Journal of Petrology Advance Access originally published online on April 15, 2005
Journal of Petrology 2005 46(8):1661-1687; doi:10.1093/petrology/egi030
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Near-Ultrahigh Pressure Processing of Continental Crust: Miocene Crustal Xenoliths from the Pamir
1 DEPARTMENT OF GEOLOGICAL SCIENCES, UNIVERSITY OF CALIFORNIA, SANTA BARBARA, CA 93106-9630, USA
2 DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF BUCHAREST, BUCHAREST, 70139, ROMANIA
3 GEOLOGICAL INSTITUTE OF THE TAJIK ACADEMY OF SCIENCE, 734063, DUSHANBE, TAJIKISTAN
4 INSTITUT FÜR GEOWISSENSCHAFTEN, TECHNISCHE UNIVERSITÄT BERGAKADEMIE FREIBERG, 09599 FREIBERG, GERMANY
5 DEPARTMENT OF EARTH SCIENCES, 685 COMMONWEALTH AVENUE, BOSTON UNIVERSITY, BOSTON, MA 02215, USA
6 DEPARTMENT OF GEOSCIENCES, UNIVERSITY OF ARIZONA, TUCSON, AZ 85721, USA
7 DEPARTMENT OF GEOLOGY, UNIVERSITY OF GEORGIA, ATHENS, GA 30602, USA
8 GEOLOGICAL AND ENVIRONMENTAL SCIENCES, STANFORD UNIVERSITY, STANFORD, CA 94305-2115, USA
RECEIVED JUNE 24, 2004; ACCEPTED FEBRUARY 16, 2005
 |
ABSTRACT |
|---|
TOP
ABSTRACT
INTRODUCTIONXenoliths of subducted crustal origin hosted by Miocene ultrapotassic igneous rocks in the southern Pamir provide important new information regarding the geological processes accompanying tectonism during the Indo-Eurasian collision. Four types have been studied: sanidine eclogites (omphacite, garnet, sanidine, quartz, biotite, kyanite), felsic granulites (garnet, quartz, sanidine and kyanite), basaltic eclogites (omphacite and garnet), and a glimmerite (biotite, clinopyroxene and sanidine). Apatite, rutile and carbonate are the most abundant minor phases. Hydrous phases (biotite and phengite in felsic granulites and basaltic eclogites, amphiboles in mafic and sanidine eclogites) and plagioclase form minor inclusions in garnet or kyanite. Solid-phase thermobarometry reveals recrystallization at mainly ultrahigh temperatures of 1000–1100°C and near-ultrahigh pressures of 2·5–2·8 GPa. Textures, parageneses and mineral compositions suggest derivation of the xenoliths from subducted basaltic, tonalitic and pelitic crust that experienced high-pressure dehydration melting, K-rich metasomatism, and solid-state re-equilibration. The timing of these processes is constrained by zircon ages from the xenoliths and 40Ar/39Ar ages of the host volcanic rocks to 57–11 Ma. These xenoliths reveal that deeply subducted crust may undergo extensive dehydration-driven partial melting, density-driven differentiation and disaggregation, and sequestration within the mantle. These processes may also contribute to the alkaline volcanism observed in continent-collision zones.
KEY WORDS: xenolith; high-pressure; subduction; Pamir; Tibet
|
INTRODUCTION |
|---|
Exposures of ultrahigh-pressure metamorphic rocks tell us much about the subduction or foundering and exhumation of continental margins or fragments. These ultrahigh-pressure terranes always undergo significant recrystallization and overprinting during their exhumation, however, such that details about the chemical and physical processing of subducted crustal material at high pressures and high temperatures are invariably erased. Only xenoliths erupted from ultrahigh-pressure settings can provide this important information, but such rocks are exceptionally rare. We report here on the petrology, physical properties, and geochronology of a Miocene xenolith suite from the Pamir that has captured near-ultrahigh pressure metamorphism in the act. The xenoliths yield insight into the geodynamic effects of the melting, metasomatism, and melt injection that takes place during profound subduction or foundering of continental crust. They also provide information on the deep crust beneath the southern Pamir and Tibet and, together with geological data for Tibetan xenoliths (Hacker et al., 2000
), suggest a similar first-order crustal structure for the Pamir and Tibet. |
DUNKELDIK ULTRAPOTASSIC SUITE |
|---|
The Dunkeldik magmatic field is part of a poorly defined Cenozoic metamorphic and magmatic belt that stretches from the Chinese Pamir through the southern and central Pamir of Tajikistan (Fig. 1a) (Vlasov et al., 1991
; Hubbard et al., 1999; Schwab et al., 2004). It contains basalts, diorites, granodiorites, and leucogranites with Eocene to late Miocene crystallization ages (Ratschbacher et al., in preparation). The Dunkeldik field (Fig. 1b and c) lies between the Late Triassic–Early Jurassic, south-facing Jinsha–Tanymas suture and the Late Jurassic–Early Cretaceous, north-facing Rushan–Pshart–Bangong–Nujiang suture (Schwab et al., 2004) in an area transected by partly active faults related to the Karakoram fault zone (Strecker et al., 1995). In the Dunkeldik valley, a series of 
11 Ma (see below), alkaline volcanic–hypabyssal complexes erupted deep crustal and mantle xenoliths (Dmitriev, 1976). The xenolith-bearing igneous suite includes ultrapotassic (K2O/Na2O = 4·7; 7·4 wt % K2O; n = 9) dikes and subvolcanic bodies that range in composition from alkali basalt to trachyte, syenite, and carbonatite (Dmitriev, 1976); xenoliths are found only in the alkali basalts. Abundances of the various xenolith types (n > 300) are as follows: 38% eclogite and biotite eclogite, 19% phlogopite pyroxenite and glimmerite [in the sense of Gregoire et al. (2002)], 15% biotite–garnet clinopyroxenite, 8% biotite–garnet granulite, 6% kyanite–garnet granulite, 4% mafic garnet granulite, 3% phlogopite–garnet websterite, and 7% other rocks (Lutkov, 2003).
