The Calcite-Dolomite Solvus Temperature and T-X(CO 2 ) Evolution in High-Grade Impure Marble from Thongmön Area, Central Himalaya: Implications for Carbon Cycling in Orogenic Belts

: Impure dolomitic marble from the Great Himalayan Sequences (GHS) in Thongmön area, central Himalaya, is ﬁrst systematically reported here concerning its petrographic features, textural relations, and ﬂuid evolution. The Thongmön impure marble is characterized by the assemblage of calcite + dolomite + forsterite + spinel + phlogopite + clinohumite ± diopside ± retrograde serpentine. Three groups of calcite and dolomite occurring both as inclusions and in the matrix were identiﬁed: group I is represented by relatively magnesium-rich calcite (Cal) (Cal I :X Mg = 0.10–0.15) and almost pure dolomite (Dol) (Dol I :X Mg = 0.47–0.48), corresponding to the Cal-Dol solvus temperatures of 707–781 ◦ C; group II is characterized by vermicular dolomite exsolutions (Dol II :X Mg = 0.45–0.46) in Mg-rich calcite and Mg-poor calcite (Cal II :X Mg = 0.05–0.08) adjacent to Dol II , and the recorded solvus temperatures are 548–625 ◦ C; group III is represented by nearly pure calcite (Cal III :X Mg = 0.003–0.02) and Ca-rich dolomite in the matrix (Dol III :X Mg = 0.33–0.44). Isobaric T-X(CO 2 ) pseudosection at a peak pressure of 15 kbar in the system K 2 O-CaO-MgO-Al 2 O 3 -FeO-SiO 2 -H 2 O-CO 2 suggests that the peak ﬂuid composition of the Thongmön forsterite marble is restricted to X(CO 2 ) < 0.04 at T > 780 ◦ C due to a potential inﬁltration event of H 2 O-rich ﬂuid. Alternatively, the forsterite marble is a retrograde product subordinated to the GHS exhumation process, and its ﬂuid composition is relatively CO 2 -rich (0.6 < X(CO 2 ) < 0.8 at 5 kbar, 750 ◦ C) at a nearly isothermal decompression stage. In either case, we suggest that the carbon ﬂux contributed by metacarbonate rocks in an orogen setting to the global carbon cycling must be considered.


Introduction
Metamorphic fluid is crucial for promoting metamorphic reactions, mass transport, and partial melting [1,2], and its composition has a great impact on the element cycling in subduction zones [3][4][5]. In particular, metacarbonate rocks generally yield various compositions of CO 2 -bearing fluid due to carbonation and decarbonation reactions. In high-grade metamorphic regions, investigation of metamorphosed carbonate-bearing rocks from continental collision belts has demonstrated that fluid compositions have a great difference at different metamorphic stages, varying from nearly pure H 2 O to relative CO 2 -rich compositions [4,[6][7][8][9], which may provide crucial information for the metamorphic evolution and deep carbon cycling during continental subduction and exhumation [4,6,[9][10][11].
The carbon cycling between the biosphere, ocean, and the interior of the Earth jointly controls global climate change and furnishes our life with a livable planet [12][13][14][15][16][17][18]. Carbonation and decarbonation processes determine the carbon sequestration and outgassing in subduction-collision settings, e.g., the Himalaya [19][20][21][22], the North American Cordillera [23][24][25], the European Alps [10,[26][27][28], the Southern Alps of New Zealand [16], The Ama Drime is a prominent N-S striking dome separated from the GHS by two opposite-dipping shear zones, the western margin Ama Drime Detachment (ADD) and the eastern margin Nyönno Ri Detachment (NRD). The western ADD is described as a 100-300 m-thick normal-sense detachment system and primarily consists of metacarbonate rocks and quartzite [51], while the eastern NRD offsets the position of the STDS right-laterally [52]. To the south of the ADM, the GHS can be divided into the Upper and the Lower Greater Himalayan Sequence termed as the UGHS and the LGHS, respectively [53]. The UGHS is mainly composed of paragneisses which were widely intruded by the Miocene leucogranite [54]. The LGHS mostly consists of strongly deformed metapelites and orthogneiss rocks with local partial melting at the early stage of retrograde metamorphism. After the anatexis, penetrative ductile shear displacements and foldings were developed [55].  [39]). Abbreviations are as in Figure 1.  The Ama Drime is a prominent N-S striking dome separated from the GHS by two opposite-dipping shear zones, the western margin Ama Drime Detachment (ADD) and the eastern margin Nyönno Ri Detachment (NRD). The western ADD is described as a 100-300 m-thick normal-sense detachment system and primarily consists of metacarbonate rocks and quartzite [51], while the eastern NRD offsets the position of the STDS right-laterally [52]. To the south of the ADM, the GHS can be divided into the Upper and the Lower Greater Himalayan Sequence termed as the UGHS and the LGHS, respectively [53]. The UGHS is mainly composed of paragneisses which were widely intruded by the Miocene leucogranite [54]. The LGHS mostly consists of strongly deformed metapelites and orthogneiss rocks with local partial melting at the early stage of retrograde metamorphism. After the anatexis, penetrative ductile shear displacements and foldings were developed [55].  [39]). Abbreviations are as in Figure 1.  [39]). Abbreviations are as in Figure 1.
