Multiphase-solid fluid inclusions in HP-LT eclogite facies rock (Zavkhan Terrane, Western Mongolia): evidence for the evolution from saline to hypersaline fluids during metamorphism in subduction zone

Fluid inclusions in high- and ultrahigh-pressure metamorphic rocks provide direct information on the composition of the fluids that evolved during metamorphism and fluid-rock interactions in deep subduction zones. We investigate the fluid inclusions in the Khungui eclogite of the Zavkhan Terrane, Central Asian Orogenic Belt. Fluid inclusions are observed in garnet and quartz in the eclogite samples that underwent metamorphism during subduction. The primary fluid inclusions in quartz are composed of liquid and vapor with high salinities (15.7–16.4 wt.% NaCl eq.), whereas the secondary fluid inclusions in quartz are classified as: relatively high salinity (Type I:12.5–16.3 wt.% NaCl eq.) and low salinity (Type II:6.7–10.6 wt.% NaCl eq.). The garnet shows compositional zoning from Ca-poor cores to Ca-rich rims, and the rims that grew during the eclogite-stage metamorphism (2.1–2.2 GPa at 580–610 °C) preferentially contain numerous primary fluid inclusions. The primary fluid inclusions in garnet are commonly bi-phases (liquid and vapor); however, some are multiphase-solid fluid inclusions composed of fluids (liquid and vapor) and combinations of several minerals (halite, quartz, apatite, calcite, biotite, chlorite, and actinolite). Bi-phase fluid inclusions preferentially occur in the inner parts of the Ca-rich garnet rim, whereas multiphase-solid fluid inclusions occur along the margins of the Ca-rich rim. We hypothesize that the multiphase-solid fluid inclusions are formed via interactions between trapped fluids, trapped minerals, and the host garnet during exhumation. By combination of FIB–SEM and synchrotron X-ray CT analyses, the detailed occurrences, volumes, and compositions of the solid phases in the fluid inclusion was analyzed. We then conduct mass balance analysis to reconstruct accurate fluid compositions using data from the FIB–SEM and synchrotron X-ray CT images of the multiphase-solid fluid inclusion. The results of these analyses reveal that (1) fluid changed from an H2O-dominated saline fluid (13–16 wt. % NaCl eq.) at the prograde to the earlier eclogite stage to H2O–CO2-dominated hypersaline fluid at later eclogite stage (~ 32 wt. % NaCl eq., 7.3 wt. % CO2 and ~ 19 molal dissolved cations); (2) a variety of mineral assemblages in multiphase-solid fluid inclusions are produced by post-entrapment reactions between the trapped hypersaline fluid, trapped minerals and the fluid host mineral. The evolution of fluids from saline to hypersaline during the eclogite facies stage is probably caused by the formation of hydrous minerals (i.e., barroisite) under a near-closed system.

The presence of multiphase-solid fluid inclusions in the HP and UHP rocks confirms that multi compounds are present in deep fluids within subduction zones.Such fluid inclusions with variable salinities (~ 13-60 wt.% NaCl eq.) have been reported in UHP metamorphic rocks in metamorphic belts such as the Dabie-Sulu orogenic belt in China (Shen et al. 2003;Ferrando et al. 2005b, a;Zhang et al. 2005Zhang et al. , 2008)), the Tauern Window, and the Dora-Maira massif in the European Alps (Selverstone et al. 1992;Maffeis et al. 2021), and the Western Gneiss Region in Norway (Svensen et al. 1999;Svensen and Museum 2001).Mineralogical and petrological studies of multiphase-solid inclusions in UHP metamorphic rocks have revealed that most minerals present in the fluids contain a diverse range of major elements (Si, Al, Mg, Fe, Ca, Na, and K) and trace elements (Ti, Zr, Ba, and high field strength elements), that form silicate minerals such as mica, amphibole, and epidote (Frezzotti and Ferrando 2015).Such inclusions rarely contain silica minerals, diamonds, phosphates, carbonates, sulfates, or chlorides (Stöckhert et al. 2001;van Roermund et al. 2002).
Several studies have tried to reconstruct the compositions of the trapped fluids observed as multiphase-solid fluid inclusions in HP and UHP rocks found in various regions, for example, the Dabie-Sulu orogenic belt in China (Xiao et al. 2000;Fu et al. 2001;Ferrando et al. 2005a, b;Zhang et al. 2005;Frezzotti et al. 2007;Jin et al. 2023), the Dora-Maira massif in Italy (Philippot et al. 1995;Ferrando et al. 2009Ferrando et al. , 2017;;Maffeis et al. 2021), the Kokchetav massif in northern Kazakhstan (De Corte et al. 1998;Hwang et al. 2003;Dobrzhinetskaya et al. 2005;Korsakov et al. 2011;Mikhno et al. 2014;Stepanov et al. 2014Stepanov et al. , 2016) ) and the Erzgebirge massif in Germany (Hwang et al. 2001(Hwang et al. , 2006;;Stöckhert et al. 2001Stöckhert et al. , 2009;;Dobrzhinetskaya et al. 2003b).However, quantitative reconstruction of fluid compositions from the multiphase fluid inclusions is not always easy due to two possible reasons.First, the formation of solid phases in fluid inclusions may not be due to precipitation from aqueous fluids or crystallization from hydrous silica melts (Ferrando et al. 2005a), but may have formed during reactions with host minerals after fluid trapping (Philippot and Selverstone 1991;Selverstone et al. 1992;Scambelluri et al. 1998;Svensen and Museum 2001;Perchuk et al. 2005Perchuk et al. , 2008;;Korsakov et al. 2011;Korsakov 2018).Therefore, detailed studies on the microstructural relationships of these rocks are required to determine how these minerals were formed and whether any reactions occurred within the inclusions.Second, previous attempts to quantitatively reconstruct fluid compositions were mostly based on 2D-observations using optical microscopes, which generated significant errors in the volume estimates of the individual phases used for mass balance analyses (Ferrando et al. 2005a;Frezzotti et al. 2007).Recently, 3D-phase analyses of the multiphase fluid-and void-bearing inclusions has become possible using micro-Raman imaging, which can even be used on high-grade metamorphic rocks (e.g., Dabie-Sulu; Jin et al. 2023); however, the spatial resolution of Raman imaging is limited to several hundred nanometers, which makes it difficult to analyze small (< 10 μm) inclusions that are common in metamorphic rocks.
In this paper, we report the results of our investigation of fluid inclusions in the Khungui eclogite, found in the northern part of the Central Asian Orogenic Belt (CAOB).Numerous primary and secondary fluid inclusions are present in the garnet and quartz of this eclogite.Moreover, various types of multiphase-solid fluid inclusions occur in compositionally zoned garnets, the rims of which correspond to the eclogite stage of metamorphism.Based on the textures, mineralogy, and microthermometry of fluid inclusions, we determine the variations in salinity and fluid composition in the Khungui eclogite at various stages of its metamorphic history.Furthermore, this is the first report of using 3D-Xray CT images of multiphase-solid fluid inclusion in garnet to reconstruct the fluid compositions during the eclogite stage of metamorphism.We use these new findings to speculate on the possible mechanisms underlying the mineralogical variation in these multiphase-solid fluid inclusions.

