This sample consists of garnet (10%), staurolite (20%), biotite (25%), muscovite (10%), plagioclase (5%), and quartz (30%) and accessory minerals, including zircon, graphite, and ilmenite (Figures 4A, B). Garnet occurs as euhedral to subhedral porphyroblasts with grain sizes of 0.4–1.1 mm in diameter (Figures 4A, B), with minor quartz inclusions. Staurolite is present as a subhedral porphyroblast with grain sizes of 0.4–1.0 mm and is generally in contact with garnet (Figures 4A, B); it displays a yellow-to-black color due to abundant graphite inclusion. Biotite and muscovite occur in the matrix and define the foliation (Figure 4A). Plagioclase occurs as relatively small crystals in the matrix, and quartz occurs as small grains in the matrix or as mineral inclusions in garnet (Figures 4A–C). As a result, the peak metamorphic mineral assemblage is inferred to be garnet–biotite–muscovite–plagioclase–staurolite–ilmenite–quartz–H₂O.

Mineral compositions and the mole fractions of end-members for sample RB12-19 are given in Supplementary Table S1. The garnet is almandine-rich (X _Alm = 0.72–0.83), with low concentrations of spessartine (X _Sps = 0.07–0.18), pyrope (X _Prp = 0.04–0.09), and grossular (X _Grs = 0.04–0.06). The zoning profile of a representative garnet grain shows obvious compositional variation from core to rim (Figure 5A), with increasing almandine and decreasing spessartine but relatively flat pyrope and grossular, which are consistent with distributions of Fe, Mg, Ca, and Mn on garnet compositional maps (Figure 6A). Staurolite has homogeneous compositions, with X _Mg of 0.08–0.10. Biotite and muscovite also exhibit homogeneous compositions, with X _Mg of 0.40–0.41, Ti of 0.10–0.12 cations per formula unit (cpfu), and Si of 3.09–3.13 cpfu, respectively. Plagioclase displays no compositional variation between different grains, with X _An of 0.27–0.31.

Sample RB12-24 is composed of garnet (10%), staurolite (10%), biotite (20%), muscovite (5%), plagioclase (20%), and quartz (35%) and accessory minerals, including zircon, graphite, and ilmenite (Figures 4C,D). Garnet occurs as subhedral porphyroblasts with grain sizes ranging from 0.5 to 2.7 mm (Figure 4D), larger than garnet in sample RB12-19 (Figures 4A,B); it contains mineral inclusions of quartz (Figure 4D). Staurolite is present as subhedral porphyroblasts with grain sizes of 0.3–1.4 mm and displays yellow-to-black color due to abundant graphite inclusion (Figures 4C,E). Both garnet and staurolite are wrapped by a continuous foliation delineated by biotite and muscovite in the matrix (Figures 4D,E). Plagioclase occurs as small grains in the matrix, and quartz occurs as small grains in the matrix or as mineral inclusions in garnet (Figures 4D,E). Therefore, the peak metamorphic mineral assemblage is inferred to be garnet–biotite–muscovite–plagioclase–staurolite–ilmenite–quartz–H₂O.

Mineral compositions and the mole fractions of end-members for sample RB12-24 are given in Supplementary Table S2. The garnet is almandine-rich (X _Alm = 0.75–0.87), with low pyrope (X _Prp = 0.03–0.04), spessartine (X _Sps = 0.01–0.08), and grossular (X _Grs = 0.09–0.14) (Figure 5B). Compared to those of the sample RB12-19, the pyrope is lower, whereas the grossular is higher (Figures 5A,B). The compositional profile of a representative garnet porphyroblast shows moderate variation from the core to the rim (Figure 5B), involving increasing almandine and decreasing spessartine but flat pyrope and grossular. These are consistent with distributions of Fe, Mg, Ca, and Mn on garnet compositional maps (Figure 6B). Staurolite is uniform in composition, with X _Mg ranging from 0.07 to 0.08. Biotite and muscovite also have homogeneous compositions, with X _Mg of 0.33–0.35, Ti of 0.10–0.12 cpfu, and Si of 3.08–3.12 cpfu, respectively. Plagioclase displays moderate compositional variation between different grains and is more calcic than that in sample RB12-19, with X _An of 0.38–0.53.

