Gondwana Research 20 (2011) Contents lists available at ScienceDirect. Gondwana Research. journal homepage:

Transcription

1 Gondwana Research 0 (0) 4 Contents lists available at ScienceDirect Gondwana Research journal homepage: New constraints on UHT metamorphism in the Eastern Ghats Province through the application of phase equilibria modelling and in situ geochronology F.J. Korhonen a,, A.K. Saw b, C. Clark a, M. Brown c, S. Bhattacharya b a Department of Applied Geology, Curtin University, GPO Box U, Perth, WA 4, Australia b Geological Studies Unit, Indian Statistical Institute, Kolkata, India c Laboratory for Crustal Petrology, Department of Geology, University of Maryland, College Park, MD 04, USA article info abstract Article history: Received 4 September 0 Received in revised form May 0 Accepted 4 May 0 Available online May 0 Handling Editor: M. Santosh Keywords: Granulite Monazite Phase equilibria modelling SHRIMP UHT metamorphism U Pb geochronology High Mg Al granulites from the Sunki locality in the central portion of the Eastern Ghats Province record evidence for the high-temperature peak and retrograde evolution. Peak metamorphic phase assemblages from two samples are garnet+orthopyroxene +quartz+ilmenite+melt and orthopyroxene +spinel + sillimanite+melt, respectively. Isochemical phase diagrams (pseudosections) based on bulk rock compositions calculated in the chemical system Na O CaO K O FeO MgO Al O SiO H O TiO Fe O (NCKFMASHTO) and Al contents in orthopyroxene indicate peak UHT metamorphic conditions in excess of 0 C and. kbar. Microstructures and the presence of cordierite interpreted to record the post-peak evolution show that the rocks underwent decompression and minor cooling from conditions of peak UHT metamorphism to conditions of ~00 C at ~. kbar. In situ U Pb isotope analyses of monazite associated with garnet and cordierite using the Sensitive High Resolution Ion Microprobe (SHRIMP) yield a weighted mean 0 Pb/ U age of ca. 0 Ma, which is interpreted to broadly constrain the timing of high-temperature monazite growth during decompression and melt crystallization at ~00 0 C and. kbar. However, the range of 0 Pb/ U monazite ages (from ca. 4 Ma to Ma for one sample and ca. 4 Ma to Ma for the second sample) suggest protracted monazite growth during the high-temperature retrograde evolution, and possibly diffusive lead loss during slow cooling after decompression. e results of the integrated petrologic and geochronologic approach presented here are inconsistent with a long time gap between peak conditions and the formation of cordierite-bearing assemblages at lower pressure, as proposed in previous studies, but are consistent with a simple evolution of a UHT peak followed by decompression and cooling. 0 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.. Introduction e Eastern Ghats along the eastern coast of Peninsular India expose a deep crustal section of a Proterozoic orogenic belt. Recent work has divided the Eastern Ghats Belt into four major crustal units: the Late Archaean Jeypore and Rengali Provinces, the Late Palaeoproterozoic Ongole Domain, and the Meso-Neoproterozoic Eastern Ghats Province (Fig. ; Dobmeier and Raith, 00). e rocks preserve a complex history of high-grade metamorphism and intense deformation suggestive of several phases of crustal reworking during a prolonged evolution (Chetty, 0). Previous petrologic studies in the Eastern Ghats Province (e.g. Lal et al., ; Kamineni and Rao, ; Sengupta et al., 0; Dasgupta et al., ; Sen et al., ; Mohan et al., ; Mukhopadhyay and Bhattacharya, ; Shaw and Arima, ; Bose et al., 000; Rickers et al., 00; Bhattacharya and Kar, 00; Sarkar et al., Corresponding author. Tel.: + ; fax: +. address: f.korhonen@curtin.edu.au (F.J. Korhonen). 00; Bose et al., 00; Das et al., 00; Bose and Das, 00; Nasipuri et al., 00) have identified evidence for ultra-high temperature (UHT) metamorphism with peak conditions estimated to be greater than 0 C and kbar, although there remains considerable debate about whether the pressure(p) temperature (T) time (t) path is clockwise or counterclockwise. A polyphase history of near-isobaric cooling following peak metamorphism, and subsequent reheating and decompression has also been proposed in many of these studies, although the P T estimates for the retrograde events vary. Geochronological investigations have attempted to constrain the timing of these events in order to characterize the polyphase history of Eastern Ghats Province. Estimates for the timing of UHT metamorphism range from ca. 400 Ma (U Pb in zircon, and Pb Pb in feldspar Jarick, ; reviewed in Simmat and Raith, 00) to ca. 0 0 Ma ( U Pb chemical ages from monazite Simmat and Raith, 00). Evidence for high-temperature metamorphism and felsic magmatism between ca. 00 and 0 Ma has been well documented in the Eastern Ghats Province (U Pb in zircon Grew and Manton, ; Rb Sr, Sm Nd whole rock and mineral isochrons, U Pb in zircon Shaw et al., ; U Pb in zircon and monazite Mezger 4-X/$ see front matter 0 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved. doi:./j.gr

