322). The rock was crushed and sieved to obtain a 150–250 µm fraction. Mineral concentrates were extracted using a Frantz magnetic separator and then purified by careful handpicking under a binocular microscope to obtain clinopyroxene, garnet and plagioclase separates, avoiding inclusion-bearing grains or small impurities derived from symplectitic intergrowths. Sm–Nd and Lu–Hf analyses of the garnet clinopyroxenite layer were carried out on the same mineral and whole-rock splits. Details of sample digestion and ion exchange chromatography were presented by Anczkiewicz et al. (2004). Mass spectrometry procedures follow Thirlwall & Anczkiewicz (2004). Total procedure analytical blanks for Hf and Nd were <30 pg. All elements were analysed in a static, hard extraction mode using the Royal Holloway IsoProbe®. All errors are 2SE and relate to the last significant digits, except for age errors, which are 95% confidence level. All measurements were conducted on a single day to minimize correction for secular variation in the static 176Hf/177Hf of JMC47. 147Sm/144Nd errors are 0·3%. Reproducibility of the Aldrich Nd standard on the day of analyses was 143Nd/144Nd = 0·511388 ± 11 (2SD, n = 4).Daily variations in 143Nd/144Nd ratios were normalized to 143Nd/144Nd = 0·511421. 176Lu/177Hf errors are 0·5%; the JMC475 standard on the day of analysis yielded 0·282173 ± 6 (2SD, n = 3). Daily variations in 176Hf/177Hf ratios were normalized to 176Hf/177Hf = 0·282165. Mass bias corrections were made using 146Nd/144Nd = 0·7219 and 179Hf/177Hf = 0·7325.
|
MINERAL CHEMISTRY |
|---|
Garnet pyroxenites
Garnet compositions cover the range Prp53–45Alm29–38Grs11–20, with only minor spessartine component (
1·5 mol %, Table 1, Fig. 5). Cr2O3 contents do not exceed 0·25 wt %. The low TiO2 (0·10–0·20 wt %) and Na2O concentrations (0·02 wt %, confirmed by unpublished LA-ICP-MS analyses that yielded Na 120 ppm) are comparable with those of the diamond-free Group II eclogites defined by McCandless & Gurney (1989). The bulk composition of the symplectitic intergrowths in retrogressed pyroxenites was determined by electron microprobe and is a pyrope-rich garnet composition (Table 1). However, the analyses contain an excess of MgO relative to garnet stoichiometry and high amounts of Na2O, which may be related to involvement of olivine and primary clinopyroxene in the garnet breakdown reaction, respectively.
|
|
The oldest clinopyroxene (Cpx I) has up to 12·6 wt % Al2O3 and 2·7 wt % Na2O. Recalculation of structural formulae gives high proportions of the Ca-Tschermak component (15–20 mol %) and moderate jadeite contents (10–16 mol %). The structural formulae do not display any cation deficiency requiring the presence of a Ca-Eskola component. Similar clinopyroxene compositions were reported for Group II garnet pyroxenites from Beni Bousera (Kornprobst et al., 1990
). The different generations of retrograde clinopyroxene have lower amounts of the Ca-Tschermak and jadeitic component than Cpx I and define a negative correlation in a plot of Na2O–Al2O3 (Fig. 6a). The latest pyroxenes (Cpx IV) have the lowest amounts of the Ca-Tschermak and jadeitic component. The composition of the coarse clinopyroxene porphyroclasts characterized by plagioclase exsolution (Cpx IIb) is highly variable, both for Al2O3 and Na2O contents. In particular, Na2O concentrations vary over a wide range (1·3–0·1 wt %), even within the same crystal, with the lowest values occurring close to the plagioclase lamellae and patches.
|
The Mg/(Mg + Fe2+) ratios of clinopyroxene and orthopyroxene from the garnet clinopyroxenites span a wide range (0·73–0·90 and 0·71–0·88, respectively). Cr2O3 varies in the range 0·05–0·30 wt % in both pyroxenes. The lowest Mg/(Mg + Fe2+) values are associated with the lowest Cr contents in both pyroxenes. The forsterite content of olivine (61–85 mol %) exhibits a rough positive correlation with the Mg/(Mg + Fe2+) values of the pyroxenes.
