Introduction

Garnet crystals in metamorphic rocks provide a great amount of valuable information on their nucleation and growth histories, given their common textural and / or chemical zoning (e.g., Andersen 1984; George and Gaidies 2017; George et al. 2018; Rice and Mitchell 1991; Rice et al. 2006; Yogi et al. 2024). Pressure-temperature (P-T) trajectories of metamorphic rocks are commonly inferred using equilibrium thermodynamics and garnet chemical zoning (e.g., Gaidies et al. 2006; Spear 1993).

However, disequilibrium crystallization (e.g., overstepping) of porphyroblasts may be important in many metamorphic rocks (e.g., Carlson et al. 2015a; Jamtveit and Andersen 1992; Pattison et al. 2011; Spear et al. 2014; Wilbur and Ague 2006). Evidences for such disequilibrium processes have been particularly emphasized in the formation of garnet with sector zoning (e.g., Andersen 1984; Jamtveit and Andersen 1992; Wilbur and Ague 2006). While texturally and / or chemically sector-zoned garnet crystals have been reported in rocks worldwide (e.g., Burton 1986; Castellanos et al. 2004, 2016; Gaidies et al. 2021, 2024; Jamtveit and Andersen 1992; Rice and Mitchell 1991), commonly in rocks with abundant graphite (e.g., Burton 1986; Gaidies et al. 2024; Rice and Mitchell 1991), their formation mechanisms remain controversial. Models of their formation include (i) growth under supersaturation conditions or from reaction overstepping (Andersen 1984; Jamtveit and Andersen 1992; Wilbur and Ague 2006); (ii) matrix displacement growth (Burton 1986; Rice and Mitchell 1991; Rice et al. 2006); (iii) growth controlled by external crystal morphology (Stowell et al. 2011) or differential element incorporation between different lattice planes (Kohn 2004; Kohn et al. 2024). Therefore, understanding the origin of sector-zoned garnet is crucial for future studies on growth mechanisms of sector-zoned porphyroblasts and the interplay of thermodynamics and kinetics during garnet growth.

In this work, we report and discuss the results of a comparative study of texturally and chemically sector-zoned garnet crystals in metapelitic rocks from the Danba dome in the eastern Tibetan Plateau (SW China) (Fig. 1A). We use detailed petrography, bulk-rock and mineral geochemistry, garnet crystallographic orientation analyses and thermodynamic calculations to shed light on the petrogenesis of these sector-zoned garnet crystals in the rocks studied.

Fig. 1
figure 1

(A) Geological map of the Danba dome with main structural boundaries and isograds marked (modified from Huang et al. 2003a; Roger et al. 2003; SBGMR 1991). Liq − migmatite zone. Abbreviations for Neoproterozoic orthogneisses: QGL − Qingganglin, NCP − Niuchangping, CNC − Chunniuchang, BW − Bawang. (B) Sample 21DB103 occurred as compositional lenses in the garnet-mica schist. (C) Field occurrence of sample 21DB104

Full size image

Geological background and sample collection

The Songpan-Ganzê Orogenic Belt (SGOB) at the eastern margin of the Tibetan Plateau has been ascribed to the closure of the Paleo-Tethys Ocean by concurrent convergence of North China, Yangtze and Qiangtang Blocks (e.g., Huang et al. 2003a; Huang et al. 2003b; Pullen et al. 2008; Roger et al. 2003, 2004). The Triassic flysch sequence comprises a large proportion of the SGOB, and metamorphic rocks that exhumed from middle- to lower-crustal levels are restricted to the Danba dome and the Longmenshan area (Airaghi et al. 2017; Huang et al. 2003a; SBGMR 1991; Weller et al. 2013).

The Danba domal structure is featured by Late Sinian to Paleozoic metasedimentary rocks overlying several Neoproterozoic granitic gneisses and granites representing the basement of the Yangtze Block (Fig. 1A; Billerot et al. 2017; Huang et al. 2003b; Jolivet et al. 2015; SBGMR 1991; Zhou et al. 2008). Systematic changes of index minerals (biotite, garnet, staurolite, kyanite and sillimanite) in metapelitic rocks denote a Barrovian sequence in the Danba dome (e.g., Billerot et al. 2017; Cheng et al. 2009; Huang et al. 2003a, b; SBGMR 1991; Weller et al. 2013), reaching migmatite-grade at the core of the dome (Fig. 1A). Geochronological results suggest that the metapelites from the staurolite to the migmatite zones were metamorphosed from c. 205 Ma to c. 155 Ma (Huang et al. 2003a; Li et al. 2024; Weller et al. 2013). Except for the Gezong granite, other Neoproterozoic basement rocks also record Mesozoic ages (Billerot et al. 2017; Jolivet et al. 2015; Zhou et al. 2008). The metamorphic P-T conditions of the same zone obtained by different authors vary but nonetheless depict a medium P/T metamorphic gradient (~ 3.0-8.4 kbar / ~500–725 ℃; Billerot et al. 2017; Cheng et al. 2009; Huang et al. 2003b; Li et al. 2024; Palin et al. 2016; Weller et al. 2013). Billerot et al. (2017) ascribed such discrepancies to the various sample positions or crystallization time along the P-T paths experienced by these rocks.

In this study, two metapelitic rocks were collected from the kyanite zone. Whereas sample 21DB103 (30°51′41.51″N, 101°55′50.16″E) is a garnet-biotite-plagioclase gneiss collected from a compositional lens interbedded with weathered garnet-mica schist (Fig. 1B), sample 21DB104 (30°51′38.14″N, 101°55′47.14″E) is a staurolite-kyanite-bearing garnet-mica schist (Fig. 1C) collected approximately 120 m away from sample 21DB103. Mineral abbreviations in the text, figures and tables are after Whitney and Evans (2010).

