Using as a case study a granulite from the Kerala Khondalite Belt, India, we show that a former anatectic melt can be preserved as tiny (<25 μm) droplets within refractory minerals, in this case garnet. The melt is either fully crystallized as a Qtz-Ab-Kfs-Bt cryptocrystalline aggregate ("nanogranite"), or completely glassy in inclusions <15 μm. Both nanogranite and glassy inclusions have a peraluminous, ultrapotassic granitic composition that, in this case, does not correspond to a "minimum melt" and points to high melting temperatures, in agreement with the ultrahigh-temperature origin of the rock. This discovery indicates that peritectic minerals, growing during incongruent melting reactions, act as hosts for inclusions of anatectic melt, and that in the general case of slow cooling of the crust these inclusions will occur as nanogranite. Exceptionally, in the smallest inclusions, glass may be present due to inhibition of crystallization. Our results extend the frontiers of petrological and geochemical research in crustal melting, as the composition of natural anatectic melts can be directly analyzed rather than assumed. 20 μm 100 μm

The occurrence of crystallized and glassy melt inclusions (MI) in high-grade, partially melted metapelites and metagraywackes has opened up new possibilities to investigate anatectic processes. The present study focuses on three case studies: khondalites from the Kerala Khondalite Belt (India), the Ronda migmatites (Spain), and the Barun Gneiss (Nepal Himalaya). The results of a detailed microstructural investigation are reported, along with some new microchemical data on the bulk composition of MI. These inclusions were trapped within peritectic garnet and ilmenite during crystal growth and are therefore primary inclusions. They are generally isometric and very small in size, mostly £15 lm, and only rarely reaching 30 lm; they occur in clusters. In most cases inclusions are crystallized (nanogranites) and contain a granitic phase assemblage with quartz, feldspar and one or two mica depending on the particular case study, commonly with accessory phases (mainly zircon, apatite, rutile). In many cases the polycrystalline aggregates that make up the nanogranites show igneous microstructures, e.g. granophyric intergrowths, micrographic quartz in K-feldspar and cuneiform rods of quartz in plagioclase. Further evidence for the former presence of melt within the investigated inclusions consists of melt pseudomorphs, similar to those recognized at larger scale in the host migmatites. Moreover, partially crystallized inclusions are locally abundant and together with very small (£8 lm) glassy inclusions may occur in the same clusters. Both crystallized and partially crystallized inclusions often display a diffuse nanoporosity, which may contain fluids, depending on the case study. After entrapment, inclusions underwent limited microstructural modifications, such as shape maturation, local necking down processes, and decrepitation (mainly in the Barun Gneiss), which did not influence their bulk composition. Re-homogenized nanogranites and glassy inclusions show a leucogranitic and peraluminous composition, consistent with the results of partial melting experiments on metapelites and metagraywackes. Anatectic MI should therefore be considered as a new and important opportunity to understand the partial melting processes

With less than two decades of activity, research on melt inclusions (MI) in crystals from rocks that have undergone\ud crustal anatexis – migmatites and granulites – is a recent addition to crustal petrology and geochemistry.\ud Studies on this subject started with glassy inclusions in anatectic crustal enclaves in lavas, and then progressed\ud to regionally metamorphosed and partially melted crustal rocks, where melt inclusions are normally crystallized\ud into a cryptocrystalline aggregate (nanogranitoid).\ud Since the first paper on melt inclusions in the granulites of the Kerala Khondalite Belt in 2009, reported and\ud studied occurrences are already a few tens. Melt inclusions in migmatites and granulites show many\ud analogieswith theirmore common and long studied counterparts in igneous rocks, but also display very important\ud differences and peculiarities,which are the subject of this review. Microstructurally, melt inclusions\ud in anatectic rocks are small, commonly 10 μm in diameter, and their main mineral host is peritectic garnet,\ud although several other hosts have been observed. Inclusion contents vary from glass in enclaves\ud that were cooled very rapidly from supersolidus temperatures, to completely crystallized material in\ud slowly cooled regional migmatites. The chemical composition of the inclusions can be analyzed combining\ud several techniques (SEM, EMP, NanoSIMS, LA–ICP–MS), but in the case of crystallized inclusions the\ud experimental remelting under confining pressure in a piston cylinder is a prerequisite. The melt is\ud generally granitic and peraluminous, although granodioritic to trondhjemitic compositions have also\ud been found.\ud Being mostly primary in origin, inclusions attest for the growth of their peritectic host in the presence of\ud melt. As a consequence, the inclusions have the unique ability of preserving information on the composition\ud of primary anatectic crustal melts, before they undergo any of the common following changes in their way\ud to produce crustal magmas. For these peculiar features, melt inclusions in migmatites and granulites, largely\ud overlooked so far, have the potential to become a fundamental tool for the study of crustal melting,\ud crustal differentiation, and even the generation of the continental crust

Important advances have been made during the last 15 years in the study of melt inclusions in minerals from migmatites and granulites. Pioneer work on high temperature metapelitic anatectic enclaves in peraluminous dacites from SE Spain has shown that droplets of granitic melt can be trapped by minerals growing during incongruent melting reactions, and that the composition of such trapped melts can be representative of that of the bulk melt in the system during the anatexis of the rock. Therefore melt inclusions may represent samples of embryionic anatectic granite. In most cases, these melt inclusions define microstructures that are typical of primary entrapment, and show little or no evidence of melt crystallization upon cooling. Rather, the melt solidified to glass due to very fast cooling associated with the submarine extrusion of the dacites. Hence inclusions can readily be analyzed for major and trace elements by conventional methods such as the electron microprobe or by laser ablation-inductively coupled plasmamass spectrometry.Based on the results from these quite unusual anatectic enclaves, one would expect similar melt inclusions to be present also in more conventional, slowly cooled, regionally metamorphosed migmatite and granulite terranes. As a matter of fact, recent investigations confirm this hypothesis. Tiny (<25 μm) inclusions containing a cryptocrystalline aggregate of quartz, feldpars, biotite and muscovite have been found in garnet from the metapelitic granulites of the Keraka Khondalite Belt, as well as in garnet and ilmenite from metapelitic and quartzo-feldspathic migmatites from the Alps, Ronda and the Himalayas. Due to the grain-size, texture and chemical/mineralogical composition, these inclusions are called "nanogranites" and are interpreted to represent a crystallized inclusion of anatectic melt. Exceptionally and spatially associated with the nanogranites, inclusions containing glass have also been observed. In general, the preparation of the samples and analysis of these inclusions in migmatites and granulites require more sophisticated techniques than those applied to inclusions in xenoliths and enclaves, but the information on the composition of crustal anatectic melts can also be obtained.Since its discovery, new occurrences of nanogranite are being reported, or can be inferred from re-assessment of literature data, from migmatites and granulites worldwide. These former melt inclusions open new perspectives both for the microstructural approach to partially melted rocks and for the chemical characterization of natural crustal melts.Citation: 2011. Melt inclusions in migmatites and granulites. In: (Eds.)