|
|
ANALYTICAL TECHNIQUES |
|---|
Textural analysis was accomplished through optical microscopy and back-scattered electron microscopy. Mineral modes were determined by point counting with an optical microscope (Table 1). Mineral compositions were determined using a University of California, Santa Barbara, five-spectrometer Cameca SX-50 electron microprobe operated at 15 kV and 15 nA, and a University of Arizona four-spectrometer Cameca SX-50 operated at 15 kV and 20 nA, both calibrated with natural and synthetic mineral standards (Table 2). Table 3 gives bulk-rock compositions calculated from mineral modes and mineral compositions. Major- and trace-element compositions were measured at Boston University using the remains of thin-section billets. This is not ideal because the samples are contaminated with epoxy from thin-section impregnation, and some include bits of host lava. Billets were first abraded with silicon carbide to remove as much as possible of the epoxy and weathered surfaces, and then powdered in alumina. Major elements were measured by inductively coupled emission spectrometry (ICP-ES), following LiBO2 fusions, and by ICP mass spectrometry (ICP-MS), following HF–HNO3 digestions [Table 3; methods described by Kelley et al. (2003)
]. Elements measured by both techniques (e.g. Ba) indicate excellent agreement, within 2% relative. Differences among other elements (e.g. Ti and Zr) indicate the presence of rutile and zircon, respectively, which were not dissolved in the ICP-MS preparation. For this reason, Ti, Nb, Ta, Zr, and Hf determined by ICP-MS are not reported. Agreement in Y measured by both techniques demonstrates that the undissolved zircon had <5% effect on the heavy rare earth element (HREE) concentrations determined by ICP-MS. 40Ar/39Ar ages of biotite from two xenoliths were measured at Stanford University using analytical techniques detailed by Hacker et al. (1996).
|
|
|
|
XENOLITH DESCRIPTIONS |
|---|
The xenoliths are as large as 20–30 cm in diameter (granulites reach 50 cm), foliated, and contain millimeter-size minerals; some are composites composed of more than one rock type (Table 1; Fig. 2). Five types of xenolith were studied in detail: sanidine eclogites, felsic granulites, mafic eclogites, glimmerite, and a phlogopite–garnet websterite. The phlogopite–garnet websterite is picritic in composition, contains orthopyroxene, clinopyroxene, garnet, phlogopite, pyrrhotite and apatite, and is probably of crustal origin based on mineral
18O values of 6·6–7·1; this will be discussed in a separate publication (Luffi et al., in preparation). The glimmerite, which consists of phlogopite, clinopyroxene, and sanidine, may be a fragment of metasomatized mantle, but the remainder of the xenoliths that we studied are unambiguously crustal and contain mineral assemblages indicative of ultrahigh temperatures and near-ultrahigh pressures. The mafic eclogites consist of omphacite, garnet and trace rutile and apatite, plus relict amphibole, plagioclase and biotite as garnet inclusions, whereas the sanidine eclogites include the above phases plus sanidine, kyanite, quartz, and minor relict plagioclase. The felsic granulites contain garnet, kyanite, quartz, and alkali feldspar, with minor graphite and rutile. All contain trace zircon and monazite.
|
Interpreting the histories of the xenoliths depends, in part, on the relative rates of xenolith ascent, thermal conduction and mineral reaction. Spera (1984)
calculated ascent rates of 0·1–3 m/s for alkali basalts bearing centimeter- to decimeter-scale xenoliths with densities of 3·3 g/cm3. Such ascent rates carry xenoliths to the surface from depths of 50–100 km in periods of 5–300 h. Some of the Pamir xenoliths have even higher densities (see below), implying more rapid ascent and shorter transport time. The characteristic thermal diffusion distance for a sphere with thermal diffusivity 10–6 m2/s is 0·2–1·8 m for periods of 5–300 h (Carslaw & Jaeger, 1959), implying that the Pamir xenoliths were heated significantly by the magma during transport to the surface. Rates of grain-boundary chemical diffusion in silicates are considerably slower, however, with 10–20 µm thick grain-boundary coronae developing between silicates (e.g. Yund, 1997; Milke et al., 2001) in 25–50 h experiments at 1000–1100°C in the presence of an H2O-rich fluid. Rates of Fe and Mg volume diffusion in garnet (Ganguly et al., 1998) and clinopyroxene (Dimanov & Sautter, 2000) are among the slowest phenomena of interest, and imply diffusive lengthscales of no more than 1 µm over the same timeframe at 1100°C. Na–K interdiffusion in alkali feldspar is 1–2 order(s) of magnitude faster (Christoffersen et al., 1983). Thus, the effects of entrainment and transport of the Pamir xenoliths within the host magma are expected to be limited to fracturing, melting, growth of grain-boundary coronae tens of microns thick, K-feldspar zoning at the scale of tens of microns, and garnet Fe–Mg zoning on the micron scale. The xenoliths do display textural and chemical changes at these scales resulting from interaction with magma or fluids, both before and during eruption. Those changes that may have occurred before eruption, based on the grain size of the reaction products and the equilibrated appearance of the textures include:
- the decomposition and recrystallization of phlogopite and the growth of sanidine + oxide minerals ± glass (Fig. 2a and b);
- the crystallization of K-feldspar and carbonate spherules (Fig. 2c).
