Metamorphic Evolution of Garnet-Bearing Ultramafic Rocks in the Hujialin Area, Sulu Ultrahigh-Pressure Orogenic Belt, Eastern China

Abstract

The Rizhao Hujialin area is located in the central Sulu ultrahigh-pressure orogenic belt, where garnet clinopyroxenite is exposed in the upper part of an ultramafic rock complex and serpentinized dunite is exposed in its lower part. Based on textural criteria, the garnet clinopyroxenites were divided into three types: Equigranular garnet, porphyroclastic garnet, and megacrystic garnet pyroxenites. The garnet clinopyroxenites have convex-upward chondrite-normalized rare earth element patterns, large positive Pb anomalies, and depletion of high-field-strength elements (e.g., Nb, Zr, and Ti), suggesting a mantle source protolith overprinted by fluid metasomatism. Petrographic, mineral chemistry, phase equilibrium modeling, and zircon U–Pb geochronology data show that the evolutionary stages of the Hujialin garnet clinopyroxenites were as follows: Stage I: formation of the magmatic protoliths; stage II: formation of megacrystic garnet pyroxenite accompanying subduction; stage III: formation of porphyroclastic or equigranular garnet clinopyroxenite with a mineral assemblage of garnet + clinopyroxene + ilmenite + humite accompanying initial exhumation at ~215.0 ± 5.7 Ma; stage IV = progressive cooling and decompression associated with the crystallization of water-bearing minerals such as clinochlore and pargasite at 206 Ma; and Stage V = late epidote amphibolite-facies retrograde metamorphism producing a mineral assemblage of garnet + clinopyroxene + amphibole + chlorite + epidote + ilmenite at ~180–174 Ma associated with fluid activity in shear–tensional fractures and/or pores. The P-T conditions of the peak metamorphism were estimated at 4.5 ± 0.5 GPa and 800 ± 50 °C. Retrograde metamorphism recorded conditions of 1.0 GPa and 710 ± 30 °C during the exhumation and cooling process. The mineral transformation from early high-Al clinopyroxene to garnet and to late diopside records the general metamorphic evolution during subduction and exhumation, respectively. One zircon U–Pb analysis presents the Palaeoproterozoic age of 1817 ± 40 Ma, which is coeval with widespread magmatic and metamorphic events in the North China Craton.

1. Introduction

Following the first discovery of the ultrahigh-pressure (UHP) SiO₂ polymorph coesite in crustal metamorphic rocks [1,2], many other UHP metamorphic indicators have been found in eclogites or garnet peridotites from UHP terranes worldwide [3,4,5,6,7,8,9,10]. UHP metamorphic rocks record evidence of lithospheric subduction and exhumation processes driven by plate collision or mantle convection, and provide natural samples to investigate different types of interactions that occur between the lower crust and mantle [11,12,13,14,15,16,17].

The Triassic Dabie–Sulu collision zone located between the Sino–Korean boundary and Yangtze Craton is the world’s largest UHP orogenic belt, with an areal extent of >30,000 km². Pyroxenites are ubiquitous in the Dabie–Sulu UHP orogenic belt, and may represent an important source of magma in the upper mantle. Therefore, the petrogenesis, tectonic setting, and geodynamic evolution of orogenic belt pyroxenites have become a focus of research. Garnet pyroxenites in the Dabie–Sulu UHP orogenic belt comprise two types: (1) lenses or thin layers in mantle peridotite, such as garnet peridotites in Xugou and Yangkou areas [18]; (2) parts of cumulate igneous bodies that originated from magmatic differentiation, which comprise mainly crustal mafic–ultramafic rocks, such rocks as in the Bixiling area [8].