The Ama Drime is a prominent N-S striking dome separated from the GHS by two opposite-dipping shear zones, the western margin Ama Drime Detachment (ADD) and the eastern margin Nyönno Ri Detachment (NRD). The western ADD is described as a 100-300 m-thick normal-sense detachment system and primarily consists of metacarbonate rocks and quartzite [51], while the eastern NRD offsets the position of the STDS right-laterally [52]. To the south of the ADM, the GHS can be divided into the Upper and the Lower Greater Himalayan Sequence termed as the UGHS and the LGHS, respectively [53]. The UGHS is mainly composed of paragneisses which were widely intruded by the Miocene leucogranite [54]. The LGHS mostly consists of strongly deformed metapelites and orthogneiss rocks with local partial melting at the early stage of retrograde metamorphism. After the anatexis, penetrative ductile shear displacements and foldings were developed [55].
The Thongmön outcrop generally comprises well-foliated metapelite with minor magmatic augen gneiss, graphitic schist, marble, and calc-silicate, and mafic pods enclosed within metapelite were also found [38]. Cottle et al. (2009) [38] suggested that the Thongmön metapelites, as well as enclosed mafic granulite, have undergone a high-temperature retrograde overprint during isothermal exhumation, making it difficult to retrieve earlier metamorphic P-T conditions, and that the Fe-Mg garnet-biotite exchange thermometry combined with THERMOCALC calculation on the Thongmön metapelitic gneiss gave a re-equilibration condition of~750 • C at 5 kbar. Nevertheless, well-preserved eclogites [39] and granulitized eclogite lenses [56] within well-foliated garnet-bearing gneiss have been investigated recently in terms of their metamorphic history and evolution of the P-T path. Li et al. (2019) [39] proposed that the peak condition of the Thongmön eclogites proceeded at >21 kbar, >750 • C in the Miocene. In addition, a recent study documented that metapelitic rocks in the Thongmön area experienced HP granulite-facies metamorphism and then anatexis during exhumation [40]. The metamorphic conditions of these anatectic metapelites, hosting layers and lenses of (retrograde) eclogites and impure marbles ranging in thickness from several to hundreds of meters, have been estimated in the range of 13-16 kbar, 740-760 • C during prograde metamorphism and at about 13-14 kbar, 820-840 • C during a near isobaric heating stage followed by a heating decompression to 10-11 kbar, >860 • C [40].

Sample Description
Several samples of impure marble from Thongmön ( Figure 2) in the GHS of the central Himalayas were collected. These metacarbonates are well exposed and strongly deformed due to simultaneously or later structural folding and thrusting activities. The metapelitic rocks account for~90% of the unit volume; metacarbonate layers and lenses are on the order of 0.5-up to 10-m-thick. Occurring as distinct grey-whitish intercalations or lenses within the metapelites in Thongmön, impure marbles are very easy to recognize and could have gone through the same metamorphic evolution as the metapelites. According to the types and relative contents of silicate minerals in metacarbonates, these impure marbles were classified as forsterite-marble, diopside-marble, and tremolite-marble. The assemblage dolomite + calcite + forsterite + spinel + diopside + phlogopite + apatite ± rutile (titanite) is representative of forsterite-marble. Diopside-marble consists of calcite, quartz, clinopyroxene, calcic plagioclase, K-feldspar, apatite, and rutile (titanite). Tremolite-marble contains a similar mineral assemblage as the diopside-marble but without clinopyroxene. Tremolite-marble is generally considered a retrograde product of diopside-marble [4,57]. These different mineral assemblages reflect different bulk compositions rather than the differences in the equilibration P-T conditions [58]. This study focuses on the forsterite marble with the aim of discussing the metamorphic decarbonation reactions and fluid evolution.
The forsterite-marble (sample DR18-55) was collected from the Thongmön area (Xigaze, China), central Himalaya, and appears as massive and banded rocks within the strongly deformed metapelites (Figure 3). The sample location is a few kilometers southwest away from that of Li et al. (2019Li et al. ( , 2021 [39,40]. The Thongmön forsterite-marble mainly consists of calcite, dolomite, forsterite, spinel and minor phlogopite, clinohumite, and diopside. Calcite (Cal) is the most abundant carbonate in the forsterite-marble and ranges from 10 to 50 µm in size ( Figure 4). Under the microscope, some domains of calcite appear cloudy due to the presence of abundant and irregular fine-grained dolomite exsolutions ( Figure 5). However, some calcite grains in the matrix almost contain no dolomite inclusions or exsolutions (Figure 5a,b). Pure Mg-calcite domains are rare. Veinlets of almost pure calcite (X Mg < 0.003) can be found not only along silicate boundaries but also crosscutting some grains of the matrix calcite and dolomite (Figure 4a,c). In addition, rounded calcite inclusions, generally 5-20 µm, also occur in spinel, phlogopite, and forsterite (Figures 4c-f and 5a,b).          lamellae and coarser dolomite exsolutions were also commonly found in the matrix calcite. In some cases, the matrix dolomite is crosscut by the veinlets of Mg-poor calcite or almost pure calcite along the cracks and grain margins ( Figure 4c). The irregular contact boundaries between calcite and dolomite indicate a reaction relationship (see Section 6.1). (Mg-)calcite + dolomite polyphase inclusions in spinel and forsterite suggest the preliminary existence of aragonite during prograde metamorphism [58,60].