Both of these textures could in principle have formed by reaction with an externally derived magma, but we argue below that the K-feldspar formed from in situ melting. Textural and chemical changes that probably occurred during—or soon before—eruption include:
- symplectic breakdown of omphacite in the presence of melt to low-Na clinopyroxene + plagioclase ± melt and amphibole + plagioclase (Fig. 2b–e and j);
- breakdown of kyanite to corundum + quartz and spinel + feldspar symplectite (Fig. 2c);
- glass, K-feldspar, and carbonate minerals formed from melt intruded along grain boundaries and along curved transgranular cracks (Fig. 2d–h);
- replacement of garnet rims in contact with melt by biotite + K-feldspar symplectite and spinel + K-feldspar + quartz in felsic granulites and mafic eclogites and by amphibole + spinel in sanidine eclogites (Fig. 2e and g);
- growth of micron-scale, bladed K-feldspar grains from interaction of melt with existing alkali feldspar (in felsic granulites and sanidine eclogites);
- potassium enrichment of alkali feldspar rims as a result of melt infiltration (in felsic granulites and sanidine eclogites);
- sodium enrichment of alkali feldspar as a result of breakdown of omphacite (in sanidine eclogites and mafic eclogites).
The sanidine eclogite xenoliths underwent the most alteration, experiencing almost all of the effects described above. The mafic eclogite sample is characterized only by biotite dehydration-melting reactions and melt intrusion along grain boundaries and cracks (Fig. 2l). The felsic granulite xenoliths are affected only at their margins and along a few grain boundaries and cracks (Fig. 2g and h).
|
PETROGRAPHY, MINERAL COMPOSITIONS AND BULK-ROCK COMPOSITIONS |
|---|
Sanidine–biotite eclogite 337A is a heterogeneous rock with garnet + clinopyroxene knots wrapped by biotite + sanidine rich zones (Fig. 2b, h and l). A foliation defined by elongate biotite plus some clinopyroxene and sanidine is folded at the millimeter to centimeter scale. Garnet is alm54prp28grs17sps01 (alm, almandine; prp, pyrope; grs, grossular; sps, spessartine) and shows rimward changes of –2 mol % grossular at constant Mg-number [= Mg/(Mg + Fe); Fig. 3a]. Clinopyroxene has unusually high TiO2 contents of 0·8–1·1 wt %, and rare matrix grains show rimward zoning in jadeite (NaAlSi2O6) content from 23 to 40 mol %; grains included within garnet show uniformly high jadeite contents (Fig. 3b). Matrix biotite has very high TiO2 contents of 6–8 wt % and is more magnesian (Mg-number 55) than inclusions in garnet (Mg-number 51). Alkali feldspar grains—weakly zoned in the range or57–68ab31–42an01–02 (or, orthoclase; ab, albite; an, anorthite)—occur in low-strain zones in the rock matrix and as inclusions in garnet. The rock is transected by micron-scale carbonate (Ca52Mg27Fe21–Ca52Mg41Fe09) + or99 alkali feldspar filled fractures (Figs 2h and 3c). Nearly all garnet and omphacite rims have micron-scale symplectites developed as a result of this—although a few garnet–omphacite boundaries and all garnet–inclusion grain boundaries not intersected by carbonate + feldspar veins appear unreacted (Fig. 2h). The omphacite coronae are formed of sodic augite and an18 plagioclase. All matrix biotites have partially decomposed to or93 alkali feldspar + hercynite–magnetite spinel where in contact with the carbonate + alkali feldspar veins (Fig. 2b). Garnet and clinopyroxene decomposition is more advanced in zones rich in decomposed biotite, suggesting that biotite dehydration melting accelerated or triggered garnet and clinopyroxene breakdown. The high TiO2 and K2O, low SiO2 and Al2O3, and steep REE pattern (Table 3; Fig. 4) of 337A are consistent with a potassic alkali basalt (leucite–basanite) protolith.