The peak phase assemblage is represented by the quinivariant field grt–bt–ms–pl–st–ilm–qz–H₂O, which has a wide P–T range, with upper pressure and temperature of 670°C and 7 kbar, respectively (Figure 7A). The chlorite-in and plagioclase-out assemblage field boundaries mark the low-temperature limit of this field. The upper temperature and pressure limits are the sillimanite-in and rutile-in assemblage field boundaries, respectively.

Figure 7B shows calculated isopleths of X _Sps and X _Grs for garnet, X _Mg for staurolite, and X _Ti for biotite. In the peak phase assemblage field of grt–bt–ms–pl–st–ilm–qz–H₂O, the X _Sps in garnet decreases significantly as pressure increases, and X _Ti in biotite increases as temperature increases. Isopleths of X _Ti(Bt) of 0.10–0.12 are consistent with Ti contents of biotite (Supplementary Table S1). The plotted isopleths of X _Grs for garnet are close to vertical and decrease as temperature increases. The X _Mg in staurolite ranges from 0.08 to 0.10 (Supplementary Table S1), with corresponding isopleths plotted in the phase assemblage fields of grt–ms–chl–pl–st–ilm–qz–H₂O and grt–bt–ms–chl–pl–st–ilm–qz–H₂O. As no prograde and retrograde metamorphic mineral assemblages were observed, isopleths of X _Sps and X _Grs for garnet, X _Mg for staurolite, and X _Ti for biotite were used to constrain the P–T evolution. Plagioclase is stable in the studied sample, and thus, mineral assemblages during the early prograde metamorphic stage should contain plagioclase. Therefore, the X _Mg isopleths of staurolite in the mineral assemblage fields that contain plagioclase can be used to constrain the P–T conditions in the early stage of prograde metamorphism, although the defined P–T range is relatively large (Figure 7B). X _Sps and X _Grs in garnet and X _Ti in biotite constrain the peak metamorphic P–T conditions to be ∼630°C at ∼5.8 kbar in the peak mineral assemblage field of grt–bt–ms–pl–st–ilm–qz–H₂O. As a result, a prograde P–T path that evolves roughly from ∼540°C at ∼4 kbar to ∼630°C at ∼5.8 kbar was defined for the sample RB12-19 (Figure 7B).

The peak phase assemblage of grt–bt–ms–pl–st–ilm–qz–H₂O is represented by a quinivariant field, which occurs between 552 and 668°C and 4.7–6.9 kbar (Figure 8A). The sillimanite-in assemblage field boundary marks the upper-temperature limit of this field, and the rutile-in assemblage field boundary is the upper-pressure limit. The low temperature and pressure limits are the chlorite-in and muscovite-out assemblage field boundaries, respectively.

Figure 8B shows calculated isopleths of X _Sps and X _Grs for garnet and X _Ti for biotite. The isopleths of X _Sps in garnet are horizontal and decrease as pressure increases, and the isopleths of X _Ti in biotite are vertical and increase as temperature increases, in the peak phase assemblage field (grt–bt–ms–pl–st–ilm–qz–H₂O). Ti contents of biotite correspond to isopleths of X _Ti(Bt) of 0.10–0.12 (Supplementary Table S2). In the phase assemblage field of grt–bt–ms–pl–chl–ilm–qz–H₂O, the isopleths of X _Sps in garnet are close to horizontal and decrease as temperature increases. The plotted isopleths of X _Grs for garnet are near vertical and decrease as temperature increases. Similar to sample RB12-19, isopleths of X _Sps and X _Grs for garnet and X _Ti for biotite were used to constrain the P–T evolution. The X _Sps and X _Grs in the garnet core and mantle have crossover points in the mineral assemblage field grt–bt–ms–chl–pl–ilm–qz–H₂O, which constrain the P–T conditions in the early stage of prograde metamorphism to be ∼540°C at ∼4.4 kbar (Figure 8B). In addition, the crossover points of X _Sps and X _Grs in the garnet rim fall into the peak phase assemblage field of grt–bt–ms–pl–st–ilm–qz–H₂O, which is close to or within the P–T range defined by isopleths of X _Ti for biotite (Figure 8B). As a result, a prograde P–T path that evolves from ∼540°C at ∼4.4 kbar to ∼620°C at ∼6.3 kbar was acquired for the sample RB12-24 (Figure 8B).