Spinel does not show appreciable compositional variation between the different textural domains. It is highly aluminous and Cr-poor (Cr2O3 1·5 wt %); the Mg/(Mg + Fe2+) values vary between 50 and 75 (Table 1). The brown amphibole occurring in the olivine–plagioclase-bearing domains is kaersutite (Table 1).
The composition of plagioclase (Table 1) is controlled by textural position. Plagioclase from the Opx–Spl–Pl symplectites around garnet (Pl Ia) has high Ca contents (An = 68–85 mol %). Plagioclase lamellae within clinopyroxene (Pl Ib) and plagioclase grains in recrystallized two-pyroxene domains (Pl II) have 45–50 mol % anorthite.
Vein amphibole and amphibole overgrowths on clinopyroxene in the pyroxenitic layers is Mg-hornblende with high Al2O3 (up to 10·6 wt %) and Na2O (1·6–2·2 wt %) contents, and low TiO2 concentrations (0·13–0·25 wt %); significant amounts of Cl (0·15 wt %; see Table 4) were detected. The associated chlorite is clinochlore, with Al2O3 and Mg/(Mg + Fe2+) values of about 19 wt % and 0·88, respectively.
Websterites
Relict garnet has high pyrope (78 mol %, Fig. 5) and Cr2O3 contents (0·7 wt %), similar to the garnets from mantle lherzolites worldwide (Pearson & Nixon, 1996). Clinopyroxene porphyroclasts can have high amounts of Al2O3 (10–11 wt %) and Ca-Tschermak substitution (up to 18 mol %), whereas the jadeite component is low (6 mol %), probably as the result of exsolution of Na-rich plagioclase (An = 58 mol %; Fig. 6b). These clinopyroxenes are compositionally similar to Cpx IIb from the garnet clinopyroxenite, although they have higher Mg-number [Mg/(Mg + Fe2+)] and Cr2O3 concentrations (Table 2). Neoblastic clinopyroxenes, especially those of the fine-grained plagioclase-rich mylonitic bands, have low Al2O3 and Na2O (Table 2, Fig. 6b). The Mg-numbers of both clinopyroxenes and orthopyroxenes (0·82–0·91 and 0·85–0·89, respectively) partially overlap those of the peridotites. Cr2O3 contents of clinopyroxenes range from 0·20 to 0·60 wt %. Amphibole is kaersutite with low K2O (0·3 wt %) and moderate Cr2O3 (1·0–1·3 wt %). Spinel has Cr-number [Cr/(Cr + Al)] and Mg/(Mg + Fe2+) of 0·02–12 and 0·72–0·78, respectively. The green spinel in the symplectites and in the two-pyroxene assemblage has the lowest Cr-number. The highest Cr-numbers are observed in the greenish brown spinel from the olivine-bearing retrograde domains (Fig. 7). Primary and secondary olivines have high forsterite contents (Fo = 89–90 mol %).
|
|
Peridotites
The peridotitic olivine is Fo89. Porphyroclastic spinel-facies clinopyroxene has high Al2O3, Na2O and TiO2 contents (6·7–8·6 wt %, 1·2–2·2 wt % and 0·6–1·0 wt %, respectively; see also Fig. 8a). Its Mg-number ranges from 0·89 to 0·92 and the Cr2O3 concentrations are 0·5–1·1 wt % (Fig. 8a, Table 3). The amounts of Al2O3 and Na2O in porphyroclastic clinopyroxene decrease moderately from core to rim. The lowest Al2O3 and Na2O contents (1·4–5·0 wt % and
0·6 wt %, respectively) are shown by neoblastic plagioclase-facies clinopyroxene. Cr2O3 shows no significant variations between porphyroclastic and neoblastic clinopyroxenes (Fig. 8b).