2 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 N INDIA MS Z Ba y Z NS va da Go Jeypore of Be n ga l Z VS Fig. Bhubaneshwar ri 0 km EGP intruded by voluminous plutons of megacrystic, garnet- and orthopyroxene-bearing granitoids. e study area around Sunki in the central portion of the Eastern Ghats Province (Fig. ) exposes granulite facies metamorphic rocks including pelitic migmatite (khondalite), calcsilicate rock, mafic granulite, charnockite enderbite, and orthopyroxene-free granite. Evidence of three phases of folding is preserved in this granulite migmatite granite complex (Saw, 00). Migmatitic high Mg Al granulite (±sapphirine) occurs as small bodies within the dominant metapelitic granulite (khondalite). Orthopyroxene-free granite is closely associated with the metapelitic granulite and commonly has an intrusive relationship with it; enclaves of metapelitic granulite and charnockite enderbite are common in the granite. Ri ft Proterozoic basins. Sample descriptions Eastern Ghats Province (EGP) Major alkaline bodies rust contact Cuddapah Basin Udaigiri Domain Vinjamuru Domain Nellore Two samples of migmatitic high Mg Al granulite with discrete melanosomes and leucosomes ranging in thickness from a few millimeters to several centimeters were investigated in this study (Fig. ). Ongole Domain Jeypore Province Rengali Province Chennai.. Petrography Study area (Sunki) 0 E Fig.. Geologic outline of the Eastern Ghats Belt showing location of the study area in the Eastern Ghats Province (modified from Dobmeier and Raith, 00). MSZ, Mahanadi Shear Zone; NSZ, Nagavalli Shear Zone; VSZ, Vamsadhara Shear Zone. and Cosca, ; U Pb chemical ages from monazite Simmat and Raith, 00; U Pb zircon Das et al., 0), and is interpreted as a second granulite facies event following the early UHT metamorphic event. Based on these constraints, the current model for the hightemperature evolution of the Eastern Ghats Province invokes a polyphase history with an early UHT metamorphic event no later than ca. 0 Ma followed by near-isobaric cooling to ~0 00 C (see Dasgupta and Sengupta, 00; Simmat and Raith, 00). e granulites were subsequently reworked and remelted during a second high-temperature event at ca Ma, reaching conditions of 0 C and kbar, followed by near-isothermal decompression to ~ kbar. e polyphase nature of the proposed model involves cooling and reheating, requiring a renewed heat source. An additional consequence of a polyphase model is that melting and melt loss during the early UHT metamorphic event would reduce the fertility of the rocks for melting during the subsequent metamorphic event (e.g. Brown and Korhonen, 00; Korhonen et al., 0). In this paper, we present the results of an integrated petrologic and geochronologic study of two migmatitic high Mg Al granulite samples from the Sunki locality in the Eastern Ghats Province. e aim of this study is to test the polyphase model for the metamorphic evolution of the Eastern Ghats Province as recorded in the Sunki locality. Microstructural observations, information from isochemical phase diagrams (pseudosections) based on melanosome compositions calculated in the chemical system NaO CaO KO FeO MgO AlO SiO HO TiO FeO (NCKFMASHTO), and mineral chemistry are integrated with in situ U Pb dating of monazite. e results constrain the P T conditions of peak and retrograde metamorphism and establish the timing of granulite facies metamorphism, and show that the P T t path for the two samples is consistent with a single UHT event. Petrographic observations from melanosomes are separated into those relating to the inferred peak metamorphic phase assemblage and those relating to the post-peak metamorphic evolution. Minerals that comprise the peak assemblage occur as coarse grains, which are commonly isolated from each other by coronae or intergrowths of finer-grained minerals that represent post-peak reaction products.... Sample SK Peak metamorphic phase assemblage. e peak metamorphic phase assemblage in the melanosome of this sample is characterized by coarse-grained garnet and orthopyroxene (Fig. a). e garnet porphyroblasts are up to mm in size and contain inclusions of quartz, biotite, and ilmenite, and rare K-feldspar (Fig. b). Some garnet grains contain very rare inclusions of sillimanite and sapphirine rimmed by cordierite, and intergrown or included spinel with magnetite exsolution and minor sapphirine in the garnet. Orthopyroxene porphyroblasts are less abundant than garnet, and are typically smaller in size. ey host exsolved oxide lamellae that range in size from about μm to submicron, which are likely ilmenite hematite (cf. Korhonen and Stout, 004). Inclusions of. Geological setting e Eastern Ghats Province (Fig. ) comprises deformed and metamorphosed metasedimentary, basic, and enderbitic granulites Fig.. Field photograph of typical migmatitic high Mg Al granulite from the Sunki locality.

3 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 (a) (b) Kfs Kfs Qtz 00 μm 00 μm (c) (d) Ilm Ilm 400 μm 400 μm Fig.. Photomicrographs from sample SK--0. (a) Coarse-grained garnet and orthopyroxene porphyroblasts interpreted to be part of the peak assemblage, separated by small recrystallized grains of garnet, orthopyroxene and quartz (in box). Orthopyroxene porphyroblasts host oxide exsolution lamellae. (b) Garnet porphyroblasts with inclusions of quartz, biotite, and rare K-feldspar. (c) Orthopyroxene in direct contact with garnet lacks oxide exsolution lamellae and commonly contains abundant inclusions of biotite, in contrast to adjacent garnet and peak orthopyroxene porphyroblasts (see Fig. a). Aggregates of polygonal cordierite surround garnet. limanite in the lower left of the photograph is the edge of a resorbed grain in cordierite. (d) Cordierite rimming garnet, note the grain boundary melt film in the lower right of the photograph (in box). biotite; cordierite; garnet; Ilm ilmenite; Kfs K-feldspar; Mt magnetite; orthopyroxene; Pl plagioclase; Qtz quartz; sillimanite; Spl spinel; Spr sapphirine. embayed biotite are also present. Modally minor quartz occurs as small rounded grains throughout the melanosome and as inclusions in garnet. Ilmenite grains with exsolved hematite lamellae occur in garnet-rich domains, either as inclusions in garnet porphyroblasts (Fig. c) or as grains in cordierite near garnet (Fig. d). Melt is also inferred to be part of the peak metamorphic phase assemblage. Based on these observations, the peak metamorphic assemblage is interpreted to be garnet + orthopyroxene + quartz + ilmenite + melt.... Post-peak metamorphic phase assemblage. In the melanosome of this sample, small recrystallized grains of garnet, orthopyroxene and quartz commonly occur along the line of the former grain boundaries between peak garnet and orthopyroxene porphyroblasts (Fig. a). Some coarse grains of garnet and orthopyroxene are in contact, and orthopyroxene in this microstructural setting contains abundant inclusions of biotite and lacks the oxide exsolution lamellae (Fig. b and c), in contrast with the orthopyroxene porphyroblasts. Aggregates of polygonal cordierite commonly surround garnet, and cordierite typically contains fewer inclusions of biotite as compared to garnet or orthopyroxene porphyroblasts (Fig. b, c and d). Grain boundaries of garnet in contact with cordierite may be rounded and embayed, and garnet relics in cordierite are common. limanite is sparse, but is observed as small rounded or ragged grains within cordierite (Fig. c) and as rare inclusions rimmed by cordierite hosted in garnet. Biotite occurs as coarse laths and aggregates around garnet, orthopyroxene, and cordierite, and less commonly as rounded inclusions in these phases. ese observations suggest that the postpeak assemblage is characterized by the recrystallization of garnet and the growth of orthopyroxene, sillimanite, cordierite, and biotite.... Sample D--S... Peak metamorphic phase assemblage. e peak metamorphic phase assemblage in the melanosome of this sample comprises abundant elongate orthopyroxene porphyroblasts (Fig. 4a and c), rare grains of spinel (Fig. 4a and b), and sparse isolated coarse-grained sillimanite (Fig. 4c). Coarse-grained orthopyroxene contains few inclusions, and the spinel grains have abundant exsolved magnetite. Melt is inferred to be part of the peak metamorphic phase assemblage.... Post-peak metamorphic phase assemblage. In the melanosome of this sample, the peak phases typically have ragged and embayed edges, suggestive of resorption, and are commonly surrounded by cordierite or intergrowths of fine-grained sillimanite and orthopyroxene with varying amounts of rounded biotite (Fig. 4). e finegrained sillimanite in this microstructure is optically continuous with the coarse-grained sillimanite (e.g. Fig. 4c), and may be intergrown with ragged fine-grained orthopyroxene or interspersed with isolated and resorbed orthopyroxene (Fig. 4d). Cordierite is abundant and occurs both as coarse grains (Fig. 4b, c and d) and as fine-grained

4 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 (a) (b) + Spl Spl Spl + 00 μm (c) (d) 400 μm + + Kfs ± Pl μm 400 μm Fig. 4. Photomicrographs from sample D--S. (a) Coarse orthopyroxene porphyroblasts and spinel with magnetite exsolution rimmed by cordierite. Intergrowths of fine-grained sillimanite and orthopyroxene are commonly replaced by cordierite. (b) Spinel rimmed by sillimanite, which is optically continuous with sillimanite in the sillimanite orthopyroxene intergrowths. e intergrowths occur within cordierite. (c) Coarse-grained sillimanite in the lower middle of the photograph is optically continuous with sillimanite orthopyroxene intergrowths. (d) Fine-grained sillimanite in the center right of the photograph is intergrown with ragged fine-grained orthopyroxene, whereas fine-grained sillimanite in the upper portion of the photograph is interspersed with isolated and resorbed orthopyroxene. Cordierite surrounds the sillimanite orthopyroxene intergrowths, and occurs both as coarse grains and as fine-grained polycrystalline aggregates with K-feldspar and minor plagioclase. Biotite occurs as small rounded grains within the sillimanite orthopyroxene intergrowths and in cordierite, and surrounding peak sillimanite and orthopyroxene porphyroblasts. polycrystalline aggregates with K-feldspar and minor plagioclase (Fig. 4d). Both types of cordierite may surround the peak phases and the sillimanite orthopyroxene intergrowths, which are rounded and resorbed in this setting (Fig. 4b and c). e cordierite K-feldspar aggregates may also contain embayed coarse cordierite grains. Spinel is isolated from orthopyroxene by a rim of cordierite (Fig. 4a), whereas other spinel grains are rimmed by sillimanite (Fig. 4b). e sillimanite rims in this microstructure are continuous with finegrained sillimanite orthopyroxene intergrowths that separate spinel from cordierite. Biotite occurs mainly as small rounded grains within the sillimanite orthopyroxene intergrowths, as well as coarsegrained aggregates surrounding peak sillimanite and orthopyroxene porphyroblasts (Fig. 4d). Rare biotite also occurs in the coarse-grained cordierite and in the cordierite K-feldspar intergrowths. e observed microstructures are interpreted to record the continuous growth of post-peak sillimanite with biotite, likely at the expense of orthopyroxene, followed by the growth of cordierite. e fine-grained cordierite Kfeldspar intergrowths are interpreted to post-date the growth of the coarse-grained cordierite... Mineral chemistry Biotite and spinel compositions were determined with a JEOL JXA-00 wavelength dispersive electron microprobe at the Univer- sity of Maryland, Nanoscale Imaging Spectroscopy and Properties (NISP) Laboratory. e compositions of orthopyroxene, garnet and cordierite were obtained with a JEOL JXA-00M Superprobe at the Institute Instrumentation Centre, Indian Institute of Technology Roorkee, India, and with a CAMECA SX 0 Electron Probe Microanalyzer at the Geological Survey of India, Kolkata. Quantitative point analyses were collected using an accelerating voltage of kv, a beam current of 0 na, and a μm beam diameter. Representative compositions of key minerals are listed in Tables and. Ferric iron contents for orthopyroxene, garnet and spinel were determined using the stoichiometric method of Schumacher (). Ferric iron contents for orthopyroxene were also calculated using the ideal analysis approach of Powell and Holland (00). is approach considers the consequence of varying the proportion of iron in an analysis that is considered to be ferric within the analytical uncertainty of a mineral analysis using a least squares calculation (see also Carson and Powell, ; Štípská and Powell, 00). e minimum and maximum Fe + contents constrained by this approach and the resulting Al contents [y() = xal,m()] are provided in Table. Orthopyroxene grains show minimal compositional zoning and are relatively Mg and Al rich, with XFe (=Fe +/[Fe + + Mg]) in sample SK--0 ranging from 0. to 0., and in sample D--S ranging from 0.0 to 0.4 (Table ). e y() values calculated using the ideal analysis approach have mean values between 0. and 0.,