A total of 19 spot analyses were carried out on biotite in sample RB12-19. During a single spot analysis, spots 1, 6, and 7 yielded multiple intervals with obviously different ratios of ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr, which imply biotite may have inhomogeneous Rb–Sr isotopic compositions. Thus, these spot analyses obtained more than one effective data (Table 2). The Rb–Sr isotopic data of 3, 6.2, 7.2, 8, 14, 16, 17, and 18 were excluded during age calculation as these data are scattered on the ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr isochron diagram and cannot be well-included into an isochron (Figure 9A). The remaining ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr ratios are 3.7942–131.847 and 0.7062–0.7866, respectively, which yielded an isochron age of 37.17 ± 5.66 Ma (n = 16, MSWD = 1.5).

Similar to sample RB12-19, 19 spot analyses were carried out on biotite in sample RB12-24. Spot 17 yielded multiple intervals with obviously different ratios of ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr, which obtained two effective datasets (Table 2). The ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr ratios of biotite in sample RB12-24 range from 0.763 to 82.637 and 0.711 to 0.729, respectively (Table 2). On the ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr isochron diagram, these data yielded an isochron age of 5.27 ± 3.10 Ma (n = 20, MSWD = 1.9) (Figure 9B). Obviously, this age is younger than that of sample RB12-19, although samples RB12-19 and RB12-24 were collected at the same outcrop, which may be due to limited spot analyses for each sample. In fact, sample RB12-19 has two Rb–Sr isotopic trends (Figure 9A). The first trend yielded an age of 37.17 ± 5.66 Ma, whereas the second trend did not yield believable age due to the lack of low ⁸⁷Rb/⁸⁶Sr data (Figure 9A). We found that if the Rb–Sr data within the second trend of sample RB12-19 were pooled together with the data of sample RB12-24, they could give a young age of 4.99 ± 1.30 Ma (not shown), with a small error.

Previous dating on Barrow-type metamorphic rocks in the NHGD generally yielded Oligocene to Miocene ages (Stearns et al., 2013; Wang et al., 2018). For example, LA–ICP–MS monazite U/Th-Pb analysis constrained the timing of metamorphism to be 29–14 Ma for grt + bt ± st ± ky ± sil schists from the Kangmar and Mabja gneiss domes (Stearns et al., 2013), and SHRIMP monazite U/Th-Pb analysis yielded ages of ca. 18 Ma for grt + bt + st ± ky ± sil schists in the Yardoi gneiss dome (Wang et al., 2018). These ages were interpreted as the timing of peak Barrow metamorphism recorded in the NHGD (Stearns et al., 2013; Wang et al., 2018), whereas prograde metamorphism occurred as early as 54–49 Ma based on garnet Lu–Hf analysis (Smit et al., 2014). In addition, monazite U/Th-Pb geochronology of the Gianbul dome in the GHC yielded both Eocene (37–33 Ma) and Miocene (26–22 Ma) ages, which were interpreted as the timing of prograde Barrovian metamorphism and doming driven by upper-crustal extension and positive buoyancy of decompression melts, respectively (Horton et al., 2015).