Spinel (Spl) is colorless, rounded in shape, and 10-50 µm in size. Some spinel contains polyphase inclusions. Spinel is replaced by calcite and a very fine-grained aggregate of Mgand Al-rich hydroxides during a late stage of metamorphic history (Figure 5e). In most cases, forsterite is well preserved in the matrix in contact with dolomite, calcite, and other silicates, but a few forsterite grains are partly replaced by thin veins of serpentine, suggesting that some hydration reaction occurred during retrograde metamorphism ( Figure 5f). Spinel, dolomite, and calcite can also be found as inclusions in forsterites (Figures 4d and 5b,e). Phlogopite (Phl) is common both in the matrix and as inclusions (Figures 4a and 5a). In addition, phlogopite inclusions were found in zircon and spinel (Figures 5a and 6). Diopside (Di) is very rare in most forsterite-bearing impure marbles and commonly occurs as a corona surrounding forsterite ( Figure 4a). Clinohumite occurs as irregular granular-to tabular-shaped grains and is colorless to pale yellow (Figure 4b), indicating its low TiO 2 contents.
Minerals 2022, 12, x FOR PEER REVIEW 7 of 19 lamellae and coarser dolomite exsolutions were also commonly found in the matrix calcite. In some cases, the matrix dolomite is crosscut by the veinlets of Mg-poor calcite or almost pure calcite along the cracks and grain margins ( Figure 4c). The irregular contact boundaries between calcite and dolomite indicate a reaction relationship (see Section 6.1).
(Mg-)calcite + dolomite polyphase inclusions in spinel and forsterite suggest the preliminary existence of aragonite during prograde metamorphism [58,60]. Spinel (Spl) is colorless, rounded in shape, and 10-50 µm in size. Some spinel contains polyphase inclusions. Spinel is replaced by calcite and a very fine-grained aggregate of Mg-and Al-rich hydroxides during a late stage of metamorphic history (Figure 5e). In most cases, forsterite is well preserved in the matrix in contact with dolomite, calcite, and other silicates, but a few forsterite grains are partly replaced by thin veins of serpentine, suggesting that some hydration reaction occurred during retrograde metamorphism (Figure 5f). Spinel, dolomite, and calcite can also be found as inclusions in forsterites ( Figures  4d and 5b,e). Phlogopite (Phl) is common both in the matrix and as inclusions (Figures 4a  and 5a). In addition, phlogopite inclusions were found in zircon and spinel (Figures 5a and 6). Diopside (Di) is very rare in most forsterite-bearing impure marbles and commonly occurs as a corona surrounding forsterite ( Figure 4a). Clinohumite occurs as irregular granular-to tabular-shaped grains and is colorless to pale yellow (Figure 4b), indicating its low TiO2 contents.

Analytical Methods
Whole-rock major element compositions were determined by X-ray fluorescence spectrometry (XRF, PrimusII, Rigaku, Osaka, Japan) and inductively coupled plasmamass spectrometry (ICP-MS, Agilent 7700e, Wuhan Sample Solution Analytical Technology Co., Ltd., Wuhan, China). Mineral composition analysis and back-scattered electron (BSE) imaging were performed with an electron microprobe (EMP, JEOL JAX-8100) at the Key Laboratory of Orogenic Belts and Crustal Evolution, Peking University, China. The analyses were operated in wavelength dispersion mode (WDS) with a 10 nA beam current and a 15 kV accelerating voltage. A minimum beam diameter of 1-3 µm and a defocused beam of ~10 µm were used for silicate and carbonate minerals, respectively. To reduce the measurement error and improve the data reproducibility, carbonates containing < 10 µm exsolution lamellae were measured using a broad beam of diameter 50 µm for obtaining the average chemical composition. Fluorine contents in clinohumite and phlogopite were also analyzed under 15 kV accelerating voltage and 40 nA beam current conditions with a beam diameter of ~20 µm. SPI standards were utilized and raw data were reduced with the Phi-Rho-Z (PRZ) correction [61].
Raman analysis of mineral inclusions in zircon and other minerals was performed on a Renishaw-RW 1000 Raman microprobe adopting the 514.5 nm line of an Ar-ion laser at the School of Earth and Space Sciences, Peking University. The analytical procedures followed the description in [62]. The accumulation time varied from 5 to 10 s and the laser spot size was focused to 2 µm. The estimated spectral resolution was 1.0 cm −1 , and the

Analytical Methods
Whole-rock major element compositions were determined by X-ray fluorescence spectrometry (XRF, PrimusII, Rigaku, Osaka, Japan) and inductively coupled plasma-mass spectrometry (ICP-MS, Agilent 7700e, Wuhan Sample Solution Analytical Technology Co., Ltd., Wuhan, China). Mineral composition analysis and back-scattered electron (BSE) imaging were performed with an electron microprobe (EMP, JEOL JAX-8100) at the Key Laboratory of Orogenic Belts and Crustal Evolution, Peking University, China. The analyses were operated in wavelength dispersion mode (WDS) with a 10 nA beam current and a 15 kV accelerating voltage. A minimum beam diameter of 1-3 µm and a defocused beam of 10 µm were used for silicate and carbonate minerals, respectively. To reduce the measurement error and improve the data reproducibility, carbonates containing <10 µm exsolution lamellae were measured using a broad beam of diameter 50 µm for obtaining the average chemical composition. Fluorine contents in clinohumite and phlogopite were also analyzed under 15 kV accelerating voltage and 40 nA beam current conditions with a beam diameter of~20 µm. SPI standards were utilized and raw data were reduced with the Phi-Rho-Z (PRZ) correction [61].
