Introduction

UHT metamorphism in pelitic rocks is identified by diagnostic mineral assemblages, such as sapphirine plus quartz and orthopyroxene plus sillimanite (e.g. Harley 2008; Kelsey 2008). The spinel plus quartz assemblage is widely reported in UHT metapelites worldwide, but is commonly regarded to be non-diagnostic for UHT metamorphism (e.g. Harley 2008), because its stability range can be influenced by the incorporation of elements such as Zn and Cr (Nichols et al. 1992; Guiraud et al. 1996). Nevertheless, spinel plus quartz assemblage is a possible UHT advocate when spinel has exsolved lamellae of ferrotitanium oxides or directly contacts with them, because in these cases, spinel can be miscible with magnetite, ulvöspinel and so on (e.g. Kelsey 2008; Clarke et al. 1989; Waters 1991; Sengupta et al. 1999). For example, spinel with exsolved titanomagnetite in granulites from the Eastern Ghats was reported to record metamorphic temperatures of 1060–1140 °C (Das et al. 2017). On this account, spinel compositions and exsolution behaviour may be useful for constraining the temperatures of UHT metamorphism, but this potential is insufficiently investigated. Moreover, the enlargement of spinel stability field caused by elements, such as Zn and Cr, may be insignificant if their contents are very low. The factors that primarily influence the stability of spinel also need to be addressed.

UHT granulites in the Jining complex, Khondalite Belt, North China Craton (NCC) have attracted much attention in the last decade. Most of these UHT granulites are characterized by spinel-bearing (sapphirine-absent) assemblages such as that in the Xuwujia area. The peak temperatures of the spinel-bearing granulite therein were estimated as ~1120 °C by the occurrence of ferro-pseudobrookite (Wang et al. 2019) at odds with ~1020 °C determined by two-feldspar thermometry (Jiao and Guo 2011) and above 960 °C by phase equilibria modelling (Li and Wei 2018; Huang et al. 2019). As peak temperature is crucial for evaluating the heat budget and understanding the geodynamic nature of UHT metamorphism, more data are required to resolve the discrepant estimates of peak temperature. On the other hand, an almost uniform mean age of ~1.92 Ga for the UHT metamorphism in the Jining complex was widely reported based on zircon U–Pb dating (Santosh et al. 2007a; Li and Wei 2018; Lobjoie et al. 2018; Li et al. 2019), but the mean age was obtained from wide and continuous age spans on the concordia curve and its geological meaning is ambiguous. Seeing that UHT metamorphism with post-peak slow cooling (generally >30 Myr) or fast cooling (mostly <10 Myr) was interpreted to have formed in distinct tectonic regimes (e.g. Kelsey and Hand 2015; Harley 2016), Wang et al. (2020) constrained a duration of ~40 Myr for supra-solidus cooling stage of the UHT metamorphism. As this outcome is based on dating of a single sapphirine-bearing sample, its universality and internal consistency still need an assessment.

The present study will focus on three targets. They are: (i) to retrieve the metamorphic evolution especially the peak temperatures of the spinel-bearing (sapphirine-absent) granulite in the Xuwujia area using phase equilibria modelling, (ii) to address the stability and compositional variation of spinel during UHT metamorphism, and (iii) to assess or authenticate the timescale of the UHT metamorphism in the Jining complex using Monte Carlo simulation.

Geologic setting

The Khondalite Belt in the NCC was regarded to result from the collision between the southern Ordos Block and the northern Yinshan Block at ~1.95 Ga (Fig. 1a, b; e.g. Zhao et al. 2005, 2012). It exposes in the Helanshan–Qianlishan, Daqingshan–Ulashan and Jining areas from west to east (Fig. 1b). Recent studies suggest that the Khondalite Belt may extend southward as the basement beneath the Ordos Block (Wang et al. 2014, 2017) and eastward to the Huaian complex (Zhai and Santosh 2011; Zhou et al. 2017; Liao and Wei 2019).

Fig. 1
figure 1

Geologic sketch maps. a Tectonic sketch of the North China Craton (after Zhao et al. 2005). b Geologic sketch showing the distribution of the Khondalite Belt and its sub-units (after Zhao et al. 2005). c Geologic map of the Jining complex (modified after Guo et al. 1999), showing the lithologic distribution and the localities of UHT granulites

Full size image

The Jining complex is the eastern segment of the Khondalite Belt, where the lithology is dominated by pelitic granulite/gneiss, peraluminous granite and minor meta-gabbronorite (Fig. 1b, c). A growing body of pelitic granulites were reported to record UHT conditions, most of them containing spinel and just a few containing sapphirine (e.g. Liu and Li 2007; Santosh et al. 2007b; Jiao and Guo 2011; Zhang et al. 2012; Liu et al. 2012; Shimizu et al. 2013; Yang et al. 2014; Li and Wei 2016, 2018; Lobjoie et al. 2018; Li et al. 2019). Metamorphic zircon in the UHT granulite yields mean ages of 1.91–1.92 Ga (Santosh et al. 2007a; Li and Wei 2018; Lobjoie et al. 2018; Li et al. 2019) and ~ 1.88 Ga (Yang et al. 2014). The P–T path for the UHT metamorphism was initially inferred to be anticlockwise with a pressure increase at the pre-Tmax stage (Santosh et al. 2009; Liu et al. 2011; Shimizu et al. 2013), while was corrected to be clockwise with heating and decompression at the pre-Tmax stage (Yang et al. 2014; Li and Wei 2016, 2018; Lobjoie et al. 2018; Li et al. 2019). The peraluminous granites commonly contain voluminous garnet, and display crystallization ages of 1.94–1.90 Ga (Zhong et al. 2007; Wang et al. 2018; Huang et al. 2019). They are regarded to relate with extensive anatexis of the metapelites (Shi et al. 2018; Wang et al. 2018). The meta-gabbronorite crystallized at 1.96–1.92 Ga (~1.93 Ga), and metamorphosed at ~1.86 Ga (Peng et al. 2010).

