Petrogenesis of Early Cretaceous adakites from the Liaodong Peninsula: insight into the lithospheric thinning of the North China Craton

Abstract

Lithospheric thinning occurred in the North China Craton (NCC) that resulted in extensive Mesozoic magmatism, which has provided the opportunity to explore the mechanism of the destruction of the NCC. In this study, new zircon U–Pb ages, geochemical and Lu–Hf isotopic data are presented for Early Cretaceous adakitic rocks in the Liaodong Peninsula, with the aim of establishing their origin as well as the thinning mechanism of the NCC. The zircon U–Pb data show that crystallization occurred during 127–120 Ma (i.e. Early Cretaceous). These rocks are characterized by high Sr (294–711 ppm) content and Sr/Y ratio (38.5–108), low Yb (0.54–1.24 ppm) and Y (4.9–16.4 ppm) contents, and with no obvious Eu anomalies, implying that they are adakitic rocks. They are enriched in large-ion lithophile elements (e.g. Ba, K, Pb and Sr) and depleted in high-field-strength elements (e.g. Nb, Ta, P and Ti). These adakitic rocks have negative zircon ϵ _Hf(t) contents (−28.9 to −15.0) with two-stage Hf model ages (T _DM2) of 3004–2131 Ma. Based on the geochemical features, such as low TiO₂ and MgO contents, and high La/Yb and K₂O/Na₂O ratios, these adakites originated from the partial melting of thickened eclogitic lower crust. They were in an extensional setting associated with the slab rollback of the Palaeo-Pacific Ocean. In combination with previous studies, as a result of the rapid retracting of the Palaeo-Pacific Ocean during 130–120 Ma, the asthenosphere upwelled and modified the thickened lithospheric mantle, which lost its stability, resulting in the lithospheric delamination and thinning of the NCC.

1. Introduction

The North China Craton (NCC) is one of the oldest known cratons in China, and has a unique evolutionary history. Based on the research of mantle xenoliths from the Palaeozoic kimberlites and Cenozoic basalts, the NCC underwent reactivation and dramatic lithospheric thinning (> 100 km) (Griffin et al. 1998; Fan et al. 2000; Menzies et al. 2007; Zhu et al. 2011; Wu et al. 2019). Although some understanding of lithospheric thinning in the NCC has been obtained by petrology, geochemistry and geophysics, the initial timing and geodynamics mechanism are still controversial (Wu et al. 2005 a, 2008; Zhai et al. 2007). The initial timing of thinning and destruction of the NCC ranges from Late Triassic (Duan et al. 2014) to Late Jurassic (Jiang et al. 2010) and Early Cretaceous (Wu et al. 2008). Regarding the geodynamics mechanism, there exist delamination (Deng et al. 1994, 2007; Gao et al. 2002, 2004; Xu et al. 2006 a, b; Windley et al. 2010) and thermo-chemical erosion (Menzies et al. 1993; Xu et al. 2004; Zheng et al. 2006) models.

Adakites were first proposed by Defant & Drummond (1990) on the basis of their geochemical features (e.g. high Sr/Y and La/Yb ratios; low Y and Yb contents). Scholars proposed that adakites may have originated from either: partial melting of hot subducted oceanic crust (Defant & Drummond, 1990); partial melting of the thickened lower crust (Muir et al. 1995); or assimilation and fractional crystallization (AFC) processes associated with basaltic magma (Castillo et al. 1999), the reaction of delaminated lower crust with mantle peridotite (Gao et al. 2004) or the mixing of basaltic and felsic magmas (Guo et al. 2007). The various origins of adakites play a critical role in understanding the growth and evolution of the crust (Guo et al. 2006). Large-scale Mesozoic magmatism is one type of geological evidence of the lithospheric thinning and destruction of the NCC (Wu et al. 2005 a; Xu et al. 2009; Yang et al. 2009). It is noteworthy that a large number of adakitic rocks (c. 175–110 Ma) have been identified from the Mesozoic magmatic activity, distributed along the edge of the NCC (Fig. 1a; Zhang & Wang, 2001; Xiong et al. 2011). Because the age and distribution of adakites are consistent with the thinning range of the NCC, the age, source and origin of adakites can provide an important window into the processes and mechanisms of lithospheric thinning and destruction of the NCC.