Methods

TESCAN integrated mineral analysis

Phase identifications and modal proportions, as well as elementary panorama maps for a thin section of sample 21DB103 were obtained using a TESCAN Integrated Mineral Analyzer (TIMA) mineralogy system at the Key Laboratory of Orogenic Belts and Crustal Evolution, School of Earth and Space Sciences, Peking University (Beijing, China). The TIMA comprises a TESCAN Mira Schottky field emission scanning electron microscope with four silicon drift energy dispersive (EDS) detectors arranged at 90° intervals around the chamber. The measurements were performed with high-resolution liberation analysis mode, and the backscattered electron (BSE) images were obtained to determine individual particles and boundaries between distinct preliminary phases. A rectangular mesh of measurements on every distinct phase was obtained with X-ray spectra for a total of 448 scanned fields. The spectroscopic data were matched to mineral definition files, allowing for mineral identification and mapping. The measurements were performed at 25 kV beam voltage and 10.9 nA current, with a spot size of approximately 117.8 nm. The working distance was set at 15.0 mm and the field size was adjusted to 1,500 μm with a pixel size of 3.0 μm.

X-ray micro-computed tomography

Sample 21DB104 was scanned using a SkyScan 1173 high-resolution X-ray micro-computed tomography (XR-µCT) scanner at Carleton University (Ottawa, Canada). The analytical conditions were set at 130 kV and 61 µA with an exposure time of 1350 ms, and a resolution of about 22 μm. A 0.25 mm brass filter was applied to reduce beam-hardening. The reconstruction of 3D data from the 2D attenuation slices was done using SkyScan NRecon software. Numerical segmentation was conducted with the Blob3D software (Ketcham 2005). Blob3D was also employed to locate the geometric centers and to calculate the volumes of all garnet grains. The crystal size frequency distribution (CSD) was calculated assuming a spherical garnet geometry. Two garnet crystals (grt 1 and grt 2) were centrally cut along their long axes to obtain cross-sections containing the geometric centers of the respective grains in this sample, whereas other garnet crystals in this study were randomly sectioned.

Bulk-rock chemical composition analysis

The fresh rocks were crushed into powders and made into fused pieces for bulk-rock chemical composition analyses at the Australian Laboratory Services (ALS), Guangzhou (China). The major elements of the samples were analyzed using ALS method code ME-XRF26F (XRF method), with both precision (relative deviation) and accuracy (relative error) below 5%. The trace elements were analyzed using ALS method code ME-MS81g (ICP-MS method), with both precision (relative deviation) and accuracy (relative error) below 10%. The effective bulk compositions (EBCs) of the quartz-rich and quartz-poor zones of sample 21DB103 were calculated based on the modal abundances and average compositions of the minerals in the observed Grt + Bt + Pl + Qz + Rt assemblage.

Electron probe microanalysis

X-ray maps of Mg, Fe, Ca, Mn of garnet, BSE images and mineral compositions were obtained by a JEOL JXA-8230 electron probe microanalyzer (EPMA) at School of Resource and Environmental Engineering, Hefei University of Technology (Hefei, China). The analytical conditions were 15 kV accelerating voltage and 20 nA beam current with electron beam size of 5 μm and a counting time of 10–20 s. Major components of centrally sectioned garnet crystals (grt 1 and grt 2) of sample 21DB104 were obtained using a JEOL JXA-8230 EPMA at the University of Ottawa (Ottawa, Canada). The analytical conditions were 15 kV accelerating potential, 40 nA beam current, 5 μm beam diameter, and 10–20 s counting time. Matrix corrections were done by the ZAF program.

Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analysis

Trace element analyses of garnet crystals were conducted by LA-ICP-MS at the Nanjing FocuMS Technology Co. Ltd., Nanjing (China). The laser ablation system consists of an Agilent 7900 ICP-MS coupled with a 193 nm ArF excimer laser (PhotonMachines Analyte HE). During the laser ablation process, the He gas was used as carrier gas (He-flow rate was 0.3 L/min) and the Ar gas as the complemental gas (Ar-flow rate was 0.87 L/min) to regulate sensitivity. The laser ablation energy was 2–3 J/cm2 and the repetition rate was 10 Hz. The resolution of the maps is 40–65 μm per spot and the sample was simultaneously moved with a speed of 70–80 μm/s, depending on the grain sizes. Background signals were measured for 20 s before and 20 s after each line. The reference material NIST SRM 610 was analyzed for about 30 s at the beginning and the end of each mapping analysis for data calibration. Analytical results of several external standard materials (BCR-2G, BHVO-2G and BIR-1G) from the same laboratory are given in Supplementary Table S1 for reference. Data processing was performed using the MATLAB-based program LaIcpMsSoftware (LIMS) (Wang et al. 2017). For each raster and every element, the average background was subtracted from its corresponding raster, and the rasters were compiled into a 2D image displaying background / drift corrected intensity for each element (Wang et al. 2017).

Electron back-scattered diffraction (EBSD) analysis

EBSD analysis was applied to garnet crystals of sample 21DB103 to reveal the crystallographic orientations. Data were acquired using a field-emission scanning electron microscope (FEI-SEM Quanta450) equipped with the Nordlys F + EBSD camera commercialized by Oxford Instruments at the Continental Dynamics Lab in the Institute of Geology, Chinese Academy of Geological Sciences (Beijing, China). The analytical conditions were set as follows: an acceleration voltage of 20 kV, a beam current of ~ 217.3 µA, a stage tilt of 69.8°, a spot size of 6.0, a step size of 8–20 μm and a working distance of 18–20 mm. Automatic indexing of the EBSD patterns was achieved by using the AZtec 2.4 software. Acquired data were processed using HKL CHANNEL 5 software (Oxford Instruments, UK). Measurements with a mean angular deviation (MAD) greater than 0.75 were removed, and neighboring measurements with misorientations < 10° were considered parts of the same grain.

Metamorphic P-T estimates

Conventional geothermobarometry

The garnet-biotite (GB) thermometer (Holdaway 2000) coupled with the garnet-biotite-plagioclase-quartz (GBPQ; Wu et al. 2004), garnet-biotite-aluminosilicate-quartz (GBAQ; Wu 2017) and garnet- aluminosilicate-plagioclase-quartz (GASP; Holdaway 2001) barometers were applied to estimate the near peak P-T conditions, using chemical compositions of the garnet with lowest Fe# [= Fe2+ / (Mg + Fe2+)] (Florence and Spear 1991) and average plagioclase and biotite compositions.

