Ultrahigh-Temperature Mafic Granulites in the Rauer Group, East Antarctica: Evidence from Conventional Thermobarometry, Phase Equilibria Modeling, and Rare Earth Element Thermometry

Abstract

The Rauer Group in East Antarctica is a typical high- to ultrahigh-temperature (HT–UHT) granulite-facies terrane. As UHT metamorphism has been recognized only in Mg–Al-rich pelitic granulites from the Mather Paragneiss, the regional extent of UHT metamorphism remains uncertain, which has hindered our understanding of the genesis and tectonic setting of UHT metamorphism in the Rauer Group. In this study, representative samples of mafic granulite were selected from Archean crustal domains to constrain the peak metamorphic conditions and P–T path and to assess the regional extent of UHT metamorphism in the Rauer Group. Integrated results from mineral reaction histories, thermobarometry, and phase equilibria modeling indicate a multi-stage clockwise P–T evolution for mafic granulites involving pre-peak compression, heating to UHT peak conditions, post-peak near-isothermal decompression under UHT conditions, and subsequent decompressional cooling. The pre-peak prograde history is based mainly on the inclusion assemblage of clinopyroxene + plagioclase + amphibole + quartz + ilmenite ± orthopyroxene ± k-feldspar within porphyroblastic garnet and clinopyroxene and records the transformation from a quartz-present to quartz-absent system. The UHT peak conditions are well constrained at 930°C–1030°C and 10.6–12.8 kbar on the basis of the stability field of the observed peak assemblage of (orthopyroxene–quartz)-free garnet + clinopyroxene + plagioclase + amphibole + ilmenite + melt, as well as measured mineral compositions, including the high Ti content in amphibole (Ti = 0.38–0.42 p.f.u.), the anorthite content of coarse-grained plagioclase cores (X_An = 0.35–0.42), and the grossular content in garnet (X_Grs = ~0.21) in P–T pseudosections. The peak T conditions are consistent with thermometric estimates in the range of 930°C–1030°C obtained from garnet–clinopyroxene, garnet–orthopyroxene, and Ti-in-amphibole thermometers, and are slightly lower than estimates (1020°C–1120°C) obtained from thermometers based on rare earth elements. The near-isothermal decompression under UHT conditions can be divided into two stages. The early stage is recorded by coronae of orthopyroxene + plagioclase around clinopyroxene and core–mantle/rim anorthite-increasing zoning in plagioclase. The late stage is identified from symplectites of orthopyroxene + plagioclase ± amphibole around porphyroblastic garnet, which were formed at the expense of garnet at 915°C–950°C and 7.6–8.2 kbar as inferred from the amphibole–plagioclase thermometer. The subsequent decompressional cooling to fluid-absent solidus conditions (~875°C and ~6.5 kbar) is indicated by the growth of biotite, which formed at the expense of symplectic minerals, reflecting back-reaction of melt with symplectite minerals. The peak UHT metamorphic conditions and clockwise P–T path of the studied mafic granulites from the Archean crustal domains are similar to those of Mg–Al-rich pelitic UHT granulites from the Mather Paragneiss. The UHT conditions recorded by the mafic granulites, combined with previously identified isolated UHT localities in the Rauer Group, imply that UHT metamorphism in the Rauer Group occurred over a much wider region than previously thought and probably extends over the whole Archean crustal domain. Our findings have general significance in understanding the regional extent of other UHT granulite-facies terranes worldwide.

INTRODUCTION

Ultrahigh-temperature (UHT) metamorphism is an extreme type of regional crustal metamorphism and involves elevated temperatures in excess of 900°C (Harley, 1998a, 2008, 2021; Kelsey, 2008; Kelsey & Hand, 2015). Although UHT metamorphism has been recognized at nearly 70 localities globally, most UHT occurrences are isolated or relatively restricted in spatial extent (Kelsey & Hand, 2015; Harley, 2016, 2021). This local preservation of UHT metamorphism and the difficulty involved in recognizing such metamorphism have hampered our understanding of its nature and extent. Generally, the recognition of UHT metamorphism is based on diagnostic mineral assemblages, such as sapphirine + quartz, orthopyroxene + sillimanite ± quartz, and osumilite in Mg–Al-rich granulites (Harley, 1998a, 2008, 2021; Kelsey, 2008; Kelsey & Hand, 2015). Unfortunately, such Mg–Al-rich rocks are volumetrically rare in nature (Chinner & Sweatman, 1968; Kelsey, 2008; Kelsey & Hand, 2015). For UHT granulites without diagnostic mineral assemblages, peak temperatures have generally been underestimated owing to the modification of mineral assemblages and compositions during post-peak cooling and melt crystallization (White & Powell, 2002; Korhonen et al., 2013; Li & Wei, 2016). Moreover, rapid Fe–Mg diffusion during slow cooling may further modify the compositions of Fe–Mg-bearing minerals (Frost & Chacko, 1989; Pattison et al., 2003).

The relative rarity of diagnostic UHT assemblages and the modification of mineral assemblages and compositions present challenges in constraining the peak temperature and regional extent of UHT metamorphism. Recently, multiple thermobarometric methods have been shown to be effective for determining extreme metamorphic temperatures, including Al-in-orthopyroxene thermometry (e.g. Harley & Motoyoshi, 2000; Sajeev & Osanai, 2004; Harley, 2008; Clark et al., 2019; Dharmapriya et al., 2020), ternary feldspar thermometry (e.g. Hokada, 2001; Jiao & Guo, 2011; Ague & Eckert Jr., 2012; Ague et al., 2013; Cai et al., 2020; Wang et al., 2022; Zou et al., 2022), Ti-in-quartz/zircon and Zr-in-rutile thermometry (e.g. Baldwin et al., 2007; Jiao et al., 2011; Ague & Eckert Jr., 2012; Mitchell & Harley, 2017; Hart et al., 2018; Zou et al., 2022), Ti-in-amphibole thermometry (e.g. Liao & Wei, 2019; Liao et al., 2021), rare earth element (REE)-in-clinopyroxene–orthopyroxene–plagioclase thermometry (e.g. Yang & Wei, 2017; Liu & Wei, 2018; Ray et al., 2021; Gou et al., 2022), and phase equilibria modeling using isopleths of slow-diffusing components (e.g. Li & Wei, 2016; Feisel et al., 2018; Gou et al., 2018; Liu & Wei, 2018; Clark et al., 2019; Liao & Wei, 2019; Wang et al., 2020; Dharmapriya et al., 2020). With the use of these methods, an increasing number of UHT rocks lacking diagnostic mineral assemblages have been identified (Kelsey & Hand, 2015; Harley, 2021). These effective UHT identification methods have thus provided opportunities to re-examine the regional extent of UHT metamorphism in relatively isolated UHT localities or areas of proven restricted extent.

The Rauer Group is a granulite-facies terrane (i.e. the Rauer Terrane; Harley, 2003) on the Prydz Bay coast, East Antarctica (Fig. 1), and consists predominantly of reworked Archean and Mesoproterozoic felsic orthogneiss with layered paragneiss and mafic granulites that have undergone HT–UHT granulite-facies metamorphism (Harley, 1988, 1998b; Kinny et al., 1993; Kelsey et al., 2003a; Tong & Wilson, 2006). On the basis of the occurrence of orthopyroxene–sillimanite–quartz- and sapphirine-bearing assemblages, UHT metamorphism has been recognized in Mg–Al-rich rocks from the Mather Paragneiss, which crop out only on Mather Peninsula and at Short Point (Harley & Fitzsimons, 1991). Peak P–T conditions have been constrained in the range of 950°C–1050°C and 9.5–12.0 kbar using both conventional thermobarometers and a P–T petrogenetic grid (Harley & Fitzsimons, 1991; Harley, 1998b; Kelsey et al., 2003a; Tong & Wilson, 2006). Mineral reaction textures suggest a clockwise P–T path along near-isothermal decompression under UHT conditions (Harley & Fitzsimons, 1991; Harley, 1998b; Tong & Wilson, 2006) or decompressional cooling (Kelsey et al., 2003a). The other ‘normal’ granulites, consisting of orthogneiss, mafic granulite, and paragneiss (i.e. the Filla Paragneiss), occupy the majority of the Rauer Group. These granulites are characterized by a typical granulite-facies mineral assemblage and have been considered to have undergone regional ‘normal’ granulite-facies metamorphism with peak temperatures of ~850°C as estimated using conventional thermobarometric estimations and phase relations (Harley, 1988; Harley & Fitzsimons, 1991; Harley & Buick, 1992; Buick et al., 1993; Harley & Fitzsimons, 1995). However, some studies have argued on the basis of field relationships and petrological analysis that the Filla Paragneiss has a shared metamorphic history with the Mather Paragneiss and that UHT metamorphism may be more widespread in the Rauer Group than previously thought (Dirks & Wilson, 1995; Sims, 1999; Kelsey et al., 2003a; Tong & Wilson, 2006). Consequently, it remains unclear whether the ‘normal’ granulites in the Rauer Group record UHT conditions and whether the UHT metamorphism is recorded only in the Mather Paragneiss or affected the entire Rauer Group.