5 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 Table Representative orthopyroxene, garnet, and spinel compositions. Sample SK--0 SK--0 D--S D--S D--S D--S SK--0 SK--0 SK--0 D--S D--S D--S D--S Mineral Spl Spl Spl Spl Setting Rim Core Core Rim Rim Core Analysis / 4/ 40/ 4/ 4/ / / / 4/ / / 4/ /4 Wt.% SiO TiO Al O Cr O FeO ZnO....0 MnO MgO CaO Na O K O Total Cations Oxygen Si Ti Al Cr Fe + * Fe Zn Mn Mg Ca Na K X Fe Ideal analysis Fe + ** min Fe + ** max Fe + ** mean y() min y() max y() mean , not applicable; Fe + * determined using the stoichiometric method of Schumacher (); X Fe =Fe + /(Fe + +Mg) based on the Schumacher () method; Fe + ** minimum (min), maximum (max) and mean cations and resulting y() [=x Al,M ()] determined by ideal analysis of Powell and Holland (00), see text for further discussion. although the minimum and maximum range of possible y() range from 0.0 to 0.. Garnet in sample SK--0 is an almandine pyropedominated solid solution (Table ) with X Fe of Grossular and spessartine contents are low ( cations per formula unit (pfu), Table ). Garnet shows minor Mg Fe zonation with rims corresponding to slightly higher X Fe compared to the cores. Spinel in sample D--S is primarily a spinel hercynite solid solution, with a minor gahnite component (Zn~ cations pfu; Table ). e spinel grains commonly host exsolved magnetite lamellae. Cordierite in both samples is Mg-rich, with X Fe in sample SK--0 ranging from 0. to 0.4, and in sample D--S ranging from 0. to 0. (Table ). e grains show minimal compositional zoning and have low alkali content. Biotite in sample SK--0 has high Ti and F contents, with TiO ranging from. to. wt.% (=0.0 to 0.0 cations pfu) and F ranging from.0 to. wt.% (Table ). X Fe values range from 0. to 0., although most grains in sample SK--0 have values between 0. and 0.. ere is no systematic compositional variability between biotite in different microstructural settings. Biotite in sample D--S has slightly lower Ti, but higher F contents as compared to sample SK--0, with TiO ranging from. to 4. wt.% (=0.0 to 0. cations pfu) and F ranging from. to. wt.%, (Table ). X Fe values range from 0. to 0.4. Similar to sample SK--0, there is also no systematic compositional variability between biotite in different microstructural settings in this sample. 4. Monazite geochronology Individual monazites were identified in thin section and imaged using a backscattered electron (BSE) detector. in section fragments were cast in mm epoxy disks with chips of the India monazite standard Ind- (0 Ma, 0 Pb/ U=0.0), and coated with a thin membrane of gold that produced a resistivity of 0 Ω across the disk. Monazite elemental mapping for Ce,,, U and Pb was undertaken using the JEOL Superprobe at James Cook University in order to visualize variations in monazite chemistry (Fig. ). An accelerating voltage of kv and a current of 0 na with a step size of 0. μm in both the x and y directions were used for electron probe mapping. Maps were processed in NIH Image using the Fire lookup table, and all maps were processed to the same degree so color intensities are directly comparable between grains (Fig. ). U Pb isotopic measurements were carried out using the SHRIMP-II at the John de Laeter Centre for Mass Spectrometry, Western Australia, and analyzed with ~0. na O primary beam focused on to ~ μm spots, a -scan duty cycle, and a mass resolution of ~000 (a more detailed description of the Curtin SHRIMP procedure for monazite analysis is provided by Foster et al., 000). Isotopic age data for the two samples are given in Table ; uncertainties given for individual analyses (ratios and ages) are σ values. e calculated 0 Pb/ U ages are reported with σ uncertainties. A Tera Wasserburg concordia plot for all the data is shown on Fig..

6 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 Table Representative cordierite and biotite compositions. Sample SK--0 SK--0 SK--0 D--S D--S D--S D--S SK--0 SK--0 SK--0 SK--0 D--S D--S D--S Mineral Setting Core Rim Core Rim Core Rim Rim In In In In In In l Analysis / / / / / 4/ / / / / / / / / Wt.% SiO TiO Al O Cr O FeO ZnO MnO MgO CaO Na O K O F Cl Total Oxygen Si Ti Al Cr Fe Zn Mn Mg Ca Na K F Cl Total X Fe In [phase], inclusion in [phase]; l, sillimanite orthopyroxene intergrowth;, not applicable; Fe as total Fe; X Fe =Fe/(Fe +Mg). 4.. Monazite age results for sample SK--0 Forty-nine monazite analyses from five grains were performed on sample SK--0 (Table, Fig. a). Four of the grains are closely associated with fractures in garnet, and one grain (SK-M) is an inclusion in cordierite. e grains show compositional zoning in with depletion in thin overgrowths and relative enrichment in the core regions (Fig. ). e -poor domains are typically associated with a slight increase in content, whereas variations in Ce, Pb, and U are less pronounced (Fig. ). Of the 4 analyses, were greater than % discordant and were not included in the age calculations (gray ellipses, Fig. a). e remaining analyses yielded 0 Pb/ U ages ranging between ca. 4 Ma and 4 Ma (Fig. a). e spread of ages does not correlate with petrographic setting or systematic differences in U or contents. e overgrowths were typically too thin to analyze, although the -poor domain in grain SK-M typically yielded discordant and younger ages as compared to the -rich domain. Twenty-three of the analyses with a range of ages from ca. 4 Ma to Ma define a weighted mean 0 Pb/ U age of ± Ma (σ, MSWD=0.4; Fig. c), with the remaining analyses showing a steady decrease in age. e spread of ages younger than ca. Ma is interpreted as evidence for diffusive lead loss at high temperatures (e.g. Smith and Giletti, ; Catlos et al., 00). 