Fig. 1. (a) Distribution of Mesozoic adakites (ages mainly 175–110 Ma) in the eastern NCC (after Zhang & Wang, 2001; Xiong et al. 2011); and (b) geological map and distribution of Mesozoic magmatic rocks in the Liaodong Peninsula (modified after Wu et al. 2005 b; Yang et al. 2007 a).

The Liaodong Peninsula is one of the most important parts of the NCC, which also underwent complex magmatic activity and dramatic lithospheric thinning (Fig. 1b; Zhu et al. 2011). Studies shown that a large number of Mesozoic magmatic rocks, such as A-type granite, mafic rocks, calc-alkaline I-type rocks and adakites, are distributed in the Liaodong Peninsula (Wu et al. 2008; Liu et al. 2011, 2013; Wang et al. 2015). These Mesozoic rocks indicate that it is an important place to study the thinning of the NCC. Previous studies on these rocks have obtained some periodic results (Yang et al. 2004, 2012; Wu et al. 2008; Duan et al. 2014). However, the geodynamics setting of Early Cretaceous magmas still remains controversial (Sun et al. 2007; Xiao et al. 2010; Zhu et al. 2011; Zhang, 2013). Recently, we discovered a set of Early Cretaceous magmatic rocks with adakitic features in Liaodong Peninsula (Fig. 2). In this paper, we use major- and trace-element compositions, zircon U–Pb ages and Hf isotopes to restrict their magma sources and origin. Our aim is to reveal the tectonic dynamic setting of the Early Cretaceous magmatic rocks in the Liaodong Peninsula, and provide new evidence for the mechanism of lithospheric thinning and destruction of the NCC.

Fig. 2. Geological sketch map of the Pulandian region with sample locations.

2. Geological setting and sample descriptions

The NCC is triangular in shape, with the Central Asian Orogenic belt, and the Qinling–Dabie and Sulu high–ultrahigh-pressure metamorphic belts, located to its north, south and east, respectively (Fig. 1a; Zhao et al. 2005; Liu et al. 2020). The NCC was formed as a result of the collision between eastern and western blocks along the central orogenic belt during the Palaeoproterozoic Era (c. 1.85 Ga; Zhao et al. 2005; Zhai & Santosh, 2011), after which the NCC remained relatively stable and formed sedimentary basins and thick sedimentary rocks. The NCC then began to activate during the Mesozoic Era, which resulted in the thinning and destruction of the lithosphere, and the formation of a large number of ore deposits, magmatic rocks and extensional structures (Wu et al. 2005 a; Zhu et al. 2012; Zhu & Xu, 2019).

Located in the eastern part of the NCC (Fig. 1a), the Liaodong Peninsula can be divided into three tectonic units: the Archean Liaobei block in the north, the Archean Liaonan block in the south and the Palaeoproterozoic Jiao-Liao-Ji orogenic belt (JLJOB) in the middle (Liu et al. 1992; Wu et al. 2005 b). The basement rocks mainly consist of Early Archean tonalite, trondhjemite and granodiorite (TTG) suites in the Liaobei block; Late Archean diorite, tonalite and granodiorite in the Liaonan block; and Palaeoproterozoic Liaohe Group in the JLJOB (Lu, 2004; Wu et al. 2005 b). These basement rocks were then covered by Mesoproterozoic–Palaeozoic sedimentary strata (Yang et al. 2007 b). The reactivation of the NCC resulted in the widespread distribution (c. 20 000 km²) of Mesozoic intrusive rocks in the Liaodong Peninsula (Fig. 1b; Yang et al. 2007 b). These Mesozoic magmatic rocks, consisting of syenite, monzogranite, granodiorite and diorite, were mainly deposited during three episodes: (1) Late Triassic (mainly 230–210 Ma); (2) Jurassic (mainly 180–155 Ma); and (3) Early Cretaceous (mainly 131–106 Ma) (Li, 2019). In addition, minor Mesozoic mafic dykes also occur in in the Liaodong Peninsula (Fig. 1b).

The Pulandian area covers an area of about 17 km² and is distributed in the western part of the Liaodong Peninsula (Fig. 1b). The Precambrian basement is mainly composed of Neoarchean gneissic complex (e.g. biotite plagioclase gneiss) and Mesoproterozoic quartz diorite (Fig. 2). The intrusive rocks are mainly composed of intermediate-acidic rocks, which include monzodiorite, granodiorite, porphyritic monzogranite and monzogranite. In this study, these intermediate-acidic rocks were collected in the Pulandian area for research (Fig. 2).