Samples and methods

Sample collection and preparation

The samples in this study were collected from the Xuwujia outcrop (Fig. 1c). The rocks are migmatites with schlieren, comprising felsic leucosomes and mafite-rich mesosomes (Fig. 2a), which reflects extensive melting during high-grade metamorphism (also see Jiao and Guo 2011; Li and Wei 2018). A representative spinel-bearing (sapphirine-absent) pelitic granulite sample 15XWJ02 from the mesosomes was selected for petrologic analyses. The sample was trimmed away from stale portions and surface contaminants to prepare thin-sections in thicknesses of 0.02–0.03 mm. The thin-sections were carbon coated for electric conduction in electron probe micro-analyser (EPMA) work.

Fig. 2
figure 2

a Photograph showing the field occurrence of UHT granulite at the Xuwujia area (Li and Wei 2018). b Photograph showing the texture of the spinel-bearing granulite sample 15XWJ02

Full size image

Mineral analyses

Analyses for the chemical compositions of minerals were performed by means of a JEOL-8230 EPMA. The system was operated in wavelength-dispersive X-ray spectrometry mode, with an acceleration voltage of 15 kV and a beam current of 10 nA. For all elements, the Kα line has been utilized. The beam was focused to a diameter of 1–2 μm (10 μm for densely and tinily exsolved spinel mantle), and counting times were 10–15 s. The following reference materials [from the “53 minerals standards for microanalysis” set provided by Structure Probe Inc. (SPI) Supplies, West Chester, PA] were used for calibration: sanidine for K, diopside for Ca and Mg, rutile for Ti, jadeite for Na, Al and Si, chromium oxide for Cr, rhodonite for Mn, hematite for Fe, and nickel silicide for Ni. Data were reduced using the PAP (Phi-Rho-Z) correction method (Pouchou and Pichoir 1985). More analytical details were described elsewhere (Li et al. 2018). The image analyses for exsolved lamellae were performed using ImageJ software, to reintegrate pre-exsolved homogeneous mineral compositions. The area proportions were measured by particles summation. The mass densities used to calculate weight fractions of exsolved lamellae were 5.20 g/cm3 for magnetite, 3.55 g/cm3 for spinel, 2.67 g/cm3 for plagioclase and 2.57 g/cm3 for K-feldspar. Mineral formulae and site occupancies were determined based on assumed fixed oxygen amounts per formula unit, with trivalent iron calculated by stoichiometric charge balance.

Calculations

The model system NCKFMASHTO (Na2O–CaO–K2O–FeO–MgO–Al2O3–SiO2–H2O–TiO2–Fe2O3) was chosen for phase equilibria modelling. The calculations were performed using THERMOCALC 3.40 (Powell and Holland 1988) with dataset ds62 (Holland and Powell 2011). Mineral activity–composition relationships are the same as those in White et al. (2014) and references therein, with simplified stoichiometric formulae shown in Table 1. The bulk-rock composition was taken from Wang et al. (2019) and normalized in the model system. The H2O content was adjusted to ensure the final-phase assemblage occurring just above the solidus (e.g. Korhonen et al. 2011, 2012), where hydrous mineral (biotite) is in the observed mode (around 3–5 wt%). The Fe3+/Fe value was properly adjusted using T–XFe3+ diagram (see further below), ensuring that ilmenite occurs as the predominant phase of Ti − Fe oxide, and the reintegrated Fe3+ content in spinel can be reproduced further along the path.

Table 1 Simplified stoichiometric formulae of mineral models used for phase equilibria modelling
Full size table

As a large quantity of zircon age data for UHT granulites in the Jining complex have been accumulated, the Monte Carlo simulation (a probability simulation) was performed to constrain the duration of the UHT metamorphism, because this method can avoid the statistical bias resulted from taking all dates into account and can well define an accurate and credible timescale of geologic events (e.g. Svensen et al. 2012). We collected 260 zircon 207Pb/206Pb dates for the UHT granulite/gneiss (Santosh et al. 2007a; Yang et al. 2014; Li and Wei 2016, 2018; Lobjoie et al. 2018; Li et al. 2019). Among them, 86 dates were rejected through 12 times of iterative consistency check, which include those with abnormally large deviations and those recognized to be outliers. We generated 50,000 simulations of the duration. In each simulation, two random ages from the rest 174 data are assigned to calculate an age difference.