Raman analysis of mineral inclusions in zircon and other minerals was performed on a Renishaw-RW 1000 Raman microprobe adopting the 514.5 nm line of an Ar-ion laser at the School of Earth and Space Sciences, Peking University. The analytical procedures followed the description in [62]. The accumulation time varied from 5 to 10 s and the laser spot size was focused to 2 µm. The estimated spectral resolution was 1.0 cm −1 , and the spectrum reproducibility is better than 0.05 cm −1 . Calibration was conducted using synthetic silicon and spectra were fitted using the Peakfit software (v4.04, Aisn Software Inc., Wilmington, DE, USA).
Phlogopite grains in the matrix usually contain Mg-calcite inclusions with a maximum X Mg value of 0.115. The Mg/(Mg + Fe) ratio in phlogopite is generally >0.99 and the TiO 2 content can be up to 0.57 wt%.
Clinopyroxene is nearly pure diopside. Clinohumite (Chu) occurs as granoblastic crystals associated with forsterite and spinel. The TiO 2 contents of clinohumite range from 1.54 to 1.96 wt% and the fluorine contents range from 1.68 to 2.18 wt% (Table 2). A certain amount of fluorine can stabilize the humite group minerals to higher temperatures [63,64].

Calcite-Dolomite Solvus Geothermometer
The calcite-dolomite solvus thermometer is a widely used geothermometer for marbles [65][66][67]. Anovitz and Essene (1987) [65] provided an empirical calibration of the solvus thermometer in the range of 473-1173 K with ±10 • C uncertainty. Recently, Hermann et al. (2016) [66] performed an experimental constraint on the reaction of dolomite + aragonite to Mg-calcite at 2-6.5 GPa, 660-940 • C, emphasizing the important role of an experimental calibration of this reaction in interpreting subducted impure marbles. This thermometer has been applied to inclusions that are located far from the margin of the host minerals and not crosscut by late-stage calcite or serpentinized veins [65].
The X Mg (Mg/(Mg + Ca + Fe)) compositions of Mg-calcite inclusions (Cal I ) in spinel, forsterite, and phlogopite range from 0.097 to 0.153, corresponding to the solvus temperatures of 707-781 • C, which is at least 100 • C higher than the recorded temperature of the matrix Mg-calcite (Cal I ). The X Mg compositions of Mg-poor calcite in the matrix (Cal II ) are from 0.056 to 0.080, corresponding to the solvus temperature of 548-625 • C. Calcite grains (Cal III ) with lower X Mg values (0.002-0.02) imply that the solvus temperatures are below 400 • C, which may be attributed to later retrograde metamorphism. Thus, the compositions of these carbonates cannot be used to calculate and evaluate the metamorphic temperatures. Pure calcite inclusions in spinel and forsterite were not found, but Mg-calcite and polyphase dolomite + Mg-calcite inclusions were included in spinel. An Mg-calcite inclusion (X Mg = 0.153) in spinel constrains a minimum temperature of 781 • C for the growth of this spinel grain. This value is the highest temperature recorded by Mg-calcite composition in the Thongmön impure marble.
We analyzed the compositions of carbonate inclusions in forsterite without serpentinization to undertake a further evaluation of temperature. The X Mg values of monophase Mg-calcite and composite carbonate inclusions consisting of dolomite and Mg-calcite inclusions in forsterite are from 0.12 to 0.14, corresponding to the calcite-dolomite solvus temperature of 720-760 • C, which is lower than that of the maximum 781 • C recorded in spinel. The composition of carbonate inclusions in forsterite indicates that the temperature was at least >720 • C during forsterite growth. Calcite inclusion with X Mg = 0.115 in phlogopite gives the corresponding solvus temperature of 707 • C, which is slightly lower than that recorded in spinel and forsterite. The lower temperature record was possibly influenced by the infiltration of later external fluids along phlogopite cleavage.
Notably, various compositions, shapes, and sizes of carbonate inclusions possibly recorded different temperatures at different metamorphic stages. Thus, to obtain the peak metamorphic temperatures, it is necessary to carefully select those characteristic carbonate inclusions that have homogenous and unmixing features. In this study, the chemical composition of carbonate does not show distinct zoning patterns in terms of calcium and magnesium. The absence of zoning features can be attributed to the relatively higher temperature or/and longer crustal residence time of impure marble in the Tibetan-Himalayan orogenic belt. Experimental work on the cation diffusion in carbonate minerals indicates that the chemical heterogeneities will be erased by the thermal diffusion if heating temperatures are high enough and/or the duration is long [68].
We constructed P/T-X(CO 2 ) pseudosections for impure marble in the H 2 O-CO 2 saturated system; thus, hydrous phases in equilibria can be calculated and shown in such a phase diagram. The reasons for constructing such a phase diagram were (1) to illustrate a series of metamorphic decarbonation reactions occurring during peak and nearly isothermal decompression metamorphism and (2) to constrain the fluid evolution at certain P-T conditions of the Thongmön impure forsterite marble. The fluid evolution can be qualitatively constrained by isobaric P/T-X(CO 2 ) pseudosections. Fixed pressures of 15 and 5 kbar were used for T-X(CO 2 ) pseudosection calculation, following the previous estimate of pressure on metapelites at peak and retrograde stages which are spatially related to the impure marble [38,40].
