Mesoarchean melt and fluid inclusions in garnet from the Kangerlussuaq basement, Southeast Greenland

The present work reports the first anatectic melt in clusions found so far in the Mesoarchean basement in East Greenland. Using optical microscope observations and MicroRaman spectroscopy, we show that garnets in metasedimentary migmatite contain primary polycrys talline aggregates which can be confidently interpreted as former droplets of anatectic melt, i.e. nanogranitoids. In some cases, they coexist with coeval fluid inclusions under conditions of primary fluid-melt immiscibility. The re-evaluation of the metamorphic pressure and temperature conditions with up-to-date phase equi libria modelling, combined with the identification of nanogranitoids and fluid inclusions, suggests metamor phic peak equilibration and partial melting in presence of a COH-fluid at T ~1000°C and P > 7 kbar. To date, this is the oldest verified occurrence of nanogranitoids and fluid-melt immiscibility during garnet growth in a par tially molten environment.

stagnant-lid tectonics to mobile-lid tectonics (Palin et al. 2020). Because of its exceptional exposure, age and pristine aspect, the Mesoarchean lithologies exposed in the Uttental Plateau (Fig. 1a) represent an ideal natural laboratory to investigate crustal reworking processes and the remobilisation of volatile elements (e.g. H, C, N) in active continental margin during the onset of plate tectonics.

Geological settings
The Mesoarchean basement exposed in the Kangerlussuaq Fjord, Southeast Greenland, represents a part of the Nagssugtoquidian fold belt which has escaped Proterozoic reworking (Nicoli et al. 2018 and references therein). The basement was first described by Wager (1934), but remains overall relatively unexplored since then. The most extensive tectono-metamorphic study in the area of interest was conducted by Kays et al. (1989) who focused on the structural history and geochemical diversity of the different lithologies around the Watkins Fjord and the Skaergaard Intrusion, notably in the Uttental Plateau (Fig. 1a). More recently Holwell et al. (2013) have investigated gold-bearing quartz veins in exhumed supracrustal units 35 km north-northeast of the Skaergaard Intrusion. They described the presence of COH-fluids inclusions in quartz (300-350°C) indicating of the presence of important fluid fluxes associated with devolatilisation reactions during retrograde metamorphism. The basement comprises a fully exposed, typical Archean assemblage of felsic intrusions, TTG and grey gneisses, inter-layered with plurimetric lenses of metasediments and amphibolites, mostly concordant with the main NE-SW trending foliation ( Fig. 1b) (Kays et al. 1989). According to Kays et al. (1989) the supracrustal sequence can be divided in four categories: (1) biotite-garnet ± cordierite schists metapelites; (2) garnet ± hornblende ± orthopyroxene-bearing quartzites; (3) metabasic amphibolites and (4) ultramafic rocks pods. The presence of garnet porphyroblasts overprinting the foliation of some of the lithologies suggests that the regional metamorphic event, 650-700°C and 3-4 kbar, occurred in a relative static environment. The only geochronological data available in the area are whole rock 207 Pb/ 206 Pb and Rb-Sr analyses on the felsic basement, which give a TTG emplacement age and a regional metamorphic age of 2980 ± 20 Ma and 2860 ± 40 Ma respectively (Leeman et al. 1976;Kays et al. 1989).

Methods and sample petrography
The samples were investigated using a polarised light optical microscope (both in reflected and transmitted light). FI and former MI (i.e. nanogranitoids, Bartoli et al. 2016) were measured in eight garnets on one thin and one thick sections. The inclusions were analysed via Micro-Raman spectroscopy using a HORIBA Jobin-Yvon Confocal LabRAM HR 800 at the University of Potsdam (Germany), with grating 300 lines/mm, slit width 100 μm, confocal hole 200 μm (λ = 532 nm, laser power on sample: 2-3 mW). Spectra were acquired in the range 100-4000 cm −1 using 3 accumulations of 30 s each, with spectral resolution of 10 cm −1 . Garnet composition was constrained by EMPA using a JEOL JXA-8200 with an acceleration voltage of 15 kV, 15 nA beam current and a probe diameter of 2 μm. The sample 589590 investigated in this work is a garnet-bearing stromatic migmatite (i.e. described as garnet-quartzite in Kays et al. 1989) sampled from one of the metasedimentary lenses during a field expedition led by the University of Cambridge in 2017. The sampling site is far enough from the Skaergaard Intrusion to prevent any contact metamorphism overprinting (Bufe et al. 2014). The main mineral assemblage consists of garnet, biotite, feldspar, quartz and minor oxides, chlorite, apatite and zircons. Quartz and feldspar are segregated into leucocratic bands (i.e. leucosome - Fig. 1c, Fig. 2a) where feldspar shows serrated grain boundaries, undulatory exsolution and recrystallisation. The melanocratic parts instead mainly consists of garnet and partially chloritised biotite. The alternance of felsic and mafic bands defines the main fabric. Myrmekite occurs at the contact between the melanosome and the leucosome (Fig. 2b). We identified two types of garnets: large xenoblastic garnets (Grt 1 ), up to 5 mm in size, and small idioblastic garnets (Grt 2 ) overprinting the main fabric (Fig. 2c), both displaying locally abundant MI and FI (see following paragraphs).