Phase equilibria modelling

The phase diagrams were calculated using Theriak/Domino (version 11.02.2015; de Capitani and Petrakakis 2010) in the MnO-Na2O-CaO-K2O-FeO-MgO-Al2O3-SiO2-CO2-H2O-TiO2 chemical system using the thermodynamic database of Holland and Powell (1998). Fe2O3 was not considered given the negligible Fe3+ in the main mineral assemblages and the presence of pyrite in sample 21DB104. A COH-phase produced by the equilibration of H2O and excess graphite was modelled as the metamorphic fluid based on Connolly and Cesare (1993). The activity-composition models used are as follows: ternary feldspar (Baldwin et al. 2005), garnet (Holland and Powell 1998), ilmenite is a ternary ideal solid solution, chloritoid follows White et al. (2000) but is extended to cover the Mn-bearing end-member, chlorite is modelled as outlined by Gaidies et al. (2008), muscovite (Coggon and Holland 2002), staurolite follows Holland and Powell (1998) but is extended to include Mn-bearing solutions, biotite follows White et al. (2005) without its Ti-bearing end-member. For comparison, a phase diagram for sample 21DB104 was calculated in the MnO-Na2O-CaO-K2O-FeO-MgO-Al2O3-SiO2-H2O-TiO2 model system without graphite and assuming a pure H2O fluid. Whereas the EBCs of the two zones of sample 21DB103 were employed to calculate the respective phase diagrams, the bulk-rock composition of sample 21DB104 obtained by XRF analysis was used for calculation.

Results

Petrography

Sample 21DB103

Sample 21DB103 mainly consists of garnet, plagioclase, biotite, and quartz, with minor muscovite, chlorite and negligible sillimanite (Fig. 2). The accessory minerals include rutile, ilmenite, apatite, and allanite. Graphite is mainly distributed along grain boundaries or in garnet interiors across the sample (Fig. 2A). In the following, the slide is divided into two parts based on the contents of quartz, namely the quartz-rich zone and the quartz-poor zone (Fig. 2B). Note that this is a relatively rough division due to the gradual variation across the two zones and the large sizes of garnet grains. The estimation of mineral modes of each zone is based on the TIMA results and listed in Table 1.

Fig. 2
figure 2

The slide panorama of garnet-biotite-plagioclase gneiss sample 21DB103. (A) Plane polarized transmitted light (PPL) scan. (B) TIMA scan and minerals presented in the slide. The dashed line in B denotes the rough boundary between the quartz-rich zone and quartz-poor zone. Grt 1 to Grt 3 and Bt 1 to Bt 3 in the boxes of B are the analyzed garnet and biotite crystals mentioned in the text

Full size image
Table 1 Mineral proportions (vol %) of sample 21DB103 obtained by TIMA analysis
Full size table

In the quartz-rich zone, garnet porphyroblasts (~ 12%) exhibit apparent sector zoning, and are wrapped by the fine-grained matrix composed of biotite (~ 16%), plagioclase (~ 48%), and quartz (~ 23%). The largest garnet crystal (Grt 1) is c. 3.8 mm in diameter. Tiny apatite grains (< 1%) are widely distributed in this zone (Fig. 3A). The abundance of graphite is high in this zone, and graphite is concentrated along the grain boundaries and garnet sector edges (Fig. 2A). Chlorite (< 1%) is spatially associated with garnet and mainly observed in the garnet interior (Fig. 2B). Rutile, muscovite, allanite, sillimanite and ilmenite are rare in this zone (< 1%).

Fig. 3
figure 3

Microphotographs of sample 21DB103. (A) BSE image and X-ray Ca map showing widely distributed apatite in the quartz-rich zone. (B) Rutile rim replaced by ilmenite in the quartz-poor zone. (C) Biotite grain surrounded by several ilmenite crystals in the quartz-poor zone. (D) Quartz and sillimanite aggregates occurred at the triple junction point or grain boundaries of plagioclase

Full size image

In the quartz-poor zone, garnet porphyroblasts (~ 10%) are euhedral to subhedral, and are wrapped by the relatively coarse-grained matrix of biotite (~ 25%) and plagioclase (~ 63%). The largest garnet crystal in this zone is c. 4.2 mm in diameter, and garnet crystals in this zone have a relatively regular morphology (e.g., Grt 2 in Fig. 2B). Compared to the quartz-rich zone, the main differences in this zone are: (i) quartz constitutes only ~ 1% of the mineral assemblage; (ii) relatively coarse-grained plagioclase and biotite form the primary matrix; (iii) there is a relatively higher abundance of rutile and muscovite; (iv) regular garnet morphologies are more common (although some crystals also exhibit sectors); and (v) apatite and graphite are less abundant (Fig. 2). Ilmenite is observed to either replace rutile (Fig. 3B) or is spatially associated with biotite (Fig. 3C). Sillimanite is spatially associated with quartz and mainly distributed at the grain boundaries between plagioclase grains (Fig. 3D).

Sample 21DB104

Sample 21DB104 mainly comprises garnet, biotite, muscovite, kyanite, plagioclase and quartz (Fig. 4A), with minor staurolite, as well as accessory rutile, allanite, ilmenite, apatite, zircon and pyrite. Graphite is pervasive throughout the rock, and is present in both the matrix and the garnet interior. Staurolite and tabular kyanite grains exhibit ragged morphologies (Fig. 4A and B). Biotite and muscovite flakes mark the matrix foliation, whereas plagioclase and quartz are interstitial phases in the matrix (Fig. 4A). Oriented plagioclase inclusions are preserved in the garnet interior (Fig. 4A and C), with many tiny quartz grains within or surrounding them. Quartz inclusions in garnet also display oriented features (Fig. 4C), with some quartz rods perpendicular to the garnet faces showing garnet-quartz intergrowth texture in the radial sector-zoned garnet (Fig. 4D). Garnet porphyroblasts are generally anhedral to subhedral and enclose abundant inclusions of graphite, quartz, plagioclase, allanite, rutile and ilmenite with a thin inclusion-poor rim (Fig. 4A). Most garnet porphyroblasts exhibit sector zoning, characterized by either radial sectors (Fig. 4E) or irregular inclusion-poor sectors (Fig. 4A) in the interior. The reconstructed 3D volume of radial sector-zoned garnet comprises at least six sectors (Fig. 4E) in the interior and closed faces at the surface (animation, Supplementary File S1 shows the crystal in Fig. 4E in video). The CSD of garnet crystals in this sample is depicted in Fig. 4F.