Mafic granulite is an important rock type in granulite-facies terranes on account of its high sensitivity to P–T variations and its suitability for estimating the P–T conditions and path of metamorphism. Mafic granulites are widely distributed in the Rauer Group, where they occur as boudins or enclaves in both felsic orthogneiss and paragneiss (Fig. 2a and b). As these mafic granulites are variably deformed and occur essentially parallel to felsic orthogneiss and paragneiss fabrics in high-strain zones (Fig. 2c and d) (Harley, 1987, 1988; Sims et al., 1994; Mikhalsky et al., 2019), their metamorphic evolution should be representative of that of the host rocks. In this study, we focus on mafic granulites from the Archean crustal domain in the Rauer Group, for which we present an integrated thermobarometric investigation involving detailed petrography, mineral chemistry, conventional thermobarometry, phase equilibrium modeling, and REE-based thermometry. Our results reveal that the mafic granulites have undergone UHT metamorphism with peak conditions of 930°C–1030°C and ~12 kbar, and a clockwise P–T path. These newly estimated UHT conditions indicate that UHT metamorphism in the Rauer Group occurred over a much wider region than previously thought and probably extends over the whole Archean crustal domain.

GEOLOGICAL SETTING

The Rauer Group is exposed along the northeastern coastline of Prydz Bay and comprises an Archean crustal domain, a Mesoproterozoic crustal domain, and the Mather Paragneiss (Fig. 1). The Archean domain, which occupies the eastern and southern Rauer Group, consists mainly of tonalitic to granitic orthogneisses with ages of 3470–3270 and 2840–2800 Ma (Kinny et al., 1993; Harley et al., 1998). The Mesoproterozoic domain, which dominates the Rauer Group, comprises mafic–intermediate–felsic intrusive rocks with ages of 1400–1000 Ma (Kinny et al., 1993; Alexeev et al., 2016; Mikhalsky et al., 2019; Liu et al., 2021) that are interlayered with paragneisses (i.e. the Filla Paragneiss; Sims et al., 1994; Harley et al., 1998). The Mather Paragneiss appears at limited outcrops as thin and laterally discontinuous horizons from Mather Peninsula to Short Point in the eastern Rauer Group and is composed of Mg–Al-rich metapelite, garnet/orthopyroxene–sillimanite quartzite, orthopyroxene-bearing leucogneiss, and forsteritic marble (Harley & Fitzsimons, 1991; Harley, 1998b; Kelsey et al., 2003a; Tong & Wilson, 2006). Although the depositional age of the Mather Paragneiss is uncertain (Wang et al., 2008), detrital zircon age data indicate deposition during the Neoproterozoic (Kelsey et al., 2008; Hokada et al., 2016). Zircon and monazite age data suggest that the Archean domain and Mather Paragneiss underwent a single late Neoproterozoic–Cambrian (i.e. Pan-African age; 590–500 Ma) metamorphic cycle, whereas the Mesoproterozoic domain underwent both early Neoproterozoic (i.e. Grenvillian age; 1000–900 Ma) and late Neoproterozoic–Cambrian high-grade metamorphism (Kinny et al., 1993; Hensen & Zhou, 1997; Harley et al., 1998, 2009; Sims et al., 2001; Kelsey et al., 2003b, 2007, 2008; Harley & Kelly, 2007; Hokada et al., 2016; Liu et al., 2021).

The Mather Paragneiss preserves orthopyroxene + sillimanite and sapphirine-bearing diagnostic mineral assemblages and has undergone UHT metamorphism with peak conditions of 950°C–1050°C and 9.5–12.0 kbar (Harley & Fitzsimons, 1991; Harley, 1998b; Kelsey et al., 2003a, 2007; Tong & Wilson, 2006). Harley & Fitzsimons (1991) and Harley (1998b) provided evidence for a clockwise P–T path involving isothermal decompression under UHT conditions. In contrast, Kelsey et al. (2003a, 2007) argued that the post-peak thermal evolution was dominated by decompression–cooling. The age of UHT metamorphism remains ambiguous. Monazite U–Th–Pb isotope ages and the ca. 600 Ma Sm–Nd garnet–whole-rock isochron suggest that UHT metamorphism occurred prior to 590–580 Ma and was an independent event separate from the Pan-African tectonic event (545–510 Ma) (Hensen & Zhou, 1997; Harley et al., 2009; Hokada et al., 2016). However, microstructure-controlled in situ U–Th–Pb monazite dating results suggest a peak metamorphic age of ~540 Ma (Kelsey et al., 2003b, 2007).

The Archean and Mesoproterozoic crustal domains were intruded by several generations of abundant mafic dikes. These dikes have been metamorphosed into mafic granulites and variably deformed, and are essentially parallel to felsic and composite gneiss fabrics in high-strain zones (Harley, 1987; Harley & Fitzsimons, 1991; Sims et al., 1994; Harley et al., 1995). The protolith age of the mafic granulites remains unknown. On the basis of geochemistry, isotopic dating, cross-cutting relationships, and orientation, the protolith age of the mafic granulites has been inferred as Archean (~2.8 Ga) or Mesoproterozoic (ca. 1400–1000 Ma) (Harley, 1987; Sims et al., 1994; Harley et al., 1998; Alexeev et al., 2016; Mikhalsky et al., 2019; Liu et al., 2021). Harley (1988) systematically summarized the mineral assemblages and reaction textures of the mafic granulites, and using conventional thermobarometers obtained peak metamorphic conditions of 860 ± 40°C and 6.0–8.5 kbar along a clockwise P–T path involving isothermal decompression.

ANALYTICAL METHODS

Bulk-rock compositions were measured by X-ray fluorescence analysis (PW4400) at the National Research Center for Geoanalysis, Chinese Academy of Geological Sciences, Beijing, China. Relative standard deviations of these analyses are better than 5%. FeO contents were determined by Fe²⁺ titration, and Fe₂O₃ contents were calculated by difference. Loss on ignition (LOI) was calculated as the weight loss on heating to 105°C, and H₂O⁺ was measured as the loss over the interval 105°C–1050°C. Results for selected samples are listed in Table 1.

Mineral chemical compositions were determined using a JEOL JXA–8100 wavelength-dispersive electron microprobe at the Institute of Geology, Chinese Academy of Geological Sciences, Beijing, China. Analyses were conducted using an accelerating voltage of 15 kV, a beam current of 2 × 10⁻⁸ A, and a peak counting time of 10 s. A beam diameter of 2 μm was used for most minerals except for thin symplectic minerals (1 μm). Natural and synthetic minerals were used as standards, and ZAF corrections were applied. Ferric iron in garnet and clinopyroxene was determined by charge balance. The molecular formula for amphibole was calculated after Holland & Blundy (1994), with corrections given by Dale et al. (2000). Representative mineral analyses are presented in Tables 2–6. A summary of mineral chemistry data from two types of mafic granulites is presented in Table 7. Mineral abbreviations used in this study are after Whitney & Evans (2010).