The monzodiorite (sample no. Yd2005) is grey-white in colour and contains plagioclase (50–55%), biotite (15–20%), K-feldspar (5–10%), quartz (5–10%) and minor hornblende (c. 5%) (Fig. 3a–c). Typical polysynthetic twinning can be seen in plagioclases (Fig. 3c). The granodiorite (sample nos Yd2006 and Yd2010) is medium-grained and comprises plagioclase (45–50%), K-feldspar (20–25%), quartz (15–20%) and minor hornblende and biotite (c. 5%) (Fig. 3d–i). Small amounts of fine-grained biotite and hornblende are found in the margins of large-grained plagioclase and quartz. Sample Yd2010 contains minor perthite (Fig. 3i). The porphyritic monzogranite (sample no. Yd2009) has porphyritic texture, consisting of phenocryst (c. 15%) and matrix (c. 85%) (Fig. 3j–l). The phenocryst mainly consists of quartz, K-feldspar, plagioclase and biotite (Fig. 3l). The matrix is composed of fine-grained plagioclase, quartz and minor biotite.

Fig. 3. Hand specimens and crossed-polarized light micrographs of rocks from Liaodong Peninsula from (a–c) monzodiorite (sample Yd2005); (d–f) granodiorite (sample Yd2006); (g–i) granodiorite (sample Yd2010); and (j–l) porphyritic monzogranite (sample Yd2009). Hbl – hornblende; Bt – biotite; Per – perthite; Pl – plagioclase; Qz – quartz.

3. Analytical methods

3.a. Zircon U–Pb dating

Zircons were separated from samples by magnetic and heavy-liquid separation methods at Langfang Hongxin Geological Exploration Technology Service Co. Ltd, Hebei Province, China. Zircons were observed and imaged to reveal their internal structure under cathode-luminescence (CL). Zircon laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) U–Pb dating was conducted at the Laboratory of Mineral Resources Evaluation in Northeast Asia, Ministry of Nature Resources of China, Changchun (LMRENA), using a 193 nm ArF laser ablation system and Agilent 7900 ICP-MS. The denudation frequency and spot diameter of laser were 7 Hz and 32 μm, respectively. Standard zircon 91500 was adopted as the external standard for age calibration. These 91500 zircons yield a ²⁰⁶Pb/²³⁸U age range 1060.9–1069.1 Ma, with weighted mean 1065.0 Ma, which is consistent with the recommended values of 1064 Ma for 91500 zircons within analytical errors. NIST610 was used as an external standard and ²⁹Si as an internal standard to normalize zircon trace-element contents. The isotope ratio was calculated using the LA-ICP-MS DATECAL program (Liu et al. 2008). Pb correction methods according to Andersen (2002) were followed. The calculated age and Concordia diagrams were calculated using the Isoplot3 program (Ludwig, 2003).

3.b. Major- and trace-element analysis

Whole-rock major- and trace-element geochemical analyses were undertaken at LMRENA. Altered surfaces were removed from all samples and were ground to 200 mesh. The X-ray fluorescence (XRF) and ICP-MS were used for major- and trace-element testing, respectively. The trace-element analytical samples were prepared by dissolving in mixed acid (HF + HNO₃) in Teflon bombs. The national standards GBW07103 and GBW07105 were adopted for element content correction. The analytical precision of major- and trace-element testing was 1% and 5%, respectively.

3.c. Zircon Hf isotopic analyses

Zircon in situ Lu–Hf isotope analyses were conducted using a Neptune-plus multicollector (MC) ICP-MS and NewWave UP213 laser ablation system at Yanduzhongshi Geological Analysis Laboratory, Beijing. The denudation frequency and spot diameter of the laser were 8 Hz and 50 μm, respectively. Standard zircons (e.g. GJ-1, Mud Tank, Penglai and 91500) were treated as precision control (Yuan et al. 2008; Li et al. 2010). The testing steps and calibration methods used are described by Wu et al. (2006) and Guo et al. (2012).