Fig. 4
figure 4

Microphotographs (A-D), reconstructed 3D volume (E) and CSD (F) of garnet crystals of sample 21DB104. (A) An overview showing the main mineral assemblages and their features. (B) Close view of the occurrences of staurolite and kyanite. (C) Oriented plagioclase grain in the garnet interior with inclusions of apatite and abundant quartz. (D) Garnet-quartz intergrowths in the radial sector-zoned garnet. (E) Reconstructed 3D volume of garnet showing six sectors (highlighted with white dashed line) in its interior. (F) CSD as determined from the extracted garnet volume, with 15 size classes. The red stars below grt 1 and grt 2 indicate which size classes they are from

Full size image

Mineral chemistry

Sample 21DB103

The representative mineral compositions are presented in Supplementary Table S2.

Major and trace element distributions in garnet

Compositional profiles, X-ray maps of Mg, Fe, Ca and Mn, as well as trace element maps for Grt 1 and Grt 2 (Fig. 2B) are depicted in Figs. 5 and 6, respectively. The X-ray elemental panorama maps of the whole slide are presented in Supplementary Fig. S1, whereas the compositional data and other compositional maps from the LA-ICP-MS analysis are given in Supplementary Tables S3–S4 and Figs. S2–S3, respectively.

Fig. 5
figure 5

Crystal sketch, compositional profiles, and selected major (by EPMA) and trace (by LA-ICP-MS) element maps for Grt 1 (Fig. 2B) from the quartz-rich zone of sample 21DB103. The four sectors are highlighted as gray domains, whereas the sector edges are marked as green domain in the sketch. The vertical dashed lines in compositional profiles represent different petrographic locations of Grt 1: VC = visual core, SI = sector inner part, SC = sector center, IR = inner rim, VR = visual rim

Full size image
Fig. 6
figure 6

Crystal sketch, compositional profiles, and selected major (by EPMA) and trace (by LA-ICP-MS) element maps for Grt 2 (Fig. 2B) from the quartz-poor zone of sample 21DB103. The inclusion-rich domains where the neighboring garnet exhibits compositional fluctuations are highlighted as orange fields in the sketch. The vertical dashed lines in compositional profiles represent different petrographic locations of Grt 2: C = core, M = mantle, IR = inner rim, OR = outmost rim

Full size image

For garnet (Grt 1) in the quartz-rich zone (Figs. 2B and 5), the distribution of XCa [= Ca / (Ca + Mg + Fe2+ + Mn)] is generally homogeneous (0.16–0.18) in the four sectors, but XCa slightly decreases towards the sector boundaries (0.13–0.14). XMn [= Mn / (Ca + Mg + Fe2+ + Mn)] increases from the centers of each sector and then decreases slightly at the sector boundaries (0.07→0.13→0.12). XFe [= Fe2+ / (Ca + Mg + Fe2+ + Mn)] and XMg [= Mg / (Ca + Mg + Fe2+ + Mn)] exhibit asymmetric zoning patterns in each sector, specifically with higher compositional gradients at the junction location of sector edges (visual core) than at the peripheral sector faces (visual rim) of the four sectors. Nevertheless, XFe and Fe# decrease towards the sector boundaries (XFe: 0.68→0.60; Fe#: 0.92→0.80), whereas XMg displays a contrary trend (0.06→0.15).

Li, V, Cr, Gd, Tb, Y and heavy rare earth elements (HREE, from Dy to Lu) of Grt 1 are enriched at the sector boundaries (Fig. 5). On the contrary, Ti and Nb are depleted at the sector boundaries and enriched in the sector interiors. Although Sc is enriched at the sector boundaries, there are even higher Sc (and maybe also Yb and Lu) concentrations in the regions close to the visual core of each sector’s interior. Co is depleted in the regions close to the visual core but enriched at the positions close to the peripheral faces of each sector, whereas the intervening Co concentrations are retained at the sector edges and faces (Fig. 5).

In the quartz-poor zone, all the four endmembers of garnet (Grt 2) show almost symmetric distributions with slight fluctuations in the region surrounding the inclusions (Fig. 6). XMn exhibits “bell-shaped” zoning patterns from core (0.17) to mantle and then increases from the mantle (0.05) to the rim (0.10). XCa increases slightly from core (0.16) to mantle (0.18) and then decreases towards the rim (0.11). XFe exhibits elevated values from core (0.61) to mantle (0.69), then drops to the inner rim (0.64) with a slight upwards inflexion at the outermost rim (0.66). XMg increases from the core (0.05) to the inner rim (0.14) and then decreases slightly at the outmost rim (0.13). On the contrary, Fe# drops from the core (0.92) to the inner rim (0.82) and then shows an upwards inflexion at the outmost rim (0.83). Similar to the major elements, the trace elements of Grt 2 exhibit almost symmetric distributions with the exception of the regions surrounding the inclusions (Fig. 6). Y and HREE as well as Li, Sc, Ti, Nb, Gd and Tb are enriched in the core domain whereas V, Cr and Co are depleted. The contents of Li and Ti decrease continuously from the core to the rim, whereas Sc, Nb, Gd, Tb, Y and HREE are also enriched at the rim domain, attributed to garnet resorption.

Other minerals

Plagioclase in the quartz-poor zone is almost compositionally homogeneous, with XAn (= Ca / (Ca + Na + K)) = 0.42–0.44. In the quartz-rich zone, however, some of the smaller plagioclase grains contain Ca-rich cores (X-ray Ca map in Fig. 3A; XAn = 0.60), compared to lower Ca rims (XAn = 0.46–0.48). Nevertheless, plagioclase in the quartz-poor zone has distinguishable lower XAn than in the quartz-rich zone.