Geological setting
The CAOB is a giant Paleozoic orogenic belt that extends east-west from the Caspian Sea to the northern west Pacific Ocean (Fig. 1a).Most researchers argue that four main subduction systems were involved in the development of the CAOB, during the Neoproterozoic and the early, middle, and late Paleozoic eras (Safonova et al. 2017).The timing of these events was confirmed by the ages of the HP and UHP metamorphic rocks in the northern and southern parts of the CAOB (Bayarbold et al. 2022).Based on protolith compositions and typically exhumed HP and UHP rocks, oceanic plate subduction was dominate in the CAOB, but the continental subduction can explain the tectonic setting of a few HP and UHP metamorphic rocks, such as the Kokchetav massif and Muya eclogites (Zhang et al. 2012;Skuzovatov et al. 2019).Direct evidence of deep fluids in these subduction zones has been limited in the metamorphic rocks of the CAOB, except for the Kokchetav massif (De Corte et al. 1998;Hwang et al. 2006;Korsakov and Hermann 2006;Korsakov et al. 2011) and western Tianshan (Gao and Klemd 2001;Yoshida et al. 2018).
Mongolia is situated in the central part of the CAOB (Şengör et al. 1993).In western Mongolia, the Zavkhan Terrane is bordered by the Sangelin metamorphic terrane to the north, the Neoproterozoic-early Cambrian Lake island to the west, and the Dariv metamorphic terrane to the south (Fig. 1a; Badarch et al. 2002).The Zavkhan Terrane is a single cratonic terrane (Badarch et al. 2002;Bold et al. 2016), however, Kozakov et al. (2021)  The Khungui eclogite is found in the southeastern part of the Zavkhan Terrane's Mélange zone (Fig. 1b, c), and is classified into high-and low-grade rocks (Kovach et al. 2021).The high-grade rocks consist of felsic gneiss, quartzite, garnet amphibolite, and eclogite, and have been intruded by a complex of late Paleozoic alkaline intrusive rocks.The Khungui eclogite forms lenses or interlayers within the felsic gneisses and quartzites of the high-grade rocks found in the Based on petrological studies and mineral analysis of the Khungui eclogite, Bayarbold et al. (2022) identified three metamorphic stages: (i) prograde, (ii) eclogite, and (iii) decompression.The mineral assemblage of the eclogite facies occurs as inclusions within the high-Ca rims of the garnet, and consists of omphacite, barroisite, phengite, rutile, epidote, and quartz (Fig. 2a).Based on the Cpx-Grt (Ravna 2000) and Zr-in-rutile (Tomkins et al. 2007) thermometers, and Grt-Cpx-Ph barometer (Ravna and Terry 2004), the peak pressures during metamorphism of the Khungui eclogite were estimated to be 2.1-2.2GPa, at 580-610 °C (Bayarbold et al. 2022).Moreover, the trace element characteristics of the Khungui eclogite indicate that the protolith is a continental arc basalt, which differs from the protoliths of other HP and UHP metamorphic rocks in the CAOB (i.e., in Mid ocean ridge basalt (MORB) and Ocean island basalt (OIB); Bayarbold et al. 2022).

Electron probe micro-analysis (EPMA)
The chemical composition of the minerals in the samples were determined using an electron microprobe analyzer (EPMA; JEOL JXA-8200) at Tohoku University, Japan.The acceleration voltage used was 15 kV, beam current 12 nA, and beam diameter 1 μm.Natural and synthetic standards (wollastonite for Ca and Si, rutile for Ti, eskolaite for Cr, hematite for Fe, manganosite for Mn, periclase for Mg, albite for Na, feldspar for K, halite for Cl, and fluorite for F) were used for calibration.The counting time for the peaks and background of all the major elements were 10 and 5 s, respectively, whereas for the minor elements Cl and F they were 30 and 15 s, respectively.X-ray mapping using EPMA was conducted with an accelerating voltage of 15 kV, beam current 120 nA, dwell time 150-200 ms, and beam diameter 1 μm.

Fluid inclusion microthermometry
A doubly polished thick section (~ 100 µm thick) of the Khungui eclogite sample was prepared for fluid inclusion analysis.Microthermometry was performed at Tohoku University using a Linkam THMS600 heating/freezing stage, with an operating temperature range of -180 to 600 °C.The homogenization and ice melting temperatures were determined at a heating rate of 1 °C/min and 0.2 °C/min, respectively.The salinity of the fluid inclusions was measured in NaCl wt.% equivalent (eq.) using the final melting temperature of ice (Bodnar 1993).The fluid inclusion isochores were determined using the equation described by Bodnar (2003).