|
|
Orthopyroxenes have Mg-numbers of 0·89–0·91. The cores of orthopyroxene porphyroclasts (Table 3) are Al2O3-rich (4·7–5·7 wt %) and contain variable amounts of CaO (0·6–1·4 wt %). The CaO and Al2O3 contents are lower in porphyroclast rims (0·5–0·7 wt % and 3·5–4·4 wt %, respectively). Orthopyroxene neoblasts have the lowest concentrations of Al2O3 and CaO (1·3–2·7 wt % and
0·5 wt %, respectively). As a whole, the orthopyroxene analyses show a rough negative correlation between Al2O3 and CaO (Fig. 9).
|
Spinel shows (Fig. 7) variable Mg-number and Cr-number. This variability has been also found in the same sample and is probably related to the recrystallization under plagioclase-facies conditions. The lowest Cr-numbers are found in the cores of the rare porphyroclasts not rimmed by plagioclase, whereas increasing values occur towards the plagioclase coronas or in small relict crystals in plagioclase-bearing domains. TiO2 contents in spinel are invariably lower than 0·25 wt %.
The brown amphibole porphyroclasts belonging to the spinel-facies assemblage range in composition from kaersutite to Ti-pargasite, according to the nomenclature of Leake et al. (1997). They have variable TiO2 contents (3·4–5·3 wt %), even at the thin-section scale (Table 3, Fig. 10), and low to moderate K2O concentrations (0·05–0·30 wt %).
|
Plagioclase is invariably anorthite-poor. The highest anorthite contents (48–56 mol %) were found in plagioclase coronas around spinel. The neoblastic grains in the mylonitic matrix have about 45 mol % anorthite and the plagioclase exsolution in the clinopyroxene porphyroclasts is albite-rich (An35).
Pale green amphibole from the veins is Mg-hornblende, which spans a wide range of Al2O3 (4·3–10·4 wt %) and Na2O (0·9–2·1 wt %) contents (Fig. 10). Vein amphibole has TiO2 concentrations between 0·14 and 0·75 wt %, and small amounts of Cl (0·18 wt %). Small prismatic tremolite crystals occur locally in the veins, probably as a later Mg-hornblende replacement (Table 4, Fig. 10). The colourless chlorite forming coronas around spinel (Table 4) is penninite.
|
The composition of serpentine (Table 4) varies in relation to the microstructural site of nucleation. Serpentine replacing olivine has low Al2O3 and Cr2O3 (<0·1 wt %), and relatively high NiO (0·15–0·38 wt %). On the other hand, fibrous serpentine after orthopyroxene has high Al2O3 (3·8–5·0 wt %) and Cr2O3 (0·17–0·44 wt %) contents, similar to those observed in the primary mineral, thus suggesting a relative immobility of these elements during serpentinization (Bonatti & Hamlyn, 1981
; Agrinier et al., 1988; Mével, 2003). FeO contents are variable in both types of serpentine pseudomorphs (4·9–7·8 wt %). The different veins belonging to the F2 stage have low Al2O3 (1·2–2·8 wt %) and highly variable FeO contents (1·9–8·8 wt %). The lowest FeO values were observed for the serpentine with undulose extinction, which is probably older than the Fe-rich veins. Serpentine from the matrix of the F3 jigsaw breccia has low to moderate Al2O3 (1·7–3·8 wt %), coupled with relatively high FeO. Non-pseudomorphic serpentine from both F2 and F3 stages commonly displays negligible Cr and Ni. Chlorine contents of the different serpentine types vary over a relatively wide range from 0·05 to 0·31 wt %, without a systematic relation with the nucleation site. |
GEOTHERMOBAROMETRY |
|---|
Conventional thermobarometry methods have been applied to minerals occurring in domains formed at different stages and distinguished on the basis of textural relations. Different cation exchange thermometers have been applied to both pyroxenites and enclosing peridotitic rocks to assess their thermal evolution, whereas pressure estimates have been obtained only for the garnet-bearing rocks (Table 5). To avoid the uncertainties related to Fe3+ recalculations from microprobe analyses, all iron in pyroxenes and garnet is assumed to be Fe2+. The computed P–T estimates should, therefore, be considered as maximum values.