Three biotite grains were analyzed from the two zones (Fig. 2B). Bt 1 from the quartz-rich zone contains 1.47–1.82 wt% TiO2 (Ti = 0.08–0.10 a.p.f.u.) and has a Fe# of 0.38–0.39. In the quartz-poor zone, two biotite grains with different occurrences in the matrix were analyzed. Bt 2, surrounded by several ilmenite crystals (Fig. 3C), has similar TiO2 content of 1.29–1.73 wt% (Ti = 0.07–0.10 a.p.f.u.) and higher Fe# (0.39–0.43) than those of Bt (1) Bt 3, with no ilmenite aggregates surrounding it, yields higher TiO2 content (1.78–2.18 wt%; Ti = 0.10–0.12 a.p.f.u.) and similar Fe# (0.41–0.42) to Bt (2) Overall, biotite in the quartz-rich zone has distinguishable lower Fe# than in the quartz-poor zone.

Sample 21DB104

The representative mineral compositions are presented in Supplementary Table S5. Four garnet crystals of sample 21DB104 were analyzed, and the garnet compositional profiles are presented in Fig. 7. The major and trace element maps for grt 1, grt 3 and grt 4 are shown in Figs. 8, 9 and 10, and the compositional data and other compositional maps from LA-ICP-MS analysis are given in Supplementary Tables S6-S8 and Figs. S4-S6, respectively.

Fig. 7
figure 7

Compositional profiles for the four analyzed garnet crystals of sample 21DB104. The location of the profile for grt 2 is shown in the insert, whereas the location of the profiles for grt 1, grt 3 and grt 4 are shown in Figs. 9, 10 and 11, respectively. Crystals grt 1 and grt 2 are central sectioned crystals from size classes 1 and 7 (Fig. 4F) of the scanned volume. The vertical dashed lines in compositional profiles of grt 1 represent different petrographic locations of largest crystal grt 1: C = core, IR = inner rim

Full size image
Fig. 8
figure 8

Crystal sketch and selected LA-ICP-MS maps for major and trace elements of the central section of the largest garnet crystal grt 1 of sample 21DB104. The details of the sketch are as follows: the gray domains represent where prograde growth zoning was preserved, whereas the yellow domain indicates where the garnet compositions were affected by retrograde modification; two types of oriented plagioclase inclusions are marked in blue patches; four groups of the sectors at the core region distinguished by the elemental distributions are labelled as 1, 2, 3 and 4 and highlighted with different colors; traverse A to B denotes the location of the compositional profile in Fig. 7A

Full size image
Fig. 9
figure 9

Microphotograph, crystal sketch, and selected major (by EPMA) and trace (by LA-ICP-MS) element maps for grt 3 of sample 21DB104. The details of the sketch are as follows: the radial sectors are highlighted in orange and they are not grouped in this crystal due to its random sectioning and consequently fuzzy chemical zoning; within the radial sectors, many quartz rods perpendicular to the garnet faces are marked as blue dashed lines; traverse A to B denotes the location of the compositional profile in Fig. 7C.

Full size image
Fig. 10
figure 10

Microphotograph, crystal sketch, and selected major (by EPMA) and trace (by LA-ICP-MS) element maps for grt 4 of sample 21DB104. The details of the sketch are as follows: the irregular sectors are highlighted as orange fields and labelled as 1, 2 and 3 to distinguish them; two types of oriented plagioclase inclusions are marked in blue patches; apatite inclusion within the type 2 plagioclase inclusion is highlighted in purple and allanite inclusion within the sector 2 is marked in yellow; traverse A to B denotes the location of the compositional profile in Fig. 7D

Full size image
Major element distributions in garnet

The XMn and Fe# of grt 1 (Fig. 8), the centrally sectioned, largest garnet crystal of the scanned volume, show apparent “bell-shaped” zoning patterns (Fig. 7A), indicative of growth zoning (e.g., Woodsworth 1977). The slight upwards inflexions of XMn and Fe# from inner rim to rim are interpreted to be retrograde modifications (e.g., Spear and Florence 1992). Whereas the upwards inflexions of Fe# are observed at the rims of the other three crystals, upwards inflexions of XMn are not (Fig. 7). The XCa zoning also displays a “bell-shaped” pattern, decreasing from core to rim. Contrary to XCa, XMn and Fe#, the XFe and XMg increase from core to rim, with a platform of XFe close to the rim. The central sectioned grt 2 (without sector zoning, size class 7; Fig. 7B) and random sectioned grt 3 (with radial sectors; Fig. 9) share generally similar compositional trends to the grt 1. In the random sectioned grt 4 with irregular sectors (Fig. 10), the major element distributions are predominately controlled by the inclusion-poor sectors, resulting in asymmetric zoning patterns. Specifically, XMn and XCa are enriched in these sectors, whereas XMg and XFe are depleted in these sectors (Fig. 10). Major elements in garnet crystals with radial sectors (grt 1 and grt 3) also show radial zoning patterns (Figs. 8 and 9).

Trace element distributions in garnet

Crystals grt 1, grt 3 and grt 4 were chosen for trace element analysis. In the largest crystal (grt 1), trace elements including Li, P, Sc, Ti, Cr, Ni, Zn, Nb, light rare earth elements (LREE, from La to Nd), Y and HREE exhibit apparent radial sector zoning patterns, whereas V, Co, Rb and Zr display fuzzy sector zoning. Four groups of sectors based on the trace element distributions are identified (sector 1 to sector 4 in Fig. 8). Sector 1 is featured by enrichments of Li and major elements Ca and Mn and by depletions of Sc, Cr, Y, LREE and HREE as well as the major element Mg. Sector 2 is characterized by enrichments of Sc, Cr, Y, LREE and HREE and by depletion of Li. Sector 3 is marked by a medium concentration of Cr (compared to sectors 1 and 2), enrichments of Li, Sc, Ti, Y, Nb and HREE as well as the major elements Ca and Mn, and depletions of P, Ni, Zn and HREE as well as the major element Mg. Sector 4 is depleted in Li and Ti, whereas concentrations of other elements are unclear due to the resolution of the maps. In addition to these sector zoning features, Li, Sc, Ti, Tb, Y and HREE are also enriched in the core (Fig. 8), whereas the enrichment of some of these elements at the outermost rim is attributed to garnet resorption and back diffusion. Na, V, and Ga decrease continuously from the core to the rim, whereas Co exhibits a contrary trend. Cr exhibits not only radial sector zoning but also displays patchy sectors. There are low Nb and Ti concentrations at the core enclosed within the higher Nb and Ti concentration sectors, and Nb and Ti decrease from the sectors to the rims. Radial sector-zoned grt 3 exhibits similar trace element zoning patterns (Fig. 9), but these are not as well developed, likely reflecting the non-central section through this crystal.