Combined FIB and X-ray CT analysis
To determine the volume proportion of each phase, highresolution three-dimensional images of multiphase-solid fluid inclusion in garnet were obtained using Synchrotron radiation nanoscopic X-ray CT (SR-XCT).A small particle containing fluid inclusion was selected from a thin section of the eclogite sample, for SR-XCT observation using a scanning electron microscope (SEM) with a focused ion beam (FIB), as described by Yoshida et al. (2016).A specific area of the sample (20 × 20 µm 2 ) was cut out to depth of 20 µm and held at the tip of a tungsten needle.The FIB analysis was performed using a Helios NanoLab G4 system (Thermo Scientific) at the Japan Agency for Marine-Earth Science and Technology (JAMSTEC), Japan.The analyses employed a Ga + gun at 30 kV voltage and 0.09-20 nA current.
SR-XCT measurements were performed on the NW2A beamline of the Photon Factory Advanced Ring Synchrotron facility at the Institute of Materials Structure Science, High Energy Accelerator Research Organization (KEK), Japan (Niwa et al. 2019).The voxel size and X-ray beam energy were maintained at 24.2 nm and 7118 eV, respectively.The X-ray CT measurements were performed by rotating the test stage (− 90° to 90°).Consequently, 1441 radiographs were obtained with an exposure time of 2 s (total scan time of approximately 1 h).
FIB-SEM analysis provides the opportunity to directly analyze the coexisting solid phases in a fluid inclusion by excavating the inclusions (Yoshida et al. 2018).After the SR-XCT observations, the composition of the solid phases in the multiphase-solid fluid inclusion was determined using Energy-Dispersive X-ray Spectrometry (EDS).The microsample held on the tungsten needle was machined using the FIB system at 30 kV voltage and 9.1 nA current until the multiphase-solid fluid inclusion reached the surface.The EDS analysis was performed at an acceleration voltage of 15 kV and beam current of 1.6 nA.The X-ray spectra were measured such that the exposed surface became horizontal.

Microanalysis using Raman spectroscopy
Raman spectroscopy was used to identify the fluids and solids in the garnet and quartz inclusions.The analyses were performed using a HORIBA XploRA Plus Confocal Raman Microscope at the Graduate School of Environmental Studies, Tohoku University, Japan.The laser beam was positioned using a built-in Olympus BX optical polarizing microscope.The spectrum acquired after excitation with a 532 nm solid-state wavelength was measured with a CCD detector cooled to -70 °C by a Peltier cooling system.The analyses were performed in a backscattered geometry, and the scattered light was focused using a 100 × lens.The spectra were obtained in the range of 100-4200 cm -1 .The spectrometer was calibrated using a pure silicon line (520.7 cm -1 ) and an intrinsic laser line (0 cm -1 ).The spectral resolution of the diffraction grating was 1800 g/mm.Measurements were carried out with 100-300 μm apertures.The counting time and number of counting cycles within a spectral window varied from five cycles of 5 s to twenty cycles of 20 s.Detailed analytical condition of 2D Raman mapping is mentioned in Table S1.The scanning step was set at 0.5 × 0.5 μm.The exposure time was set to 3 s, with three acquisitions per point.

Petrography
Five eclogite samples are analyzed; the mineral assemblages of the samples are presented in Table 1.The mineral abbreviations used in this study are obtained from Whitney and Evans (2010).The composition of the main minerals of the Khungui eclogite has already been documented (Bayarbold et al. 2022).Therefore, we focus our study on fluid inclusion-hosted minerals and minerals found in fluid inclusions.All chemical and representative microprobe data for the minerals are listed in Table 2.

Ti-bearing minerals
The primary Ti-bearing minerals are Rt, ilmenite (Ilm), and titanite (Ttn).These minerals occur together in the matrix, whereas only Rt and Ttn occur as inclusions in the Ca-rich rims and Ca-poor cores of the Grt, respectively (Figs. 3a,  4a).In the matrix, Rt is commonly surrounded or overlapped by Ilm and Ttn; however, the order of crystallization is unclear in some instances (Fig. S2a).The Rt in the matrix contains inclusions of zircon (Zr) and Ttn.

Apatite
The Ap occurs in the matrix and as inclusions in Grt.The Ap in the matrix is subhedral with a grain size of 50 μm, which commonly occurs with Ti-bearing minerals (Fig. S2a).In addition, Ap shows a compositional zoning in Cl, which is characterized in the matrix by an increased amount from the core (Cl = 0.10 wt.%) to the rim (Cl = 0.12 wt.%) (Table 2; Fig. S2b).Inclusions of Ap commonly occurs with Omp in the Ca-rich rims of Grt (Fig. S2b).Ap (Cl = 0.25 wt.%) also occurs within the exposed multiphase-solid fluid inclusions in the Grt (Fig. 4b).The Cl content of Ap varies with its occurrence; for example, matrix Ap has a much lower Cl content (0.10-0.12 wt.%) than the Ap inclusions in Grt (0.25 wt.%) (Table 2).

Characteristics of fluid inclusions
Detailed analysis of one eclogite sample (ZM10) reveals various types of fluid inclusions (FIs) in both Grt and Qz (Fig. 5; Table 1); however, the other key minerals (Omp, Amp, and Pl) of the Khungui eclogite do not contain FIs.