The T-X(CO 2 ) pseudosection at 15 kbar (Figure 7) is dominated by large tri-and four-variant fields, with few narrow two-variant fields. Calcite/aragonite is always stable in the whole region of interest. Dolomite is absent in Bt + Spl + Chu + Cal, Bt + Fo + Spl + Chu+ Cal, and Bt + Fo + Spl + Cpx+ Cal stability fields. Regardless of the composition of the fluid, muscovite can be stable below 700 • C, whereas biotite appears at higher temperatures on the relatively low X(CO 2 ) side. Spinel can only be found in a narrow stability field at X(CO 2 ) < 0.2, and the stability field of clinohumite is limited to the extremely H 2 O-rich side. Forsterite together with biotite, dolomite, spinel, and calcite occur at 650 • C < T < 850 • C and within the field of X(CO 2 ) < 0.12. Diopside can be found on the CO 2 -poor side at lower temperatures and on the CO 2 -rich side at higher temperatures. In addition, in order to trace the fluid evolution during isothermal exhumation, we also constructed the T-X(CO 2 ) pseudosection at a fixed pressure of 5 kbar and P-X(CO 2 ) pseudosection at a fixed temperature of 750 • C (Figure 8). The Bt + Dol + Fo + Spl + Cal stability field at 5 kbar extends to be wider than that at 15 kbar. The X(CO 2 ) of the fluid in the Thongmön impure marble can reach up to 0.8 at 5 kbar, 750 • C. Similarly, the P-X(CO 2 ) diagram at 750 • C indicates that the fluid compositions evolve to be CO 2 -richer during isothermal exhumation.
pressure on metapelites at peak and retrograde stages which are spatially related to the impure marble [38,40].
The T-X(CO2) pseudosection at 15 kbar (Figure 7) is dominated by large tri-and fourvariant fields, with few narrow two-variant fields. Calcite/aragonite is always stable in the whole region of interest. Dolomite is absent in Bt + Spl + Chu + Cal, Bt + Fo + Spl + Chu+ Cal, and Bt + Fo + Spl + Cpx+ Cal stability fields. Regardless of the composition of the fluid, muscovite can be stable below 700 °C, whereas biotite appears at higher temperatures on the relatively low X(CO2) side. Spinel can only be found in a narrow stability field at X(CO2) < 0.2, and the stability field of clinohumite is limited to the extremely H2O-rich side. Forsterite together with biotite, dolomite, spinel, and calcite occur at 650 °C < T < 850 °C and within the field of X(CO2) < 0.12. Diopside can be found on the CO2-poor side at lower temperatures and on the CO2-rich side at higher temperatures. In addition, in order to trace the fluid evolution during isothermal exhumation, we also constructed the T-X(CO2) pseudosection at a fixed pressure of 5 kbar and P-X(CO2) pseudosection at a fixed temperature of 750 °C (Figure 8). The Bt + Dol + Fo + Spl + Cal stability field at 5 kbar extends to be wider than that at 15 kbar. The X(CO2) of the fluid in the Thongmön impure marble can reach up to 0.8 at 5 kbar, 750 °C. Similarly, the P-X(CO2) diagram at 750 °C indicates that the fluid compositions evolve to be CO2-richer during isothermal exhumation.  lines with orange filled band represent the range of peak temperature at 15 kbar evaluated by the calcite-dolomite solvus thermometer from Anovitz and Essene (1987) [65] and Hermann et al. (2016) [66]; (b) the same T-X(CO2) pseudosection as (a) but with a narrower fluid range from pure H2O to X(CO2) =0.2. The white dashed arrow points to the internally buffered T-X(CO2) path compatible with both the observed peak assemblage and the corresponding decarbonation Reactions (2)-(4). See more details in the text. 6. Discussion

Different Generations of Carbonates and Mineral Reaction History in the Impure Marble
A common method of retrieving the P-T history of metamorphic rocks is to gather information from the well-preserved inclusions in refractory minerals and from the metamorphic reaction textures [74,75]. This method is categorically evidenced by the description of coesite and diamond inclusions in felsic granulite formed at UHP conditions [76]. Carbonate inclusions in impure marble, which is less prone to be influenced by late retrograde metamorphism, can not only provide crucial information on the temperature of formation of the host minerals [4,58,66] but also constrain the mineral reaction history. Thus, carbonate inclusions are extraordinary candidates for extracting metamorphic evolution information on high-grade metamorphic marbles.