Results of MI and FI investigation
Inclusions of different types, mostly fluid and nanogranitoids, were identified in garnet from the investigated sample. They form clusters in the inner part of the host, a typically primary distribution which support their interpretation as primary fluids trapped during garnet growth (see e.g. Ferrero et al. 2016), rather than infiltrated in cracks during a subsequent stage of rock evolution. FI and nanogranitoids can be distinguished in these samples exclusively based on their optical features under microscope observations. Whereas they are both generally isometric in shape (Fig. 2e,f), FI are generally up to 40 μm in diameter, very dark in appearance under transmitted light due to the presence of graphite and CO 2 (see below) -under crossed polars, carbonates can be generally distinguished due to their extreme birefringency. Nanogranitoids are generally smaller (up to 15 μm), have lighter colour in transmitted light and under crossed polars appear to be formed by an aggregate of birefringent phases. MI and FI occur in both types of garnets, Grt 1 and Grt 2 , which show different microstructural features (Fig. 2c). Large garnets (Grt 1 ) contain monomineralic inclusions of quartz, feldspar and biotite, along with scattered nanogranitoids randomly distributed in the garnet, a limited number of fluid inclusions and rutile needles (Fig. 2f). Small garnets (Grt 2 ) have clusters composed of mainly FI (up to 30 μm, Fig. 2d,e) along with a minor amount of nanogranitoids, at their core. In both gar- net types FI contain CO 2 + CH 4 + pyrophyllite +graphite +carbonates (mainly siderite, Fig. 3a,d), with quartz often present along with minor phlogopite. Neither N 2 or liquid H 2 O were identified within the FI during Raman investigation. Nanogranitoids in the two different types of garnet show different phase assemblages. In Grt 1 they contain an assemblage consisting of quartz/cristobalite + kokchetavite + kumdykolite ± phlogopite (Fig. 3b,c). In Grt 2 , they mostly consist of quartz +chlorite, with H 2 O locally present in one inclusion along with K-feldspar (Fig. 3e,f). Overall, the statistic distribution of the phases identified via Micro-Raman spectroscopy indicates that ~60% of the phases in Grt 1 are characteristic of nanogranitoids (mainly quartz, phlogopite, cristobalite) or are accessory minerals (zircon, apatite, monazite, rutile) (Fig. 3g). Conversely, Grt 2 contains both nanogranitoids and FI, but its inclusion population is clearly dominated by the latter. About 85% of the spectra acquired on Grt 2 show mineral phases associated to FI (carbonates, pyrophyllite, graphite) and volatiles species (CO 2 , CH 4 ), with the rest representing phases commonly found in nanogranitoids (e.g. quartz, feldspar and mica). Microprobe analyses (Table  1) show that both garnet types are in average Alm-and Prp-rich in the inclusion-bearing core with slightly variable Grs component (Grt 1 : Alm 70.8 Prp 23.4 Sps 1.9 Grs 3.9; Grt 2 : Alm 71.2 Prp 23.1 Sps 2.0 Grs 3.7 ), which becomes significant in the light of the results of the thermodynamic modelling on these rocks (see below).