Three sectors are identified in the irregularly sector-zoned grt 4 (Fig. 10). Sectors 1 and 2 share similar trace element patterns, featured by depletions in Li, Na, Co, Tb, Y and HREE and by enrichment in Ti. Sector 3, close to the visual core, exhibits a distinct trace element pattern. Specifically, sector 3 is enriched in Li, Sc, Y and HREE.

Other minerals

Biotite contains 1.27–2.01 wt% TiO2 (Ti = 0.07–0.11 a.p.f.u.) and has a Fe# ranging between 0.40 and 0.46. Staurolite contains high Fe and low Mg contents and has a Fe# ranging between 0.81 and 0.83. Muscovite has Na = 0.13–0.15 a.p.f.u. and Si = 3.02–3.07 a.p.f.u. Plagioclase inclusions in garnet have almost uniform XAn of 0.36–0.38, whereas plagioclase grains in the matrix exhibit varying XAn ranging from 0.33 to 0.42, with an average of 0.38.

Plagioclase inclusions enclosed within grt 1 and grt 4 exhibit varying trace element features among different grains (Figs. 8, 10, S4, S6) and are classified into two types. One type (type 1) is enriched in P, V, Y, middle rare earth element (MREE, Sm to Tb) and HREE, whereas another (type 2) is depleted (V, Y, MREE and HREE) or poorer (P) in these elements. Contrasting patterns emerge in the Co enrichment for the two types of plagioclase inclusions within grt 1 and grt 4. Notably, Co exhibits an enrichment in type 1 plagioclase inclusions enclosed within grt 1 (Fig. 8), while it demonstrates an enrichment in type 2 plagioclase inclusions enclosed within grt 4 (Fig. 10).

Bulk-rock chemistry

The bulk-rock compositions from both XRF analyses and EBC calculations are presented in Table 2 and plotted in the AFM projection (Fig. S7; Thompson 1957). Most elements of the EBCs of sample 21DB103 are comparable to the XRF results, whereas the EBCs of both zones have relatively lower FeO, CaO, and MnO and higher Na2O. The discrepancies could be attributed to the extrapolation of phase modes from 2D to 3D (e.g., Palin et al. 2016) and / or neglected minor phases. In the AFM projection, both EBCs and XRF results of sample 21DB103 show higher “A” values than the average metapelite (Symmes and Ferry 1991) and the worldwide median pelite (Forshaw and Pattison 2023), whereas the bulk composition of sample 21DB104 locates close to the average metapelite (Symmes and Ferry 1991) and the worldwide median pelite (Forshaw and Pattison 2023). The bulk-rock concentrations of most trace elements are similar between the two samples (Table 2).

Table 2 Effective bulk compositions of the two zones in sample 21DB103, and major and trace element bulk compositions of the two studied samples. FeOtotal represents ferrous iron in the EBC results and ferric iron in the XRF results
Full size table

Crystallographic orientation of garnet

The crystallographic orientation maps for both sector-zoned and regular garnet crystals of sample 21DB103 obtained by EBSD analysis are largely homogeneous (Fig. S8), indicating that each of them is characterized by a single crystallographic orientation. These results suggest that both regular (Grt 2) and sector-zoned (Grt 1 and Grt 3) garnet crystals of sample 21DB103 are single grains, excluding the possibility of coalescence of multiple subgrains.

Metamorphic P-T conditions

Conventional geothermobarometry

The near peak P-T conditions obtained by the GBPQ-GB geothermobarometers for the quartz-poor and quartz-rich zones of sample 21DB103 are 7.1 ± 1.2 kbar / 605 ± 25 ℃ and 7.1 ± 1.2 kbar / 602 ± 25 ℃, respectively. The near peak P-T conditions obtained by GBPQ-GB, GBAQ-GB and GASP-GB geothermobarometers for sample 21DB104 are 6.1 ± 1.2 kbar / 604 ± 25 ℃, 6.8 ± 1.8 kbar / 587 ± 25 ℃, and 6.0 ± 0.8 kbar / 604 ± 25 ℃, respectively. These P-T conditions are almost identical, as may be expected given the close proximity of the sample locations (Fig. 1A).

Metamorphic phase equilibria and P-T conditions of initial garnet growth

The system compositions used for the phase equilibria modelling are given in Supplementary Table S9. The metamorphic phase equilibria diagrams for the two samples are presented in Fig. 11.

Fig. 11
figure 11

Phase equilibrium diagrams for the studied samples. (A) quartz-rich zone of sample 21DB103. (B-C) Quartz-poor zone of sample 21DB103. (D) Sample 21DB104 with COH fluid and graphite. (E) Sample 21DB104 with pure H2O as fluid phase and without graphite. (F) Summary of the P-T conditions of initial garnet growth in 21DB103 and 21DB104 (from D). Garnet isopleths of core composition or composition with lowest XMg are shown and garnet stability field are marked in gray. The dashed red line in E is the garnet-in curve from D

Full size image
Quartz-rich zone of sample 21DB103

As shown previously (e.g., Atherton 1964; Mahar et al. 1997; Symmes and Ferry 1992; White et al. 2014), a small increase in bulk MnO content significantly lowers the P-T conditions of the garnet-in reaction. For the quartz-rich zone of 21DB103, garnet-bearing assemblages are predicted across the entire phase diagram and down to 1 kbar / 450 ℃. The garnet composition with the lowest XMg (location “SI” of the profile in Fig. 5) was employed to estimate the P-T conditions of initial garnet growth. This garnet composition can be predicted with our equilibrium model at ~ 5.0 kbar / ~540 ℃ in an assemblage with plagioclase, ilmenite, chlorite, muscovite, biotite, quartz, graphite and COH-fluid, given that the observed isopleths intersect in a small P-T area (Fig. 11A).