4. Analytical results

4.a. Zircon U–Pb ages

The zircon U–Pb results are provided in online Supplementary Table S1 (available at http://journals.cambridge.org/geo). Zircon grains are grey, subhedral to euhedral, 50–150 μm long and with aspect ratios of 1:1 to 2:1. Zircons from samples Yd2006, Yd2009 and Yd2010 all display typical oscillatory zoning in cathodoluminescence (CL) images, whereas no obvious oscillatory zoning was observed in zircons from sample Yd2005 (Fig. 4). Zircons from sample Yd2005 show variable Th (7–778 ppm) and U (51–303 ppm) contents, and Th/U ratios of 0.11–2.57. A total of 23 concordant analyses yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 123.3 ± 1.6 Ma (Fig. 5a). Eighteen zircons from sample Yd2006 have Th content of 67–346 ppm, U content of 48–266 ppm and Th/U ratios of 0.71–1.44, and yielded a ²⁰⁶Pb/²³⁸U weighted mean age of 125.5 ± 1.5 Ma (Fig. 5b). Sixteen zircons from sample Yd2009 display Th content of 68–690 ppm, U content of 115–410 ppm and Th/U ratios of 0.34–2.07, giving a mean age of 120.0 ± 1.9 Ma (Fig. 5c). Twenty concordant zircon analyses from sample Yd2010 yielded Th content of 166–760 ppm, U contents of 194–463 ppm and Th/U ratios of 0.72–1.34, with a mean age of 127.1 ± 1.2 Ma (Fig. 5d). Their ages range from 127–120 Ma, suggesting that they were formed during the Early Cretaceous Epoch. All zircons have high Th/U ratios (> 0.1), suggesting a magmatic origin (Hoskin & Schaltegger, 2003).

Fig. 4. Representative CL images of zircons from the samples. Solid red circles and blue circles indicate the locations of in situ U–Pb and Hf analyses spot, respectively.

Fig. 5. Zircon U–Pb concordia and weighted mean diagrams of the Pulandian samples.

4.b. Major- and trace-element geochemistry

The major- and trace-element composition results are provided in online Supplementary Table S2. All samples have low loss on ignition (LOI; 0.29–1.11 wt%). Sample Yd2005 has low SiO₂ contents (55.93–56.16 wt%), high MgO contents (3.63–4.13 wt%) and high Al₂O₃ contents (16.78–17.58 wt%), giving Mg no. values of 48.24–50.77. Its K₂O and Na₂O contents are 2.48–2.58 wt% and 3.55–3.77 wt%, respectively, with low K₂O/Na₂O ratios (0.69–0.70). The samples demonstrate high-K and subalkaline features (Fig. 6a, b). In contrast, samples Yd2006, Yd2009 and Yd2010 show high SiO₂ (63.77–70.80 wt%), high Al₂O₃ (14.13–17.38 wt%) and low MgO (0.36–1.78 wt%) contents, with Mg no. values of 21.18–46.83. Their K₂O and Na₂O contents are 2.05–5.37 wt% and 4.08–4.51 wt%, respectively, with variable K₂O/Na₂O ratios (0.45–1.31). The K₂O contents of these Pulandian plutons show positive correlation with the SiO₂ contents (Fig. 6b), whereas Fe₂O₃^T, TiO₂, P₂O₅, MgO, CaO and Al₂O₃ contents display negative correlations (Fig. 7a–f). The granodiorite (samples Yd2006 and Yd2010) shows high-K and subalkaline characteristics, whereas the porphyritic monzogranite (sample Yd2009) demonstrates shoshonitic and alkaline properties (Fig. 6a, b).

Fig. 6. (a) SiO₂ and K₂O + Na₂O correlation (TAS) diagram (after Wilson, 1989); and (b) SiO₂ and K₂O correlation diagram (after Peccerillo & Taylor, 1976). Data from Wu et al. (2005 a, b); Wang et al. (2007); Yang et al. (2007 a, b).

Fig. 7. Harker diagrams display the major-element variations of the Pulandian plutons.

Samples have total rare earth element (REE) contents of 74–190 ppm and show enrichment in low REE (LREE) with high (La/Yb)_N ratios (13.5–57.5). They have low Yb contents (0.54–1.24 ppm), low Y contents (4.9–16.4 ppm), high Sr contents (294–711 ppm; with the exception of one sample with 188 ppm content) and Sr/Y ratios of 38.5–108. In the chondrite-normalized REE pattern (Fig. 8), they display no clear Eu anomalies (Eu/Eu* = 0.79–1.21). The primitive-mantle-normalized spidergram (Fig. 8) reveals that these samples are enriched in large-ion lithophile elements (LILEs; e.g. Ba, K, Pb and Sr) and depleted in high-field-strength elements (HFSEs; e.g. Nb, Ta, P and Ti).