The Sulu UHP terrane forms the eastern part of the Dabie–Sulu collision zone between the North China and Yangtze cratons, and consists of UHP, very high-pressure (VHP), and HP zones separated by ductile shear zones (Figure 1). These rocks are unconformably overlain by Jurassic clastic strata and Cretaceous volcanic–sedimentary rocks and intruded by post-orogenic Mesozoic granites [19,20]. Coesite-bearing eclogite and related ultramafic rocks usually occur in the surrounding gneisses as lenses or layers, which were initially regarded as being allochthonous. However, coesite inclusions in zircons are also found in gneisses from the Sulu UHP terrane, indicating that the whole orogenic belt experienced UHP metamorphism [21]. Studies of garnet clinopyroxenite in the Hujialin area of the Sulu UHP orogenic belt were carried out in order to understand the petrogenesis and metamorphic evolution of these rocks [8,22,23,24,25,26,27,28]. Hiramatsu and Hirajima [22] divided the metamorphic evolution of the Hujialin garnet clinopyroxenites into three stages, and calculated a peak metamorphic temperature of 740 °C with pressures above 2.5 GPa. Chen et al. [23] suggested there were at least two episodes of exsolution, based on the microstructure of amphibole and ilmenite exsolutions in clinopyroxene from the garnet clinopyroxenites, and obtained peak metamorphic pressures of 8 GPa. Zhang et al. [18,24] considered that the Hujialin garnet clinopyroxenites may have formed at 1400 °C and P > 15 GPa. Yang [26] conducted a detailed study of Ca-rich garnet clinopyroxenites from the Hujialin area, and proposed that the Ca-rich garnet clinopyroxenite might have been derived from spinel clinopyroxenite. Zhao et al. [29] also studied the rocks in this area and determined that the Hujialin garnet clinopyroxenites experienced four stages of metamorphic evolution, and obtained a zircon U–Pb age of 215 ± 2 Ma for the UHP metamorphism. However, the nature of the protolith, peak metamorphic conditions, and metamorphic evolution of the Hujialin garnet clinopyroxenite are still controversial. In this paper, we present detailed petrographic, mineralogical, zircon age dating, and phase equilibrium modeling results for the Hujialin garnet clinopyroxenites, and discuss their petrogenesis and metamorphic history.

2. Geological Setting

The Sulu UHP orogenic belt is the eastern part of the Triassic Dabie–Sulu orogenic belt, and is bounded by the Tanlu Fault in the west. The Jiashan–Xiangshui and Wulian–Yantai faults constitute its southern and northern boundaries adjacent to the Yangtze and Sino–Korean plates, respectively. The Hujialin area is situated in Rizhao City, Shandong Province, and is located in the middle of the Sulu UHP–HP orogenic belt. The rock unit (i.e., the No. 8 rock body) is ca. 6 km long and trends NNW–SSE. This rock unit is cut by a NE–SW-trending fault (Figure 2). The collected samples are all from the northern rock unit, and garnet pyroxenite can be observed in the upper part of the quarry outcrop in contact with dunite in the lower part. The transverse contact relationship between peridotite and pyroxenite described in previous studies [18,26] is only observed locally. In generally, the pyroxenite lies above the dunite and they tectonically constitute a synform (Figure 3a).

3. Analytical Methods

Whole-rock major element contents were analyzed using an X-ray fluorescence (XRF) spectrometer at the Shandong Provincial Bureau of Geology and Mineral Resources Analysis Technology, Shandong, China. Sixteen major and minor elements were determined by wavelength dispersive XRF spectrometry (method GB/T 14506.28-2010) (Table S1).

Garnet and other rock-forming minerals were analyzed with a JEOL JXA-8230 electron probe micro-analyzer (EPMA) at the State Key Laboratory of Marine Geology, Tongji University, Shanghai, China. The operating conditions were: 15 kV accelerating voltage; 10 nA beam current; 0–5 μm beam diameter. Natural and synthetic mineral standards (SPI) and ZAF corrections were used to calibrate all the quantitative analyses.

U–Pb dating and trace element analyses of zircon were simultaneously conducted by laser ablation–inductively coupled plasma–mass spectrometry (LA–ICP–MS) at the Wuhan Sample Solution Analytical Technology, Wuhan, China. Detailed operating conditions for the LA–ICP–MS system and data reduction are the same as those described by Zong et al. [31]. Laser sampling was performed using a GeolasPro laser ablation system that consists of a COMPexPro 102 ArF excimer laser (wavelength = 193 nm; maximum energy = 200 mJ) and a MicroLas optical system. An Agilent 7700e ICP–MS instrument was used to acquire the ion signal intensities. Helium was used as a carrier gas. Argon was used as the make-up gas, and mixed with the carrier gas via a T-connector before entering the ICP. A signal smoothing device is included in this laser ablation system. The spot size and frequency of the laser were set to 32 µm and 6 Hz, respectively. Zircon 91500 and glass NIST610 were used as external standards for U–Pb dating and trace element calibration, respectively. Each analysis incorporated a background acquisition of ca. 20–30 s, followed by 50 s of data acquisition during ablation. The Excel-based software ICPMSDataCal was used to perform off-line data selection, integration of background and ablation signals, time-drift corrections, and quantitative calibrations for the trace element analysis and U–Pb dating [32]. Concordia diagrams and weighted-mean calculations were performed using Isoplot/Ex_ver3 [33].