Results

Petrography and mineral chemistry

The spinel-bearing UHT granulite consists of quartz (Qz), perthite, plagioclase (Pl), garnet (Grt), sillimanite (Sil), spinel (Spl), orthopyroxene (Opx), biotite (Bt), cordierite (Crd) and Ti − Fe oxides (abbreviations of mineral names after Whitney and Evans 2010). Garnet and spinel with subordinate biotite and cordierite are randomly distributed in the felsic matrix (Fig. 2b). Representative results of chemical analyses of minerals are presented in Table 2.

Table 2 Chemical compositions (EPMA results) and calculated formulae for minerals in sample 15XWJ02
Full size table

Garnet occurs in variable size of 0.2–1.5 mm across. It is surrounded by a symplectite of orthopyroxene plus cordierite, and has spinel inclusions limited to its rim domains (Fig. 3a). It also includes exsolved rutile needles oriented in three directions (Fig. 3a). Garnet contains Xalm of 0.54–0.57, Xprp of 0.39–0.42, Xsps of ~0.01 and Xgr of 0.03–0.04 (where Xalm = Fe2+/(Fe2++Mg + Ca + Mn), Xprp = Mg/(Fe2++Mg + Ca + Mn), Xsps = Mn/(Fe2++Mg + Ca + Mn) and Xgr = Ca/(Fe2++Mg + Ca + Mn)) in the cores, while Xalm of 0.57–0.59, Xprp of 0.37–0.38 and very similar Xsps and Xgr in the rims. It shows conspicuous core-to-rim zoning of increasing Xgr and Xalm but decreasing Xprp (Fig. 4a, b). Xsps shows an increase in the rim.

Fig. 3
figure 3

Photomicrographs and BSE images of minerals and textures in sample 15XWJ02. a Garnet contains spinel inclusions in its rim domain and is surrounded by the symplectite of orthopyroxene plus cordierite. b Sillimanite overgrows around the intergrown spinel and ilmenite, and develops cordierite coronae locally. c Perthite contains inclusions of spinel and plagioclase, and biotite flakes are aggregated around perthite. d Symplectite of spinel plus cordierite replaces sillimanite. e Some coarse spinel grains display zoning texture. f A perthite grain shows dense exsolusion lamellae in the core and sparse exsolusion spots in the rim. g Orthopyroxene overgrows around biotite flakes. h Ilmenite contains mottles of rutile, magnetite overgrows around ilmenite, and titanomagnetite parallelly exsolves in ilmenite. i Sillimanite has inclusions of spinel, rutile and ilmenite. Abbreviations of mineral names after Whitney and Evans (2010)

Full size image
Fig. 4
figure 4

Mineral compositional diagrams. a Zoning profiles of Xalm, Xprp, Xgr and Xsps for garnet. b Xgr–XMg diagram showing the compositional variation from core to rim in garnet. c XMg–XAl diagram showing the compositional variation and zoning characteristic of spinel. d Box plots of Cr2O3 and ZnO contents for spinel (Data of ZnO are from Li and Wei (2018), Li (2019) and Wang et al. (2019)). e Xan–Xor diagram for measured perthite, plagioclase and reintegrated K-feldspar, with isotherms at 0.85 GPa based on the models of Elkins and Grove (1990). Xalm = Fe2+/(Fe2++Mg + Ca + Mn); Xprp = Mg/(Fe2++Mg + Ca + Mn); Xsps = Mn/(Fe2++Mg + Ca + Mn); Xgr = Ca/(Mg + Fe2++Ca + Mn); XMg = Mg/(Mg + Fe2+); XAl = Al/(Al + Fe3++Cr); Xan = Ca/(Ca + Na + K); Xor = K/(Ca + Na + K)

Full size image

Spinel generally coexists with Ti − Fe oxides, and occurs as inclusions in garnet, sillimanite and feldspars (Fig. 3a−c), as symplectite with cordierite surrounding sillimanite (Fig. 3b, d), or as single grains in the matrix. The different types of spinel show similar XAl [= Al/(Al + Fe3++Cr)) of 0.93–0.96 but variable XMg (= Mg/(Mg + Fe2+)] of 0.35–0.39 in the inclusions, 0.39–0.47 in the symplectites and 0.37–0.40 in the matrix grains (Fig. 4c). Occasionally, some coarse grains in the matrix are zoned, with wide trellised exsolution lamellae of magnetite in the core, very tiny magnetite lamellae in the mantle and lamellae-free in the rim (Figs. 3e and 5). Image analyses suggest that magnetite lamellae occupy 6.9–7.8 vol% of the core domains, allowing reintegrated compositions with XAl of 0.87–0.88 and XMg of 0.36–0.38 (Fig. 4c), distinct from the measured host compositions with XAl of 0.93–0.95 and XMg of 0.39–0.41. The measured compositions show XAl of 0.90–0.94 and XMg of 0.37–0.39 in the mantle, and XAl of ~0.94 and XMg of 0.38–0.41 in the rim. That is, these zoned spinel grains are characteristic of increasing XAl from core to rim (Fig. 4c). Moreover, spinel contains Cr2O3 of 0.27–0.70 wt% and ZnO of 0.82–2.14 wt% (Fig. 4d), with XCr [= Cr/(Al + Fe3++Cr)] below ~0.01 and XZn [= Zn/(Mg + Fe2++Zn)] of 0.02–0.05.