Fig. 8. Chondrite-normalized REE and primitive-mantle-normalized trace-element patterns for the samples. Normalized data for normalization and plotting after Sun & McDonough (1989). K-rich adakites data from Wang et al. (2007).

4.c. Zircon Hf isotopic compositions

The results of the zircon Hf isotopic analysis are provided in online Supplementary Table S3. Sample Yd2005 yielded relatively uniform ¹⁷⁶Hf/¹⁷⁷Hf ratios of 0.282275 to 0.282176, with ϵ _Hf(t) values of −18.5 to −15.0 and two-stage model ages (T _DM2) of 2351–2131 Ma. Sample Yd2006 zircon ϵ _Hf(t) values ranged from −27.2 to −25.4, with ¹⁷⁶Hf/¹⁷⁷Hf ratios of 0.281977–0.281927 and T _DM2 ages of 2899–2789 Ma. Sample Yd2009 yielded zircon ϵ _Hf(t) values ranging from −28.9 to −21.1, with ¹⁷⁶Hf/¹⁷⁷Hf ratios of 0.282107–0.281882 and T _DM2 ages of 3004–2511 Ma. The ϵ _Hf(t) values of sample Yd2010 ranged from −24.6 to −22.1, and its corresponding ¹⁷⁶Hf/¹⁷⁷Hf ratios and T_DM2 ages ranged over 0.282071–0.282000 and 2937–2580 Ma.

5. Discussion

5.a. Rock type and petrogenesis

Although the samples yielded very low LOI (0.29–1.11 wt%), the effect of element migration must be excluded before the geochemical characteristics of the rocks can be interpreted. Studies have shown that the major elements, HFSEs and REEs can remain constant during weak alteration, whereas LILEs can migrate more easily (Alirezaei & Cameron 2002); LILEs are therefore good indicators to gauge the effect of weak alteration on the samples. Some ratios (e.g. K/Sr and La/Th) can be used to monitor the mobility of the LILEs (Rudnick et al. 1985). All sample have low to medium K/Sr ratios (26–233) and low La/Th ratios (3.0–7.1), suggesting that the weak alteration did not modify their element compositions (Xin et al. 2019). The geochemical data of the selected samples, such as major- and trace-element content, therefore reflects their inherited magmatic source characteristics.

All samples in this study are characterized by high Sr contents (294–711 ppm), high Al₂O₃ contents (14.13–17.58 wt%), high Sr/Y ratios (38.5–108), low Yb contents (0.54–1.24 ppm), low Y contents (4.9–16.4 ppm) and no clear Eu anomalies (Eu/Eu* = 0.79–1.21), displaying typical adakitic geochemical characteristics (Defant & Drummond, 1990). In addition, most samples fall within the classification of adakite (Fig. 9a, b). These rocks have K₂O/Na₂O ratios (0.48–1.31) higher than for those of typical adakitic rocks (c. 0.4; Moyen, 2009), which are similar to K-rich adakites from Dabie orogen (Fig. 10a; Wang et al. 2007). The trace-element pattern of the rocks is also similar to that for the K-rich adakite rocks (Fig. 8); the samples can therefore be identified as K-rich adakitic rocks.

Fig. 9. (a) Sr/Y versus Y diagram (after Defant et al. 2002); and (b) (La/Yb)_N versus Yb_N diagram (after Martin, 1999). Data from Zhang & Wan (2001) and Xiong et al. (2011).

Fig. 10. (a) K₂O/Na₂O versus K₂O and (b) La/Yb versus La (after Wang et al. 2007).

Adakites are formed during various tectonomagmatic processes, such as partial melting of subducted oceanic slab (Defant & Drummond, 1990; Wang et al. 2007); AFC associated with basaltic magmas (Wareham et al. 1997; Castillo et al. 1999); magma mixing of basaltic magmas and crustal-derived felsic magmas (Guo et al. 2007; Xu et al. 2012); partial melting of thickened lower continental crust (Atherton & Petford, 1993; Petford & Atherton, 1996; Wang et al. 2005); and partial melting of delaminated lower continental crust (Gao et al. 2004; Hou et al. 2004).