Fluid inclusions in quartz
The Qz grains in the matrix contain both primary and secondary FIs.The primary FIs (< 5 μm) in matrix Qz consist of an aqueous solution, vapor, and rarely halite (Hl) crystals (Fig. 5a).HI crystals mostly occur in irregularly shaped primary FIs, whereas the HI-free primary FIs in Qz had ellipsoidal shapes (Fig. 5a).Analyses of the photomicrographs (Fig. 5a) reveal that the Hl-bearing primary FIs in Qz are composed of 3 vol.% vapor, 20 vol.% Hl, and 77 vol.% liquid; the HI-free primary FIs are composed of 10-19 vol.% vapor, and 81-90 vol.% liquid.Secondary FIs in Qz are subdivided into two types based on their occurrence.Type I secondary FIs are aligned along the healed cracks with random orientations and consist of two phases at room temperature: a vapor and an aqueous solution (Fig. 5a).Type II secondary FIs occur in close association with micro-vein (20-50 µm wide) is composed of K-feldspar (Kfs) and calcite (Cal).The micro-vein (Kfs + Cal) intersects the matrix minerals such as Grt, Omp, Amp, and Qz (Fig. S1a, c).Type II secondary FIs are aligned at low angles to the micro-veins (Fig. S1b) and composes of liquid and vapor phases.The sizes of primary and secondary FIs in the Qz varies from 1 to 4 μm.
Raman spectroscopy shows that the liquid parts of the FIs in Qz exhibit a broad peak in the range of 3100-3600 cm -1 , indicating the presence of H 2 O. Peaks corresponding to gases such as CO 2 and CH 4 are not detected in the vapors from the HI-free primary FIs and Type I and Type II secondary FI in Qz (Fig. 5b).Detecting the vapor within HI-bearing FI in Qz using Raman spectroscopy is difficult to determine due to its tiny size (< 0.5 μm).Therefore, it is still uncertain whether vapor hosts gas phases or not (Fig. 5a).

Fluid inclusions in garnet
The Grt crystals contain only primary FIs; these primary FIs are classified into two types: bi-phase and multiphase-solid FIs.The most common primary FIs in Grt are bi-phase FIs (1.5-5.5 μm), which contain a vapor and an aqueous fluid without minerals (Figs.3e, 5c; Table 3).Multiphase-solid FIs are relatively rare, typically larger than 9 μm (Figs.3e, 5e, 6; Table 3), and show irregular shapes; they contain a vapor and an aqueous liquid with several solid phases (Figs. 6 and 7).No other volatiles compounds (N 2 , CO 2 , or CH 4 ) are detected in either the bi-phase or multiphase-solid FIs of Grt.The bi-phase FIs (D = 0.07-0.37)are presented in both the Ca-poor Grt core and the inner part of the Ca-rich Grt rim, whereas the multiphase-solid FIs (D = 0.42-0.53)are preferentially observed at the outer part of the Ca-rich Grt rim (Figs.3e, f, h; Fig. S3).'D' is the normalized value of the absolute distance between the center and crystal edge of Grt.
The vol. % for the vapor and liquid phases in the bi-phase FI are estimated to be 9.7-22.1% and 77.9-90.3%,respectively.Moreover, the bi-phase FIs exhibit tubular shapes (Figs.3e and 5c).

Microthermometry
Microthermometry is performed on HI-free primary and secondary FIs in Qz and primary FIs in Grt (Table 3).However, the amount of microthermometric data obtained in this study is limited by: (1) the small size of the common FIs (most FIs are smaller than 5 μm) and (2) the appearance of large solid phases along with fluids in the inclusions.

Fluid inclusions in quartz
Most primary FIs and all secondary FIs in Qz contain two phases, liquid, and vapor, at room temperature (Fig. 5a; Fig. S1).The ice melting temperature (T mt ) of the primary FIs in Qz ranged from − 11.7 to − 12.5 °C, corresponding to a salinity of 15.7-16.4wt.% NaCl eq. (Fig. 8a; Table 3).The rare primary FIs in Qz include cubic Hl crystals (Fig. 5a), indicating that the salinity of the primary FI is higher than 26.3 wt.% NaCl eq.(Roedder 1984).The eutectic melting temperature for three Hl-free primary FIs in Qz is − 32.8 °C, which is similar to the eutectic melting temperature of the MgCl 2 -H 2 O system (Table 3) (Davis et al. 1990;Samson and Walker 2000).The homogenization temperatures (T h ) of vapor and liquid in HI-free primary FIs ranged from 136 to 196 °C (Fig. 8a, b; Table 3).
Type I and II secondary FIs in Qz show contrasting salinity ranges (Fig. 5a).The T mt values of the type I secondary FIs ranged from -8.7 to -12.4 °C, corresponding to a salinity of 12.5-16.3wt.% NaCl eq., whereas the T mt values of type II secondary FIs range from -4 to -7.1 °C, corresponding to a salinity of 6.4-10.6 wt.% NaCl eq. (Fig. 8a; Table 3).T h of type I secondary FIs in Qz range from 117 to 181 °C.The eutectic melting temperature of 36 °C for the type II secondary FI in Qz may contain measurement errors due to its geometry and size.However, this low eutectic melting behavior may be attributed to the solute of divalent cations such as Mg or Fe (Table 3;Borisenko 1977;Davis et al. 1990).