4. Petrography and Mineral Chemistry

The rock samples can be divided into three types: Equigranular garnet, porphyroclastic garnet, and megacrystic garnet pyroxenites (Figure 3b). The porphyroclastic garnet pyroxenite is randomly distributed in a narrow strip in the fine-grained garnet pyroxenite, and garnet megacrysts are only found sporadically in the western part of the rock unit. These three types of garnet clinopyroxenite have a similar mineral composition, including clinopyroxene, garnet, ilmenite, magnetite, and minor retrograde amphibole, chlorite, and spinel. Some samples also contain olivine, humite, and clinochlore.

Garnet generally occurs as irregularly shaped megacrysts or porphyroclasts in the pyroxenites, but in the matrix small euhedral grains are also scattered amongst the equigranular diopside aggregates (Figure 4a–d). In some of the porphyroclastic clinopyroxene, the excluded garnet lamellae can be seen (Figure 4e). The garnet consists mainly of grossular and pyrope components (Table S2; Figure 5a). There is a subtle change in the composition of the end-members in the porphyroclastic grain in the studied sample (Figure 6). On one side of the garnet cross-section, the grossular component increased sharply in the rim, the pyrope component decreased simultaneously, and the almandine and andradite components did not change significantly, similar to observations reported by Hiramatsu and Hirajima [22]. The core and rim of the porphyroclastic garnet shown in Figure 6 have average compositions of Prp_0.35–0.29Alm_0.22–0.15Gro_0.50–0.43Spe_0.07–0.03; the outermost layer has an average composition of Prp_0.33–0.27Alm_0.14–0.12Gro_0.46–0.41 Spe_0.04–0.03. These characteristics indicate that the outermost layer experienced metamorphic replacement rather than a metamorphic growth process from core to rim. This is consistent with the porphyroclastic garnet typically being in relic form (Figure 4c). In addition, the Si content of porphyroclastic garnet is generally >3.0 apfu. Zhang et al. (2003) obtained a similar content for garnet, particularly in a sample of megacrystic garnet-bearing clinopyroxenite. The high garnet Si contents imply that the Hujialin garnet clinopyroxenite experienced UHP metamorphism [18].

Clinopyroxene occurs mainly as porphyroclast, matrix and inclusions in the megacrystic garnet. (Figure 4a,c–e). The chemical compositions of clinopyroxene are listed in Table S3. In the Wo–En–Fs diagram, data for all clinopyroxene is plotted in the diopside field (Figure 5b). The fine-grained clinopyroxene in the matrix is chemically zoned, and the diopside has a constant X_Mg (Mg/[Mg + Fe²⁺]) of 0.91–0.99, and contains a small amount of Al₂O₃ (0.86–3.1%) and Na₂O (0.42–0.98%; Table S3). The porphyroclastic clinopyroxene is diopside with X_Mg of 0.92–0.95, and an A1₂O₃ content that decreases from the core (up to 3.1%) to the rim (ca. 1.5%). Some of the clinopyroxene surrounded by chlorite in the matrix shows component zoning. For example, X_Mg decreases from the core (0.94) to rim (0.79–0.83), and A1₂O₃ contents increase from the core to rim. Al₂O₃ contents of the clinopyroxene inclusions in the porphyroclastic garnets (2.6–3.1%) are distinctly higher than those of the porphyroclastic (1.45–1.87%) and matrix clinopyroxene (0.74–1.82%). In a Si–Al diagram, the three types of clinopyroxene have different data patterns. Data for clinopyroxene inclusions tend to show a negative correlation, whereas the other data, especially in the matrix, are relatively clustered, implying that clinopyroxene in the matrix is equilibrated (Figure 7).