Fig. 5
figure 5

Zoning texture of the spinel. Photomicrograph (a) and BSE image (b) showing the wider trellised exsolution lamellae of magnetite in the core, very tiny magnetite lamellae in the mantle and lamellae-free rim

Full size image

Sillimanite occurs as overgrowths commonly surrounding spinel and ilmenite (Fig. 3b). Sillimanite itself is surrounded by symplectite of spinel plus cordierite (Fig. 3d).

Perthite occurs as anhedral grains of 0.5–1.5 mm across, and occasionally has inclusions of plagioclase and spinel (Fig. 3c). It shows dense exsolution lamellae in the core and sparse exsolution spots in the rim (Fig. 3f). In the core domain, the exsolution lamellae are 5–10 μm wide and occupy 6.1–17.6 vol% by image analyses. Therein, the host sanidine has Xor [= K/(Ca + Na + K)] of 0.64–0.80 and Xan [= Ca/(Ca + Na + K)] of 0.02–0.07, and the lamellae have Xan of 0.37–0.48 and Xor of 0.01–0.11, which allows reintegrating a supra-solvus K-feldspar (Kfs) with Xor of 0.56–0.72 and Xan of 0.06–0.11 (Fig. 4e). In the rim domain, the host sanidine has Xor of 0.83–0.89 and Xan below 0.01, while the exsolved spots are too small to be analyzed.

Plagioclase occurs as anhedral crystals with variable sizes of 0.1–1 mm. It occasionally displays as inclusions in perthite or contains inclusions of spinel (Fig. 3c). Matrix plagioclase has Xan of 0.38–0.39, but the grains enclosed in perthite have slightly higher Xan of 0.40–0.42 (Fig. 4e).

Biotite occurs as aggregated flakes with irregular and cuspate boundaries in the matrix (Fig. 3a−c, g), commonly surrounding garnet or perthite and occasionally surrounded by orthopyroxene. It shows noticeable TiO2 contents of 4.55–5.64 wt%, and XMg of 0.65–0.69.

Cordierite occurs as coronae or symplectite with orthopyroxene (around garnet) or spinel (around sillimanite; Fig. 3a–d). It has a uniform composition with XMg of 0.86–0.88.

Orthopyroxene occurs as symplectite with cordierite around garnet (Fig. 3a) or overgrows around biotite (Fig. 3g). It contains Al2O3 of 3.41–5.48 wt% and XMg of 0.63–0.66.

Ti − Fe oxides are mainly ilmenite (Ilm) with small amounts of rutile (Rt), magnetite (Mag) and titanomagnetite (Ti-Mag). Ilmenite contains exsolved lamellae of titanomagnetite and mottles of rutile along the titanomagnetite boundaries (Fig. 3h), which may imply the formation of tiny rutile during the redox exsolution of titanomagnetite through open system precipitation (Wang et al. 2019; Proyer et al. 2013). Magnetite occurs as exsolutions in spinel (Figs. 3e and 5) and as overgrown rims around ilmenite (Fig. 3h), indicating its later formation. Titanomagnetite occurs as 5–10 μm wide parallel exsolution lamellae in ilmenite (Fig. 3h). Rutile occurs as inclusions in sillimanite (Fig. 3i) and as mottles in ilmenite.

Based on the textural relationships described above, the peak assemblage is inferred to include silicate melt (Liq) and spinel likely together with garnet, and the post-peak evolution may involve the sequential appearance of plagioclase, perthite, biotite and the symplectitic or coronary assemblages of cordierite, orthopyroxene and spinel (Fig. 3a−d, g). Sillimanite should be absent in the peak assemblage and formed before biotite, probably growing together with the feldspars (Fig. 3b, c).

Phase equilibria

Pressure–temperature evolution

A P–T pseudosection for sample 15XWJ02 was calculated over 0.4–1.2 GPa and 750–1200 °C (Fig. 6a). It is contoured with the Xan(Kfs), Xan(Pl), Xgr(Grt) and XAl(Spl) isopleths for the relevant assemblages. The inferred peak assemblage including spinel and garnet is modelled to be stable at temperatures over ~1125 °C at 0.8–0.9 GPa, covering the rock’s dry solidus (~1125 °C at 0.8–0.9 GPa; Fig. 6a). The observed post-peak sequential appearance of plagioclase, perthite/sillimanite and biotite is well consistent with the cooling evolution at pressures of 0.8–0.9 GPa. It includes (i) the formation of plagioclase plus quartz at ~1125 °C followed by K-feldspar at ~1110 °C, as a result of the crystallization of unsegregated melts (Fig. 6b); (ii) the growth of sillimanite and garnet at the expense of spinel and quartz (Spl + Qz = Grt + Sil; R1 in Fig. 6b) from ~1110 °C to ~1085 °C, accompanied with further melt crystallization; and (iii) the growth of biotite at the expense of melt, garnet and K-feldspar (Grt + Kfs + Liq = Bt + Sil + Pl + Qz; R2 in Fig. 6b) from ~870 °C to the fluid-absent solidus. Along the cooling process, the substantial melt crystallization that consumes over 80% melts (60 to 10 mol%) occurs at 1030–1110 °C (C in Fig. 6b). The above metamorphic evolution can be well supported by the contours of mineral compositions. The reintegrated XAl = 0.87–0.88 in the core of the zoned spinel can yield temperatures of ~1150 °C in the peak assemblage (Grt − Spl − Liq) or 1090–1100 °C in the sillimanite-bearing assemblage (Grt − Sil − Kfs − Pl − Qz − Spl − Liq). The reintegrated perthite composition with Xan = 0.06–0.10 yields temperatures above 1075 °C, in accordance with the voluminous growth of K-feldspar (Fig. 6). The core-to-rim Xgr increasing (0.03 → 0.04) in garnet indicates a cooling trend with a wide temperature range at 900–1100 °C and a pressure range of 0.7–1.0 GPa. Moreover, the measured Xan in plagioclase of 0.40–0.42 from the inclusions in perthite and 0.38–0.39 from the matrix grains yield temperatures fairly consistent with the initial and later growth of plagioclase during cooling.