Timing of fluid trapping for various types of fluid inclusion in the Khungui eclogite
The primary FIs in Qz and Grt isochores intersect the inferred pressure and temperature (P-T) paths of the Khungui eclogite, between 0.20 and 0.75 GPa and 580-610 °C which is much lower compared to the eclogite facies P-T conditions (Fig. 9a).This suggests that the density could have changed owing to plastic deformation of the host phase after fluid trapping (Küster and Stöckhert 1997).Therefore, in this study, we discuss the entrapment stage of FIs based on the petrographic description of the FI-hosting minerals.
The mineral inclusions (Omp, Brs, Rt, Qz, and Ep) indicate that the Ca-rich rim of Grt formed at the eclogite facies conditions (Figs. 3a, 4a), with estimated P-T conditions at 2.1-2.2GPa and 580-610 °C (Bayarbold et al. 2022).The bi-phase FIs occur in both the Ca-poor core and inner parts of the Ca-rich rim of Grt, whereas the multiphase-solid FIs occur only in the outer parts of the Ca-rich rims (Fig. 3e,  h).Such distinct spatial distributions of the two types of FIs indicate that bi-phase FIs were trapped in Grt during the prograde stage to the early eclogite-stage of metamorphism, whereas multiphase-solid FIs were trapped during the later stages of eclogite facies conditions (Fig. 9a).
In contrast to the FIs in Grt, it is difficult to constrain the timing of formation of primary FIs in Qz, because Qz is stable over the entire range of P-T conditions estimated for  14) Svensen et al. (1999Svensen et al. ( , 2001)).Dotted lines present isochore calculations of primary fluid inclusions in the studied sample.The black solid line presents a wet solidus of basalt (Kessel et al. 2005).The this sample (Fig. 2a).The primary FIs in the Qz matrix has a high salinity (15.7-16.4wt.% NaCl eq.), which is identical to the salinity of the bi-phase FIs in Grt (13-16.1 wt.% NaCl eq.), except for the additional occurrence of HI in a limited number of FIs in Qz.Therefore, HI-free primary FIs in Qz can be trapped in the same metamorphic stage as bi-phase FIs in Grt (Fig. 9a, b).The limited HI-bearing FIs in Qz could be related to the partial water leakage or water trapped simultaneously with hypersaline fluid, corresponding to multiphase-solid FIs in Grt, as described in a later section.
The observed occurrences and microthermometric characteristics (12.5-16.3wt.% NaCl eq.) indicate that the type I secondary FIs in the Qz matrix were formed after eclogite stage metamorphism (Fig. 5a; Table 3).Therefore, the first high salinity event (13-16.4wt.% NaCl eq.) of fluids occurs at greater depths (~ 70 km) that corresponds to the eclogite facies metamorphism (2.1-2.2GPa at 580-610 °C), whereas the second relatively high salinity (12.5-16.3wt.% NaCl eq.) of fluids are found during the exhumation of the Khungui eclogite (Fig. 9a).In contrast of Type I, Type II secondary FIs with micro-veins may represent fluid infiltration into the Khungui eclogites at later stages of exhumation because that fluid system of Type II secondary FIs ((Fe, Mg) Cl 2 -H 2 O) in Qz is completely different from the primary FIs (MgCl 2 -H 2 O) in Qz and FIs in Grt (Table 3).Moreover, the salinity of Type II secondary FIs in Qz is significantly lower than that of Type I secondary FIs in Qz.

Formation of the multiphase-solid fluid inclusions in garnet
The following three scenarios have been hypothesized as the origin of multiphase-solid mineral inclusions or multiphasesolid FIs, in the anhydrous minerals of HP and UHP rocks (Ferrando et al. 2005a).
In this study, the peak P-T conditions (2.1-2.2GPa at 580-610 °C; Bayarbold et al. 2022) were much lower than the second critical endpoint of the basalt water system (~ 5.2 GPa at 1050 °C; Kessel et al. 2005) or the wet solidus of the basaltic system (~ 2 GPa at 710 °C; Kessel et al. 2005) (Fig. 0.9a).In addition, the multiphase-solid FIs in Grt show a variety of mineral assemblages, including Qz, Cal, Qz + Cal, Amp + mica, Ap + Amp + mica, C a l + A m p + m i c a , Q z + A m p + C a l + H l , a n d Cal + Amp + mica + Chl + Hl, in addition to the fluid phase (Figs. 4,6,and 7;Fig. S4; Table 1).Such variations cannot be explained by the crystallization of a single-phase supercritical fluid or melt (hypothesized scenarios (i) or (ii)).
Accordingly, we consider that in scenario (iii), the insitu reactions between the coexisting minerals and aqueous fluids most likely formed the mineralogy of the multiphase-solid FIs observed in Grt.However, the variation in mineralogy (Table 1) and high solid-to-fluid volume ratio could not be explained by simple reactions between aqueous fluids and the host Grt.Therefore, we consider the origin of the observed minerals in the inclusions to be either trapped, daughter, or step-daughter mineral phases (i.e., the product of the inclusion fluid and host and/or coexisting mineral reactions; Svensen et al. 1999).
Ap and Qz commonly occur in the matrix and as mineral inclusions in the Grt (Figs. 2b, c, 3; Table 1; Fig. S4).In addition, variable proportions of Ap and Qz exist in the multiphase-solid FIs (Ap of size 3-10 μm; Qz of size 5-20 μm; Fig. 4a; Fig. S4).These occurrences indicate that Ap and Qz are trapped synchronously with fluid (Fig. 9b).However, the possibility that these minerals grew or dissolved in aqueous fluids cannot be excluded.
The Hl crystals in the multiphase-solid FIs of Grt are euhedral (Figs.5a and 6; Fig. S4), suggesting that Hl precipitated directly from the trapped fluid; the results were similar to those commonly reported for other HP and UHP metamorphic rocks (Philippot and Selverstone 1991;Selverstone et al. 1992;Svensen et al. 1999;Svensen and Museum 2001;Zhang et al. 2005;Xiao et al. 2006).
The Cal crystals in the multiphase-solid FIs in Grt are mostly anhedral (Fig. 6, Fig. S4).The SR-XCT images show that Chl and Cal are intercalated along the walls of the host Grt (Fig. 7b, c), suggesting that Chl and Cal might have formed simultaneously via the replacement of Grt.However, the Chl + Cal assemblage is not observed in the 2D view of the multiphase-solid FIs (Table 1; Fig. S4); this could be related to the cutting effect or to the reaction product (Chl) being too minuscule to be observed.
Act and Bt occur exclusively in the multiphase-solid FIs in Grt, but do not occur as matrix minerals or fluid-absent mineral inclusions in the analyzed samples (Figs. 2b,2c,3; Table 1).Therefore, we consider that these minerals were formed via hydration reactions between fluids and minerals that were simultaneously trapped in Grt.Some experimental studies have suggested that fluid-mineral interactions might be as effective and fast as high-salinity fluids, compared to low-salinity fluids (Huang et al. 2023).Thus, we conclude that the lack of reaction products or precipitates in the bi-phase FIs in Grt could be attributed to the relatively low salinity (13-16 wt.% NaCl eq.) of these FIs, compared to the hypersaline fluids to produce primary multiphase-solid FIs in Grt as described in later.In contrast to FIs in Grt, the reaction between FIs in Qz and host Qz is more restricted except for silica.Therefore, mineral phases present in HI-bearing FIs in Qz and the multiphase-solid FIs in Grt formed from hypersaline fluid might exhibit notable differences.