|
|
Eclogite 1154A has a strong foliation defined by the shape-preferred orientation of omphacite, quartz, rutile, apatite, and slightly oblate garnet grains (Fig. 2e and g). Garnets are up to 2·2 mm in diameter and have alm39prp39grs20sps01 cores (Fig. 3a) and alm39prp35grs24sps01 rims. This whole-grain zoning reflects changes of –4 mol % pyrope, +4 mol % grossular, and –2 Mg-number; Fe and Mn are unzoned, and Mg and Ca vary inversely. Omphacites are unzoned, except for sharp, 7% increases in Fe in the outer 15 µm of the grains; omphacite matrix grains and inclusions in garnet share the same composition. Fairly abundant pre-eclogitic inclusions within garnet are magnesian hastingsite (Na0·6K0·3)(Ca1·7Na0·3)(Mg1·8–2·6Fe1·7–2·3Al0·6)(Al1·9Si6·1)O22(OH1·0–2·0F0·0–1·0), plagioclase zoned from an18 cores to an36 rims, biotite with 3 wt % TiO2, quartz, and apatite. Micron-scale veins filled by carbonate and alkali feldspar transect garnet grains, and carbonates fill rare grain triple-junction pockets as large as 100 µm (similar to Fig. 2f). Garnets are surrounded by coronae of
10 µm width of subsilicic magnesian hastingsite (Na0·5K0·2)(Ca1·4Na0·6)(Mg2·2Fe1·5Al1·3)(Al2·3Si5·7)O22(OH2) + an20 plagioclase. All pyroxene grain boundaries have been penetrated by carbonate veins and formed symplectites >10 µm thick that consumed <10% of the clinopyroxene. Rutile is rimmed by ilmenite where in contact with quenched melt. The calculated bulk composition of this rock is that of a low-K tholeiitic basalt (Table 3). Sanidine–(kyanite) eclogite 1154B is a compositionally distinct part of the same xenolith. It has a moderate foliation defined by phase-ratio layering and the shape-preferred orientation of garnet, clinopyroxene, K-feldspar, minor kyanite, and rutile (Fig. 2i). Garnet grains as large as 3·5 mm are zoned with decreasing Mg-number and increasing Ca from alm44prp34grs22sps01 to alm42prp29grs28sps01; the magnitude and scale of the zoning vary from grain to grain independently of grain size. The garnet grains contain rare inclusions of an32 plagioclase and hornblende. Omphacites are homogeneous at 44 mol % jadeite and Mg-number 71. Sanidines have optically visible zoning, from or84ab15 to or74ab25. The bulk composition of 1154B is similar to that of a continental arc tonalite or andesite (Table 3).
Sanidine eclogite 1283 has a moderate foliation defined by elongate patches of clinopyroxene, plus garnet, K-feldspar, rutile, and significant apatite (Fig. 2d, f and k); 1283A and 1283B represent parts of the same xenolith with different clinopyroxene/K-feldspar ratios. Garnets in 1283B are zoned from alm45prp28grs26sps01 to alm45prp31grs24sps01, exhibiting rimward changes of +2 in Mg-number and –2 mol % grossular. They contain relatively abundant inclusions of biotite (0·3 wt % TiO2), subsilicic ferroan pargasite (Na0·4K0·5)(Ca1·9Na0·1)(Mg2·1Fe2·0Al0·9)(Al2·4Si5·6)O22(OH1·5F0·5), and carbonate (see below). All garnet external grain boundaries—and all garnet inclusion grain boundaries intersected by carbonate-filled veins—have symplectite rims of 10 µm thickness. Omphacites vary slightly in jadeite content from 44 to 46 mol % at constant Mg-number 67; individual grains are replaced 10–100% by micron-scale symplectites of plagioclase, sodic augite, and carbonate. Sanidines of composition or75ab24 have micron-scale rims of or99, are cut by micron-scale carbonate veins, and contain abundant healed fractures. Carbonate constitutes 2 vol. % of the rock and occurs in three textural types: (1) distributed thin grain-boundary films and trans-granular cracks (Fig. 2f); (2) subrounded inclusions in alkali feldspar (similar to Fig. 2c); (3) polycrystalline inclusions inside garnet (Fig. 2k). Most of the carbonates are clustered around the composition Ca0·55Mg0·30Fe0·15CO3, but a continuum from Ca0·54Mg0·28Fe0·18CO3 toward Ca0·5Mg0·5CO3 is clearly separated by a compositional gap from inclusions of pure CaCO3 (Fig. 3). The carbonate inclusions within garnet are anhedral to subhedral and consist of (1) euhedral carbonate grains with fine-scale oscillatory zoning visible with cathodoluminescence and (2) very fine-grained anhedral carbonate grains. Both forms are typically ternary Fe–Mg–Ca carbonate, but some euhedral crystals gave spot analyses of pure CaCO3. Rare barite is associated with the garnet-hosted carbonate inclusions. The bulk compositions of 1283A and 1283B are basaltic to trachybasaltic; 1283B shares an affinity to 1154B in its nearly parallel REE pattern (Table 3; Fig. 4).
Sanidine–(kyanite) eclogite 1309 has a granoblastic habit formed by garnet surrounded by clinopyroxene in a sea of K-feldspar, kyanite, quartz, and rutile (Fig. 2c and j). Minor, but widespread micron-scale transgranular and grain-boundary cracks contain carbonate. Garnets are essentially unzoned with only a minor decrease in Mg-number from alm46prp27grs26sps01 to alm47prp26grs26sps01. They contain common inclusions of an58–67 plagioclase, or71–82 alkali feldspar, quartz, and (Mg0·8Fe1·1Al0·1)(Al0·2Si1·8)O6 orthopyroxene, and are surrounded by sodic augite + an24 plagioclase coronae. Omphacites are homogeneous at 51–52 mol % jadeite and Mg-number 63–64; all show 20–100% decomposition to symplectites. Sanidines have optically visible zoning from or73–74ab25 to or56–68ab31–43. Some kyanites are surrounded by micron-scale coronae composed of corundum + quartz (Fig. 2c). The bulk composition of 1309 is that of an arc tonalite or andesite, similar to 1154B (although it is contaminated by a few vol. % of the host lava) (Table 3).