In addition, some fine-grained garnet, ilmenite, magnetite, and pargasite inclusions are also present in clinopyroxene (Figure 4e,i and Figure 8). In the studied clinopyroxene, ilmenite lamellae formed by exsolution is common, and is cut by exsolved pargasite (Figure 8b,d).

Most of the ilmenite associated with garnet and clinopyroxene is anhedral, and a small amount of ilmenite is present as exsolution lamellae within clinopyroxene (Figure 4g–i). The two groups of ilmenite can be discriminated by their Mg and Mn contents (Table S4): High-Mg and high-Mn groups. The high-Mg ilmenite is usually associated with green spinel in the matrix of the porphyroclastic garnet pyroxenite, and the high-Mn ilmenite occurs mostly in the equigranular garnet pyroxenite.

Spinel is anhedral and looks green in transmitted light. It often associates with ilmenite (Figure 4b,g). The spinels in this study have very low Cr/(Cr + Al) ratios, similar to those reported by Zhang et al. [25] (<0.2), and the Mg/(Mg + Fe²⁺) ratio is between 0.45 and 0.60 (Table S5). The Ti contents of the spinel are extremely low, and it is possible that during the retrograde process, Ti in the spinel entered the accompanying ilmenite.

The amphibole can be divided into two types (Figure 9), which are pargasite and edenite. Amphibole occurs in three different microtextural contexts: (1) inclusions in porphyroclastic garnet; (2) exsolution lamellae in clinopyroxene included in porphyroclastic garnet (Figure 8b–d); (3) small grains spatially related to garnet rims (Figure 4a). The edenite is mainly a retrograde phase within amphibole-rich bands in the pyroxenite samples (Figure 4h). The Al^IV content (1.60–1.95) of pargasite is higher than that of edenite (1.20–1.45), and both have a high Mg/(Mg + Fe²⁺) ratio (0.83–1.00). A Na–Si plot (Figure 9c) shows that the Si contents of pargasite inclusions in garnet are significantly lower than those of the pargasite in the matrix. In a K–Si plot (Figure 9d), there is a positive correlation between the K and Si contents of pargasite inclusions and matrix pargasite. These pargasites occur mainly as inclusions in garnet or in the matrix associated with clinopyroxene, and its Na₂O content (1.45–2.53%) is slightly lower than that of pargasite exsolved in clinopyroxene (2.86–3.64%), and similar to the Na₂O content (1.77–2.45%) of pargasite in the Lizard Complex lherzolite [35]. The clinopyroxene-exsolved ilmenite lamella shown in Figure 4i exhibit a strong crystallographic orientation. Figure 4a shows the exsolution texture of clinopyroxene inclusions in megacrystic garnet. The exsolved minerals are mainly amphibole and ilmenite, and the two minerals are mutually cut by each other. Chen and Xu [23] reported similar observations for clinopyroxene from the Hujialin area, whereby acicular ilmenite cuts pargasite lamellae, indicating that exsolution of ilmenite occurred later than that of pargasite. This shows that the Hujialin garnet clinopyroxenite experienced at least two metamorphic events during exhumation and decompression.

Two chlorites were found in sample RZ-1; one is a clinochlore with polysynthetic twinning (Figure 4f) and the other is retrograde chlorite (Figure 4h). There are differences in the compositions of the two chlorites (Table S6). Al₂O₃ and MgO contents of the clinochlore are slightly higher than those of the chlorite, while FeO contents are significantly lower in the clinochlore. This indicates that the two chlorite generations reflect different metamorphic events. Numerous well-formed clinochlore crystals is surrounded by small grains of garnet, clinopyroxene, ilmenite, spinel, and other minerals (Figure 4f). Clinochlore can be stable at HP and survive UHP metamorphism, as evidenced by the reaction clinochlore = forsterite + pyroxene + H₂O at pressures of up to 3.5 GPa [36,37]. Clinochlore is the primary water transporting mineral in subduction zones, and may be the only aqueous phase present above the decomposition temperature of serpentine [38]. When clinochlore eventually breaks down, it contributes a large amount of water to the subduction zone [39].