Composition of fluids trapped in the Khungui eclogite under HP condition
Based on the optical microscope, SEM, and Raman spectroscopy observations, we determine the variable mineral assemblages in the multiphase FIs of the studied samples (Figs. 4,6,and 7;Fig. S4; Table 1).Although the liquid and vapor ratios of bi-phase FIs could be accurately estimated optically (e.g., Bakker and Diamond 2006), the estimation of individual volumes of the solid-and fluid-phases in multiphase-solid FIs is not easy.
We reconstructed the composition of the initial trapped fluid in the SiO 2 -TiO 2 -FeO-MnO-MgO-CaO-Na 2 O-K 2 O-CO 2 -Cl-H 2 O system, based on the volume (cm 3 ), density (g/cm 3 ), and composition (wt.%) of the daughter (HI) and step-daughter minerals (Cal, Chl, Act, and Bt) in the multiphase-solid FIs.The compositions of the initial fluids are estimated using the following mass balance equation: Where m F i is the molar amount of element i in the reconstructed initial fluid per unit volume of the FI (mol/cm 3 fluid inclusion), M k is the molar amount of phase k per unit volume of the FI, X k i is the molar amount of element i per formula unit of phase k, r, and p in the summation indicate the reactants (Grt and Brs; as explained later) and products (Chl, Act, Bt, Cal, HI, and fluid) of the FI, respectively.M k is obtained from the observed volume fractions of phase k (F k ) and the molar volume of phase k ( V k ), using the formula (1) A mass-balance calculation is conducted on the 3D-SR-XCT data (Fig. 7c; Table 4), and a similar calculation is subsequently performed for the 2D volume estimations for comparison (Table 4).The compositions of Act and Bt in the exposed multiphase-solid FIs are determined using EPMA (Table 2; Fig. 4), whereas Cal and Hl are assumed to be pure CaCO 3 and NaCl, respectively.Chl is assumed to be Fe-dominated based on the semi-quantitative data obtained from the FIB-SEM-EDS analysis (Table 2; Fig. S5).
As discussed in the previous section, it is assumed that the trapped minerals reacted with the fluid, as depicted using the following reaction: Apart from Grt, which hosts the FIs, we assume Brs to be an additional reactant mineral, since mafic minerals (Chl, Act, and Bt) are formed in the multiphase-solid FI, and Brs is common in the matrix and as inclusions in the Ca-rich rim of Grt (Figs. 3, and 4; Table 1).As the observed liquid in the FI is saturated with HI, the composition of fluid B is assumed to be NaCl saturated at room temperature (i.e., 26.3 wt.% NaCl eq.).The compositions of the Ca-rich rims of Grt and Cl-bearing Brs inclusions (Cl = 1.01 wt.%) are used for mass-balance calculations (Table 2).
The amount of parent minerals (Grt and Brs) that reacted with fluid A are systemically varied in the range of 0-0.20 [mol/100 cm 3 FI] (Fig. 10).Increasing the amount of Grt and/or Brs resulted in a decrease in the Si, Al, Fe, Mg, and Ca contents in the initial fluid.The Na, K, Cl, and C contents of the initial fluid are barely affected by the amounts of reactants used.The amount of reacted Grt and/or Brs is limited because of the limited availability of Al, Si, and/or Mg in the bulk FI.As a result, the total amount of dissolved cations is observed to be the lowest (~ 19 molal = mol/kg H 2 O) when Grt and Brs are reacted at ratios of 0.048 and 0.087 (mol/100 cm 3 FI), respectively (Fig. 10).Because the solubility of aqueous fluids is limited, we consider this to be the most probable initial fluid composition for the studied multiphase-solid FI.
The estimated elemental concentration of the initial fluid (in molal) are ~ 0 for Si, Ti, Al, and Mn; 2.9 for Fe; 2.1 for Mg; 2.7 for Ca; 10.4 for Na; 0.5 for K; 3.8 for C; and 12.7 for Cl, corresponding to 32.2 wt.% NaCl eq.; X NaCl [= NaCl / (H 2 O + CO 2 + NaCl) in mol] = 0.18 and X CO2 [= CO 2 / (H 2 O + CO 2 + NaCl) in mol] = 0.05 (Table 4).Moreover, ~ 15% of the H 2 O in the initial trapped fluid is consumed in the reaction of Grt and Brs with Chl, Act, and Bt.
The overall results of the reaction are summarized as follows: For comparison, the representative results of the reconstructed initial trapped fluid composition based on 2D observation of the multiphase-solid FI (in molar) are 0 for Si and Al; 0.1 for Ti and Mn; 3.4 for Fe; 2.3 for Mg; 0.7 for Ca; 3.1 for Na; 1.1 for K; 1.5 for C; and 7.0 for Cl, corresponding to 22.8 wt.% NaCl eq.; X NaCl = 0.11 and X CO2 = 0.02 (Table 4).The Ti, Fe, Mn, Mg, and K concentrations in the 2D reconstructed initial fluid composition are estimated to be higher than those in the 3D reconstructed fluid composition; in contrast, the Ca, Na, C, and Cl concentrations are lower in the 2D reconstruction.In addition, not all minerals are fully recognized by 2D analysis (Table 1).This implies that the fluid compositions reconstructed using 2D images have large uncertainties.
The reconstructed fluid composition of the multiphasesolid FIs suggests that the initially trapped fluid at the eclogite stage of the Khungui eclogite was hypersaline (~ 32.2 wt.% NaCl eq.) and consisted of fluids with variable solute components (Ca, Na, K, Mg, Fe, C, and Cl).In contrast, the bi-phase FIs indicate a relatively high salinity (13-16.1 wt.% NaCl eq.) H 2 O-dominated fluid with solutes (Mg and K) is identified based on (2) 0.048Grt + 0.087Brs + fluidA = 0.077Chl + 0.058Act + 0.027Bt + 0.318HI + 0.154Cal + fluidB microthermometry data (Table 3; Davis et al. 1990;Dubois and Marignac 1997;Bakker and Baumgartner 2012).This suggests that two types of fluids with different characteristics existed during the earlier and later stages of eclogite facies metamorphism.