The three pieces (A, B, C) of kyanite–garnet granulite xenolith 1503 have granoblastic textures; their foliations are defined by the shape-preferred orientation of quartz, kyanite, alkali feldspar and porphyroblastic garnet, and locally by the orientation of graphite flakes and rutile. Garnets from these three samples are near-homogeneous alm56prp39grs04sps01 with slight rimward increases in grossular of +1 mol % at a constant Mg-number of 41. The relic phengites have Si contents as high as 3·6 atoms per formula unit of 11 oxygen atoms, and high F content (XF = F anions/2 = 0·35). Alkali feldspar are or58–70ab28–40an02, with rim enrichment to or99 where in contact with alkali-feldspar-filled veins. Rare biotite relicts are shielded in garnet grains, whereas phengite and K-feldspar are included in kyanite grains. In contrast to the eclogites, only minor alkali-feldspar or99 veins and carbonate veins are present in the granulites; healed fractures are absent. Grain-boundary symplectites are minor or absent because of the absence of clinopyroxene. Garnet, kyanite, sanidine, quartz, apatite, monazite, and zircon contain CO2-rich, peraluminous dacitic to rhyodacitic primary glass inclusions with 69·76% SiO2, 0·16% TiO2, 15·74% Al2O3, 0·7% FeO*, 0·13% MgO, 0·61% CaO, 2·05% Na2O, and 5·73% K2O and primary CO2-rich fluid inclusions (Chupin et al., 2001). The bulk compositions of these rocks are similar to Fe + Mg-rich and Na-poor pelitic schists (Patiño Douce & McCarthy, 1998) (Table 3). The major element composition of 1503C is within 20% of Taylor & McLennan's (1985) post-Archean average shale (adjusted to the same silica content to compensate for the higher quartz content of 1503C), except for Na, Ca and P, which are depleted in 1503C.
Glimmerite (or alkali minette) 2014 consists of phlogopite and clinopyroxene with minor apatite, cut by or99 alkali feldspar veinlets (Fig. 2a). Clinopyroxene has <5 mol % jadeite and has completely recrystallized to a clinopyroxene–carbonate symplectite. The phlogopites have 4–5 wt % TiO2 and significant F contents; all are recrystallized where in contact with alkali feldspar to more euhedral crystals with slightly higher Si, Mg, Ca and F and slightly lower Al, Ti, Fe, and Na, but are otherwise unchanged.
In summary, all eclogites show intrusion of carbonate and alkali feldspar along fractures that triggered, enhanced or accompanied symplectic decomposition of clinopyroxene and garnet. The xenolith compositions span a broad range, from mafic igneous rocks to felsic igneous rocks and metasediments. The association of continental arc magmatic rocks with mature sediments suggests that the bulk of the xenoliths were derived from evolved continental crust.
|
PRESSURE–TEMPERATURE CALCULATIONS |
|---|
Pre-eruption temperatures and pressures were determined by thermobarometry (Figs 5–7). Because of the widespread evidence for small-scale disequilibrium in the form of reaction coronas and mineral zoning, we made a careful assessment in each sample of whether rim and core compositions might represent equilibrium, using textural criteria such as the absence of grain-boundary reaction products and compositional criteria such as the magnitude and direction of zoning; this is discussed for each sample on a case-by-case basis below. Generally, the Pamir xenoliths have high-variance assemblages.
|
|
|
Interpreting the mineral zoning in the xenoliths requires knowing the diffusivities of elements in different phases and the operative phase transformations. The simplest rock to consider is the ‘two-phase’ eclogite 1154A, which consists of 59 vol. % omphacite, 37% garnet, 2% quartz and other trace phases; some of the 2·2 mm diameter garnet grains exhibit core-to-rim zoning of Mg (–4 mol %) and Ca (+4 mol %) against a homogeneous Fe and Mn background. As noted above, the diffusive lengthscale for Fe and Mg volume diffusion at 1100°C over the probable transport time for the xenolith is
1 µm in garnet (Ganguly et al., 1998) and less in clinopyroxene (Dimanov & Sautter, 2000)—any larger-scale zoning must therefore be the result of pre-eruption processes. It is improbable that the observed garnet zoning—decreasing Mg (and Mg-number) and increasing Ca rimward—could have developed solely through temperature-dependent Fe–Mg exchange with the omphacite, because Fe and Mg have equal charge and do not require compensation by more slowly diffusing Ca (Chakraborty & Ganguly, 1992). Alternatively, the garnet zoning could reflect net-transfer reactions among garnet, omphacite and other phases (e.g. melt). This would require an increasing jadeite content in the omphacite to compensate for the loss of Ca from the omphacite to the garnet. The jadeite content of the omphacite cannot change, however, unless another Na-bearing phase exists in the rock. We therefore conclude that the Ca zoning in the garnet reflects consumption of former plagioclase or melt, and that the Mg zoning may reflect simple charge compensation (Mg diffuses more slowly than Fe; Chakraborty & Ganguly, 1992); the fact that Mg-number decreases rimward in garnet suggests that temperature was declining during the compression-induced net-transfer reactions that pumped Ca into garnet. We ascribe the preservation of this zoning, plus the presence of low-pressure inclusions in some rocks, to shielding within garnet and to the generally slow reaction rates that are likely to have prevailed in these chiefly anhydrous rocks. Three types of reaction prove the most valuable for thermobarometry of these rocks: (1) Fe–Mg exchange between garnet and clinopyroxene; (2) net-transfer equilibria with positive P–T slopes; (3) net-transfer equilibria with negative P–T slopes. These reactions intersect within uncertainty for individual samples (Fig. 6). Because of the textural and chemical evidence that mineral rims underwent reaction with the melt, pressures and temperatures were calculated from mineral compositions interior to the rim zoning.