Olivine was rarely recognized in the studied samples, and a very small amount of olivine was found during electron microprobe analysis in the matrix or as inclusions in clinopyroxene. A small amount of humite was also found in sample RZ-1 (Figure 4b,g), with ilmenite surrounding reddish brown humite. The humite contains mainly SiO₂ (34–41%), FeO (16–23%), and MgO (40–42%) (Table S6). The Al₂O₃ and CaO contents are extremely low and the TiO₂ contents are negligible. In the Sulu UHP metamorphic terrane, Yang [40] reported garnet pyroxenite and eclogite containing Ti-clinohumite, which is mainly present between garnet and pyroxene. The garnet and orthopyroxene combined with Ti-clinohumite typically record a pressure condition of >4.5 GPa, indicating UHP conditions [41]. However, the humite TiO₂ content in the studied sample was extremely low, indicating that Ti was consumed during exhumation after UHP metamorphism by surrounding ilmenite (Figure 3g).

5. Whole-Rock Geochemistry

5.1. Major Elements

The analysis results of the major elements of the garnet pyroxenite samples in Hujialin are shown in Table S1. Samples RZ-1 and RZ-2 are collected near the megacrystic garnet pyroxenite, which is rich in garnet porphyroclasts and labeled as garnet-rich porphyroclastic pyroxenite. Samples RZ-5, 6, 8, and 14 are categoried to porphroclastic garnet pyroxenite. Sample RZ-11 is equigranular garnet pyroxenite, enriched in pyroxene. It can be seen that the three types of samples show slight differences in the composition of major elements. The content of MgO (12.98%), SiO₂ (53.88%) and TiO₂ (0.73%) in equigranular garnet pyroxenite is slightly higher than that of porphyroclastic garnet clinopyroxenite MgO (9.32–11.0%), SiO2 (42.41–51.66%) and TiO₂ (1.39–1.76%). The samples have relatively high CaO/Al₂O₃ (2.85–3.05), low MgO/(MgO + FeO^T) (0.62–0.51), and have a similar composition as large clinopyroxene crystals, plus exsolved garnet in eclogite xenoliths from kimberlite analyzed by Zhang et al. [18], which suggests their protolith may have been a pyroxenite [42,43].

5.2. Trace Elements

Trace element data for the three types of garnet clinopyroxenites are listed in Table S1 and shown in Figure 10a,b. The garnet clinopyroxenites have similar convex-upward rare earth element (REE) patterns, with light REE enrichment and heavy REE depletion. Given that heavy REEs partition strongly into garnet, the depleted heavy REE pattern of the garnet clinopyroxenites indicate that their protolith was garnet-free. Trace element patterns (Figure 10b) show that the three types of rock have some similarities, but also significant differences. The samples have large positive Pb anomalies and depletions in high-field-strength elements (HFSEs; e.g., Nb and Ti), suggestive of fluid metasomatism [28].

6. U–Pb Geochronology

Two garnet clinopyroxenite samples (18RZ-1-1 and 18RZ-1-3) were U–Pb dated, which yielded five groups of ages (Table S7, Figure 12): (1) 1824 ± 40 Ma, (2) 753.0 ± 9.8 Ma, (3) 350.0 ± 3.9 Ma, (4) 215.0 ± 5.7 Ma, and (5) 180.0 ± 2.4 Ma. Zircon cathodoluminescence images (Figure 11) and morphological characteristics identified two types of zircon. The first is compositionally unzoned with Th/U = 0.015–0.081, representing metamorphic zircon. The second has core and rim structures with a strong contrast in cathodoluminescence and Th/U = 0.57–2.20, which are magmatic zircon.

In chondrite-normalized REE diagrams, the magmatic zircon shows obvious positive Ce and negative Eu anomalies and heavy REE enrichment, while the metamorphic zircon has flat heavy REE patterns and no Eu anomalies (Figure 12b).

The first group of ages from 1824–1795 Ma, with a weighted-mean age of 1817 ± 40 Ma (MSWD = 0.2; n = 3), is coeval with magmatism that occurred during rifting of the Columbia Supercontinent in the North China Craton. This may represent the formation age of the Hujialin garnet pyroxenite, when it crystallized from a magma in the upper mantle or at the crust–mantle boundary [45]. The second group of ages with a weighted-mean of 753.0 ± 9.8 Ma coincides with the formation age of the surrounding gneisses (ca. 780–750 Ma) in the Sulu orogenic belt [46,47,48]. The third group of ages is from 371–336 Ma. It has been reported [49] that inherited magmatic zircon has a ²⁰⁶Pb/²³⁸U age of 378 ± 6 Ma in the Weihai garnet peridotite. This type of zircon is not completely recrystallized and has not experienced complete Pb loss. Such age data cannot represent the age of the protolith, and have no geological significance. The fourth group of ages is from 229–204 Ma, with a weighted-mean of 215.0 ± 5.7 Ma (MSWD = 1.7; n = 10). The ages of this fourth group are slightly younger than those of Sulu UHP metamorphism (240–220 Ma) [46,47,48,49], and likely records the timing of exhumation after UHP metamorphism. The fifth group of ages is from 180–174 Ma, and records amphibolite-facies retrogression.