Contents of REEs in clinopyroxene, orthopyroxene, and plagioclase were determined using laser-ablation–inductively coupled plasma–mass spectrometry (LA–ICP–MS) at Wuhan Sample Solution Analytical Technology, Wuhan, China. Detailed operating conditions for the LA system and the ICP–MS instrument, as well as procedures for data reduction, followed those described by Zong et al. (2017). Laser sampling was performed using a GeolasPro LA system that consists of a COMPexPro 102 ArF excimer laser (wavelength of 193 nm and maximum energy of 200 mJ) and a MicroLas optical system. An Agilent 7700e ICP–MS instrument was used to acquire ion-signal intensities. Helium was used as the carrier gas. Argon was used as the make-up gas and was mixed with the carrier gas via a T-connector before entering the ICP. A ‘wire’ signal smoothing device is included in this LA system (Hu et al., 2015). The spot size and frequency of the laser were set to 44 μm and 5 Hz, respectively, for the period of analysis. Trace-element compositions of minerals were calibrated against various reference materials (BHVO–2G, BCR–2G, and BIR–1G) without using an internal standard (Liu et al., 2008). Each analysis incorporated a background acquisition of 20–30 s followed by 50 s of data acquisition from the sample. The Excel-based software program ICPMSDataCal was used to perform off-line selection and integration of background and analyzed signals, time-drift correction, and quantitative calibration for trace-element analysis (Liu et al., 2008). Relevant data are presented in Supplementary Table 1.

PETROGRAPHY AND MINERAL CHEMISTRY

Mafic granulites from the Archean crustal domain can be subdivided into two types based on mineralogy and reaction texture: type 1 mafic granulites and type 2 mafic granulites. Type 1 granulites show a near-equigranular texture with less modal amphibole (Fig. 2e). Type 2 granulites record pronounced retrograde reaction textures with high modal amphibole (Fig. 2f). The two types of mafic granulite have similar bulk-rock compositions, with X_Mg [=Mg/(Mg + Fe²⁺)] of 43–52 and X_Ca [=Ca/(Ca + Mg + Fe²⁺)] of 0.29–0.36 (Table 1). The differences in mineralogy and reaction texture between the two types may record information relating to distinct stages of P–T evolution. The petrology and mineralogy of five representative samples, namely, two type 1 samples (TKL02-2 and TKL18-4 from Torckler Island) and three type 2 samples (MS02-3, MS04-1, and MS07-3 from Mather Peninsula) (Fig. 1), were carefully examined and are described below (Figs 3–6; Table 7).

Type 1 mafic granulite with equilibrium texture

Type 1 mafic granulites are garnet two-pyroxene granulites with a near-equigranular texture and consist of garnet (20–25 vol%), clinopyroxene (25–30 vol%), orthopyroxene (5–8 vol%), amphibole (5–8 vol%), plagioclase (30–40 vol%), and ilmenite (1–3 vol%) (Fig. 3a). Garnet occurs as xenomorphic grains measuring 0.2–0.8 mm across and commonly contains inclusions of clinopyroxene, amphibole, plagioclase, quartz, and ilmenite (Fig. 3b and e). Garnet has X_Alm [= Fe²⁺/(Ca + Mg + Fe²⁺ + Mn)] of 0.51–0.59, X_Pyp [= Mg/(Ca + Mg + Fe²⁺ + Mn)] of 0.19–0.26, X_Grs [= Ca/(Ca + Mg + Fe²⁺ + Mn)] of 0.18–0.22, X_Sps [= Mn/(Ca + Mg + Fe²⁺ + Mn)] of ~0.02, and X_Mg of 0.22–0.34, without obvious zoning except for a slight change in the outer rim (Table 2). Clinopyroxene occurs mainly as matrix grains measuring 0.2–2.0 mm across and contains numerous narrow and closely spaced exsolution lamellae of orthopyroxene (Fig. 3b and e). Clinopyroxene has a diopsidic composition with Wo [= Ca/(Ca + Mg + Fe²⁺)] of 0.45–0.49, En [= Mg/(Ca + Mg + Fe²⁺)] of 0.32–0.34, and Fs [= Fe²⁺/(Ca + Mg + Fe²⁺)] of 0.17–0.22 (Table 3). Most grains have high Ca^M2 [Ca in the M2 site] contents of 0.83–0.90 and X_Mg of 0.60–0.65, whereas the domain with few orthopyroxene exsolution lamellae in the cores of some coarse grains (2.0 mm across) shows lower Ca^M2 contents of 0.77–0.80 and X_Mg of 0.58–0.60 (Table 3; Fig. 5b), which may be very close to the composition of pre-exsolved clinopyroxene. The clinopyroxene inclusions in garnet have Ca^M2 of 0.87–0.89 and X_Mg of ~0.65 (Table 3). Orthopyroxene occurs as matrix grains measuring 0.1–0.5 mm across either adjacent to clinopyroxene or as coronae around clinopyroxene (Fig. 3c and d). Of note, the matrix orthopyroxene grains in all cases occur adjacent to clinopyroxene (Fig. 3c). The two types of orthopyroxene have similar ferrosilite compositions with En of 0.47–0.49, Fs of 0.49–0.51, Wo of ~0.02, and X_Mg of 0.48–0.52 (Table 4; Fig. 5c). Amphibole is mostly irregular in shape and exhibits straight contact boundaries with clinopyroxene and plagioclase, suggesting an equilibrium paragenesis (Fig. 3a and d). Amphibole has a ferropargasite and pargasite composition with Ti of 0.36–0.42 p.f.u. and X_Mg of 0.43–0.52, whilst a few amphibole inclusions in garnet show a pargasite composition with similar Ti contents of 0.36–0.39 p.f.u. and higher X_Mg of 0.50–0.54 (Table 5; Fig. 5d). Plagioclase occurs as coarse grains (Pl-1) measuring 0.5–2.5 mm across, fine grains (Pl-2) measuring <0.5 mm across in the matrix, and inclusions in garnet (Fig. 3b and d). Pl-1 grains have obvious compositional zoning, with increasing X_An from core (0.34–0.45) to rim (0.45–0.54) (Table 6; Fig. 6a). Pl-2 grains have X_An of 0.52–0.60, similar to the rims of Pl-1 grains (Table 6; Fig. 6b and c). The plagioclase inclusions in garnet have X_An of 0.37–0.50 (Table 6).

Textural relationships and mineral compositions allow three generations of mineral assemblage to be recognized. The first generation (M₀) is represented by preserved inclusions of clinopyroxene, plagioclase, amphibole, quartz, and ilmenite within garnet porphyroblasts, and is interpreted as a prograde assemblage. The second generation (M₁) is characterized by abundant garnet, clinopyroxene, plagioclase (Pl-1 cores), and amphibole, which show mutual contacts and granoblastic–polygonal texture, representing the peak assemblage. The third generation (M₂) is marked by the formation of orthopyroxene as coronae and matrix grains and plagioclase with higher X_An (Pl-1 rims and Pl-2). Ilmenite is present in all three generations.

Type 2 mafic granulite with reaction texture

Type 2 mafic granulites are garnet-bearing two-pyroxene granulites with complex reaction textures. The main difference between type 1 and type 2 mafic granulites is the presence in type 2 of symplectites of orthopyroxene + plagioclase + ilmenite ± amphibole around garnet, which are absent from the type 1 samples. The common assemblage is garnet (10–15 vol%), clinopyroxene (25–30 vol%), orthopyroxene (10–15 vol%), plagioclase (30–35 vol%), amphibole (15–20 vol%), ilmenite (2–5 vol%), and minor biotite.

Garnet occurs as porphyroblasts measuring 1–10 mm in diameter and commonly contains inclusions of clinopyroxene, orthopyroxene, amphibole, plagioclase, quartz, ilmenite, sphene, and apatite, with or without rutile (Fig. 4a). All garnet porphyroblasts have been replaced by orthopyroxene + plagioclase + ilmenite ± amphibole symplectites to various extents. Garnet compositions are generally X_Alm of 0.49–0.54, X_Pyp of 0.23–0.30, X_Grs of 0.19–0.21, X_Sps of ~0.02, and X_Mg of 0.27–0.38 (Table 2). Most grains have flat compositional profiles with a narrow retrograde or resorption rim. A large garnet porphyroblast (~10 mm in diameter) from sample MS02-3 exhibits weak compositional zoning characterized by slightly outwards-increasing X_Pyp and decreasing X_Alm in the core domain, and sharply outwards-increasing X_Alm and X_Sps and decreasing X_Pyp in the rim domain (Fig. 5a). This compositional variation indicates prograde garnet growth followed by retrograde re-equilibration.