Three generations of calcite can be recognized. The first generation contains the inclusion-free Mg-calcite with 0.10 < XMg < 0.15 (CalI) coexisting with pure matrix dolomite, forsterite, spinel, and phlogopite (Figure 9a). The second generation is documented by the exsolution of dolomite from the primary homogenous Mg-calcite (Figure 9b). The lathshape and vermicular intergrowths of dolomite in Mg-poor calcite with 0.05 < XMg < 0.08 (CalII) are not uncommon in the Thongmön impure marbles. The last generation is represented by the formation of almost pure calcite with XMg < 0.003 (CalIII) via an external CO2rich hydrous fluid flux (Figure 9c). Moreover, this hydration event also resulted in the growth of serpentine replacing forsterite. In the same way, three generations of dolomite can also be recognized (Figure 9). DolI (XMg > 0.48) forms coarse grains in the matrix, and smaller grains occur as mineral inclusions. The second generation (DolII) (XMg = 0.45-0.46) forms fine-grained exsolutions in Mg-poor calcite (CalII) in the matrix, showing wormy or elliptical shapes with low aspect ratios. Some DolII crystals coexist with Mg-poor calcite as polyphase inclusions within spinel and forsterite, showing tiny dolomite lamellae exsolution textures in the Mg-poor calcite host. Locally, the matrix dolomite is crosscut by the

Different Generations of Carbonates and Mineral Reaction History in the Impure Marble
A common method of retrieving the P-T history of metamorphic rocks is to gather information from the well-preserved inclusions in refractory minerals and from the metamorphic reaction textures [74,75]. This method is categorically evidenced by the description of coesite and diamond inclusions in felsic granulite formed at UHP conditions [76]. Carbonate inclusions in impure marble, which is less prone to be influenced by late retrograde metamorphism, can not only provide crucial information on the temperature of formation of the host minerals [4,58,66] but also constrain the mineral reaction history. Thus, carbonate inclusions are extraordinary candidates for extracting metamorphic evolution information on high-grade metamorphic marbles.
Three generations of calcite can be recognized. The first generation contains the inclusion-free Mg-calcite with 0.10 < X Mg < 0.15 (Cal I ) coexisting with pure matrix dolomite, forsterite, spinel, and phlogopite ( Figure 9a). The second generation is documented by the exsolution of dolomite from the primary homogenous Mg-calcite (Figure 9b). The lath-shape and vermicular intergrowths of dolomite in Mg-poor calcite with 0.05 < X Mg < 0.08 (Cal II ) are not uncommon in the Thongmön impure marbles. The last generation is represented by the formation of almost pure calcite with X Mg < 0.003 (Cal III ) via an external CO 2 -rich hydrous fluid flux (Figure 9c). Moreover, this hydration event also resulted in the growth of serpentine replacing forsterite. In the same way, three generations of dolomite can also be recognized (Figure 9). Dol I (X Mg > 0.48) forms coarse grains in the matrix, and smaller grains occur as mineral inclusions. The second generation (Dol II ) (X Mg = 0.45-0.46) forms fine-grained exsolutions in Mg-poor calcite (Cal II ) in the matrix, showing wormy or elliptical shapes with low aspect ratios. Some Dol II crystals coexist with Mg-poor calcite as polyphase inclusions within spinel and forsterite, showing tiny dolomite lamellae exsolution textures in the Mg-poor calcite host. Locally, the matrix dolomite is crosscut by the veinlets of Mg-poor calcite or almost pure calcite (Cal III ) along cracks and grain margins, and it is transformed in Dol III with lower Mg contents (X Mg = 0.33-0.44). veinlets of Mg-poor calcite or almost pure calcite (CalIII) along cracks and grain margins, and it is transformed in DolIII with lower Mg contents (XMg = 0.33-0.44). As outlined above, firstly, the dolomite (DolI) and Mg-calcite (CalⅠ) polyphase inclusions in spinel and forsterite (Figure 4c,d) imply the existence of former aragonite during prograde metamorphism (Reaction (1)). Forsterite and calcite (CalII) were formed by Reaction (2) in the presence of dolomite (DolII) and diopside, but the absence of primary diopside in the matrix or as inclusions may imply that it has been exhausted during prograde metamorphism. Moreover, the thin reaction rims of dolomite and diopside around forsterite indicate that forsterite was a stable phase at peak conditions ( Figure 4). Spinel and forsterite started to form more or less at the same time by Reaction (3). Notably, Reaction (3) will produce a K-rich aqueous fluid due to the absence of a K-bearing phase among the products. In the presence of an external H2O-rich fluid during exhumation, clinohumite and calcite (CalIII) were formed by Reaction (4). However, if the forsterite marble is a retrograde product of diopside marble, these decarbonation reactions can also be applied to the trace its fluid evolution.  [58] proposed that clinohumite can also grow by a decarbonation reaction consuming diopside and dolomite: Di + Dol + H2O → Chu + Cal + CO2. However, from petrological observation of this study, the growth of clinohumite from forsterite is testified by clinohumite rimming forsterite (Figure 4b). The high amount of fluorine in clinohumite and phlogopite indicates that an external fluorine-bearing hydrous fluid (or melt) played a crucial role in their growths. The tiny serpentinization veins through forsterite also indicate a later hydration process. However, the very low amount of clinohumite and serpentine testifies that such a process is limited.  (1)). Forsterite and calcite (Cal II ) were formed by Reaction (2) in the presence of dolomite (Dol II ) and diopside, but the absence of primary diopside in the matrix or as inclusions may imply that it has been exhausted during prograde metamorphism. Moreover, the thin reaction rims of dolomite and diopside around forsterite indicate that forsterite was a stable phase at peak conditions ( Figure 4). Spinel and forsterite started to form more or less at the same time by Reaction (3). Notably, Reaction (3) will produce a K-rich aqueous fluid due to the absence of a K-bearing phase among the products. In the presence of an external H 2 O-rich fluid during exhumation, clinohumite and calcite (Cal III ) were formed by Reaction (4). However, if the forsterite marble is a retrograde product of diopside marble, these decarbonation reactions can also be applied to the trace its fluid evolution.