Fig. 6
figure 6

a P–T pseudosection with proposed Tmax conditions and P–T path for sample 15XWJ02. The compositional isopleths of spinel, garnet, K-feldspar and plagioclase are contoured in the relevant assemblages. The dry solidus was calculated in H2O-free subsystem. b T–Mode diagram showing the variation of mineral and melt modes, and the temperature ranges of melt crystallization (C) and crucial metamorphic reactions (R1 and R2) along the cooling path at 0.85 GPa. Mineral abbreviations are after Whitney and Evans (2010). Liq = silicate melt

Full size image

The symplectitic and coronary assemblages around garnet, sillimanite and biotite (Fig. 3a–g) indicate further evolution in sub-solidus conditions. These will be addressed below.

Spinel-involving equilibria

The bulk-rock composition of sample 15XWJ02 (Wang et al. 2019) is presented herein as Table 3. Normalized molar proportions of main constituents corresponding to Figs. 6–8 are presented in Table 4.

Table 3 Bulk-rock chemical composition of sample 15XWJ02*
Full size table
Table 4 Normalised proportions (mol%) of sample 15XWJ02
Full size table

The phase equilibria modelling (Fig. 6) suggests that spinel should have been consumed through the reaction R1, and thereby could not survive as matrix grains during cooling to the fluid-absent solidus. Besides, the pseudosection in Fig. 6a cannot accommodate the higher XAl of 0.90–0.95 and the wide variation of XMg in spinel (Fig. 4c). On these accounts, the T − M(SiO2), T − XFe3+ and T–XMg pseudosections at 0.85 GPa (Figs. 7 and 8) were calculated to investigate how these components influence the stability and composition of spinel. In the T − M(SiO2) pseudosection (Fig. 7), spinel can stabilize to much lower temperatures and even down to sub-solidus conditions in the quartz-absent assemblages, suggesting that spinel can be preserved in local domains isolated from quartz. The XAl values in spinel are predicted to roughly increase as temperature decreases above the solidus. The measured XAl of 0.90–0.94 in the mantle of the zoned spinel can appear during the cooling evolution in the SiO2-unsaturated fields (Fig. 7). Also suggested in Fig. 7, both the solidus and biotite growth can move to higher temperatures above 900 °C in SiO2-unsaturated situations.

Fig. 7
figure 7

T–M(SiO2) pseudosection calculated for sample 15XWJ02 at 0.85 GPa, showing the variation the phase relations, especially the stability of spinel, with the bulk-rock or local domain M(SiO2) values. Also shown are the compositional isopleths of spinel and the minimum temperature (solid circle) indicated by XAl in the core of the zoned spinel

Full size image
Fig. 8
figure 8

T–XFe3+ (a) and T–XMg (b) pseudosections calculated for sample 15XWJ02 at 0.85 GPa, showing the variation of the stability and composition (XAl and XMg) of spinel with the bulk-rock or local domain XFe3+ and XMg values. The dashed lines labeled with sample number represent the relevant values used in Fig. 6

Full size image

The T − XFe3+ pseudosection (Fig. 8a) suggests that the spinel stability is slightly extended with increasing XFe3+, and spinel is replaced by magnetite at XFe3+ > 0.32 fields. The XAl values of spinel decrease dominantly with increasing XFe3+ and slightly with ascending temperature in the higher-T sillimanite-absent fields, while the spinel XAl decreases dominantly with descending temperature and slightly with increasing XFe3+ in the lower-T sillimanite-present fields. The T − XMg pseudosection (Fig. 8b) shows that the spinel stability is distinctly extended to lower temperatures with decreasing bulk-rock XMg, and spinel is replaced by magnetite in Fe-rich rocks with XMg < 0.33. The XMg values of spinel increase dominantly with increasing bulk-rock XMg, almost temperature independent in the higher-T sillimanite-absent assemblages, while in the lower-T sillimanite-present assemblages, the spinel XMg increases mainly with increasing temperature.