We calculated the positions of the Fe–Mg exchange equilibrium between garnet and clinopyroxene using three recent calibrations: THERMOCALC (Powell & Holland, 1988), Krogh Ravna (2000), henceforth K00, and Ganguly et al. (1996), henceforth GCT96. Figure 5 shows that agreement among the three calibrations is not excellent. Both THERMOCALC and GCT96 indicate temperatures 100°C hotter than K00 at the cooler end of the scale, and, whereas agreement between GCT96 and K00 is excellent around 1100°C, THERMOCALC gives temperatures that are 150–200°C hotter. To be conservative at the hotter end of the temperature scale, we use the GCT96 and K00 temperatures. These reactions have steep Clapeyron slopes, such that changing the assumed pressure by 0·5 GPa changes the apparent temperature by <45°C (Fig. 5).
Net-transfer reactions with positive slopes that are applicable to these rocks include albite = jadeite + quartz, plus the less well-characterized reaction garnet = Ca-Tschermak component in clinopyroxene + clinopyroxene (CaTs). We calculated the position of the latter reaction using THERMOCALC and PTGIBBS (Brandelik & Massonne, 2004) with the garnet solution models of Ganguly & Saxena (1984), Berman (1990) and Massonne (1995) and the C2/c clinopyroxene solution models of AX (http://www.esc.cam.ac.uk/astaff/holland/ax.html) and PTGIBBS, which explicitly consider jadeite, acmite, and Ca-Tschermak components. The various solution models for clinopyroxene yield calculated CaTs-based pressures that differ by >0·5 GPa, much greater than the 0·1 GPa uncertainty induced by 10% variation in CaTs activity. We calculated plagioclase activities using the average of models 3 and 4 of Holland & Powell (1992) (using Xan = 1 – Na – K and Xab = Na), incrementing the uncertainties of the activities to include the range in activities encompassed by the models. We calculated alkali feldspar activities using an average of the two solution models of Hovis et al. (1991) for disordered alkali feldspar. Five per cent uncertainty in the jadeite activity in clinopyroxene translates into ±0·3 kbar relative deviation; following Tropper et al. (1999), we assigned an absolute uncertainty of ±0·9 kbar (1
) to the position of the albite = jadeite + quartz equilibrium calculated with these activities. In kyanite-bearing samples we used the 1 pyrope + 1 grossular + 2 quartz = 3 diopside + 2 kyanite (GCKQ) net-transfer reaction, which has a negative P–T slope; we used the method of Ravna & Terry (2004), based on the garnet activity model of Ganguly et al. (1996), the clinopyroxene activity model of Holland (1990), and the Holland & Powell (1998) database. These equilibria were supplemented with reactions calculated by THERMOCALC when possible.
Sanidine–kyanite eclogite 1154B has a lower variance mineral assemblage, such that pressure and temperature can be estimated from three reactions: Fe–Mg partitioning between garnet and clinopyroxene, the albite breakdown reaction using the activity of albite in sanidine (Hovis et al., 1991), and the pyrope + grossular + quartz = diopside + kyanite net-transfer reaction (Fig. 6). Intersections among these equilibria span 24·7–28·0 kbar and 1015–1060°C.
Pressure and temperature can be estimated for the more mafic piece of the same xenolith, 1154A, by using the intersection of the garnet–clinopyroxene Fe–Mg exchange reaction and the CaTs component in omphacite in equilibrium with garnet (Fig. 6). Depending on the calibration chosen, intersections among these equilibria range from a minimum of 22·5 kbar and 1010°C to a maximum of 28·5 kbar and 1050°C. A minimum pressure of 21 kbar is given by the albite = jadeite + quartz reaction, assuming—in the absence of matrix feldspar—that aab = 1.
Pressure and temperature can be estimated for sanidine xenolith 1283B from intersection of the garnet–clinopyroxene Fe–Mg exchange reaction and the albite = jadeite + quartz reaction, using aab for sanidine (Hovis et al., 1991; Fig. 6). The intersections are at 1100°C (K00) and 28·5 kbar and 1130°C (GCT96) and 29·2 kbar (Fig. 6). Fe–Mg partitioning between garnet and biotite indicates temperatures 35°C hotter, using THERMOCALC.
Sanidine–kyanite eclogite 1309 has an identical mineral assemblage to 1154B and permits application of the same thermometers and barometers (Fig. 6). Intersections among the Fe–Mg exchange, albite-breakdown, and GCKQ reactions span a relatively narrow range of 26·5–30·5 kbar and 1035–1090°C.