4.. Monazite age results for sample D--S irty-six monazite analyses from nine grains were performed on sample D--S (Table, Fig. b). One grain (D-M4) is an inclusion in orthopyroxene, and the other eight grains are inclusions or closely associated with cordierite. In contrast to sample SK--0, the grains from sample D--S are typically homogeneous, with the exception of the monazite inclusion in orthopyroxene (Fig. ). is grain contains a core region relatively depleted in, rimmed by a domain relatively enriched in. ere is no detectable age variation between the chemical domains. Of the analyses, seven were greater than % discordant and were not included in the age calculations (gray ellipses, Fig. b). Two analyses are not shown on Fig. a; one of these analyses (D-M.) is highly discordant, and the other (D-M.) yielded a singular young age of ± Ma. e remaining analyses yielded 0 Pb/ U ages ranging between ca. 4 Ma and Ma (Fig. a), with monazite inclusions in cordierite or grains closely associated with cordierite defining a weighted mean 0 Pb/ U age of ± Ma (σ, MSWD=.). e six analyses from the monazite inclusion in orthopyroxene typically yield older ages, although the three analyses from the -poor core yield ages greater than % reversely discordant. e remaining three analyses yield a weighted mean 0 Pb/ U age of 4±4 Ma.. Phase equilibria modelling e conditions of metamorphism and the microstructural evolution of the high Mg Al granulites in the Sunki locality are investigated using isochemical phase diagrams (pseudosections) based on melanosome compositions. A comparison of the inferred peak mineral assemblage preserved in the melanosome with the predicted assemblage on a P T pseudosection allow constraints to be placed on the peak conditions of metamorphism. e inferred post-peak assemblages and retrograde evolution are investigated using changes

7 0 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 (a) 4 (b) 4 4 (c) 4 (d) (e) 4 (f) 4 (g) 4 4 (h) 4 (i) 4 (k) (j) (m) (l) 4 M 4 4 M Fig.. ttrium and thorium X-ray compositional maps and SHRIMP analytical spots (ellipses) of selected monazite grains (listed in Table ). Maps were processed in NIH Image using the Fire lookup table, and all maps were processed to the same degree so color intensities are directly comparable between grains. Scale bar is 40 μm. in predicted mineral abundance to model mineral growth and consumption as a consequence of multivariant mineral reactions. e calculations were performed using THERMOCALC. (Powell and Holland, ; updated June 00) and the internally consistent thermodynamic dataset of Holland and Powell (; dataset tcds, created in November 00). e calculations were undertaken in the chemical system Na O CaO K O FeO MgO Al O SiO H O TiO Fe O (NCKFMASHTO), which is currently the most realistic system in which melt-bearing equilibria can be calculated (White et al., 00, 00, 00). e phases considered in the modelling and the corresponding a x models include biotite and melt (White et al., 00), orthopyroxene and spinel magnetite (White et al., 00), garnet (Diener et al., 00), hydrous cordierite (Holland and Powell, ), osumilite (Holland et al., ), K-feldspar and plagioclase (Holland and Powell, 00) and ilmenite hematite (White et al., 000). e aluminosilicates, quartz and rutile are treated as pure end-member phases. e mineral abbreviations are as follows (after Kretz, ): biotite; cordierite; garnet; Ilm ilmenite (hematite proportionb0.); Kfs K-feldspar; Ky kyanite; Liq silicate liquid/melt; orthopyroxene; Pl plagioclase; Qtz quartz; Rt rutile; sillimanite; Spl spinel. e chemical composition of melanosome for each sample was determined by X-ray fluorescence spectroscopy at the Wadia Institute of Himalayan Geology (WIHG), Dehradun, India. e chemical compositions (Table 4) reflect the integrated effect of processes such as melt segregation into the melanosome and leucosome components in the migmatite and possibly loss of melt from the original protolith. erefore, these compositions are only appropriate for investigating the peak and post-peak evolution recorded in the melanosomes. e compositions contain negligible MnO (b0. wt.%, Table 4), which was not considered in the modelling. e appropriate H O content used in the modelling was evaluated for each sample using T M HO diagrams at and kbar, within the range of previous estimates of peak pressure (Dasgupta et al., ; Rickers et al., 00; Sarkar et al., 00). e H O contents range from a near-anhydrous composition (M HO =0, Table 4) to a composition with a solidus temperature below 00 C (M HO =, Table 4). Using these diagrams, the H O content was adjusted to stabilize the observed peak assemblage. Similarly, the appropriate O (for Fe + ) content to use for the P T pseudosections was evaluated using a T M O diagram at kbar. ese diagrams allow the affect of Fe O on the phase equilibria to be assessed over the range of all Fe as Fe + to all Fe as Fe +.