Adakitic rocks were originally defined as the partial melting of subducted young, hot slabs (Defant & Drummond, 1990). As the slab melt rises, it is metasomatized by mantle-derived material, resulting in rocks characterized by low K₂O/Na₂O ratios (c. 0.5) and high MgO (c. 4–9 wt%), Mg no. (> 50), Ni and Cr contents (Stern & Kilian, 1996). The Pulandian adakitic rocks have high K₂O/Na₂O ratios (mean 0.85) and low MgO content (0.36–4.13 wt%), Mg no. (mean 37.7), Ni content (mean 20 ppm) and Cr content (mean 44 ppm), which are inconsistent with the features of melt of subducted oceanic slab. The samples have low Ce/Pb ratios (mean 7.6) and obvious negative Nb and Ta anomalies (Fig. 7), indicating that they may have originated in continental crustal source rather than partial melting or AFC of oceanic basalts (Ce/Pb > 20) (Rudnick & Fountain, 1995; Plank, 2005). Moreover, no clear Eu anomalies in all the samples imply that fractional crystallization of plagioclase did not play a crucial role in the process of their formation (Wang et al. 2007). The lack of obvious indications of fractional crystallization (Fig. 10b) and of large-scale mafic rocks around Pulandian area (Fig. 2) also support the fact that these rocks were not formed from AFC of basalt. The significant negative ϵ _Hf(t) values (−33.3 to −15.0) indicate that they derived from partial melting of the ancient crust rather than magmatic mixing (Fig. 11a). The geochemical characteristics of the samples are similar and there are no mafic microgranular enclaves in the rocks (Figs 3, 6, 8); magma mixing therefore does not explain their origin. The melt formed by partial melting of the delaminated lower crust may have been upwelling, resulting in interaction with the mantle and the production of geochemical features similar to those of the melting of the subducted oceanic slab (Xu et al. 2012); the rocks therefore did not originate from the partial melting of the delaminated lower crust. Experimental petrology shows that adakitic magmas can be formed by partial melting of the thickened lower continental crust (Petford & Atherton, 1996; Xiong et al. 2005). The melts from the above model are characterized by low MgO (< 2 wt%) and Mg no. (< 50), and high SiO₂ (> 60 wt%) and K₂O (> 2 wt%) contents, which are matched by the features of all samples (see online Supplementary Table S2). As shown in TiO₂ versus SiO₂ and MgO versus SiO₂ diagrams (Fig. 12), the majority of the samples from Pulandian plot within the field of thickened lower-crust-derived adakitic rocks.

Fig. 11. (a) ϵ _Hf(t) versus age (Ma) and (b) age histogram of Mesozoic magmatic rocks of NCC. Data from Yang et al. (2004, 2007 a, b).

Fig. 12. (a) TiO₂ versus SiO₂ and (b) MgO versus SiO₂ diagrams for samples. The field of subducted oceanic crust-derived adakites are after Wang et al. (2007). The field of thickened lower crust-derived adakites are after Atherton & Petford (1993) and Muir et al. (1995), and the field of delaminated lower crust-derived adakities are after Wang et al. (2004 a, b, 2007).

Previous studies shown that while partial melting if the garnet is remained as residue, the partial melt result HREE depletion (Othman et al. 1989; Wang et al. 2007). Plagioclase has a large partition coefficient for Sr and Eu elements, and its crystallization can result in Sr and Eu negative anomalies (Nash & Crecraft, 1985). On the basis of the geochemical features (online Supplementary Table S2; Fig. 8), the residues are rich in garnet and poor in plagioclase. According to partial melting experimental research, garnet-absent and plagioclase-rich samples indicate formation pressures lower than 10 kbar, both garnet and plagioclase are present at formation pressures of 12.5–15 kbar, and garnet-rich and plagioclase-absent samples indicate formation pressures higher than 15 kbar (Skjerlie & Johnston, 1993; Sen & Dunn, 1994; Patiño Douce & Beard, 1995; Litvinovsky et al. 2000). The above-mentioned residues can therefore only form in a high-pressure environment, in accordance with the pressure of thickened lower continental crust. The source and petrogenesis of the Pulandian plutons can be further illustrated with diagrams discerning the source nature of adakites (Foley et al. 2002). In the K₂O versus K₂O/Na₂O diagram (Fig. 10a), samples fall within the metabasaltic and eclogite melt fields, similar to the adakitic granites in the Dabie Orogen. In the (La/Yb)_N versus Yb_N diagram (Fig. 9b), all plots are mainly close to the melt of eclogite. We therefore suggest that the Pulandian adakitic rocks were formed by partial melting of thickened eclogitic lower crust.