Clinopyroxene occurs as porphyroblasts in the matrix, fine grains in the matrix, and inclusions in garnet (Fig. 4a). It has diopsidic and augitic compositions with Wo of 0.40–0.49, En of 0.35–0.40, Fs of 0.14–0.23, and X_Mg of 0.62–0.74 (Table 3). Clinopyroxene porphyroblasts measure 0.5–1.5 mm across and are surrounded by orthopyroxene + plagioclase coronae, of which the majority domain has Ca^M2 of 0.82–0.90 and X_Mg of 0.63–0.74, whereas the domain with few orthopyroxene exsolution lamellae in the core has lower Ca^M2 of 0.75–0.80 and X_Mg of 0.58–0.67 (Table 3; Fig. 5b). Fine grains in the matrix are 0.1–0.5 mm in diameter and have Ca^M2 of 0.82–0.92 and X_Mg of 0.63–0.70, which are similar values to those of the most domain of the clinopyroxene porphyroblasts (Table 3; Fig. 5b). The clinopyroxene inclusions have Ca^M2 of 0.87–0.92 and X_Mg of 0.68–0.72 (Table 3). All clinopyroxene grains contain exsolution lamellae of orthopyroxene (Fig. 4b and f).

Orthopyroxene occurs as grains of 0.3–1.5 mm across in the matrix, as coronae around clinopyroxene, and as symplectite with plagioclase around garnet (Fig. 4). Orthopyroxene grains in the matrix in all cases appear adjacent to clinopyroxene, locally enclosing fine-grained clinopyroxene, suggesting that orthopyroxene grew after clinopyroxene. The three types of orthopyroxene have similar hypersthene compositions with En of 0.51–0.58, Fs of 0.41–0.48, Wo of ~0.01, and X_Mg of 0.52–0.59 (Table 4; Fig. 5c).

Amphibole occurs as coarse grains measuring 0.3–2.0 mm in the matrix, symplectite with orthopyroxene and plagioclase around garnet, and inclusions in garnet. Coarse amphibole has a ferropargasite or pargasite composition with Ti contents of 0.31–0.39 p.f.u. and X_Mg of 0.48–0.55; rims of individual grains have a lower Ti content of 0.28 and a higher X_Mg of 0.55 (Table 5; Fig. 5d). Symplectic amphibolite show ferropargasite and pargasite composition with Ti contents of 0.20–0.36 p.f.u. and X_Mg of 0.48–0.54 (Table 5; Fig. 5d). The amphibole inclusions show a pargasite composition with high Ti contents of 0.30–0.36 p.f.u. and X_Mg of 0.53–0.60 (Table 5).

Plagioclase occurs as coarse grains (Pl-1) measuring 0.5–1.0 mm across in the matrix, fine grains (Pl-2) measuring 0.1–0.5 mm across in the matrix, corona-like grains (Pl-s1) with orthopyroxene around clinopyroxene, symplectite-like grains (Pl-s2) with orthopyroxene around garnet, and inclusions in garnet. Pl-1 grains show obvious compositional zoning, with increasing X_An from core (0.45–0.56) through mantle (0.56–0.62) to thin rim (0.69–0.74), and Pl-2 grains are also zoned with increasing X_An from core (0.51–0.55) to rim (0.56–0.66); thus, Pl-1 mantles and Pl-2 cores have similar X_An (Table 6; Fig. 6). Pl-s1 and Pl-s2 grains have similar compositions with a high X_An of 0.77–0.88 (Table 6; Fig. 6), which is much higher than those for Pl-1 core, Pl-1 mantle, and Pl-2. Those anomalously high X_An values may have resulted from local enrichment of CaO (e.g. Zhang et al., 2018; Liao & Wei, 2019). The plagioclase inclusions in garnet have X_An of 0.47–0.56 (Table 6).

Biotite is present as discrete individual flaky crystals and overprints symplectite of orthopyroxene + plagioclase + ilmenite ± amphibole around garnet porphyroblasts, indicating its later formation. Biotite exhibits X_Mg of 0.54–0.56 and Ti content of 0.27–0.29 p.f.u. (Table 5).

Five generations of mineral assemblage can be identified on the basis of the textural relations and mineral compositions described above. The first generation (M₀) is represented by inclusions of clinopyroxene, orthopyroxene, plagioclase, amphibole, quartz, and ilmenite within garnet and clinopyroxene porphyroblasts. The second generation (M₁) is characterized by garnet porphyroblast + clinopyroxene porphyroblast + coarse-grained plagioclase (Pl-1 core) + amphibole (probably with rutile, as indicated by inclusions in garnet) and is considered as the peak assemblage. The third generation (M₂) is marked by the growth of orthopyroxene + plagioclase coronae around clinopyroxene and plagioclase with higher X_An (Pl-1 mantle/rim and Pl-2). The fourth generation (M₃) corresponds to the formation of orthopyroxene + plagioclase symplectites around garnet. The fifth generation (M₄) is characterized by the later growth of amphibole rims with lower Ti contents, and the formation of biotite. Ilmenite is present in all five generations.

MINERAL REACTION HISTORY

Prograde reactions

Although the major prograde history of the mafic granulites has been obliterated by the subsequent peak metamorphic stage, inclusions within garnet and clinopyroxene porphyroblasts provide information on the prograde history. The abundant inclusions of orthopyroxene + amphibole + plagioclase + quartz indicate the following prograde reactions:

Amp + Qz + Pl → Grt + Cpx + melt.

Amp + Opx + Pl → Grt + Cpx + melt.

Those reactions led to the development of the peak matrix assemblage of (orthopyroxene–quartz)-free garnet + clinopyroxene. Commonly, the orthopyroxene-free garnet + clinopyroxene + plagioclase + amphibole assemblage marks high-pressure granulite-facies metamorphism (Harley, 1989; O’Brien & Rötzler, 2003; Pattison, 2003). The absence of quartz in the matrix assemblage suggests that quartz was consumed during prograde metamorphism, as quartz commonly shows lower modal abundance compared with amphibole or plagioclase (de Waard, 1967; Hartel & Pattison, 1996).

Retrograde reactions

Complex corona and symplectite textures are developed between the garnet porphyroblasts and clinopyroxene grains, and are indicative of multi-stage decompression. The earlier retrograde reaction texture of orthopyroxene + plagioclase coronae around clinopyroxene suggests the occurrence of the following generalized reaction:

Grt + Cpx + melt → Opx + Pl.

This reaction led to the growth of orthopyroxene as coronae and matrix grains, marking the transition from high-pressure granulite-facies to medium-pressure granulite-facies metamorphism. Similar corona textures have been reported from various UHT granulite occurrences, including from the North China Craton (Cai et al., 2019; Liao & Wei, 2019), the Eastern Ghats Belt (Padmaja et al., 2021), and the East Sahara Ghost Craton (Karmakar & Schenk, 2015).

The later retrograde reaction texture of orthopyroxene + plagioclase + amphibole symplectites around garnet can be related to a combination of the following generalized reactions (e.g. Harley, 1989; Mengel & Rivers, 1991; Zhao et al., 2001; Prakash et al., 2007; Chowdhury & Chakraborty, 2019):

Grt + melt ± Cpx → Opx + Pl.

Grt + melt ± Cpx → Amp + Pl.

Grt + melt ± Cpx → Amp + Opx + Pl.

These reactions led to the formation of various symplectite assemblages at the expense of garnet and clinopyroxene. Such orthopyroxene + plagioclase + amphibole symplectites have been documented from many granulite-facies terranes and are interpreted to be a signature of near-isothermal decompression (e.g. Harley, 1989; Zhao et al., 2001; O’Brien & Rötzler, 2003; Pattison, 2003).

In addition, the formation of biotite, replacing symplectic orthopyroxene + plagioclase around garnet, is in agreement with the following hydration reaction:

Opx ± Cpx + melt → Amp + Bt + Pl.

This reaction may reflect back-reactions of the crystallizing anatectic melt with the symplectic assemblages.

P–T CONDITIONS OF METAMORPHISM

The P–T conditions of the M₀–M₄ generations of mineral assemblages were estimated using the following methods: (1) conventional thermobarometry, (2) pseudosection analyses (phase equilibria modeling) using isopleth thermobarometry, and (3) REE-based thermometry.