Metamorphic Fluid Evolution of the Thongmön Impure Marble
Proyer et al. (2008) [58] proposed that clinohumite can also grow by a decarbonation reaction consuming diopside and dolomite: Di + Dol + H 2 O → Chu + Cal + CO 2 . However, from petrological observation of this study, the growth of clinohumite from forsterite is testified by clinohumite rimming forsterite (Figure 4b). The high amount of fluorine in clinohumite and phlogopite indicates that an external fluorine-bearing hydrous fluid (or melt) played a crucial role in their growths. The tiny serpentinization veins through forsterite also indicate a later hydration process. However, the very low amount of clinohumite and serpentine testifies that such a process is limited.

Metamorphic Fluid Evolution of the Thongmön Impure Marble
We attempted to construct the P/T-X(CO 2 ) pseudosections to evaluate the fluid composition under peak and retrograde conditions, which may provide some clues on the metamorphic evolution of impure marbles. In addition, most metamorphic reactions in metacarbonates are strongly dependent on temperature and more or less subparallel to each other in P-T space [58]. In this case, constructing the P/T-X(CO 2 ) diagrams is a direct attempt.
Forsterite-spinel impure marbles are typical for high-grade granulite-facies terrains such as southern India [30,31], northeastern Greece [58,60], and central Australia [77]. Rapa et al. (2017) [41] reported that the prograde growth of titanite in the calc-silicate rocks from the Nepal Himalaya (GHS) is closely associated with the substantial release of CO 2 . In addition, Groppo et al. (2013Groppo et al. ( , 2017 [9,33] investigated the metamorphic evolution of calcsilicate rocks from the eastern Nepal Himalaya and concluded that the formation of garnet and scapolite involved simultaneous CO 2 -producing reactions with a relatively CO 2 -rich fluid composition (X(CO 2 ) max = 0.44) at peak conditions and that estimated metamorphic CO 2 flux was about 1.1-3.4 × 10 10 mol/year. However, Castelli et al. (2007) [10] reported the impure marbles from the UHP Brossasco-Isasca Unit (Dora-Maira Massif, western Alps) and suggested that the modeled fluid compositions at peak conditions point to X(CO 2 ) ≤ 0.0012 and 0.025 ≤ X(CO 2 ) ≤ 0.1 in the dolomite marble and calcite marble, respectively. In addition, Proyer et al. (2008) [58] investigated the high-grade aluminous calcite-dolomite marbles from the Greek Rhodope massif and claimed that the assemblages preserved in these marbles most probably equilibrated at a CO2~0 .01. Similarly, in the Dabie Mountains, China, the peak fluid composition of the UHP impure marble was extremely H 2 O-rich with X(CO 2 ) < 0.02 [4,6].
Forsterite-spinel impure marbles may be a product of prograde metamorphism. We take two steps to constrain the fluid composition during prograde to the peak metamorphism. Firstly, we used the mineral assemblage at peak conditions of the impure marble to outline the X(CO 2 ) range. Combined with the petrological observation and metamorphic evolution as mentioned above, the Thongmön impure marble has a peak mineral assemblage of forsterite, spinel, calcite, dolomite, and phlogopite. Thus, it is easy to see from the T-X(CO 2 ) diagram that the composition of peak fluid is X(CO 2 ) < 0.12 (Figure 7a). Following this range, we applied the acquired calcite-dolomite solvus temperature to the T-X(CO 2 ) diagram and obtained an intersection point that corresponds to an approximate fluid composition of X(CO 2 ) < 0.04 (Figure 7b). Our modeling results thus suggest that the fluid composition of the Thongmön impure marble is relatively CO 2 -poor (X(CO 2 ) < 0.04) at the peak stage. At such high temperatures (>750 • C), metamorphic decarbonation reactions should have been robust to produce a large amount of CO 2 furnishing the hydrous fluid with higher X(CO 2 ), which is not in accordance with our modeling results and petrological observations. Thus, in this case, a weak metamorphic decarbonation process is not supported, and such high H 2 O contents of fluid at peak conditions may imply that the Thongmön forsterite marble experienced significant infiltration of external H 2 O-rich fluid, which is consistent with the results from Buick et al. (1997) and Proyer et al. (2008) [58,77].
Alternatively, forsterite-bearing marble may be formed during the decompression exhumation process. In this way, the evolution of fluid composition during isothermal exhumation gradually becomes richer in CO 2 ( Figure 8). Specifically, the former diopside-bearing marble has a H 2 O-rich (X(CO 2 ) < 0.1) fluid composition at the peak stage and then gets CO 2 -richer with the formation of forsterite by the decarbonation reaction: Di + Dol = Fo + Cal + CO 2 . Due to the continuous isothermal exhumation process, the fluid composition could be partly buffered along with the producing forsterite and spinel equilibria from X(CO 2 ) = 0.2 at~9 kbar until the diopside is nearly exhausted at X(CO 2 ) = 0.5 at~7 to 6 kbar (Figure 8b). At last, it may evolve to 0.6 < X(CO 2 ) < 0.8 at 5 kbar, 750 • C assuming that the impure marble is subordinate to the retrograde evolution of the GHS [38][39][40]. Therefore, the decompression process results in an increase in X(CO 2 ) of fluid during the exhumation of the Thongmön impure marble. An increase in X(CO 2 ) of fluid in the UHP impure marbles during isothermal exhumation from the Dabie Mountains is reported by Liu et al. (2015) [4] and Omori et al. (1998) [6]. The relatively higher X(CO 2 ) in fluid during isothermal exhumation decreased again at a late retrograde stage, which could partly be attributed to the infiltration of external aqueous fluids leading to a dilution in CO 2 of the fluid. The formation of serpentinine and clinohumite also testify to the occurrence of a hydration event. In this regard, the fluid composition of the Thongmön impure marble varies greatly in the whole metamorphic evolution

Implications for the Carbon Cycling in Orogenic Belts
Global carbon cycling plays an important role in controlling the climate change and habitability of the Earth [18,[78][79][80]. Thus, up to now, numerous scientists have paid much attention to carbon flux calculations at oceanic subduction zones [5,15,17,37]. However, in continental collision zone settings, the evaluation of carbon flux has not received enough focus in recent years. Recently, Guo et al. (2021) [36] reported that the CO 2 flux of India-Asia collision-induced metamorphic decarbonation was one order of magnitude higher than that of magmatic outgassing in the Cenozoic, which implies that continent-continent collision belts can release substantial subducted carbon and act as a net carbon source [21,22,33,34,80].