5.b. Tectonic setting and geodynamic mechanism

Magmatic and volcanic activities were strong in the Liaoning Peninsula during the Early Cretaceous Epoch, which resulted in the formation of numerous intrusive rocks (Fig. 1b). The lithology of magmatic rocks in this period is mainly monzogranite, granodiorite and alkaline granite (Yang et al. 2004; Liu et al. 2019). According to the results of previous research, some Early Cretaceous A-type granites and alkaline rocks have been reported in the Liaoning Peninsula – for example, Qianshan granite (127 Ma), Taipingshan alkaline feldspar granite (129 Ma), Guanmenshan Granite porphyry (126 Ma) and quartz syenite (127 Ma) – which indicate that the Liaodong Peninsula existed in an extensional tectonic setting (Yang et al. 2007 a, b; Liu et al. 2016, 2019). Metamorphic core complexes (MCC) are also widely developed in the Liaoning Peninsula; for example, the Yinmawan (129–120 Ma) and Gudaoling (127–118 Ma) plutons were emplaced along the detachment fault of the Liaonan–Wanfu MCC (130–113 Ma), which was in an extensional tectonic environment (Guan et al. 2008; Ji, 2010). Zhu & Xu (2019) reported that the slab rollback of the Palaeo-Pacific Ocean began at c. 145 Ma, which may have resulted in the extension of the NCC. We therefore believe that the Early Cretaceous (130–120 Ma) extension was closely related to the slab rollback of the Palaeo-Pacific Ocean. The age of the Pulandian plutons is 127–120 Ma, which is nearly the same as the above pluton ages. The Pulandian plutons show LILE-enriched and HFSE-depleted characteristics, which indicate that these rocks were formed in a subduction environment. In the tectonic diagrams (Fig. 13), all samples fall within the area of arc-related environment. Combined with age and features, we believe that they were formed in an extensional setting involved in slab rollback of the Palaeo-Pacific Ocean.

Fig. 13. (a) Y versus Nb and (b) (Y + Nb) versus Rb (Pearce et al. 1984) for Pulandian sample rocks. VAG – volcanic-arc granites; syn-COLG – syn-collisional granites; ORG – orogenic ridge granites; WPG – within-plate granites.

Previous studies on mantle xenoliths indicate that the thickness of the NCC was 200 km during the Palaeozoic Era and 80–100 km during the Cenozoic Era, suggesting that significant destruction and lithospheric thinning occurred (Fan et al. 2000; Wu et al. 2019). The constructed seismic data of the NCC also support dramatic thinning (Zhu et al. 2011). In recent years, there has been more progress in determining the dynamic setting of lithospheric thinning of NCC (Wu et al. 2005 a; Zheng et al. 2007, 2018; Zhu et al. 2012; Niu et al. 2015). However, the geodynamic mechanism remains controversial and researchers are focused on two separate models: (1) rapid lithospheric delamination (e.g. Gao et al. 2004; Wu et al. 2005 b, 2008); and (2) thermo-mechanical-chemical erosion (e.g. Griffin et al. 1998; Zheng et al. 2007). Frequent magmatic activity was one of the important forms of destruction of NCC; the Liaodong Peninsula is the strongest and most typical area of destruction (Liu et al. 2011). A fuller understanding of the magmatic rocks of the Liaodong Peninsula will therefore yield knowledge of the lithospheric thinning and destruction of NCC.