Conventional thermobarometry

The peak P–T conditions of the peak mineral assemblage (M₁) of garnet + clinopyroxene + plagioclase + amphibole + ilmenite + melt were evaluated using conventional garnet–clinopyroxene–plagioclase–quartz (GCPQ) thermobarometers. The widely used calibrations of Ellis & Green (1979), Powell (1985), Krogh (1988), and Ravna (2000) for the garnet–clinopyroxene (GC) thermometer, and those of Eckert et al. (1991) and Powell & Holland (1988) for the GCPQ barometer, were applied. To determine meaningful peak P–T conditions, the compositions of garnet porphyroblasts with highest X_Pyp and of Pl-1 cores were adopted. For clinopyroxene, the compositions with low Ca^M2 representing the peak mineral composition were used. Five samples gave consistent P–T conditions of 930°C–1000°C and 11.6–13.7 kbar from differently calibrated GCPQ geothermobarometers (Table 8). The Ti-in-amphibole thermometer of Liao et al. (2021) yielded a similar peak temperature of 945°C–980°C (Table 8). We further calculated P–T conditions using the compositions of clinopyroxene with high Ca^M2, which gave a lower temperature range of 820°C–885°C (Table 8).

As the studied garnet porphyroblasts have flat compositional profiles, garnet compositions must have re-equilibrated after the growth of garnet and therefore cannot be used to calculate the P–T conditions of prograde metamorphism. Even so, coexisting amphibole–plagioclase–quartz inclusions within garnet porphyroblasts provide information on earlier metamorphic P–T conditions. The amphibole–plagioclase thermometer (with an uncertainty of ±40°C; Holland & Blundy, 1994) yielded a temperature range of 820°C–853°C, and the Al-in-amphibole geobarometer (with an uncertainty of ±0.6 kbar; Schmidt, 1992) yielded a pressure range of 7.5–7.9 kbar (Table 8), indicating medium-pressure granulite-facies metamorphism.

Complex retrograde reaction textures and wide ranges in mineral chemistry indicate that retrograde metamorphism involved a multi-stage evolution and might not have reached complete equilibrium at each metamorphic stage. However, local equilibrium can be reached in some local compositional domains. For the corona assemblage (M₂), the compositions of relatively coarse orthopyroxene cores, garnet porphyroblasts with highest X_Pyp, and Pl-1 mantles or Pl-2 cores constrain the P–T conditions of the corona assemblage at temperatures of 910°C–1030°C (calculated at 9 kbar) using the garnet–orthopyroxene (GO) thermometer of Harley (1984) and Bhattacharya et al. (1991), and at pressures of 11.6–13.7 kbar (calculated at 950°C) using the garnet–orthopyroxene–plagioclase–quartz barometer of Bhattacharya et al. (1991) (Table 8). The temperatures are consistent with the peak temperature range, whereas the pressures are overestimated and substantially exceed the stability field of orthopyroxene (see the P–T pseudosection analysis below). For the symplectite assemblage (M₃), it was assumed that local equilibrium was reached for the relatively coarse-grained symplectic amphibolite + plagioclase measuring 100–120 μm in diameter. Three amphibolite–plagioclase pairs in type 2 samples MS02-3, MS04-1, and MS07-3 gave P–T conditions of 915°C–950°C and 7.6–8.2 kbar (Table 8). The Ti-in-amphibole thermometer yielded a similar temperature range of 920°C–935°C (Table 8).

Phase equilibria modeling

Pseudosection modeling was performed for representative samples TKL18-4 (type 1 mafic granulite) and MS02-3 (type 2 mafic granulite) to accurately decipher the P–T evolution of the granulites. Phase equilibrium modeling was performed in the chemical system NCKFMASHTO (Na₂O–CaO–K₂O–FeO–MgO–Al₂O₃–SiO₂–H₂O–TiO₂–O) using the thermodynamic calculation program DOMINO (de Capitani & Petrakakis, 2010) and the internally consistent thermodynamic dataset ds62 of Holland & Powell (2011). Pseudosection construction considered the following solid-solution models: metabasite melt, clinopyroxene, and amphibole (Green et al., 2016); garnet and orthopyroxene (White et al., 2014); plagioclase (Holland & Powell, 2003); and ilmenite (White et al., 2000). Pure end-member phases included rutile, quartz, and aqueous fluid (H₂O). Bulk-rock compositions for the calculation were obtained from XRF analysis with some adjustments. P₂O₅ was assumed to be entirely bonded to CaO and was excluded together with the corresponding CaO. Mn contents are negligible and were deducted from the bulk compositions. H₂O content was determined using T–H₂O diagrams (Supplementary Fig. 1) to ensure that the final phase assemblage is stable just above the solidus (e.g. Korhonen et al., 2012). O (Fe₂O₃) content was determined using T–O diagrams (Supplementary Fig. 1) to mitigate potential contamination or oxidation during sample preparation (e.g. Korhonen et al., 2012) and the effects of fluids that may cause changes in oxygen fugacity during exhumation (e.g. Cao et al., 2011).

Sample TKL18-4

A P–T pseudosection was calculated for sample TKL18-4 over a P–T window of 4–14 kbar and 750°C–1100°C (Fig. 7a). The fluid-absent solidus occurs at temperatures of 800°C–930°C. The amphibole-out curve is similar to the solidus in shape but occurs at temperatures that are 20°C–125°C higher. Orthopyroxene appears below 10.5 kbar, and garnet disappears below 5.5 kbar. The P–T pseudosection is presented with isopleths of X_An in plagioclase, Ti in amphibole, Ca^M2 in clinopyroxene, X_Ca in garnet, X_Mg in garnet, X_Mg in clinopyroxene, and X_Mg in orthopyroxene for relevant mineral assemblages (Fig. 7a–c).

The inferred peak assemblage (M₁) of garnet + clinopyroxene + plagioclase + amphibole + ilmenite + melt is predicted to be stable in the P–T range of 870°C–980°C and 9.0–11.7 kbar. However, this P–T range is only consistent with the lower Ti content of amphibole (0.32–0.36), and the higher Ti content (0.38–0.42) cannot be plotted in the fields within the stability of amphibole (thick solid curve labelled with Amp) (Fig. 7a). Therefore, the upper stability of amphibole is perhaps underestimated. As the upper stability of amphibole can be extended if adding the components such as F or Cl in the modeling system (Tsunogae et al., 2003; Sajeev et al., 2009; Liao & Wei, 2019), or using local textural domains that are enriched in MgO due to the segregation of melts (Dong & Wei, 2021; Wei et al., 2021), we recalculated the amphibole-out curve (thick dashed curve labelled with Amp in Fig. 7a) using an alternative bulk composition with slightly higher X_Mg of 0.52, following Wu & Wei (2021). This amphibole-out curve is extrapolated into higher-temperature fields, accommodating the higher Ti plots. Then the Ti content of amphibole (0.38–0.42) and the X_An of Pl-1 cores (0.35–0.42) constrain the peak P–T conditions to 975°C–1010°C and 11.7–12.8 kbar (Fig. 7a). The inferred corona assemblage (M₂) of garnet + clinopyroxene + orthopyroxene + plagioclase + amphibole + ilmenite + melt is predicted to be stable in the P–T range of >850°C and < 9.5 kbar if the stability field of amphibole is expanded to a higher temperature range. Considering the X_An of Pl-1 rims or Pl-2 (0.52–0.60) and the Ti content of amphibole (0.38–0.42), the P–T conditions of the corona assemblage are constrained to 970°C–1010°C and 8.2–9.2 kbar (Fig. 7a).

Thus, peak UHT conditions of 975°C–1010°C and 11.7–12.8 kbar, and a post-peak near-isothermal decompression P–T path can be inferred for sample TKL18-4 (Fig. 7a). The core–rim X_An-increasing zoning profiles of Pl-1 grains and the growth of orthopyroxene as coronae and matrix grains are consistent with post-peak near-isothermal decompression under UHT conditions. The measured Ca^M2 of clinopyroxene (0.78 p.f.u.) with few orthopyroxene exsolution lamellae and X_Ca of garnet (0.19–0.21) approximately match the conditions from the peak to the solidus (Fig. 7b). The measured X_Mg of garnet (0.29–0.34) and clinopyroxene (0.58–0.62) match the peak conditions (Fig. 7c), indicating that Fe–Mg diffusion was limited during post-peak cooling. The measured X_Mg of orthopyroxene (0.48–0.52) plots at a P–T range (>900°C) that is slightly higher than the solidus conditions. The measured Ca^M2 of most exsolved clinopyroxene matches the subsolidus conditions, indicating that the exsolution of orthopyroxene in clinopyroxene occurred under subsolidus conditions.