To evaluate the amount of produced CO 2 on the orogen scale, there are still some problems to solve: (1) first of all, an estimate of the total volume of carbon-source rocks in the entire Himalayan orogen is needed [12,20,80]; (2) the duration of the CO 2 released is uncertain [41,43]; and (3) channelized CO 2 emission by fault and fracture zones [13] and/or possible carbonation reactions and CO 2 sequestration by carbon precipitated as graphite from a carbon-rich fluid complicates the estimate of the CO 2 fluxes [7][8][9]. Firstly, it is very sophisticated to determine the exact volume of carbon-source rocks because of the uneven distribution in different tectonic parts of the entire Himalaya. Moreover, the duration of CO 2 -producing reactions during prograde metamorphism is poorly constrained in the whole Himalaya ranging from 10 to 30 Ma [20,41]. As for point (3), CO 2 escape of the nonvolcanic degassing predominantly along faults and fractures and the CO 2 sequestration by graphite precipitation jointly control the carbon fluxes in Himalaya, which is worthy of more in-depth analysis and discussion in future work.
The India-Asia collision-induced metamorphic decarbonation of a subducted carbonatebearing component greatly contributes to the CO 2 fluxes during the Cenozoic [20,36,[40][41][42]. The estimated flux of CO 2 from different kinds of literature is even an order of magnitude difference, which may result from distinct sample fields, calculation models, and the degassing rate of CO 2 . For example, Kerrick and Caldeira (1999) [20] reported a metamorphic CO 2 -producing rate of 9-24 Mt/year adopting the mass loss method in the whole GHS, whereas Becker et al. (2008) [35] and Evans et al. (2008) [34] gave the estimated CO 2 outgassing rate of 40 Mt/year from the Marsyandi basin of central Nepal and 8.8 Mt/year from the Narayani basin based on the carbon isotope analysis from spring waters, respectively. Rapa et al. (2017) [40] reported an estimated value of 1.4 Mt/year < CO 2 flux < 19.4 Mt/year using multiple component phase diagrams of calc-silicate rocks from the GHS. Groppo et al. (2020Groppo et al. ( , 2022 [7,8] suggested that calcic metapelite and carbonatepoor marls should be considered as likely CO 2 -source rocks when modeling the global CO 2 degassing budget in orogenic settings. Although we did not quantify the amount of releasing CO 2 from the Thongmön impure marble during the prograde, peak, and retrograde metamorphism, our study potentially makes some contribution to the Himalayan CO 2 flux in the Eocene. It is thus significant for future work to focus more on an estimate of the total volume of the Himalaya metacarbonate rock and the duration of the CO 2 released on a regional scale, which helps to quantify the contribution of an orogenic belt to the global cycling. In this study, the fluid composition of the Thongmön forsterite marble is extremely H 2 O-rich (X(CO 2 ) < 0.04 at 15 kbar) at the peak stage, which implies that it has experienced potential infiltration of external H 2 O-rich fluid. Alternatively, the fluid composition of the Thongmön forsterite marble is CO 2 -richer (0.6 < X(CO 2 ) < 0.8 at 5 kbar) due to the strong overprint of regional retrogression. Thus, we suggest that the important role of orogenic settings in the global Earth's CO 2 tectonic outgassing should be considered.

Conclusions
This study is the first systematic work on textural relations and chemical compositions combined with metamorphic decarbonation reactions, calcite-dolomite solvus geother-mometer, and the fluid evolution on the Thongmön impure granulite-facies marble, providing new insights into the metamorphic evolutions and metamorphic degassing processes of metacarbonate rocks in the central Himalaya. Carbonate inclusion in spinel (X Mg = 0.153) gives the highest peak temperature record (>780 • C), demonstrating that the Thongmön impure marble may have gone through similar metamorphic evolution as the Thongmön metapelites. The T-X(CO 2 ) pseudosection at 15 kbar shows the extremely hydrous fluid (X(CO 2 ) < 0.04) at peak condition of the Thongmön granulite-facies forsterite marble, which possibly results from the influx of external H 2 O-rich fluid. Alternatively, after the robust retrogression overprint on a regional scale, the fluid composition of forsterite marble is CO 2 -rich (X(CO 2 ) = 0.6-0.8 at 5 kbar). Thus, in either case, we suggest that the contribution of metamorphogenetic CO 2 flux of impure marbles to global carbon cycling must be considered.