8 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 Table Summary of SHRIMP U Pb monazite results for samples SK--0 (SK) and D--S (D). Radiogenic ratios Age (Ma) Spot name Fig. U (ppm) (ppm) /U 0 Pb/ U ±% 0 Pb/ U ±% U/ 0 Pb ±% 0 Pb/ 0 Pb ±% ρ 0 Pb/ U ± σ Discord % D-M. (l), D-M., D-M., D-M.4, D-M., D-M. 4, D-M. (e) 4, D-M., D-M., D-M.4, D-M. (k), D-M. 4, D-M., D-M.4 40, D-M4. (g) 0 4, D-M4., D-M4. 0, D-M4.4, D-M4., D-M4., D-M. (i), D-M. 4, D-M. 04 0, D-M.4, D-M., D-M. (h) 4, D-M. 4, D-M. 4, D-M.4 0, D-M. (m) D-M. 4, D-M., D-M.4, D-M. 4, D-M. (m), D-M. (j) 4, SK-M. (a) 0, SK-M. 4, SK-M. 44, SK-M.4 0, SK-M., SK-M. 4, SK-M. 4, SK-M., SK-M. 4, SK-M. 4, SK-M. (b) 4 4, SK-M. 4, SK-M. 4, SK-M.4 44, SK-M., SK-M. 4, SK-M. 44, SK-M., SK-M., SK-M. 44, SK-M. 44, SK-M. 4, SK-M., SK-M. 4, SK-M. 4, SK-M. 4, SK-M.4, SK-M. 4, SK-M. 0, SK-M. 4, SK-M. 0, SK-M., SK-M. 4, SK-M4. (d), SK-M4. 0 0, SK-M4. 4, SK-M4.4 4, (continued on next page)

9 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 Table (continued) Radiogenic ratios Spot name Fig. U (ppm) (ppm) /U 0 Pb/ U ±% Age (Ma) 0 Pb/ U ±% U/ 0 Pb ±% 0 Pb/ 0 Pb ±% ρ 0 Pb/ U ± σ Discord % SK-M4., SK-M4. 0, SK-M4. 4 0, SK-M4., SK-M4. 4, SK-M4., SK-M. (f) 4, SK-M., SK-M., SK-M.4, SK-M. 4, SK-M. 4, Correction for common Pb made using the measured 04Pb/0Pb ratio; ρ=correlation between 0 Pb/ U and 0 Pb/ U ratios... Sample SK--0 e observed peak assemblage comprises garnet+orthopyroxene+ quartz+ilmenite+inferred melt. e rare inclusions of sapphirine in garnet are interpreted to be part of the prograde assemblage, which cannot be modelled using the melanosome compositions. At kbar, sillimanite is stable across the range of modelled H O contents at temperatures above 0 C (Fig. a). e sparse sillimanite observed in thin section (Fig. c) is interpreted to be post-peak phases, suggesting that the pressure estimate of kbar is likely too low. At kbar, the peak assemblage is stable at temperatures above C for M HO N0. (Fig. b). Biotite is stable at lower temperatures across the range of modelled H O contents, and at C, K-feldspar is stabilized at M HO 0.. At kbar, the temperature of the solidus decreases from ~0 C to less than 00 C at M HO contents above 0. (thick dashed line, Fig. b). Based on these constraints, a M HO value of 0. (~.4 mol% H O, Table 4; graybaron (a) SK Error ellipses are σ (b) D--S 0.0 Error ellipses are σ Pb/ 0 Pb Pb/ 0 Pb U/ 0 Pb U/ 0 Pb (c) SK--0 0 Error bars are σ (d) D--S 0 Error bars are σ In 0 Pb/ U age (Ma) Mean = ± Ma (n = ) MSWD = 0.4, probability = 0. 0 Pb/ U age (Ma) Mean = 4 ± 4 Ma (n = ) MSWD = 0.00 * * In : Mean = ± Ma (n = ) MSWD =., probability = 0. * * * * Fig.. Results of monazite geochronology. (a) Tera Wasserburg concordia diagram for analyses from sample SK--0. Error ellipses are σ. Gray-shaded ellipses with red outline are discordant analyses that were not included to calculate the weighted mean 0 Pb/ U age. (b) Tera Wasserburg concordia diagram of monazite analyses from sample D--S. (c) Plot of 0 Pb/ U ages for concordant monazite analyses from sample SK--0. Dashed box indicates analyses used to calculate the weighted mean age. e spread of younger ages are interpreted as evidence for diffusive lead loss. Error bars are σ; weighted mean age reported at σ uncertainty (gray bar). (d) Plot of 0 Pb/ U ages for monazite analyses from sample D--S. Dashed box indicates six spots from a single monazite grain included in orthopyroxene, whereas the other analyses are monazite grains associated with cordierite. Six analyses denoted by * and in red are discordant analyses that were not included to calculate the weighted mean 0 Pb/ U age. Error bars are σ; weighted mean ages reported at σ uncertainty (gray bar).

10 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 Table 4 Bulk compositions used for mineral equilibria modelling. Wt.% (from XRF whole rock analyses) Sample SiO Al O CaO MgO FeO K O Na O TiO Fe O MnO P O Total SK D--S Normalized molar proportion Sample Figure SiO Al O CaO MgO FeO K O Na O TiO O H O Total SK--0 a and b (0) and () c (0.) c () d D--S a and b (0) a and b () c (0.) c () d , not analyzed; total iron analyzed as Fe O ; (0), M (H O) =0; (), M (H O/O)=; (0.), M (O)=0.. Fig. a and b) was selected as an appropriate H O content. At higher H O contents, the temperature of the solidus is much lower than is appropriate for residual granulite facies assemblages, and at lower H Ocontents,Kfeldspar is predicted to be part of the peak assemblage. Using this H Ocontent,aT M O diagram was constructed at kbar (Fig. c). e changes in M O on Fig. c correspond to the range of possible O contents from % Fe as Fe + to 0% Fe as Fe + (Table 4). e observed peak assemblage is stable at temperatures above C between M O contents of 0. and 0. (Fig. c). Rutile is stable at lower and higher M O contents, which is not observed. Based on the predicted mineral assemblages across the range of Fe + contents, a M O value of 0. (=% Fe as Fe + ;graybaronfig. c) was selected to construct the P T pseudosection (Fig. d). is estimate is within the range of Fe + contents in similar samples from nearby localities in which FeO contents were analyzed by Fe + titration and Fe O contents calculated by difference (unpublished data). e P T pseudosection shows that the peak assemblage is stable at temperatures above 0 C and pressures above. kbar (Fig. d), although these estimates will vary slightly with changes in H O and O contents. e solidus occurs at temperatures higher than 00 C above. kbar (thick dashed line, Fig. d), which reflects the residual bulk composition after melt segregation and potentially loss of melt, and biotite is stable up to 0 C above the solidus. Plagioclase is stable at temperatures below ~00 C. e lower pressure limit of the peak assemblage is defined by the stability of sillimanite, and cordierite is stable at pressures below ~. kbar. Osumilite is not predicted in the modelling for this sample... Sample D--S e observed peak assemblage comprises orthopyroxene+spinel+ sillimanite+inferred melt. Ilmenite is predicted