Sample MS02-3

The P–T pseudosection was calculated for sample MS02-3 over a P–T window of 4–14 kbar and 750°C–1100°C (Fig. 8a). The fluid-absent solidus occurs at temperatures of 860°C–890°C, and amphibole disappears above 975°C. Orthopyroxene appears below 11 kbar, and garnet disappears below 6 kbar. The P–T pseudosection is presented with isopleths of X_An in plagioclase, Ti in amphibole, Ti in biotite, Ca^M2 in clinopyroxene, X_Ca in garnet, X_Mg in garnet, X_Mg in clinopyroxene, and X_Mg in orthopyroxene for relevant mineral assemblages (Fig. 8a–c).

The inferred peak assemblage of garnet + clinopyroxene + plagioclase + amphibole + ilmenite + melt ± rutile matches the P–T range of 900°C–1030°C and >10.4 kbar. The Ti content of amphibole (0.32–0.36) and the X_An of Pl-1 cores (0.45–0.56) constrain the peak P–T conditions to 970°C–1030°C and 10.8–12.2 kbar (Fig. 8a). The observed corona assemblage of garnet + clinopyroxene + orthopyroxene + plagioclase + amphibole + ilmenite + melt is predicted to be stable in the P–T range of 875°C–1060°C and 7.5–11.0 kbar. Considering the X_An of Pl-1 mantle or Pl-2 core (0.56–0.62) and the Ti content of amphibole (0.32–0.36), the P–T conditions of the corona assemblage are constrained to 980°C–1015°C and 8.2–9.2 kbar (Fig. 8a). The observed final assemblage, characterized by the formation of biotite and Ti-poorer amphibole rims, represents decompressional cooling to the solidus. The Ti content of biotite (~0.27), the low Ti content of amphibole rims (0.20–0.24) and the fluid-absent solidus indicate P–T conditions of ~875°C and ~6.5 kbar (Fig. 8a).

Thus, a clockwise P–T path involving post-peak near-isothermal decompression and subsequent decompressional cooling can be inferred for sample MS02-3 (Fig. 8a). The core–mantle X_An-increasing zoning profiles of Pl-1 grains and the growth of orthopyroxene as coronae and matrix grains are consistent with the evolution of post-peak near-isothermal decompression, whereas the core–rim Ti-decreasing zoning profiles of some amphibole grains are consistent with decompressional cooling. The low Ti content of amphibole rims (~0.20) may have formed in local domains with melt segregation, and recorded lower solidus. However, the higher X_An values of coronatic and symplectic plagioclases are inconsistent with the predicted values during decompression and plot in higher temperature fields (>1000°C). The anomalously high X_An values of plagioclases may result from local enrichment of CaO, possibly caused by garnet breakdown during decompression. The measured Ca^M2 of clinopyroxene with few orthopyroxene exsolution lamellae (0.75 p.f.u.) and X_Ca of garnet (0.19–0.21) are consistent with the peak conditions (Fig. 8b). The measured X_Mg of garnet, clinopyroxene, and orthopyroxene are also consistent with the peak conditions (Fig. 8c), indicating that Fe–Mg diffusion was limited during post-peak cooling. The measured Ca^M2 of most exsolved clinopyroxene are consistent with the subsolidus conditions, indicating the subsolidus exsolution of clinopyroxene.

REE-based thermometry

Recently, REE-based thermometers have been developed on the basis of parameterized lattice-strain models, including the REE-in-clinopyroxene–orthopyroxene thermometer (Liang et al., 2013), the REE-in-garnet–clinopyroxene thermobarometer (Sun & Liang, 2015), and the REE-in-plagioclase–clinopyroxene thermometer (Sun & Liang, 2017). As the diffusivity of trivalent REEs is slower than that of divalent Fe–Mg, REE-based thermometers show higher closure temperatures than can be measured by conventional thermometers and are thus more likely to reveal the peak temperatures of mantle rocks and granulites (e.g. Yang & Wei, 2017; Liu & Wei, 2018; Ray et al., 2021; Gou et al., 2022). The REE-in-garnet–clinopyroxene and REE-in-plagioclase–clinopyroxene thermometers were applied to further determine the peak metamorphic temperature of the studied mafic granulites corresponding to the peak mineral assemblage of garnet + clinopyroxene + plagioclase + amphibole. The detailed calculation method and principles followed those of Sun & Liang (2015, 2017) and Yang & Wei (2017). The core domains of adjacent garnet porphyroblasts, clinopyroxene porphyroblasts, and coarse-grained plagioclase (Pl-1) were used. The REE-in-garnet–clinopyroxene thermometer for type 1 mafic granulites yielded a peak temperature of 1022 ± 71°C for sample TKL02-2 and 1049 ± 65°C for sample TKL18-4 (Fig. 9a and b). The temperature values calculated using the REE-in-plagioclase–clinopyroxene thermobarometer for type 1 and type 2 granulites range from 1075°C to 1124°C (Fig. 9c–f), consistent with the range from the garnet–clinopyroxene thermobarometer within uncertainties.

DISCUSSION

Peak metamorphic conditions and P–T path

Peak UHT conditions

The peak P–T conditions of mafic granulites are generally underestimated owing to post-peak cooling, especially in the case of significant Fe–Mg diffusion among minerals (Frost & Chacko, 1989; Harley, 1998a; Pattison et al., 2003; Powell & Holland, 2008). Harley (1988) reported that mafic granulites in the Rauer Group underwent ‘normal’ granulite-facies metamorphism with peak temperatures of 860 ± 40°C based on conventional Fe–Mg exchange thermometers. A few extreme metamorphic temperatures of 956°C–1028°C have also been estimated for mafic granulites (e.g. sample 630; Harley, 1988) but tend to have been overlooked by most studies. Tong & Wilson (2006) reported UHT conditions of 960°C–970°C and ∼12 kbar for garnet-bearing mafic granulites occurring as boudins in the Mather Paragneiss using conventional thermobarometry. The above findings suggest that some ‘normal’ mafic granulites in the Rauer Group underwent UHT metamorphism and that their peak temperatures may have been underestimated in earlier studies (Sims & Wilson,1997; Kelsey et al., 2003a). The multiple thermobarometric methods used in the present study confirm that mafic granulites from the Archean crustal domains underwent UHT metamorphism (Fig. 10). The GCPQ geothermobarometers yielded peak conditions of 930°C–1000°C and 11.6–13.7 kbar for the mafic granulites, with the temperature range being comparable with the results (910°C–1030°C) obtained from the GO thermometer (Table 8). Results of phase equilibria modeling for typical mafic granulite samples also yielded consistent UHT conditions of 970°C–1010°C and 11.7–12.8 kbar for sample TKL18-4, and 980°C–1030°C and 10.6–12.0 kbar for sample MS02-3, based on isopleths for Ti in amphibole, X_Ca in garnet, and X_An in plagioclase, and on the stability field of the inferred peak assemblage of (orthopyroxene–quartz)-free garnet + clinopyroxene + plagioclase + amphibole + ilmenite + melt (Fig. 10). In conclusion, mafic granulites from the Archean crustal domain in the Rauer Group in all likelihood underwent UHT metamorphism. The UHT results are supported by the following lines of evidence.

(1) Mafic granulites with compositions dominated by anhydrous minerals should be produced under UHT conditions of >900°C–950°C at 3–15 kbar, based on experimental studies and phase equilibria modeling results (e.g. Rushmer, 1991; Sen & Dunn, 1994; Freise et al., 2009; Pilet et al., 2010; Palin et al., 2016; Wei et al., 2017). Our petrographic observations, as presented above, revealed that the type 1 mafic granulites have a very low amphibole content, implying UHT conditions over 900°C.

(2) Experimental studies and pseudosection modeling have shown that the Ti content of Ca-amphibole coexisting with the Ti phase (rutile, ilmenite, or titanite) is controlled by temperature and can record amphibole-forming temperatures as high as ~1000°C (Helz, 1973; Raase, 1974; Ernst & Liu, 1998; Liao et al., 2021). Hence, high-Ti (Ti > 0.28 p.f.u.) amphibole can provide strong evidence for UHT metamorphism (Liao et al., 2021). The Ti content of all amphiboles in the studied mafic granulites is greater than 0.28 p.f.u., except for individual symplectic amphiboles. These data yield a UHT range of 945°C–980°C using the Ti-in-amphibole thermometer of Liao et al. (2021) (Table 8).