|
Pyroxenites
The occurrence of clinopyroxene inclusions (Cpx I) within garnets allows us to calculate the equilibration pressure and temperature of the early garnet clinopyroxenite assemblage. We have used the geobarometric method proposed by Simakov & Taylor (2000)
, based on the solubility of the Ca-Tschermaks component in clinopyroxene coexisting with garnet, which, recently, has been successfully applied to both diamond-bearing and diamond-free eclogite xenoliths (Peltonen et al., 2002; Simakov, 2006). Temperature calculations are based on the application of the garnet–clinopyroxene Fe–Mg exchange reaction, according to the calibrations of Ellis & Green (1979), Krogh (1988), Ai (1994) and Krogh Ravna (2000). Pressures and temperatures were calculated iteratively using the clinopyroxene inclusions characterized by the highest Na2O and Al2O3 contents (sample AM322, Table 5). Using the most recent calibration of the garnet–clinopyroxene geothermometer (Krogh Ravna, 2000), we obtained T 1100°C and P 2·8 GPa. The calibrations of Ellis & Green (1979), Krogh (1988) and Ai (1994) yield slightly higher temperature (and pressure) estimates. The high Al2O3 contents of Cpx I are consistent with the geothermobarometric evaluations. Experimental work has shown that clinopyroxenes with high proportions of Ca-Tschermaks molecule form at high-pressure conditions (2·5–3·5 GPa) as segregations from basic melts (Thompson, 1974; Eggins, 1992; Johnson, 1998) and as residual phases after anhydrous partial melting of mafic rocks (Pertermann & Hirschmann, 2003; Yaxley & Brey, 2003). Jadeite contents in these magmatic or residual clinopyroxenes are typically low and comparable with those observed in the present study.
Pyroxene thermometry has been applied to the retrograde pyroxene–spinel-bearing assemblage of the garnet clinopyroxenites and websterites. The two-pyroxene geothermometer of Wells (1977) applied to small, undeformed clinopyroxene-orthopyroxene pairs (Cpx III–Opx II) of the garnet clinopyroxenites yields mean temperatures of 960°C (Table 5). In addition, the neoblastic Cpx IV–Opx III pairs from a garnet clinopyroxenite layer characterized by dynamic recrystallization of pyroxenes + spinel gave T = 920°C. Similar temperature estimates were obtained for the clinopyroxene–orthopyroxene porphyroclasts and neoblasts from the websterites. We have also applied the Ca-in-Opx geothermometer of Brey & Kohler (1990) to the olivine-bearing websterites, which has yielded temperature estimates of 950°C for the orthopyroxene porphyroclasts.
The coexistence of Ti-rich amphibole and plagioclase in the olivine-bearing domains of the retrogressed clinopyroxenites and websterites permits temperature to be estimated on the basis of the Holland & Blundy (1994) method. Temperature values of 960°C have been obtained for both rock types.
Peridotites
Temperatures of the spinel- and plagioclase-facies equilibration in the peridotites have been estimated (Table 5) applying the Ca-in-Opx geothermometer (Brey & Kohler, 1990) and the Ca–Mg exchange geothermometers of Wells (1977) and Brey & Kohler (1990). The cores of spinel-facies orthopyroxene porphyroclasts yield TCa-in-Opx estimates mainly ranging between 1000 and 1100°C, assuming P = 1·5 GPa. Core to rim compositional variations in the orthopyroxene porphyroclasts reflect cooling of 100–150°C. Two-pyroxene porphyroclast temperatures vary over a wide range, shifted to lower values (870–1040°C) than TCa-in-Opx estimates based on the core compositions of orthopyroxene porphyroclasts. The lower estimates resulting from two-pyroxene geothermometers may be a consequence of pyroxene exsolution during slow cooling. The relatively low diffusivity of Ca in orthopyroxene compared with the Ca–Mg diffusivity in clinopyroxene (see Lenoir et al., 2001) may have hampered complete re-equilibration of large orthopyroxene porphyroclasts during cooling. Temperature estimates for the plagioclase-facies equilibration range from 860 to 1030°C, on the basis of Cpx–Opx pairs.