Monte Carlo simulation results

The cleaned zircon 207Pb/206Pb dates (174 data) distribute from 1887 ± 18 to 1954 ± 23 Ma, with a mean age of 1919 ± 2 Ma [mean squared weighted deviation (MSWD) = 1.19; Fig. 9a]. The frequency histogram with density estimate shows that these ages are within one group (Fig. 9b), which indicates a prolonged metamorphic event. The frequency histogram and cumulative probability curve as functions of the duration that were calculated from the resulting age differences are shown in Fig. 9c, d. The simulation results indicate that the duration was ~40 Myr (95% confidence).

Fig. 9
figure 9

Statistic and simulated results of zircon 207Pb/206Pb dates for the UHT granulites in the Jining complex. a Weighted mean age. b Frequency distribution of the measured age values. c Frequency distribution of the calculated durations. d Cumulative probability of the durations. The data of metamorphic zircon are collected from Santosh et al. (2007a), Yang et al. (2014), Li and Wei (2016, 2018), Lobjoie et al. (2018) and Li et al. (2019)

Full size image

Discussion

Metamorphic evolution

Petrographic characteristics and phase equilibria modelling of the spinel-bearing pelitic granulite sample from the Xuwujia area suggest a metamorphic evolution dominated by cooling from a Tmax stage with extremely high temperatures of >1125 °C (at 0.8–0.9 GPa). This extreme condition is consistent with the occurrence of ferro-pseudobrookite, which indicates temperatures above 1120 °C as previously reported by Wang et al. (2019). The peak temperature is above the dry solidus of the rock, which may have resulted in over 60 mol% melts with only spinel and garnet survived (Fig. 6) even in the mesosomes of the migmatite.

The post-Tmax cooling was inferred to proceed at pressures of 0.8–0.9 GPa from (i) the isopleths of the measured Xan(Pl) and Xgr(Grt) values, and (ii) the stability of the biotite-bearing and rutile-absent final assemblage in Fig. 6a. This pressure condition is consistent with the results from previous studies in the Jining complex (e.g. Jiao and Guo 2011; Li and Wei 2018; Li et al. 2019). The cooling process from the peak temperature to the fluid-absent solidus involves three crucial retrograde metamorphic reactions (Fig. 6b), which are Liq = Kfs + Pl + Qz (C at 1030–1125 °C), Spl + Qz = Grt + Sil (R1 at 1085–1110 °C) and Grt + Kfs + Liq = Bt + Pl + Qz (R2 at 860–870 °C), corresponding to the successive appearance of plagioclase, K-feldspar, sillimanite and biotite. The temperature of R1 can well match that of the experimentally constrained univariant reaction (Fig. 10; Hensen 1986; Kelsey et al. 2004). The cooling evolution is also supported by (i) the exsolution of magnetite from spinel at 1000–1050 °C (Figs. 5; 10; Taylor-Jones and Powell 2010) and the formation of perthite from supra-solvus Ca-rich K-feldspar (Fig. 4f); (ii) the core-to-rim increasing Xgr in garnet (0.03 → 0.04; Figs. 4b and 6a) and increasing XAl in spinel (0.87–0.88 → 0.94–0.95; Figs. 4c and 7); and (iii) the rutile needles exsolved from garnet (Fig. 3a) which indicates the expulsion of titanium from garnet through open system precipitation (e.g. Proyer et al. 2013).

The post-Tmax cooling evolution may terminate at the fluid-absent solidus and is followed by a sub-solidus decompression stage, indicated by the development of symplectitic or coronary Crd − Opx − Spl assemblages around garnet, sillimanite or biotite (Fig. 3a–g), as well as by the overgrowth of magnetite around ilmenite (Figs. 3h and 6a). The formation of the symplectitic or coronary assemblages can be attributed to the reactions governed by local domains, such as Grt + Sil = Crd + Spl and Grt + Bt = Crd + Opx, which may occur at 0.4–0.6 GPa (Fig. 10).

Fig. 10
figure 10

P–T diagram showing the P–T–t evolution of the spinel-bearing UHT granulite at the Xuwujia area. Also shown are: (i) experimentally constrained crucial metamorphic reactions (solid curves; Hensen 1986; Kelsey et al. 2004; Stevens et al. 1997); (ii) the solvus of the spinel–magnetite suggested by the compatibility diagrams in Taylor-Jones and Powell (2010); and (iii) calculated dry solidus, fluid-absent solidus, K-feldspar-in (Kfs-in) line and the sliding reaction R1 in Fig. 6. The similar P–T paths with lower peak temperatures for UHT granulites at the nearby outcrops, involving Zhaojiayao (ZJY; Li and Wei 2016), Hongshaba (HSB; Li et al. 2019) and Huangtuyao (HTY; Liao and Wei 2019), are presented for comparison. The granite wet solidus was cited from Johannes and Holtz (1996). The P–T ranges for the metamorphic facies of normal granulite (NG), ultrahigh-temperature granulite (UHTG) and high-pressure granulite (HPG) were cited from Brown (2007)

Full size image

In addition, a pre-Tmax heating decompression stage can be inferred from the spinel inclusions in the rim domain of garnet (Fig. 3a; Li et al. 2019) and the survival rutile inclusions (Figs. 3i and 6a). Such a pre-Tmax evolution was also recovered from the UHT granulites at other outcrops in the Jining complex, such as Tuguiwula and Hongshaba (Li and Wei 2018; Li et al. 2019).