(3) Although the temperature estimates of 1020°C–1120°C obtained using REE-based thermometers are slightly higher than those estimated from conventional thermobarometry and pseudosection modeling, they strengthen the reliability of the inference of UHT metamorphism.

(4) Previous studies have reported UHT conditions for the mafic granulites, using conventional Fe–Mg thermometry (e.g. Sims & Wilson, 1997; Harley, 1998b ; Tong & Wilson, 2006).

Metamorphic P–T path

Our comprehensive study involving petrographic observations, mineral compositions, conventional thermobarometers, and phase equilibria modeling suggests three stages of metamorphic evolution for mafic granulites from the Archean crustal domain of the Rauer Group: pre-peak increasing pressure and temperature (prograde), post-peak near-isothermal decompression under UHT conditions, and subsequent decompressional cooling (Fig. 10).

The pre-peak prograde stage is based mainly on the inclusion assemblage in garnet. The prograde P–T conditions are approximately constrained to 820°C–855°C and 7.5–7.9 kbar using the amphibole–plagioclase thermobarometer, indicating medium-pressure granulite-facies conditions, which is consistent with the inclusion assemblage of garnet + clinopyroxene + plagioclase + amphibole + ilmenite ± orthopyroxene. The P–T conditions match the stability fields of quartz predicted by pseudosection modeling (Fig. 10). The peak conditions are well constrained to 930°C–1030°C and ~12 kbar by multiple thermobarometric methods. Therefore, a P–T vector with increasing pressure and temperature is preferred for the pre-peak prograde stage. The P–T path passes through the stability field of quartz, suggesting that quartz was consumed during prograde metamorphism, which is consistent with the observed peak assemblage of quartz-free garnet + clinopyroxene + plagioclase + amphibole + ilmenite + melt.

The post-peak near-isothermal decompression under UHT stage can be divided into two substages. The first substage is recorded by coronae of orthopyroxene + plagioclase around clinopyroxene and core–mantle/rim X_An-increasing zoning in coarse-grained plagioclase. P–T conditions for the corona assemblage are well constrained to 960°C–1010°C and 8.2–9.2 kbar for sample TKL18-4, and 980°C–1010°C and 7.9–9.2 kbar for sample MS02-3, on the basis of plots of the X_An of Pl-1 mantle and Pl-2 core together with the Ti content of amphibole. These P–T conditions are consistent with the stability field of the corona assemblage and results obtained using the GC thermometer (910°C–1030°C). The second substage is recorded by symplectites of orthopyroxene + plagioclase ± amphibole around garnet. The relatively coarse-grained amphibole–plagioclase symplectites yield P–T conditions of 915°C–950°C and 7.6–8.2 kbar using the amphibole–plagioclase thermometer and Al-in-amphibole geobarometer. From peak to symplectic assemblage, a P–T vector with near-isothermal decompression under UHT conditions is obtained (Fig. 10). This P–T path crosses the orthopyroxene-in curve, implying the formation of orthopyroxene at the expense of garnet and clinopyroxene, which is consistent with the petrographic observation that orthopyroxene in all cases appears adjacent to clinopyroxene and separates clinopyroxene from garnet porphyroblasts (Figs 3–4). Subsequently, the P–T path crosses the garnet-out curve, indicating that garnet was consumed, leading to the formation of symplectic orthopyroxene + plagioclase ± amphibole. The predicted mineral evolution is consistent with the variations of mineral proportions along the decompression path including an initial decreasing of clinopyroxene, a continuous decrease of garnet, and the increase of orthopyroxene, plagioclase and amphibole (Fig. 11).

The subsequent decompressional cooling stage is inferred from the late growth of biotite and amphibole, which is recorded only in the type 2 mafic granulites. In the P–T pseudosection for sample MS02-3, the Ti content of biotite, the low Ti content of amphibole rim and the fluid-absent solidus constrain P–T conditions to ~875°C and ~6.5 kbar, which is consistent with the stability field of the final assemblage of clinopyroxene + orthopyroxene + plagioclase + amphibole + biotite + ilmenite. Thus, from symplectic to final assemblage, a P–T path involving decompressional cooling is inferred. This P–T path crosses the biotite-in curve and fluid-absent solidus, resulting in the formation of the observed final assemblage with biotite, which is consistent with the petrographic observation of biotite replacing symplectites of orthopyroxene + plagioclase ± amphibole (Fig. 4c).

By integrating the metamorphic stages recorded by the various mafic granulites, a relatively complete P–T path can be defined, characterized by pre-peak compression and heating to peak metamorphism, post-peak near-isothermal decompression under UHT, and late decompressional cooling (Fig. 10). This clockwise P–T path is similar to those established in previous studies for Mg–Al-rich UHT granulites from the Mather Paragneiss, which show peak P–T conditions of 950°C–1050°C and ~12.0 kbar and a clockwise P–T path involving isothermal decompression (Harley, 1998b; Tong & Wilson, 2006) (Fig. 10).

Record of UHT conditions

Conventional Fe–Mg exchange thermometers may not yield reliable estimates of the peak temperatures of granulites because of the rapid Fe–Mg diffusion during cooling that obliterates peak mineral compositions (e.g. Frost & Chacko, 1989; Harley, 1989; Spear & Florence, 1992; Pattison et al., 2003; Powell & Holland, 2008). For example, most mafic granulites are estimated to record peak temperatures of 800 ± 50°C using Fe–Mg exchange thermometers (Bohlen, 1987; Harley, 1989). For UHT granulites, the underestimation of peak temperature is even more pronounced (e.g. Prakash et al., 2007; Li & Wei, 2016; Yang & Wei, 2017; Liu & Wei, 2018; Liao & Wei, 2019; Ray et al., 2021). Nevertheless, in the present study, UHT conditions were successfully estimated by GC and GO Fe–Mg thermometers without needing to back-calculate the peak mineral compositions to account for post-peak Fe–Mg exchange (Fitzsimons & Harley, 1994; Harley, 1998a). This result suggests that some garnet, clinopyroxene, and orthopyroxene grains may preserve peak or near-peak mineral compositions, and that Fe–Mg exchange was quite limited in some grains and domains. This interpretation is supported by the matching of measured X_Mg values of garnet, clinopyroxene, and orthopyroxene with peak conditions in the P–T pseudosections (Fig. 7a and 8a).

Generally, the estimation of UHT conditions using conventional Fe–Mg thermometers is dependent on (1) a lack of contacting Fe–Mg phases, (2) a large grain size of peak-stage minerals, and (3) a rapid cooling rate (Harley, 1998a and b; Brandt et al., 2003). For the UHT samples analyzed in the present study, the first point is unlikely to be satisfied, as garnet is in direct contact with clinopyroxene in type 1 mafic granulites and is surrounded by symplectic Fe–Mg phases in type 2 mafic granulites. During temperature calculation using the GC thermometer, UHT conditions (930°C–1000°C) were obtained when the highest X_Pyp of garnet and lowest Ca^M2 of clinopyroxene were used, whereas a lower temperature of 820°C–885°C was calculated when the highest Ca^M2 of clinopyroxene was used (Table 8). The constructed P–T pseudosections suggest that the highest X_Pyp of garnet and lowest Ca^M2 of clinopyroxene match peak/near-peak conditions and represent peak/near-peak mineral compositions (Fig. 7b and 8b). Of note, those near-peak mineral compositions are locally preserved in garnet and clinopyroxene porphyroblasts with large grain size; e.g. the highest X_Pyp is preserved in garnet porphyroblasts with diameters of ~10 mm, and low Ca^M2 is preserved only in clinopyroxene porphyroblasts with diameters of >4 mm. Therefore, the large grain size of peak minerals may be an important factor accounting for the local preservation of peak mineral compositions and the identification of UHT conditions. In addition, the rapid cooling rate recognized from the studied samples may have favored the preservation of UHT conditions. For sample TKL18-4, UHT was estimated by the GC Fe–Mg thermometer even though garnet and clinopyroxene occur as medium-sized grains (0.8–2.0 mm), which is consistent with the matching of the measured X_Mg of garnet, clinopyroxene, and orthopyroxene with peak conditions in the P–T pseudosection (Fig. 7c). In addition, the measured X_Mg values of the outer rim of garnet match the temperature condition of >900°C from the peak to the solidus, indicating limited Fe–Mg diffusion during post-peak cooling. Therefore, we speculate that retrograde diffusive Fe–Mg exchange was inhibited by rapid cooling of the mafic granulites during late decompression.