Pressure evolution of the peridotites is difficult to constrain owing to the lack of appropriate geobarometers in garnet-absent systems. The pressure of the garnet–spinel transition in peridotitic systems is mainly controlled by the Cr content of the spinel (O'Neill, 1981; Webb & Wood, 1986; Klemme, 2004). In fertile compositions, characterized by spinel with Cr/(Cr + Al) <0·2, garnet is stabilized towards lower pressure values. A maximum pressure estimate for the spinel-facies equilibration can be determined on the basis of the formulation of Webb & Wood (1986). For this purpose we used the composition of spinel porphyroclast cores not rimmed by plagioclase and obtained pressure values of around 1·8 GPa. This pressure value is consistent with the P–X phase relations recently considered by Klemme (2004) for a simplified system MgO–Cr2O3–Al2O3–SiO2.
|
Sm–Nd AND Lu–Hf GEOCHRONOLOGY |
|---|
The summary of the isotopic results is presented in Table 6 and Fig. 11. Age calculations were performed using Isoplot® (Ludwig, 2003
); age errors are given at the 95% confidence level.
|
|
A Sm–Nd isochron age is defined by whole-rock, plagioclase, clinopyroxene and two garnet fractions. The small scatter (MSWD = 0·26) together with the high 147Sm/144Nd values measured in garnets results in a well-constrained age of 186·0 ± 1·8 Ma (initial
Nd = +6·5). Remarkably, both garnet fractions resulted in identical (within error) isotopic ratios. The unusually high 147Sm/144Nd ratio of plagioclase is probably due to its metamorphic origin as a product of garnet + Cpx I breakdown, in a system characterized by an extreme LREE depletion (see also Tribuzio et al., 2005). Although not all analysed minerals grew in equilibrium (e.g. garnet and plagioclase), the geothermometric investigations (Table 5) show that all analysed minerals crystallized significantly above 750–800°C, the closure temperature for the Sm–Nd system in garnet (Ganguly et al., 1988; Van Orman et al., 2002). Hence, the Sm–Nd isochron is interpreted as a cooling age, i.e. the garnet pyroxenite probably passed through the Sm–Nd blocking temperature at 186·0 ± 1·8 Ma.
Lu–Hf dating of the same whole-rock and mineral fractions (excluding plagioclase) shows only 0·06 spread in 176Lu/177Hf ratios, which does not give an age with high precision. The regression line with MSWD = 0·23 defines an age of 220 ± 13 Ma (initial Hf = +4·5). Interpretation of the Lu–Hf age is more problematic, as little is known about the closure temperature for this system. It has been proposed that the closure temperature for Lu–Hf in garnet is equal to, or slightly higher (100°C) than for Sm–Nd (Scherer et al., 2001). Although few coupled Lu–Hf and Sm–Nd age determinations have been carried out on mantle garnet-bearing rocks, the available data corroborate this interpretation. Lu–Hf cooling ages older than Sm–Nd ages have been reported for garnet pyroxenite layers from Beni-Bousera (Blichert-Toft et al., 1999; Pearson & Nowell, 2004) and for a garnet pyroxenite sample in a Mirdita (Albania) ophiolite (Blichert-Toft et al., 1999), in agreement with this notion. In particular, the large difference between the Lu–Hf and Sm–Nd mineral isochron ages (203 ± 3·4 Ma and 166 ± 2 Ma, respectively) obtained for the Mirdita ophiolite sample is similar to that found for the External Liguride pyroxenite in this study. In addition, recent work on eclogite xenoliths in kimberlites (Bedini et al., 2004) yielded Lu–Hf clinopyroxene–garnet ages generally older than the corresponding Sm–Nd ages. This suggests that radiogenic Hf diffuses out of garnet more slowly than radiogenic Nd, giving fur