Possible source of fluid origin
The fluid origin of HP-UHP metamorphism has been explained by internal or external fluid sources based on oxygen isotopes, the composition of FI, and their FI host minerals (Ferrando et al. 2005b(Ferrando et al. , a, 2020;;Zhang et al. 2008;Frezzotti and Ferrando 2015).The internal fluid was attributed to the decomposition of hydrous minerals.In contrast, external fluids were more closely related to the dehydration of metasedimentary rocks or the serpentinized mantle in the subduction slab.Some studies have suggested that both fluid origins can coexist at the same metamorphic stage (Spandler et al. 2011).
The bi-phase H 2 O-dominated FI (up to 16 wt.% NaCl eq.) indicates ubiquitous occurrence of a highly saline H 2 O-dominated fluid at the prograde to eclogite facies stages in the Khungui eclogite.Although the origin of a highly saline fluid under the eclogite facies is unclear, the potential materials for high Cl content with H 2 O, that is, Cl-rich hydrous minerals have not been clarified from the prograde stages of the Khungui eclogite (Figs.2a, 3, 4; Table 1).This suggests the influence of the infiltration of the highly saline H 2 O-dominated external fluid at the prograde to the earlier eclogite facies stage.In contrast, at the later eclogite facies stage, the presence of hypersaline fluids is recorded as multiphase-solid FIs.One mechanism for the production of hypersaline fluids is the formation of hydrous minerals.If H 2 O in the above mentioned fluid is consumed for the growth of hydrous minerals (i.e., Brs) in a near-closed system, the solute concentration in the fluid increases, to produce a hypersaline fluid (Markl and Bucher 1998).The increase in Cl content at the rim of the Brs inclusions, the rim of Ap in the matrix, and CO 2 phase (Cal) detected from only multiphase-solid FIs are similarly compatible with such an increase in the solute content in the later eclogite facies (Fig. 3c, Fig. S2).
Variable sizes of the multiphase-solid FI have been commonly reported in the peak mineral assemblages of UHP metamorphic rocks (Fu et al. 2001;Ferrando et al. 2005a, b;Frezzotti et al. 2007;Zhang et al. 2008;Jin et al. 2023) (Fig. 9a).The mineral phases in multiphase-solid FIs consist mainly of hydrous (e.g., Bt, Chl, Amp, and Ep), anhydrous (e.g., Qz, zircon, and pyroxene), and carbonate minerals (e.g., Cal, dolomite, and magnesite), and rarely include sulfates (gypsum), phosphates (Ap and monazite), and oxides (Rt).Mineral observations (Act, Bt, Chl, Cal, HI, Qz, and Ap) of the multiphase-solid FI in Grt from the Khungui eclogite suggest similar aspects to the characteristics of multiphase-solid FIs in UHP metamorphic rocks.However, these multiphase-solid FIs in the UHP metamorphic rocks originated completely different mechanisms.The origin of multiphase-solid FIs in UHP metamorphic rocks was related to melting or supercritical fluids (Frezzotti and Ferrando 2015) in deep subduction zones.Furthermore, in a few cases, the multiphase-solid FIs have been recorded from HP metamorphic rocks such as the HP vein (2.6 GPa, 550 °C) in Monviso (Philippot and Selverstone 1991), eclogitic segregation (2.0 GPa, 625 °C) in the Tauren Window (Selverstone et al. 1992), and eclogite (1.5-1.7 GPa, 600 °C) in the Western Gneiss Region (Svensen and Museum 2001).The P-T conditions of the eclogite facies stage of these HP metamorphic rocks are close to the Khungui eclogite (2.1-2.2GPa, 580-610 °C).Therefore, multiphase-solid FIs in HP metamorphic rocks may have been formed by the mechanism proposed in this study.
The high salinity in the fluid from the Khungui eclogite, evident from the bi-phase aqueous inclusions, is comparable to that reported for HP metamorphic rocks, but not UHP rock.The salinity of the decompression fluid in the Khungui eclogite remains high (up to 15 wt.% NaCl eq.) (Fig. 9a), similar to the characteristics reported for the decompression stage fluids of the Makbal UHP complex in the Kyrgyz Tianshan area, CAOB (Orozbaev et al. 2015;Yoshida et al. 2016).