Quartz-poor zone of sample 21DB103

Similar to the quartz-rich zone, garnet-bearing assemblages are also predicted across the entire phase diagram when using the EBC of the quartz-poor zone of this sample. The XFe, XMn and XCa isopleths of the garnet core with lowest XMg (location “C” of the profile in Fig. 6) intersect at a narrow P-T range, centering at ~ 3.8 kbar / ~510 ℃ in an assemblage with plagioclase, ilmenite, chlorite, muscovite, biotite, quartz and graphite, in equilibrium with COH-fluid (Fig. 11B). The thermodynamic and analytical uncertainties and comparatively large interval of XMg isopleths (Fig. 11C) may explain the misfit in the prediction of this garnet component.

Sample 21DB104

The XFe, XMn and XCa contents in the core of the largest garnet (location “C” of the profile in Fig. 7A) can be predicted at ~ 4.4 kbar / ~512 ℃ in an assemblage with plagioclase, ilmenite, muscovite, chlorite, margarite, quartz and graphite, in equilibrium with COH-fluid (Fig. 11D). These P-T estimates are located ~ 0.8 kbar / ~45 ℃ (closest distance between the isopleths intersection area and the garnet-in reaction) above the predicted equilibrium P-T conditions of the garnet-forming reaction. Phase equilibrium modelling with pure H2O as fluid phase yields an intersection of garnet core isopleths at a similar pressure (~ 4.4 kbar) but higher temperature (~ 520 ℃) compared to the scenario where a COH-fluid in equilibrium with graphite has been assumed (Fig. 11E).

Discussion

Al2SiO5 phase variation and P-T conditions of initial garnet growth within c. 120 m transect

Although the two samples are collected within about 120 m, the Al2SiO5 phase in sample 21DB103 is sillimanite whereas in sample 21DB104 it is kyanite (Figs. 2, 3 and 4). This observation is not surprising because kyanite in sample 21DB104 mainly occurs as tabular porphyroblasts (Fig. 4A) and represents part of the peak assemblage, whereas sillimanite in sample 21DB103 mainly occurs as tiny needles along the plagioclase grain boundaries (Fig. 3D), indicating a strong control of local bulk composition and partial equilibrium during retrograde metamorphism (e.g., Lanari and Hermann 2021). This is supported by the observed replacement of rutile by ilmenite (Fig. 3B), indicative of decompression, corresponding to the formation of sillimanite at lower pressure than that of kyanite. In addition, tabular kyanite crystals are also found in neighboring metapelitic rocks, in which sillimanite is absent, further confirming that the studied rocks experienced kyanite-grade peak metamorphic conditions.

The near peak P-T conditions of both studied rocks may be largely underestimated due to the resorption of garnet and the Fe2+-Mg exchange between garnet and biotite during retrograde modification (e.g., Florence and Spear 1991). Although garnet in sample 21DB103 was randomly sectioned, we assume that the compositions of the sector inner part (location “SI” in the profile in Fig. 5) formed during initial growth of the garnet segments (real core) because they contain the lowest XMg, and these garnet compositions can be predicted by our phase equilibrium modelling. Consequently, sector-zoned Grt 1 in the quartz-rich zone of sample 21DB103 started to crystallize at higher P-T conditions (~ 5.0 kbar / ~540 ℃; Fig. 11A) than Grt 2 in the quartz-poor zone (~ 3.8 kbar / ~510 ℃; Fig. 11B) (Fig. 11F). These results may imply that the P-T conditions required for the initial crystallization of the sector-zoned garnet (Grt 1) exceeded that required for the initial crystallization of regular garnet, and the growth of Grt 1 postdates Grt 2 (Fig. 11F).

The initial growth of sector-zoned garnet in sample 21DB104 occurred at P-T conditions of ~ 4.4 kbar / ~512 ℃ (Fig. 11D and F) after significant overstepping (~ 0.8 kbar / ~45 ℃ above the garnet-in curve). Alternatively, the discrepancy between the estimated P-T conditions of initial garnet crystallization and the garnet-in reaction may be a result of the diffusional overprint of the garnet core composition. However, this possibility is unlikely, because the predicted XMn contents at the garnet-in reaction are significantly higher than the observed XMn (Fig. 11D). This difference in garnet composition exceeds the level of diffusional overprint that garnet of this size could experience at kyanite-grade conditions for geological reasonable timescales. Furthermore, the observed contents of XFe, XMn and XCa intersect in a small P-T area, excluding the significant modification of the garnet core composition as possible explanation. To further test if the overstepping may be an artifact of the chosen thermodynamic database and solution models, different scenarios were tested (File S2; Fig. S9), also suggesting overstepping. Hence, the garnet isograd in the Danba dome may be controlled by both thermodynamic P-T-X relations and kinetics associated with the nucleation and growth of garnet (Carlson et al. 2015a; Pattison and Tinkham 2009; Pattison et al. 2011; Spear et al. 2014; Wilbur and Ague 2006).

Role of graphite

Textural sector zoning of garnet is commonly reported in graphite-rich rocks (e.g., Burton 1986; Gaidies et al. 2024; Rice and Mitchell 1991). In this study, distinct petrographic observations within a thin section of sample 21DB103 (Fig. 2) indicate strong correlations between the presence of garnet sector zoning, the abundances of quartz and graphite, as well as the matrix grain size. Combined with the garnet-quartz intergrowth of radial sector-zoned garnet crystals in sample 21DB104 (Figs. 4D and 9), it is suggested that the presence of COH-fluid controlled by the equilibrium with abundant graphite may have lowered the solubility of SiO2 in the intergranular fluid phase (e.g., Burton 1986; Carlson et al. 2015b). Consequently, an excess of SiO2 at the site of garnet growth may have favored the incorporation of quartz rods into the growing garnet, resulting in the formation of the observed garnet-quartz intergrowths (Fig. 4D). Differences in the matrix grain sizes between the two zones of sample 21DB103 (Fig. 2) may suggest that abundant graphite hindered the matrix coarsening / Ostwald ripening process during prograde metamorphism. The systematic variations in the biotite and plagioclase compositions of the two zones may also reflect the variation in graphite abundances and its control on chemical transport rates and, consequently, the local bulk compositions and reaction kinetics.