to be stable in variable proportions over a wide range of H OandOcontentsandP T conditions, suggesting that it is likely to be a peak phase. However, ilmenite has not been observed in the studied thin section, which may be due to slight differences between the samples used for bulk composition and the studied thin section, and/or the effect of components not included in the modelling, such as MnO. At kbar, the peak assemblage (+ilmenite) is stable within a temperature range of 00 0 C at M HO contents between 0.4 and 0. (Fig. a). At 00 C, plagioclase is stabilized at M HO 0., followed by K-feldspar at M HO 0.4.Biotiteisstableat lower temperatures across the range of modelled H O contents, and cordierite is stable at temperatures below 00 at M HO 0., but increases to higher temperature with increasing M HO.etemperature of the solidus decreases from ~0 C to less than 00 C at M HO contents above 0.0 (thick dashed line, Fig. a). e T M HO diagram at kbar predicts similar results, although cordierite is not stable over the range of modelled temperatures, and the stabilities of the cordieriteabsent assemblages are shifted to higher temperatures (Fig. b). In order to stabilize the peak assemblage and maintain an elevated solidus temperature, a M HO value of 0.0 was selected (~4. mol% H O, Table 4; gray bar on Fig. a and b). Plagioclase and K-feldspar are predicted to be part of the peak assemblage at lower H Ocontents. e T M O diagram at kbar constrains the peak assemblage to temperatures above 0 C between M O contents of 0. and 0. (Fig. c), with rutile stabilized at lower and higher M O contents. Based on these observations, a M O value of 0.0 (=0% Fe as Fe + ; gray bar on Fig. c) was selected to construct the P T pseudosection (Fig. d). e P T pseudosection shows that the peak assemblage is stable at temperatures above 0 C and pressures above ~. kbar (Fig. d), although variations in H O and O contents will affect these estimates (e.g. Fig. a, b and c). In addition, ZnO is not included in the modelling, and may extend the stability of spinel to lower temperature. Biotite is stable up to 0 40 C between. and kbar, and plagioclase is stabilized to slightly higher temperatures (0 C at. kbar to 40 C at kbar; Fig. d). Between and kbar, the temperature of the solidus ranges from 0 to 00 C, and decreases to b0 C at. kbar. Cordierite is stable at pressures less than. kbar above 0 C, extending to slightly higher pressures at lower temperatures, followed by the growth of K-feldspar at lower pressures. Osumilite is not predicted in the modelling for this sample.. Discussion.. Peak metamorphic conditions e inferred peak assemblages in the high Mg Al granulites from this study correspond to UHT temperatures in excess of 0 C. e assemblages do not constrain the maximum temperature, but the upper temperature limit for UHT metamorphism must be defined by the liquidus for crustal rocks, which is unlikely to be higher than 0 C (Brown, 00). Peak pressure is more poorly constrained. e results of phase equilibria modelling for sample SK--0 predict that garnet+orthopyroxene+quartz+ilmenite+melt is stable at pressures above. kbar (Fig. d). e peak assemblage comprising orthopyroxene+sillimanite+spinel+melt in sample D--S is stable above ~. kbar and 0 C (Fig. d), but the stability of spinel may extend to lower temperatures and pressures than predicted in the modelling. e Al content in orthopyroxene can potentially preserve a record of near-peak conditions (e.g. Aranovich and Podlesskii, ; Fitzsimons and Harley, 4; Pattison and Bégin, 4; Harley,

11 4 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 (a) NCKFMASHTO (+ Qtz + Ilm) 0 kbar Liq (b) NCKFMASHTO (+ Qtz + Ilm) 0 kbar 00 Kfs Liq Liq 00 Kfs Liq Liq 0 Kfs Liq 0 Kfs Liq Kfs Liq Liq Liq Kfs 0 Kfs Kfs Pl Kfs Pl M (H O) (c) NCKFMASHTO (+ Qtz + Ilm) Liq Rt A B 0 Pl Rt + Rt Liq Pl Liq Pl A. Liq Rt B. Pl Liq Rt Liq Pl Liq M (O) Pl Liq Liq Pl Liq Pl Rt kbar Liq Rt Liq Rt Liq Rt Pl Liq Rt + Rt M (H O) (d) NCKFMASHTO (+ Qtz + Ilm) A 0. Pl Pl Pl Pl Pl Liq Liq Liq F D Liq Liq Liq 0.0 C 0. Liq B 0.4 Pl Liq Pl Liq E G H Liq Pl Liq 0. SK % Fe as Fe + 0% 0% 0% 0% 0% Fe as Fe + A. Ky Pl B. Pl C. Pl Liq D. Liq E. Liq F. Pl Liq G. Liq H. Liq Fig.. Calculated pseudosections for sample SK--0. (a) T M HO diagram at kbar. (b) T M HO diagram at kbar. e gray bar indicates M HO content used for additional modelling based on observed equilibria. (c) T M O diagram at kbar. e gray bar indicates M O content used for P T pseudosection. e rutile-in line at high and low M O contents is denoted by a heavy black line, with rutile-bearing equilibria denoted by +Rt. (d) P T pseudosection based on the appropriate H O and O contents. e field for the peak assemblage is outlined in a thick line. e dashed line on all figures is the solidus, and hatched lines on Fig. d correspond to predicted y() values. Compositions used for modelling are listed in Table 4; and mineral abbreviations are after Kretz (). a,b, 00; Kelsey et al., 00; Baldwin et al., 00), and is contoured on the P T pseudosections in terms of y()=x Al,M () (dashed lines, Figs. d and d). e maximum y() values calculated by the ideal analysis approach (see Powell and Holland, 00, for an explanation), vary from 0. in sample D--S to 0. in sample SK--0 (Table ). For sample SK--0, the isopleth for y () =0. occurs in the peak field at pressures above ~ kbar (Fig. d). e isopleth for y() =0. for sample D--S does not occur in the peak field (Fig. d), suggesting an inconsistency between the recalculated orthopyroxene compositions and the modelling results. Although the ideal analysis approach constrains the range of Fe + contents in an orthopyroxene composition using robust statistics (Powell and Holland, 00), and likely provides a more realistic estimate of the P T conditions as compared to a Fe + -free composition (e.g. Harley, 4, ; Nandakumar and Harley, 000; Kelsey et al., 00), there is a wide range in possible Fe + contents (Table ). e difference between the minimum and maximum y() values for any of the orthopyroxene compositions correspond to a