The pressure–temperature conditions of the mineral assemblage in kyanite–garnet granulite 1503 are difficult to constrain using thermobarometry because of the high-variance assemblage. Fe–Mg partitioning between garnet and biotite inclusions in 1503B indicates temperatures of 1080°C, but this is probably a maximum because Fe–Mg exchange between the rare and tiny biotite inclusions and their host garnet certainly continued after entrapment. We can, however, exploit the experiments of Patiño Douce & McCarthy (1998) on a muscovite schist to place constraints on the equilibration conditions. Those experiments demonstrate that the presence of garnet + K-feldspar + kyanite + quartz and the absence of omphacite, biotite, muscovite and plagioclase implies temperatures between 1025°C and 1060°C at 2·1 GPa (Fig. 7). They further suggest equilibration pressures between 2·1 and 3·0 GPa.
Sanidine–biotite eclogite 337A yields different results from the other samples. Fe–Mg partitioning between garnet and clinopyroxene indicates temperatures of 720–750°C (K00) or 865–880°C (GCT96) for an assumed pressure interval of 20–25 kbar. Fe–Mg exchange between garnet and biotite rims (Ganguly et al., 1996) indicates slightly higher temperatures than GCT96 (900–925°C), regardless of whether an ideal or nonideal Al–Ti mixing model is chosen for biotite (Patiño Douce et al., 1993). Because both plagioclase and quartz are absent, pressure cannot be determined quantitatively from the mineral assemblage. However, if we assume that all the xenoliths were sampled along a single geotherm, we can estimate pressure from the intersection of that geotherm with the Fe–Mg garnet–clinopyroxene exchange equilibrium. Figure 7 shows that a linear geotherm determined by the P–T conditions of the remaining xenoliths intersects the Fe–Mg garnet–clinopyroxene exchange equilibrium for 337A at 2·0–2·5 GPa.
|
ORIGIN AND EVOLUTION OF THE XENOLITHS |
|---|
This section combines textural, paragenetic and mineral compositional evidence to assess the origin and evolution of the Pamir xenoliths. We demonstrate that the xenoliths are the result of (1) high-pressure dehydration melting that produced a dense, refractory residuum and a K-rich silicate liquid; (2) carbonate metasomatism; possibly, (3) potassic metasomatism that caused phlogopite and sanidine crystallization.
Hydrous eclogite
If the protolith of 337A was an alkali basalt or a equivalent plutonic rock, the presence of abundant biotite and sanidine could simply reflect the bulk composition of the rock. Alternatively, the biotite and/or sanidine might have been introduced into the rock as a melt or fluid. The presence of sanidine and biotite inclusions in garnet requires that such introduction predated garnet growth. The lower calculated equilibration temperature and the lower inferred equilibration pressure suggest that this xenolith was extracted from a different depth in the lithosphere and has a different history from the other xenoliths.
Anhydrous eclogites
The bulk chemical compositions of samples 1154B and 1309 resemble those of quartz monzonites or biotite granodiorites to calc-alkaline tonalites. The nominal bulk compositions of mafic eclogites 1154A, 1283A, and 1283B suggest derivation from basaltic protoliths. However, the bulk compositions are so similar to the residua produced by Patiño Douce & McCarthy (1998) during experimental partial melting of calc-alkaline tonalite (Fig. 8), that they are more probably residua derived from partial melting of typical continental-margin magmatic arc rocks. According to Patiño Douce & McCarthy (1998), dehydration melting of amphibole–biotite tonalites at 19–24 kbar and T > 950°C produces a solid residuum of garnet + omphacite + sanidine + quartz—similar to the sanidine eclogite xenoliths from the Pamir. The amphibole and plagioclase preserved in garnet cores are compatible with this interpretation. The only difference, the presence of kyanite in 1154B and 1309, suggests that the protoliths could have been richer in Al2O3. The REE patterns of these rocks suggest melt loss in the presence of garnet; such a process is permissible if the protoliths had steep REE patterns, typical of Andean or Sierran intermediate arc magmas (Fig. 4), which have major element abundances broadly similar to these xenoliths.
|
The thin grain-boundary films of carbonate in 1283B are probably the result of late-stage intrusion of the xenolith by a carbonate-rich fluid. The rounded inclusions within alkali feldspar crystals (Fig. 2c) might represent crystallization of sanidine + carbonate from a carbonate-rich alkalic melt (Lee & Wyllie, 1998
). The polycrystalline carbonate inclusions inside garnet have fine-scale compositional zoning and are euhedral (Fig. 2k). Because such fine-scale zoning should have been instantly erased by diffusion at these elevated temperatures, the carbonates must have crystallized during or soon before eruption and are presumably physically connected to veins outside the garnet.
Kyanite–garnet granulites
The bulk compositions, mineralogies, and glass inclusions of samples 1503A, 1503B, and 1503C, suggest that these rocks are residua of high-pressure melting of two-mica pelitic protoliths. Rare biotite inclusions in garnet, and phengite and alkali feldspar inclusions in kyanite, suggest a feldspathic two-mica protolith. The mineral assemblage, garnet + kyanite + sanidine + quartz, is identical to that observed by Castro et al. (2000) to form from experimental dehydration melting of two-mica pelites at elevated pressures. These xenoliths cannot represent quenched high-pressure melts, as they have SiO2 and Al2O3 contents that are granitic (sensu lato), but the concentrations of alkali elements and MgO + FeO are opposite to that expected in granitic melts. The REE pattern of 1503C, with its HREE enrichment and light REE (LREE) depletion relative to average shale (Fig. 4c), is quantitatively consistent with >40% removal of a melt with a very steep REE pattern—as expected for a melt in equilibrium with garnet. Although garnet was probably present during a substantial part of this granulite's P–T–t path, it is likely that the REE pattern of the rock reflects melting with garnet in the residue near the peak P–T conditions. The calculated REE pattern slightly differs from those of glass inclusions found in quartz from similar granulite xenoliths from the Pamir (Chupin et al., 2001). The glass inclusions have similar HREE, but higher LREE concentrations; these could represent an earlier melt fraction. Alternatively, the difference could be explained by slightly different bulk compositions or melting histories.