As a result, a clockwise P–T path can be recovered for the spinel-bearing granulite at the Xuwujia area. It involves the pre-Tmax heating decompression to the Tmax stage, the post-Tmax cooling to fluid-absent solidus and the sub-solidus decompression (Fig. 10).

Timescale of the UHT metamorphism

Most of the previous zircon U–Pb dating data for UHT granulites from the Jining complex show wide and continuous age distributions on the concordia curve, with weighted mean ages of ~1.92 Ga (Santosh et al. 2007a; Li and Wei 2018; Lobjoie et al. 2018; Li et al. 2019). Nevertheless, the geological meaning of the mean value from a very wide age span is ambiguous, and the wide age span is likely ascribed to slow cooling of UHT granulites (e.g. Li and Wei 2016; Jiao et al. 2020; Laurent et al. 2018; Clark et al. 2018). Recently, Huang et al. (2019) reported that some zircon grains yield mean ages of ~1.94 Ga in the cores and ~ 1.92 Ga in the rims, and interpreted that the ~1.94 Ga age represents the prograde anatexis predated UHT metamorphism and the ~1.92 Ga age represents later UHT metamorphism. However, Wang et al. (2020) performed texture constrained in-situ dating and revealed that zircon included in biotite yields an age of ~1.90 Ga and zircon included in sillimanite and perthite yields ages of ~1.94 Ga. Meanwhile, phase equilibria modelling shows that sillimanite and perthite have grown at 1010–1120 °C and biotite formed at ~900 °C both during cooling, so it suggests a supra-solidus cooling stage at 1.94–1.90 Ga (Wang et al. 2020). In this study, Monte Carlo simulative results suggest a duration of ~40 Myr (95% confidence; Fig. 9d). Considering that zirconium behaves as incompatible element and crystallization of melts and consequent growth of zircon generally occurs during post-peak cooling (e.g. Harley et al. 2007; Kelsey et al. 2008), we interpret this duration to represent a prolonged supra-solidus cooling stage during 1.94–1.90 Ga in the Jining complex. This timing is more meaningful for a regional metamorphic event and is accordant with that (i) the UHT metamorphism took place after the ~1.95 Ga high-pressure granulite facies metamorphism (e.g. Li and Wei 2018); (ii) the garnet-bearing granites were crystallized at ~1.92 Ga as the products of the UHT metamorphism (Wang et al. 2018; Shi et al. 2018); and (iii) the gabbronorites emplacement at 1.96–1.92 Ga (Peng et al. 2010) were involved in the genesis of the UHT metamorphism (Guo et al. 2012; Li and Wei 2018; Wang et al. 2019; Huang et al. 2019).

Spinel behaviour

Like that in the studied sample, a considerable amount of spinel in the granulite can survive during cooling (Fig. 2b), although it is predicted to be consumed (Fig. 6). This is commonly interpreted by the involvement of ZnO in spinel (e.g. Yang et al. 2014; Li et al. 2019), which can enlarge the stability of spinel to lower temperatures (Sengupta et al. 1991; Nichols et al. 1992; Hand et al. 2007; Tajcmanová et al. 2009). Experimental calibration in Nichols et al. (1992) shows that an incremental XZn of 0.2 can contribute about −150 °C to the lower limit of spinel stability range. However, in the case of this study, the low XZn in spinel of 0.02–0.05 (Fig. 4d) is not enough to explain its preservation in the rock. Phase equilibria modelling suggests that spinel stability can be markedly enlarged in the local domains that are lower in SiO2 or XMg (Figs. 7; 8b) due to the heterogeneous distribution of quartz or mafites in the rock. This is accordant with the observations that spinel commonly survives in local quartz-free domains (Tajcmanová et al. 2009) or is texturally isolated from quartz (Kelsey 2008).

The two key compositional variables (XMg and XAl) of spinel are governed by both metamorphic temperatures and local domain compositions. For instance, the measured XMg of 0.35–0.41 in spinel as inclusions and matrix grains (Fig. 4c) may represent cooling temperatures below ~1090 °C at the local compositions with XMg around 0.4–0.5 (Fig. 8b), which is somewhat lower than the measured bulk-rock XMg of 0.52. This is consistent with that these spinel grains closely coexist with ilmenite (Fig. 3b, e). Besides, the measured higher XMg of 0.39–0.47 in the symplectitic spinel intergrown with cordierite should be primarily ascribed to the local Mg-rich domains, although these equilibria have not been presented in Fig. 8b. Similarly, the core-to-rim increasing XAl and variation trend of XMg in the zoned spinel may indicate a temperature decrease at local domains isolated from quartz (Figs. 4c and 7), consistent with the recovered post-Tmax cooling evolution of the rock. Therefore, the reintegrated XAl of 0.87–0.88 from the spinel core with exsolved magnetite lamellae may represent temperatures over ~1100 °C (Fig. 7), seeing that spinel grains do not coexist with any biotite. By the way, it should be noted that the compositional zoning of spinel (Fig. 4c) corresponds to the spectacular textural zoning in form of exsolution (Fig. 5), which benefits to the composition restriction because the degree of openness for cations diffusion depends on the boundary conditions (e.g. Proyer et al. 2013).