Discussion and perspectives
Our study targeting MI and FI provides new data on the evolution of the metapelites of the Uttental plateau in   (Kays et al. 1980) in the MnNCKFMASHTO system. Stability field for the preserved metamorphic assemblage Grt + Bt + Crd + Pl + Kfs + Ilm + Mag + Sil ± Qtz + Liq. The star indicates the P-T conditions at which observed abundances match modelled ones. We assumed a ± 2vol.% error on the published abundance by Kays et al. (1989). (b) Calculated P-T pseudosection for 589590 in the MnNCKFMASHTO system. The stability field for the preserved metamorphic assemblage given by Grt 1 is Grt + Pl + Ilm + Qtz + Liq ± Mag ± Kfs ± Opx. The stability field for the preserved metamorphic assemblage given by Grt 2 is Grt + Pl + Ilm + Qtz + Liq + Mag + Kfs ± Opx ± Bt. The star indicates the conditions determined for KS97.  Kays et al. (1989) Southeast Greenland by merging results from host rock petrography, Micro-Raman spectroscopy, statistical distribution of the phases in inclusions in garnet and phase equilibria modelling. The first notable outcome is that these metasediments experienced partial melting during their metamorphic history. Previous estimates indicated peak metamorphic conditions at 650-700°C and 3-4 kbar. However, these calculations were solely based on Fe-Mg exchange thermobarometers (Kays et al. 1989 and references therein) and are then likely to reflect re-equilibration in the later stage of the retrograde path. Our P-T conditions indicate peak metamorphism at 950-1000°C and > 7 kbar and retrograde conditions of ~800°C and 6 kbar. Although located further to the north, these are similar to estimates Yakymchuk et al. (2020) obtained for the Akia Terrane (ca. 3.0 Ga) in West Greenland, > 800°C and < 9 kbar. Due to the lack of deformation and the presence of tonalitic magma, these authors suggested that such metamorphic conditions might reflect growth and differentiation of the crust in a stagnant-lid tectonic regime. Still in West Greenland, Kirkland et al. (2018) also showed that the Kangerlussuaq Supracrustal Belt in the Akia terrane has undergone partial melting and ductile deformation during a prolonged metamorphic event, 820-850°C and 8-10 kbar, between 2700-2857 Ma, consistent with a convergent margin in a mobile-lid tectonic regime. Hence, there is a need for more accurate pressure-temperature-time paths on the rocks exposed in the southeastern coast of Greenland. In addition, the recent discovery of 2900-3100 Ma orthogneisses in the Paleoprotoerzoic Rinkian Belt, West Greenland (Thrane, 2021) calls for a reevaluation of the extend of terrane boundaries between East and West Greenland, as well as Canada. A comprehensive picture of the evolution of the felsic basement and the supracrustal rocks in the Uttental Plateau would play a key role in connecting orogenic events on both sides of the icecap. The second crucial point is that garnets in these migmatites contain FI and nanogranitoids. The distribution of these inclusions in both Grt 1 and Grt 2 is primary, thus suggesting that both garnet types are peritectic in nature, i.e. they grew in presence of melt, likely as direct result of the partial melting reaction (Cesare et al. 2015). Moreover, the fact that FI and nanogranitoids occur in the same clusters testifies for the growth of both garnet types to have occurred under conditions of fluid-melt immiscibility (e.g. Ferrero et al. 2014;Carvalho et al. 2019;Gianola et al. 2021), in presence of a COH-fluid.
To date, this case study presents the oldest case study of nanogranitoids as well as the first verified instance of partial melting in presence of a COH-fluid in Mesoarchean crustal rocks.
Such COH-fluid is likely to be externally derived, as an internal origin would require the presence of graphite, which is notably absent as rock-forming mineral in the studied samples (see also discussion in Carvalho et al. 2019). Here, graphite is present exclusively in the FI where it can be interpreted as the result of post-entrapment respeciation of the COH fluid (Cesare et al. 2007).
Similarly, most of the mineral phases observed in the investigated FI (i.e. pyrophyllite, carbonates, quartz and phlogopite) were not present during inclusion formation but are to be considered as stepdaughter minerals resulting from the interaction of the trapped COH-fluid and the host/garnet during cooling, in analogy with the recent work of Carvalho et al. (2020). Raman spectroscopy and petrographic characterisation of the pairs host/inclusions allowed us to also draw more detailed conclusions on the suprasolidus history of the investigated sample. The phase assemblage visible in the nanogranitoid is clearly different between Grt 1 and Grt 2 . In Grt 1 the melt crystallises to quartz/cristobalite + kokchetavite + kumdykolite ± phlogopite, whereas in Grt 2 the nanogranitoids often contain chlorite (already found in primary, crack-free nanogranitoids by ) and occasionally H 2 O. This suggests that Grt 1 and Grt 2 contain two different melts, melt 1 and melt 2 respectively. Interestingly, the association nanogranitoids + rutile needles visible exclusively in Grt 1 was recently reported by Ferrero et al. (2021 a,b) in case studies of nanogranitoids formed at T ~1000°C, possibly pointing toward a higher formation T for the pair melt 1 /Grt 1 with respect to melt 2 /Grt 2 , as already suggested by the modelling. Further studies involving experimental re-homogenisation using a piston cylinder press (see e.g. Ferrero et al. 2021 a) will clarify in detail compositions and P-T formation conditions of the two different melts. We reiterate here the message that anatectic melt and fluid inclusions are quite common in granulitic rocks but have been majorly overlooked (Nicoli, Ferrero 2021). This aspect, combined with the preservation and full exposure of the outcrops, makes the southeastern coast of Greenland a perfect geological playground for researchers aiming to understand crustal growth, reworking and volatile recycling at the dawn of plate tectonics.