7. Discussion

7.1. Metamorphic Evolution

In order to constrain the metamorphic evolution, phase equilibria modeling is necessary. In this paper, Perple_X-6.7.4 software [50,51], combined with the Holland and Powell [52] internally consistent thermodynamic database, was used to construct the P–T pseudosection. Major elements of sample RZ-1 (Table S1) were used in the Na₂O–CaO–TiO₂–FeO–MgO–Al₂O₃–SiO₂–H₂O (NCFMASHTO) system for phase equilibria modeling. Because K₂O (0.02%) and MnO (0.11%) do not control any major phase in the rock system, and MnO is only present in garnet, they were ignored in the calculations. In this model, we assumed that the fluid H₂O is saturated, and the pseudosection shown in Figure 13 was calculated using Gt (WPH), Fsp (C1) [52], Ilm (WPH) [53], Cpx (HP) [54], Amph (DHP) [55], Chl (HP), Ep (HP) and Spl (WPC) [56] solution models.

According to the petrography and mineral chemistry, the evolution of the Hujialin garnet clinopyroxenites can be divided into five stages: Stage I = protolith formation; stage II = formation of porphyroclastic garnet pyroxenite accompanying subduction; stage III = formation of porphyroclastic or equigranular garnet clinopyroxenite with a mineral assemblage of Grt + Cpx + Ilm + Hu accompanying exhumation; stage IV = progressive cooling and decompression producing a mineral assemblage of Grt + Cpx + Ilm + Chl + Amp; stage V = late epidote amphibolite-facies retrograde metamorphism producing a mineral assemblage of Gt + Cpx + Amp + Chl + Ep + Ilm.

Protolith formation has been speculatively inferred from petrographic observations of inclusions in garnet porphyroclastics. In general, previous studies have recognized cumulate spinel pyroxenite as being the protolith, which formed in the upper mantle or at the crust–mantle boundary over a P–T range of 1–1.5 GPa and >1000 °C for stage I [18,26,29].

In stage II, the exsolved garnet and ilmenite in clinopyroxene (see Figure 4e,i) are regarded as the product of subduction and increasing pressure [18,29]. From the protolith to garnet clinopyroxenite, the primary clinopyroxene must have been transformed to garnet since excluded lamellae occurred in the porphyroclastic clinopyroxene. The reduction of Al content from precusor Cpx to the matix Di is about one half, perhaps due to Al₂O₃ being consumed by garnet as the protolith is converted to megacrystic garnet clinopyroxenite. With increasing pressure, the octahedral Si–Al substitution in garnet was important. In the megacrystic garnet clinopyroxenite, garnet and pyroxene are intergrown forming a massive pegmatite-like rock (Figure 4a). The Si content of porphyroclastic garnet is high, generally >3.0 apfu, implying that the Hujialin garnet clinopyroxenite experienced UHP metamorphism. High-pressure conditions are favorable for garnet formation, due to the difficulty of the Al–Si substitution in the tetrahedral site of the Ca Tschermak component as indicated by [25,26]:

CaMgSi₂O₆ + CaAl₂SiO₆ = Ca₂MgAl₂Si₃O₁₂

(1)

Stages III and IV represent the decomposition of the megacrystic garnet pyroxenite (Figure 13). The metamorphosed ultramafic rocks underwent isothermal decompression or a fluid addition event under high stress conditions, which resulted in dynamic recrystallization. Firstly, the massive garnet was fragmented by stress, creating minute pyroxene aggregates along the edges of small grains of high-Si garnet. Newly formed diopside grains gradually replaced the garnet fragments, leaving behind pophyroclastic garnet surrounded by a clinopyroxene–garnet aggregate, which finally led to the equigranular mineral assemblage in the matrix (Figure 4d). Unlike stage II, high grossular garnet with low Si was produced in stage III, corresponding to a pressure decrease. The reaction for this is as follows [26]:

Ca₂MgAl₂Si₃O₁₂ + Mg₂SiO₄ = 2CaMgSi₂O₆ + MgAl₂O₄

(2)

The mineral assemblage of Grt + Hu + Cpx + Sp (+ Ilm) is shown in Figure 4b, i.e., Ilmenite is associated with Mg–Al spinel in the reaction, but has a high Mg content, unlike the massive spongy ilmenite scattered around the clinopyroxene, with a high Mn content which perhaps resulted from the decomposition of ilmenite-exsolved clinopyroxene.

The geothermometer for garnet–clinopyroxene mineral pairs given by Ravna (2000) [57] was used to calculate temperatures (Table S8):

$$\begin{array}{c}\mathrm{T}(°\mathrm{C})=[(1939.9+3270{\mathrm{X}}_{\mathrm{Ca}}^{\mathrm{Grt}}-1396({\mathrm{X}}_{\mathrm{Ca}}^{\mathrm{Grt}}){}^2+3319{\mathrm{X}}_{\mathrm{Mn}}^{\mathrm{Grt}}-3535({\mathrm{X}}_{\mathrm{Mn}}^{\mathrm{Grt}}){}^2+1105{\mathrm{X}}_{\mathrm{Mg}\#}^{\mathrm{Grt}}-3561\\ ({\mathrm{X}}_{\mathrm{Mg}\#}^{\mathrm{Grt}}){}^2+2324 ({\mathrm{X}}_{\mathrm{Mg}\#}^{\mathrm{Grt}}){}^3+169.4\mathrm{P} (\mathrm{GPa})/({\mathrm{lnK}}_{\mathrm{D}}+1.223)]-273\end{array}$$

(3)

This K_D = (Fe²⁺/Mg)^Grt/(Fe²⁺/Mg)^Cpx, ${\mathrm{X}}_{\mathrm{Ca}}^{\mathrm{Grt}}$ = Ca/(Ca + Mn + Fe²⁺ + Mg), ${\mathrm{X}}_{\mathrm{Mn}}^{\mathrm{Grt}}$ = Mn/(Ca + Mn + Fe²⁺ + Mg), ${\mathrm{X}}_{\mathrm{Mg}\#}^{\mathrm{Grt}}$ = Mg/(Mg + Fe²⁺).

The mineral assemblage of stage III in the calculated P–T diagram (Figure 13) is Grt + Cpx + Ilm + Hu, which is consistent with the petrographic observations and appear in the P–T domain of 4–5 GPa and at ≥750 °C. The calculated temperature of the Ravna [57] garnet clinopyroxenite Fe–Mg exchange geothermometer is 800 ± 50 °C at 4.5 GPa, and similar to the estimations from phase equilibria modeling. Therefore, the UHP metamorphic stage of the Hujialin garnet clinopyroxenite occurred at 4.5 ± 0.5 GPa and 800 ± 50 °C.

In stage IV, some water-bearing minerals such as clinochlore and amphibole were present, and the zircon U–Pb ages of 215–208 Ma are consistent with the age of granitic magmatism in Yangkou [58,59]. It is thought that deep melting of adjacent granitic rocks contributed to the metamorphism of the Hujialin pyroxenite and were directly related to exhumation. Based on the phase equilibria modeling, amphibole geobarometer of Schmidt [60], and garnet–clinopyroxene geothermometer [57], stage IV P–T condition were 1.0 GPa and 710 ± 30 °C (Table S8).

During exhumation and uplift, the garnet clinopyroxenite underwent retrograde epidote amphibolite-facies metamorphism. In sample RZ-1, some of the garnet porphyroblasts were replaced by the secondary minerals Gt + Cpx + Amp + Chl + Ep + Ilm, and some clinopyroxene was also retrograded into amphibole and chlorite aggregates. The amphibole geobarometer of Schmidt [60] and Fe–Mg exchange geothermometer of Ravna [57] yield P–T conditions for stage V of 0.5 GPa and 600 ± 30 °C.