The frequently observed coexistence of spinel and ilmenite (Fig. 3b–e) suggests that spinel has survived together with ilmenite as refractory residuum at peak temperatures. Calculation shows that the spinel can contain ulvöspinel (Fe2TiO4) up to ~3 mol% at the peak stage. However, the measured and even the reintegrated spinel composition did not contain any perceptible Ti and meanwhile, no ulvöspinel-rich phase (e.g. titanomagnetite) was found inside or around spinel. Therefore, the Ti-component should be preferentially expulsed from spinel, in the form of ilmenite, through open system precipitation (e.g. Proyer et al. 2013) during the initial cooling. This is consistent with that the ilmenite mode slightly increases at 1090–1110 °C (Fig. 6b).

Furthermore, as suggested by the orthogonal compatibility diagrams in Taylor-Jones and Powell (2010) and Wheller and Powell (2014), the solvus of the miscible spinel–magnetite phase locates between 1000 and 1050 °C and behaves almost pressure independent. Hence, spinel with exsolved lamellae of magnetite can be an indicator of UHT conditions in metapelites, likely for temperatures above 1000 °C.

Tectonic implications

The spinel-bearing UHT granulite at the Xuwujia area was revealed to evolve along a clockwise P–T path including the pre-Tmax heating decompression, post-Tmax cooling to the fluid-absent solidus, and sub-solidus decompression stages (Fig. 10). The pre-Tmax heating decompression may suggest an exhumation of the high-pressure granulite terrane, which was proposed to relate with an orogenic crustal thickening event at 1.98–1.96 Ga (Zhou et al. 2010; Li et al. 2011; Yin et al. 2011). The recovered Tmax temperature is extremely high to ~1125 °C, located above the dry solidus of the investigated granulite (Fig. 10), which caused extensive crustal anatexis and generated the schlieren migmatite (Fig. 2a) and garnet-bearing granite in the field. The post-Tmax cooling evolution occurred at a middle–lower crustal level (0.8–0.9 GPa) and lasted for ~40 Myr (1.94–1.90 Ga; Figs. 9 and 10).

The tectonic setting for the UHT metamorphism in the Jining complex was controversially proposed to be ridge subduction (e.g. Peng et al. 2010; Guo et al. 2012; Santosh et al. 2012) and post-collision mantle upwelling together with mafic magma emplacement (e.g. Zhao 2009; Li and Wei 2018). The ridge subduction scenario may not be favored on considerations that (i) it may involve the generation of paired metamorphic belts of high- and low-P/T types and adakitic rocks from oceanic slab melting (e.g. Iwamori 2000; Santosh and Kusky 2010), which have not yet been discovered in the region; and (ii) a ridge subduction triggered UHT metamorphism may occur at pressures over 2 GPa (Syracuse et al. 2010) and cool down within ~2 Myr (Iwamori 2000), which are distinct from the characters of the UHT metamorphism in the Jining complex. The model of post-collision mantle upwelling together with mafic magma emplacement seems also inappropriate because the UHT metamorphism in this scenario usually lasted somewhat shorter (mostly within 30 Myr; Harley 2016; Kelsey and Hand 2015) and has limited temperatures below ~1000 °C (Harley 2004). For instance, the HT–UHT metamorphic overprinting related to mantle upwelling resulted from delamination in the Bohemian massif shows limited temperatures (~900 °C) and short-lived duration (e.g. Faryad and Cuthbert 2020). The prolonged UHT metamorphism, such as that in the Xuwujia area, is generally regarded to be formed in the hot underbelly of an orogen with external heat source (e.g. Harley 2016). A thermal modelling suggests that a plume-induced hot mantle upwelling can result in efficient lithosphere thermal erosion and can conductively heat the overlying crust, which, however, cannot generate the extreme UHT metamorphism (>1100 °C) without the advective heating by massive magma penetrating the granulite terranes (Wang et al. 2019). This scenario is appropriate for the Xuwujia area, where the gabbronorites with crystallization ages of 1.96–1.92 Ga show a hyperthermal primary intrusive temperature of 1400 °C and an anomalously high mantle potential temperature of 1550 °C (Peng et al. 2010). Therefore, a post-orogenic mantle plume activity accompanied with the intrusion of mantle-derived magma is more possible to account for the UHT metamorphism in the Jining complex, where the extremely high temperatures (>1100 °C) may be reached with sufficient advective heating from hyperthermal intrusions at the Xuwujia and Tianpishan areas, while less heating may depress the temperatures at, for instance, the Zhaojiayao and Hongshaba areas (Figs. 1c and 10). In addition, the 1.94–1.90 Ga UHT metamorphism in the Jining complex, North China Craton is synchronous with the markedly global Orosirian (2.0–1.88 Ga) UHT metamorphic events (Brown 2007), such as those in the Lewisian complex, Scotland (e.g. Baba 2003; Hollis et al. 2006) and the Taltson zone, northwestern Canada (Farquhar et al. 1996).