In this study, in situ LA–ICP–MS biotite Rb–Sr dating yielded two metamorphic ages of 37.17 ± 5.66 and 5.27 ± 3.10 Ma for the garnet–staurolite–two-mica schist in the Ramba gneiss dome (Figure 9). As the peak metamorphic temperatures of the garnet–staurolite–two-mica schists in the Ramba gneiss dome are higher than the closure temperature of the Rb–Sr system in biotite (∼300–400°C) (Verschure et al., 1980; Willigers et al., 2004 and references therein; Scibiorski et al., 2021), the age of 37.17 ± 5.66 Ma is interpreted to represent the timing of retrograde cooling, rather than the peak metamorphism. This interpretation is obviously inconsistent with results from the Kangmar, Mabja, and Yardoi gneiss domes in the Northern Himalayas and the Gianbul dome in the GHC. However, Laskowski et al. (2016) have conducted geochronological research on HP meta-Tethyan rocks in the Lopu Range, located ∼600 km west of the city of Lhasa, yielding a garnet Lu-Hf age of 40.4 ± 1.4 Ma and five Ar-Ar phengite ages between 39 and 34 Ma, which were interpreted as the timing of prograde metamorphism and exhumation to mid-crustal depths (∼25 km) and concomitant retrogression in the Himalayan orogen, respectively. Khanal et al. (2021) presented new monazite petrochronology for the Kathmandu Klippe in the central Nepalese Himalayas and revealed that Eocene prograde metamorphism and partial melting occurred at 44–38 Ma and 38–35 Ma, respectively (Figure 10). Although the meta-Tethyan rocks in the Lopu Range underwent HP metamorphism and the Kathmandu Klippe belong to the upper or uppermost Greater Himalayan Crystallines, these results support that part of the middle-to-lower crustal rocks in the Himalayan orogen underwent exhumation and were not overprinted by Oligocene to Miocene peak Barrow metamorphism.

The age of 5.27 ± 3.10 Ma is significantly younger than the crystallization ages (ca. 44 Ma and ca. 28 Ma) of granitic rocks that intruded into the margin of the Ramba gneiss dome (Liu et al., 2014, 2019). However, this age is slightly younger than the crystallization age (ca. 8 Ma) of the leucogranite pluton occupying the core of the dome (Liu et al., 2014) but is similar to ⁴⁰Ar/³⁹Ar cooling ages (ca. 6 Ma) of the leucogranite pluton (Guo et al., 2008). Therefore, the age of 5.27 ± 3.10 Ma is considered to represent the cooling age of the dome, following the emplacement of the ca. 8 Ma leucogranites.

In this study, we have determined P–T paths for the garnet–staurolite–two-mica schists in the Ramba gneiss dome (Figures 7B, 8B). Sample RB12-19 recorded a prograde P–T path that evolves roughly from ∼540°C at ∼4 kbar to ∼630°C at ∼5.8 kbar (Figure 7B). The prograde P–T path for sample RB12-24 evolves from ∼540°C at ∼4.4 kbar to ∼620°C at ∼6.3 kbar (Figure 8B), which displays a slightly higher temperature than those of sample RB12-19 at similar pressure (Figures 7B, 8). When the results of the two samples are combined, it is clear that they experienced an obviously early heating burial path (Figure 10). It can be noted that sample RB12-24 recorded a nearly isobaric heating path (Figure 8B), and thus, a clock-wise P–T evolution is inferred to the garnet–staurolite–two-mica schist in the Ramba gneiss dome (Figure 10).