As has been shown previously (e.g., Greenwood 1967; Kerrick 1974), the presence of CO2 in the metamorphic fluid phase can lower the P-T conditions of garnet-forming reactions. This is further attested in this study by comparing the garnet-forming reaction in the phase diagrams of sample 21DB104, which is lowered by ~ 5–15 ℃ in the presence of graphite and COH-fluid compared to a graphite-free system (Fig. 11E). However, the observed garnet core compositions can be predicted by our equilibrium models at P-T conditions significantly exceeding the garnet-in reactions in both scenarios (Fig. 11D and E, S9), indicating that the petrological record of reaction overstepping developed in our data is stronger than the influences of graphite and thermodynamic database on equilibrium phase relations.

Origin of element distribution and growth history of sector-zoned garnet

Chemical sector zoning in garnet has been suggested to be linked with variations in the incorporations of different elements into certain crystallographic faces (e.g., Kohn 2004; Kohn et al. 2024; Stowell et al. 2011). In this case, four groups of sectors manifested by different elemental incorporations in grt 1 from sample 21DB104 (Fig. 8) indicate that there should be at least 4 faces retaining differential element incorporation properties, even for the immobile Cr that tends to be enriched in remnant phases (Burns 1993; George et al. 2018; Yang and Rivers 2001). However, there are usually only two types of garnet faces in a 2D thin section (e.g., trapezohedral {211} and dodecahedral {110}; Kohn 2004; Stowell et al. 2011). Therefore, variations in element incorporations of different lattice planes alone does not explain the observed chemical sector zoning in grt 1 of sample 21DB104 (Fig. 8). Hence, an alternative mechanism may have operated to account for the differential elemental distributions in the same crystallographic face. Nevertheless, any incorporations and substitutions among these elements should maintain electroneutrality.

The chemical sector zoning is even retained for the fast-diffusing component Mn (e.g., Chakraborty and Ganguly 1992; Fig. 8), thus rapid growth is required for the development and preservation of the observed chemical sector zoning of Mn. The preservation of concentric chemical zonation and poorly developed sectors in garnet crystals in the quartz-poor zone of sample 21DB103 (Fig. 6, S1) indicate that they experienced a different crystallization mechanism than the sector-zoned garnet.

The distinct trace element concentrations of the two types of plagioclase inclusions in garnet of sample 21DB104 (Figs. 8, 10, S4, S6) indicate that the different plagioclase grains likely replaced different phases. Combined with petrographic observations (tiny apatite is hosted by type 1 plagioclase inclusions whereas allanite is observed in the neighboring garnet interior of type 2 plagioclase inclusions; Figs. 4C, 8 and 10), it is inferred that type 1 (enriched in P, V, Y, MREE and HREE compared to type 2) formed after apatite whereas type 2 (similar enrichment in LREE to type 1, especially in La and Ce) replaced allanite. Similar pseudomorphic textures of allanite replaced by plagioclase have been reported in pelitic schists from the Appalachians (Corrie and Kohn 2008).

Although the four sectors of Grt 1 in sample 21DB103 share similar zoning patterns, EBSD results repudiate the possibility of polycrystal coalescence (Fig. S8). Increasing XMg from the sector inner part to the sector faces and edges of Grt 1 in sample 21DB103 indicate a later growth of the observed sector boundaries (Fig. 12A). This hypothesis can also explain the higher XMg gradient close to the visual core. In Grt 1 of sample 21DB103, the similarity of element concentrations between sector edges and faces indicates there are still vacancies along the sector edges and they are not filled after the development of sectors (scenario 1 in Fig. 12A); or alternatively, the outermost enclosed faces have been resorbed and led to exposure of internal sectors and their edges (scenario 2 in Fig. 12A). Consequently, the sector edges would be affected during retrograde metamorphism, exhibiting similar features to the sector faces (visual rim).

Fig. 12
figure 12

Schematic diagrams illustrating the origin of observed sector zoning in the studied samples. (A) Growth history of Grt 1 in sample 21DB103. Numbers in the sectors indicate earliest growth for segment 1 and latest growth for segment 3 for each sector. (B) Reconstructed 3D volume of largest crystal grt 1 in sample 21DB104. White arrows denote reentrants or holes at the crystal surface. (C) Possible origin of irregular sector zoning by fluid infiltration in sample 21DB104

Full size image

The major and trace element distributions (Figs. 8 and 9), as well as the reconstructed 3D volumes (Fig. 12B; File S1) suggest that the radial sector-zoned garnet in sample 21DB104 developed an early flower-shaped morphology with several sector arms, followed by a later dodecahedral growth. However, the preservation of irregular sectors in grt 4 of sample 21DB104 and the decoupling between highest Mn and highest Y and HREE contents (Fig. 10) imply its formation through different mechanisms and may exclude the coalescence of numerous nuclei (Whitney et al. 2008). The reentrants or holes observed at the surface of the reconstructed 3D volume of the largest garnet in sample 21DB104 (indicated by white arrows in Fig. 12B) may represent the results of local resorption (Fig. 12C). Such resorption might have reopened the internal vacancies that formed during early flower-shaped growth, allowing for fluid infiltration and chemical adjustments in the vicinity of these reopened vacancies (pathways). This hypothesis may be a possible explanation for the formation of the irregular patchy sectors in grt 4 of sample 21DB104.

Conclusions

Graphite plays an important role in changing fluid composition and reducing the solubility of SiO2 to grow sector-zoned garnet, as well as impeding matrix coarsening during prograde metamorphism. Sector-zoned garnet grew later and at higher P-T conditions than regular garnet in the same thin section, indicating possible reaction overstepping for the formation of sector-zoned garnet. This is supported by the results obtained from isopleth thermobarometry applied to the core of the largest garnet, revealing that it started to crystallize far (~ 0.8 kbar/~45 ℃) from equilibrium. Development of sector-zoned core and dodecahedral faces of garnet may be related to rapid growth with changes in crystal morphology, rather than by coalescence of polycrystals. Plagioclase inclusions with distinct trace element concentration patterns indicate that they replaced different phases, while irregular element distributions in sector-zoned garnet may have developed through fluid infiltration and local chemical adjustments.