Compared with Fe and Mg, some components, such as, Ca in garnet, Ca in plagioclase, and Ti in amphibole, have much slower diffusion rate (Spear & Florence, 1992; Spear, 1995; Li & Wei, 2016), and have the potential to extract real peak conditions the rock reached. Pseudosection modeling of the studied mafic granulites show the X_Ca-in-garnet values of ~0.21 are roughly in agreement with the peak metamorphic conditions (Fig. 7b and 8b). The X_An zoning in matrix plagioclase perfectly record UHT information involving peak UHT conditions and post-peak near-isothermal decompression under UHT conditions (Fig. 7a and 8a). Moreover, all amphiboles except for individual symplectic amphibole are estimated to have UHT conditions of 945°C–980°C using the Ti-in-amphibole thermometer (Table 8), which is consistent with the matching of the measured Ti content of amphibole with UHT conditions in the P–T pseudosection (Fig. 7a and 8a). Therefore, the grossular content in garnet, anorthite content in plagioclase, and Ti content in amphibole are significant in determining peak conditions for UHT mafic granulites, and have been successfully applied to constrain the peak metamorphic conditions of UHT metamorphism (e.g. Li & Wei, 2016; Dong et al., 2018; Liu & Wei, 2018, 2020; Liao & Wei, 2019).

Regional extent of UHT metamorphism

The extent of UHT metamorphism can be local or regional, and is critical to understanding the genesis and heat source of metamorphism (Harley, 2004; Kelsey, 2008; Kelsey & Hand, 2015; Harley, 2021). The occurrence of UHT metamorphism on a local scale is usually related to HT anorthositic or mafic intrusions and represents local thermal anomalies (e.g. Berg, 1977; Kars et al., 1980; Arima & Gower, 1991; Dasgupta et al., 1997; Westphal et al., 2003; Peng et al., 2010; Kooijman et al., 2011; Guo et al., 2012; Mitchell et al., 2014). In contrast, the occurrence of UHT metamorphism on a regional scale requires a strong thermal perturbation over a large and tectonically significant scale (e.g. Clark et al., 2011; Kelsey & Hand, 2015; Li et al., 2019). The extent of UHT metamorphism in the Rauer Group has proved to be contentious. As Mg–Al-rich UHT granulites have been found only in the Mather Paragneiss, some studies have suggested that the UHT metamorphism is recorded only in this unit and represents a tectonic event distinct from the Pan-African event (Harley et al., 2009; Hokada et al., 2016). However, other studies have considered that the Filla Paragneiss has a shared metamorphic history with the Mather Paragneiss on the basis of field relationships and petrological analysis (Dirks & Wilson, 1995; Sims, 1999; Kelsey et al., 2003a, 2007). Recently, UHT conditions of >910°C and ~9 kbar have been estimated for samples of the Filla Paragneiss from Torckler and Lunnyj islands using Al-in-orthopyroxene thermobarometry, Zr-in-rutile thermometry, and calculated compositional isopleths (Hart et al., 2018; Clark et al., 2019) (Fig. 1). Our estimations of UHT conditions were obtained from mafic granulites of the Archean crustal domain on Torckler Island and Mather Peninsula, and suggest that the entire Archean crustal domain may have undergone UHT metamorphism. Integrating previous results with those of this study, we propose that UHT metamorphism may have affected the whole Rauer Group or at least extended over a much wider region than previously thought. If this is the case, the heat sources for UHT metamorphism in the Rauer Group are radiogenic crustal heating or elevated mantle heat flow, not local thermal anomalies (Tong & Wilson, 2006; Wang et al., 2007). Meanwhile, the shared UHT metamorphism recorded in the Archean crustal domain, the Mesoproterozoic crustal domain, and the Mather Paragneiss suggests that UHT metamorphism may be related to the juxtaposition of different crustal components of the Rauer Group. Zircon and monazite age data suggest that the juxtaposition occurred at late Neoproterozoic–Cambrian (Kinny et al., 1993; Kelsey et al., 2003b; Hokada et al., 2016; Liu et al., 2021). Consequently, the UHT metamorphism in the Rauer Group may be a consequence of the Gondwana assembly. Furthermore, the late Neoproterozoic–Cambrian UHT metamorphism was also revealed from the Larsemann Hills, implying a much wider distribution of UHT metamorphism in the Prydz Bay region (Wang et al., 2022).

Although UHT metamorphism is now widely accepted as a common type of deep crustal metamorphism (Harley, 1998a, 2008, 2021; Kelsey, 2008; Kelsey & Hand, 2015), the scattered distribution of UHT localities has hampered inferences of the regional extent of metamorphism and the understanding of its genesis. Moreover, the relationship between the scattered UHT granulites and the dominant ‘normal’ granulites in most granulite-facies terranes remains unclear. The identification of UHT metamorphism in mafic granulites and the Filla Paragneiss without diagnostic UHT assemblages (Kelsey et al., 2003a, 2007; Hart et al., 2018; Clark et al., 2019; this study) suggests that UHT metamorphism of the Rauer Group occurred over a much wider region than that indicated by Mg–Al-rich UHT granulites and is thus more common than previously considered. Similar cases have been reported from the Khondalite Belt, North China Craton, where UHT conditions have been obtained from ‘normal’ granulites without diagnostic UHT assemblages, such as Fe-rich pelitic granulites and mafic granulites (Jiao & Guo, 2011; Liu et al., 2012; Li & Wei, 2016, 2018; Gou et al., 2018; Lobjoie et al., 2018; Li et al., 2019; Liao & Wei, 2019; Zou et al., 2022), greatly expanding the known extent of regional UHT metamorphism. Therefore, generally, UHT metamorphism may be far more widespread than currently recognized, and an increasing number of UHT localities is expected from the anticipated wider application of effective UHT identification methods for ‘normal’ granulite-facies rocks.

CONCLUSIONS

In this study, mafic granulite samples from the Archean crustal domain in the Rauer Group, East Antarctica, were investigated using detailed petrography, mineral chemistry, conventional thermobarometry, phase equilibrium modeling, and REE-based thermometers. The main conclusions of the study are as follows:

1. Multiple thermobarometric methods, including conventional thermobarometry, pseudosection analyses, and REE-based thermometry, reveal that mafic granulites from the Archean crustal domain in the Rauer Group underwent UHT metamorphism with peak metamorphic conditions of 930°C–1030°C and ~12 kbar.

2. The mafic granulites underwent three stages of metamorphic evolution: pre-peak compression and heating to peak conditions, post-peak near-isothermal decompression under UHT conditions, and late decompressional cooling. The inferred clockwise P–T path is similar to that of Mg–Al-rich UHT granulites from the Mather Paragneiss.

3. The UHT conditions estimated for the studied mafic granulites indicate that UHT metamorphism in the Rauer Group may have occurred over a much wider region than previously thought and probably extends over the whole Archean crustal domain. UHT metamorphism in general may be far more widespread than is currently recognized.

FUNDING

This research was financially supported by the National Natural Science Foundation of China (grants 41941004, 41802064 and 41530209), the Fundamental Research Funds of the Chinese Academy of Geological Sciences (grant DZLXJK202003), and the Geological Investigation Project of the Chinese Geological Survey (grant DD20190579).

SUPPLEMENTARY DATA

Supplementary data for this paper are available from the Journal of Petrology online.

ACKNOWLEDGEMENTS

We thank Zhao Liu for assistance during fieldwork of the 2016–2017 Chinese National Antarctic Research Expedition. We also thank Xiaohong Mao for support with electron microprobe analyses. The authors thank Tapabrato Sarkar, Chunjing Wei, and one anonymous reviewer for their thoughtful and instructive reviews. Professor Sarah Sherlock is thanked for her constructive comments and for handing the paper.

REFERENCES