12 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 (a) NCKFMASHTO (+ Ilm) Spl Pl Kfs Liq Pl Kfs Qtz Spl Pl Kfs Liq Pl Kfs Liq Pl Kfs Qtz Liq Spl Pl Liq Spl Pl Liq Pl Liq A Pl Qtz Liq kbar Spl Liq G C Pl Liq D F E B (b) NCKFMASHTO (+ Ilm) Spl Pl Kfs Liq Spl Pl Kfs Liq Pl Kfs Liq Pl Kfs Qtz Pl Kfs Qtz Liq Spl Pl Liq Pl Qtz Liq Spl Pl Liq kbar Spl Liq Spl Pl Liq Pl Liq 0 Pl Qtz A. Pl Qtz Liq B. Pl Liq C. Spl Pl Liq D. Spl Liq E. Spl Liq F. Spl Liq G. Spl Liq Pl Qtz Pl Qtz Liq Pl Qtz M (H O) 0 Ky Pl Qtz Ky Pl Qtz Liq M (H O) (c) NCKFMASHTO (+ Ilm) % Fe as Fe + kbar + Rt Spl Pl Liq + Rt Spl Pl Liq Spl Liq Pl Liq Pl Qtz Liq Ky Pl Qtz Liq M (O) 0% 0% 0% 0% 0% Fe as Fe + (d) NCKFMASHTO (+ Ilm) Ky Pl Qtz Liq 0. Pl Qtz Pl Qtz Liq Pl Qtz B Pl Kfs Qtz C Pl Liq Spl Pl Liq A Spl D Pl Spl Liq Liq E F H I Pl Kfs Liq G 0.0 Pl Kfs Mag Liq Spl Pl Liq Spl Pl Liq 0.0 Pl Kfs Mag Qtz Pl Mag Pl Mag Qtz Liq K A. Pl Qtz Liq B. Pl Kfs Qtz C. Pl Kfs Qtz Liq D. Pl Liq E. Pl Kfs Liq F. Spl Pl Liq G. Spl Pl Kfs Liq H. Spl PlLiq D--S 0. J I. Spl Liq J. Spl Liq K. Spl Pl Mag Liq Fig.. Calculated pseudosections for sample D--S. (a) T M HO diagram at kbar. (b) T M HO diagram at kbar. e gray bar indicates M HO content used for additional modelling based on observed equilibria. (c) T M O diagram at kbar. e gray bar indicates M O content used for P T pseudosection. e rutile-in line at high and low M O contents is denoted by a heavy black line, with rutile-bearing equilibria denoted by +Rt. (d) P T pseudosection based on the appropriate H O and O contents. e field for the peak assemblage is outlined in a thick line. e dashed line on all figures is the solidus, and hatched lines on Fig. d correspond to predicted y() values. Compositions used for modelling are listed in Table 4; and mineral abbreviations are after Kretz (). temperature difference of over 00 C (Figs. d and d), demonstrating the dependence of calculated temperature on Fe +. In addition, the presence of exsolved oxide lamellae (Fig. a), inferred to be ilmenite hematite, indicates that the integrated orthopyroxene compositions during peak metamorphism would have had higher Fe + contents than the measured compositions, which would correspond to lower y() values. ere are also uncertainties associated with the modelling (Holland and Powell, ; Powell and Holland, 00). For example, the isopleth for y()=0. in the garnet+orthopyroxene+quartz+ ilmenite+melt peak field (Fig. a) has a σ P uncertainty of 0. kbar and σ T uncertainty of C. Given these uncertainties, the P T constraints based on Al content in orthopyroxene must be evaluated cautiously, although the temperatures estimated from the Al content in orthopyroxene are consistent with UHT metamorphism in the Sunki locality... Mineral assemblage evolution and P T path In order to investigate the significance and possible P T implications of the observed mineral assemblages and microstructures that record the retrograde evolution, the P T pseudosections are contoured for the

13 F.J. Korhonen et al. / Gondwana Research 0 (0) 4 molar proportions of key phases (Figs. and ). e relative changes in the predicted mineral abundances (~molar proportions) are used to model mineral growth and consumption associated with multivariant mineral reactions, and can be used to link the observed microstructures with a possible P T trajectory (cf. Kelsey et al., 00; Diener et al., 00). e granulite facies assemblage preserved in the rock is inferred to reflect conditions at the solidus, with little modification following crystallization of any trapped melt (cf. Powell and Downes, 0; Brown, 00; White and Powell, 00; Štípská and Powell, 00; Diener et al., 00; Brown and Korhonen, 00; Korhonen et al., 0). e approach followed assumes that the melanosome composition has not been modified through open-system processes after the metamorphic peak, including any melt loss or melt/fluid influx. Further, segregation of the neosome into melanosome and leucosome components is likely to have occurred during prograde melt production. erefore, the melt predicted to be present during the post-peak evolution is interpreted to have been retained in the melanosome, so that the equilibration volume is represented by the analyzed composition. We also do not consider domainal compositions or the effects of diffusion or mineral growth rates or reaction kinetics on the outcome (cf. Vernon et al., 00). Despite these limitations, calculated phase diagrams within a realistic chemical system can be used to investigate the stability of mineral assemblages and changes in mineral proportions, and may provide a useful framework to interpret the evolution of a rock.... Sample SK--0 e P T pseudosection for sample SK--0 is contoured for mol% garnet (Fig. a), orthopyroxene (Fig. b), biotite (Fig. c), cordierite (dotted lines, Fig. c), and sillimanite (Fig. d). Petrographic observations reveal that orthopyroxene in direct contact with garnet Sample SK--0 (a) 0.4 (b) (c) (d) Fig.. Calculated contours for mineral proportions (in mol%) for sample SK--0. (a) Garnet; isopleth range is from 0.0 to 0. at an interval of (b) Orthopyroxene; isopleth range is from 0.0 to 0.4 at an interval of (c) Solid lines correspond to biotite; isopleth range is from 0.0 to 0. at an interval of Dotted lines correspond to cordierite; isopleth range is from 0.0 to 0.4 at an interval of (d) limanite; isopleth range is from 0.0 to 0.0 at an interval of 0.0. e mineral assemblages for all fields are shown on Fig. d. Garnet-absent equilibria are denoted by a heavy black line and light gray shading; labelled.