7.2. Protolith and Structural Model

Based on petrographic studies, it has been proposed that the Hujialin garnet clinopyroxenite is formed from spinel clinopyroxenite through the exsolution of Ca-rich garnet, magnetite, ilmenite, and Fe spinel in augite [8,18,24,26,61]. The original spinel clinopyroxenite was formed by magmatic differentiation near the crust–mantle boundary [26,28]. The Hujialin garnet clinopyroxenite has high CaO and CaO/Al₂O₃, low Na, light REE enrichment, heavy REE depletion, and low Cr, Co, and Ni contents, consistent with a cumulate clinopyroxenite origin. Most of these data define a compositional trend from high-Mg cumulates to gabbros [62]. The significant depletions in incompatible trace elements (Zr, Y, and heavy REEs) in the garnet clinopyroxenite may also be related to the cumulate nature of the protolith [26]. The light REE enrichment can be attributed to fluid interaction (Figure 10a), and the low Ni and Cr contents and Mg^# values to early olivine crystallization in the magma [28].

The garnet clinopyroxenite samples show relative depletion of HFSEs (e.g., Nb, Zr, and Ti) and relative enrichment of large-ion lithophile elements (e.g., Rb, Sr, Ba, and Pb), which are typical geochemical features of subduction magmatism [63,64,65,66,67,68]. Most subduction-related UHP peridotites are interpreted to be from the lithospheric mantle wedge, above the pre-subduction zone [69,70], which typically exhibits enrichments of oceanic plate-derived elements [68,71,72,73,74,75]. According to field observations, and petrological and geochemical data, the Hujialin garnet clinopyroxenites are cumulates formed by melt crystallization in the mantle wedge of an ancient subduction zone. Zhang et al. [18] considered that the protoliths were formed in a subduction zone during Triassic continental collision and were recrystallized to form majorite at mantle depths (>450 km), and then underwent decompression exsolution as it ascended to shallow mantle depths [8,18,24,25,60]. The formation age of 1817 ± 40 Ma may correspond to the rifting of the Columbia Supercontinent and the widely distributed mafic dike swarms in the North China Craton [45]. We propose a five-stage tectonic model for the Hujialin garnet clinopyroxenite block in the Sulu UHP orogenic belt (Figure 14). The first stage (1817 ± 40 Ma) involved the rifting of Columbia, resulting in complex crust–mantle processes. The second stage (753.0 ± 9.8 Ma) involved the continental rifting of Rodinia and the generation of large-scale granitic rocks along the margin of the Yangtze Craton (i.e., the present-day gneisses) [19]. The third stage involved subduction of the Yangtze Plate and collision with the North China Craton, which triggered asthenospheric wedge flow and brought the cumulate ultramafic rocks into the subduction channel [9]. The fourth stage (215–206 Ma) involved subduction of granitic continental crust into the mantle, partial melting of the granitic rocks, and exhumation of UHP metamorphic rocks to crustal level. Finally, the UHP orogenic belt was overprinted by retrograde metamorphism.

8. Conclusions

The Hujialin garnet clinopyroxenites had a cumulate spinel clinopyroxenite protolith, with a formation age of 1817 ± 40 Ma. The early subduction process was accompanied by garnet and ilmenite exsolution in pyroxene, which led to the pyroxene becoming enriched in Ca and Mg. Early and late-stage pyroxene have very different Al₂O₃ contents, due to the transformation of clinopyroxene to garnet.

The protolith of garnet clinopyroxenite was carried into the subduction channel by mantle wedge flow, and experienced UHP metamorphism at age of 215.0 ± 5.7 Ma and 4.5 ± 0.5 GPa/800 ± 50 °C. The retrograde minerals were derived from a variety of HP mineral assemblages, which was formed by the decomposition of garnet, which was accompanied by the formation of abundant ilmenite and Ca-rich pyroxene. The cooling and decompression of subducted garnet clinopyroxenites associated with the retrogressive metamorphism at the age of 206 Ma and 1.0 GPa and 710 ± 30 °C.

The enrichment of light REEs and LILEs and depletion of heavy REEs and HFSEs support the hypothesis that the protolith garnet clinopyroxenites were derived from the mantle wedge that had been contaminated by fluids from the subducting crust. The mantle wedge is becoming an important source of fluid and gold for the orogenic gold deposits of the giant Jiaodong gold province in the North China Craton [76].