Glimmerite
Glimmerite 2014 is interpreted to represent the result of interaction between a H2O-rich potassic melt (or fluid) and a ferro-magnesian host rock. The apparent chemical equilibrium between sanidine and phlogopite, plus the absence of mafic neoblasts along the biotite–feldspar interfaces, makes it clear that the sanidine was not produced by dehydration melting of the biotite. Such potassic melts or fluids might be derived from the asthenosphere (Menzies et al., 1987; Gregoire et al., 2002) or by the high-pressure breakdown of fluorinated biotite in felsic granulites (Becker et al., 1999); the absence of mantle xenoliths in this suite implies the latter. It is also possible that 2014, with its 80% modal biotite, represents a cumulate crystallized from a K-rich melt. The very steep REE pattern requires derivation from a source containing garnet.
|
ERUPTION AGE |
|---|
Biotite from xenoliths 337A and 2014 were dated via stepwise degassing in a resistance furnace at Stanford University. The samples yielded slightly different ages of 11·52 ± 0·22 Ma (2
) and 11·17 ± 0·14 Ma, respectively (Fig. 9). As both samples were at temperatures well above Ar retention in biotite prior to eruption, these ages are unambiguously eruption ages. The differences between the apparent ages of the two xenoliths might indicate slightly different eruption times or may be geologically meaningless.
|
|
DEEP CRUST BENEATH THE SOUTHEASTERN PAMIR |
|---|
The thermobarometric calculations and mineral parageneses detailed above suggest that most of the xenoliths investigated equilibrated at high pressures and temperatures, P = 25–28 kbar and T = 1000–1100°C. The rocks span the entire range of probable crustal compositions—basaltic to tonalitic to pelitic—in the grossest sense. We have interpreted the textures and minerals within the xenoliths to be the result of partial melting and enrichment metasomatism at high pressure; most of the minerals present represent residua, whereas others, specifically sanidine and biotite, in some samples represent quenched melt or fluid either derived locally or introduced from a distance. These xenoliths were carried to the surface in a volatile-rich magma at
11 Ma, which may have formed by melting of enriched subcontinental upper mantle (e.g. Turner et al., 1996; Miller et al., 1999)—as has been suggested for the Neogene granitoids of the south Karakoram (Mahéo et al., 2002).
Laser ablation multicollector ICP-MS dating of zircons was performed on two of these xenoliths by Ducea et al. (2003). Sanidine–(kyanite) eclogite 1309, here interpreted as a calc-alkaline monzonite to tonalite originating from dehydration melting of amphibolite–biotite tonalite, underwent igneous crystallization at 63–88 Ma and contains inherited zircons with ages of 132 Ma, 195 Ma, and 250 Ma. The pelitic protolith of kyanite–garnet granulite 1503A either was deposited after 57 Ma or underwent high-grade zircon growth at 57–84 Ma; older zircons are 146–170 Ma, 412–465 Ma, 890 Ma and 1·4 Ga. The early Paleozoic and Precambrian zircon ages prove that the xenoliths were derived from Gondwanan crust (e.g. DeCelles et al., 2000); the Mesozoic ages preclude derivation of the xenoliths from Indian crust, as India does not contain magmatic rocks of Mesozoic age (with the exception of minor rift-related alkalic volcanism; Hodges, 2000). The zircon ages are compatible with derivation of the xenoliths from crust of the type exposed at present in a 300 km wide belt in the Karakoram–Kohistan–Ladakh and Hindu Kush–southern Pamir blocks (e.g. Gaetani, 1996; Fraser et al., 2001; Schwab et al., 2004); these blocks are Gondwanan, and contain Jurassic to Late Cretaceous plutons produced by prolonged northward subduction of Tethys. The southern Pamir contain granodiorites and two-mica monzogranites with Proterozoic to Cretaceous zircon ages (Schwab et al., 2004) identical to those found in the xenoliths (Ducea et al., 2003). The Kohistan arc, especially its southern, deeply exhumed part, contains variously metamorphosed calc-alkaline to tholeiitic gabbros, diorites, and granodiorites with ages of 82–99 Ma (e.g. Searle, 1991; Schaltegger et al., 2002). The Karakoram and Hindu Kush also include voluminous siliciclastic rocks (mostly shales) of various metamorphic grades that underwent several high-strain thickening events before and after the India–Asia collision (e.g. Gaetani, 1996; Zanchi & Gritti, 1996; Fraser et al., 2001). Granulite-facies, Barrovian metamorphism in the Karakoram block is pre-37 Ma, and 25 Ma to Recent (Fraser et al., 2001; Mahéo et al., 2002). This is similar to the prograde high-temperature metamorphism of xenolith 1503A, which contains 50 and 34 Ma monazites and 15 Ma zircon rims (Ducea et al., 2003).
|
DEEP CRUST BENEATH THE PAMIR AND TIBET |
|---|