Here, we compared this P–T path with those retrieved from the Mabja and Yardoi gneiss domes in the northern Himalaya, the hinterland GHC, and the foreland Kathmandu Klippe (Figure 10). The migmatite sample of the sillimanite zone in the Mabja gneiss dome has a peak metamorphic condition of 8.2 kbar/705°C (Lee et al., 2004), which is higher in both pressure and temperature than those of the garnet–staurolite–two-mica schist in the Ramba gneiss dome (Figure 10). The garnet–staurolite–kyanite schist in the Yardoi gneiss dome has a peak metamorphic mineral assemblage of grt–bt–ms–pl–st–ky–ilm–qz, occurring in a narrow P–T condition of 7.2–8.0 kbar and 640–645°C, and the prograde P–T condition was constrained to be ∼540°C at ∼4.6 kbar using compositional isopleths X _Mg (0.07) and X _Mn (0.17) of the garnet core (Ding et al., 2016b). Combined with the retrograde metamorphic process constrained using the occurrence of sillimanite and biotite in the shear bands and the presence of chlorite, Ding et al. (2016b) obtained a clockwise P–T path for the garnet–staurolite–kyanite schist in the Yardoi gneiss dome, which is similar to that constrained by Wang et al. (2018). This P–T path is characterized by a prograde process with both P and T increasing and a retrograde process with an early nearly isothermal decompression and late cooling and decompression. It is obvious that the garnet–staurolite–two-mica schist in the Ramba gneiss dome has a similar prograde P–T condition to that of the schist in the Yardoi gneiss dome but a lower pressure condition at P _max (Figure 10). In this study, the retrograde P–T evolution was not supported by the petrological evidence and was only inferred. The P–T path of the Kathmandu Klippe in the central Nepalese Himalayas is clockwise, with peak P–T conditions of 730–760°C and up to 10.5 kbar, which is higher in both pressure and temperature than those of the garnet–staurolite–two-mica schist in the Ramba gneiss dome but is similar to those from the GHC hinterland (Figure 10) (Iaccarino et al., 2015).

Ultrahigh pressure (UHP) metamorphic rocks at Kaghan Valley and Tso Morari in the west of the northern Himalayan belt have peak metamorphic ages of 46.2 ± 0.7 Ma and ca. 47–43 Ma, respectively (Kaneko et al., 2003; Donaldson et al., 2013). These suggest deep subduction of the Indian continent at ca. 47–43 Ma. The crustal thickening in the Tibetan Himalaya is broadly synchronous with the Eocene collision between the Indian and Asian plates (Smit et al., 2014). Following the crustal thickening, the deeply buried Indian continental crust underwent a long-lived partial melting (Wang et al., 2013; Zhang et al., 2015), which formed the ca. 48 to 8 Ma granites or leucogranites in the northern Himalayan belt (Aikman et al., 2008; Zhang et al., 2012; Liu et al., 2014; Zeng et al., 2011; Gao et al., 2012; Wu et al., 2015; Zeng and Gao, 2017 and references). In the Ramba gneiss dome, the partial melting is reflected by three epidotes (ca. 44 Ma, ca. 28 Ma, and ca. 8 Ma) of granitic magmatism (Liu et al., 2014, 2019). On the basis of the similarities in Eocene metamorphic conditions and timing of anatexis in the different Eocene units across the Himalayas, Khanal et al. (2021) concluded that an early-stage anatectic response to crustal thickening might be more common than previously thought.

The onset of extensional tectonics of the STDS occurred as early as ca. 36–30 Ma, which marks the initial thinning of thickened crust in the northern Himalayan belt (Zhang et al., 2012; La Roche et al., 2016). In the Ramba gneiss dome, the STDS is represented by the first episode of deformation with top to-NNW sliding, indicated by S–C fabric and NNW-convergent tight folds on the northwestern flank of the dome (Guo et al., 2008; Zhang et al., 2012). The clock-wise P–T evolution of the garnet–staurolite–two-mica schist in the Ramba gneiss dome, combined with the age of 37.17 ± 5.66 Ma from the in situ LA–ICP–MS biotite Rb–Sr dating, is consistent with the Eocene crustal thickening (Figure 10).

The ca. 8 Ma leucogranites occupy the core of the Ramba gneiss dome (Liu et al., 2014) and have ⁴⁰Ar/³⁹Ar cooling ages of ca. 6 Ma (Guo et al., 2008). The Yadong–Gulu rift, on the east of the dome, was active during the interval ca. 11–5 Ma, with its peak activity at ca. 8 Ma (Pan and Kidd, 1992; Harrison et al., 1995; Zhang et al., 2012), which coincides with the diapir of the ca. 8 Ma leucogranite pluton, formed the shape of the Ramba gneiss dome (Guo et al., 2008; Zhang et al., 2012; Liu et al., 2014).