Conclusions
(1) Garnets and quartz in the Khungui eclogite of the Zavkhan Terrane, western Mongolia, contain various types of fluid inclusions and multiphase-solid fluid inclusions trapped at the eclogite facies conditions and decompression stages.Both bi-phase aqueous inclusions and multiphase-solid fluid inclusions, including halite, are recognized as primary inclusions in garnet and quartz, exhibiting a high salinity of 13-16.4wt.% NaCl eq. and > 26.4 wt.% NaCl eq., respectively.Relatively high salinity aqueous fluids (< 16 wt.% NaCl eq.) in quartz were also present during the subsequent decompression.
(2) Primary multiphase-solid fluid inclusions in garnet contain a variety of mineral assemblages, primarily calcite, amphibole, and mica, as well as rare halite, quartz, chlorite, and apatite.(3) Precise mass balance analyses were conducted based on the estimates of the volumes of all phases in the multiphase-solid fluid inclusions in garnet, using synchrotron X-ray CT data.About ~ 15% of H 2 O was consumed during the reactions between the trapped fluids and minerals.(4) Based on the fluid inclusion observations, the following deep fluid history was suggested for the Khungui eclogite: (i) H 2 O-dominated fluids with high salinity (13-16.4wt.% NaCl eq.) and certain solute compounds (Na, Mg, and K) infiltrated the rocks during prograde to the earlier eclogite facies stages, and (ii) the desiccation of aqueous fluids took place at the later stage of the eclogite facies metamorphism due to the growth of hydrous minerals and hypersaline fluids (~ 32.2 wt.% NaCl eq.) with variable solute components (Fe, Mg, Ca, Na, C, and Cl) were produced at later stage of the eclogite facies metamorphism, probably due to the growth of hydrous minerals, and (iii) during exhumation, in-situ fluids-mineral reaction occurred in the multiphase-solid fluid inclusions involving the host minerals, as well as other trapped minerals, such that various mineral assemblages in the multiphase-solid fluid inclusions were dependent on the initial trapped minerals in such inclusions.
identified several subzones and tectonic blocks within this terrane based on the lithological characteristics and structural patterns of the metamorphic and magmatic rocks.These subzones and blocks (from west to east) are: Zavkhan-Khungui zone, Durvuljin block, Mélange, Urgamal and Zavkhan-Mandal zones, the Erdene-Khairkhan and Bayannur blocks.The geological setting and lithological characteristics of the magmatic and metamorphic rocks in these zones and blocks are summarized by Bold et al. (2016), Kovach et al. (2021), and Kozakov et al. (2021).

Fig. 1 a
Fig. 1 a Location of the Central Asian Orogenic Belt (Şengör et al. 1993) and tectonostratigraphic terrane map of western Mongolia modified from Badarch et al. (2002).The red line represents the Mongolian Main Lineament, and the blue lines represent terrane boundaries.The red star shows the location of the study area.b Main tectonic zones of the Zavkhan Terrane (Kovach et al. 2021).The red dashed lines represent subzones boundaries in the Zavkhan Terrane, and the black dashed line shows the boundary between high-and lowgrade rocks in the mélange zone.c Sample locations and geological

Fig. 3
Fig. 3 X-ray element mapping of garnet and fluid inclusions.a Element (Mg, Ca, Ti, Mn, and Fe) maps and backscattered electron image of the garnet.White boxes indicate the location of fluid inclusions in the garnet.Black and white dotted lines show the border of the core and rim of the garnet.The red line on the backscattered electron image shows the position of the garnet composition profile given in (b).b Composition profile across the center of the studied garnet (Fe, Ca, Mg, and Mn).The black dashed lines separate the garnet

Fig. 4
Fig. 4 The exposed multiphase-solid fluid inclusion in garnet.a Backscattered electron image of the euhedral garnet and included minerals.b Four element maps (Mg, Ca, Fe, and Mn) of the garnet (excluding other minerals) and backscattered electron image with five element maps (Na, F, K, Fe, and Mg) for the exposed multiphasesolid fluid inclusion.Black dotted lines indicate the border of the core and rim of the garnet.White dashed line presents boundary of solid phases.c Na (M4 site) vs Al (T site) diagram for amphiboles.The matrix amphibole composition used was cited from Bayarbold

Fig. 5 a
Fig. 5 a Primary/secondary fluid inclusions in quartz.b Representative Raman spectra for a mixed primary fluid inclusion in quartz.c Primary bi-phase fluid inclusion in garnet.d Representative Raman spectra for mixed primary bi-phase fluid inclusion in garnet.e Repre-

Fig. 6
Fig. 6 2D observation of a multiphase-solid fluid inclusion in garnet.a Photomicrograph of the multiphase-solid fluid inclusion in the garnet (cross polarized light).b Distribution of different phases within the fluid inclusion, based on the detailed Raman image.c Raman spectra for each phase of the multiphase-solid fluid inclusion in garnet.L H2O pure water, V H2O water vapor, Amp amphibole, Cal calcite, Grt garnet

Fig. 8
Fig. 8 Frequency histograms of the microthermometric measurements.a For ice melting temperatures (T mt (°C)).b For homogenization temperatures (T h (°C)).FI fluid inclusion, MS.FI multiphasesolid fluid inclusion, Type I S.FI Type I secondary fluid inclusion, and Type II S.FI Type II secondary fluid inclusion

Fig. 10
Fig. 10 An example of the reconstructed fluid composition based on 3D data of the multiphase-solid fluid inclusion, as determined by SR-XCT analysis.The variation in solute concentration in the reconstructed fluid is shown with respect to the volumes [mol/100 cm 3 ] of

Table 4
Estimated vol.% of all solid phases in a primary multiphasesolid fluid inclusion (MS.FI) in garnet and fluid compositions estimated by mass balance (in molal)

)
Volume % estimated of solid phases in MS.FI