During the Jurassic Period, the intrusive rocks in Liaodong Peninsula were dominated by quartz diorite, granodiorite and monzogranite, etc., most of which belong to high-K calc-alkaline granite (Fig. 6b; Yang et al. 2004, 2007 a, b; Wu et al. 2005 b; Sun et al. 2005). There were also many Jurassic adakities (Fig. 9), derived from the partial melting of the thickened lower crust (Gao et al. 2004; Wu et al. 2005 b; Zhu et al. 2012). In addition, the negative Lu–Hf isotopes also supported the theory that Jurassic granites are mainly derived from the partial melting of ancient thickened lower crust (Fig. 11a). The existence of thickened lower crust is the premise of lithospheric delamination. The rock types of Early Cretaceous intrusions in Liaodong Peninsula are complex, mainly composed of I-type granite, A-type granite, alkaline and mafic rock (Fig. 6; Sun & Yang, 2009; Lan et al. 2011). According to previous studies on geochemistry and Lu–Hf isotopes (−30 to +5; Fig. 11a), rocks from this period may have originated from different magma sources such as depleted mantle, enriched lithospheric mantle or ancient lower crust (Wu et al. 2019), possibly the result of intensive crust–mantle interaction caused by asthenosphere upwelling after lithospheric delamination of the NCC. The wide distribution of magmatic rocks (Fig. 11b) and extensional tectonics over a short period (c. 130–120 Ma) in the Liaodong Peninsula (Liu et al. 2005) is consistent with the rapid lithospheric delamination characteristics. We therefore suggest that the lithospheric delamination model is a better geodynamic mechanism for the thinning and destruction of the NCC than thermo-mechanical-chemical erosion.

Several possible trigger factors for lithospheric delamination have recently been proposed, including: (1) thickened lower crust (Gao et al. 2004, 2009); (2) hydrous fluids or melts related to Palaeo-Pacific subduction (Gao et al. 2009; Zhu et al. 2011; Zhu & Xu, 2019); and (3) upwelling asthenosphere (Deng et al. 2007; Liu et al. 2020). First, thickened lower crust is a crucial pre-condition to lithosphere delamination. The thickened lower crust undergoes a phase transformation of the eclogite, which makes it denser than the lithospheric mantle (Kay & Kay, 1993). The Jurassic subduction of the Palaeo-Pacific Ocean caused the crust of the NCC to thicken (Fig. 14a; Wu et al. 2005 b; Zhu & Xu, 2019). The Jurassic and Early Cretaceous adakites in the Liaodong Peninsula provide evidence for the thickened crust of the NCC (Yang et al. 2004; Wu et al. 2005 b; Zhang et al. 2010; this study). The thickened crust (eclogitic layer) could not directly sink because of the refractory and buoyant nature of the NCC lithospheric mantle (Deng et al. 2007). Secondly, the lithospheric mantle was metasomatized by hydrous fluids or melt produced by the subduction of the Palaeo-Pacific subduction (Niu et al. 2015). At c. 145 Ma the Palaeo-Pacific Ocean started to rollback, resulting in asthenosphere upwelling (Fig. 14b; Zheng et al. 2018; Zhu & Xu, 2019). At c. 130–120 Ma, the Palaeo-Pacific Ocean reached its maximum rate of subduction slab rollback, transitioned into a high-angle subduction and produced a stagnation slab in the mantle zone (Fig. 14c; Zhu & Xu, 2019). Because of the slower slab rollback and weak asthenosphere upwelling at 145 Ma, we believe that the refractory and thick lithospheric mantle of the NCC experienced a small-scale partial melting (Zheng et al. 2016, 2018). At c. 130–120 Ma, the stagnation slab began to dehydrate, changing the physical properties of the overlying lithospheric mantle (e.g. decreasing its viscosity) and causing its stability to be disrupted (Zhu et al. 2011). Under the combined action of these processes, the lithosphere of the NCC underwent delamination. This resulted in a rapid upwelling of large amounts of asthenosphere, providing more heat and accelerating partial melting of the lithospheric mantle. During this period, a large number of rocks of different sources and properties were generated (e.g. I-/A-type granite, adakites and rock; Yang et al. 2004; Wu et al. 2005 b; this study). Meanwhile, the widespread extensional tectonics and metamorphic core complexes appear during this stage.

Fig. 14. Lithospheric evolution process of the NCC during the Mesozoic Era. (a) Jurassic: the Palaeo-Pacific Ocean subducted, causing the NCC crust to thicken (Wu et al. 2005 b). (b) c. 145 Ma: the subducting Palaeo-Pacific slab began to rollback with the asthenosphere upwelling (Zhu & Xu, 2019). (C) Early Cretaceous (c. 130–120 Ma): the Palaeo-Pacific Ocean reached its maximum rate of subduction slab rollback, transitioned into a high-angle subduction and produced a stagnation slab in the mantle zone. The stagnation slab began to dehydrate, changing the physical properties of the overlying lithospheric mantle, which caused its stability to be disrupted and its subsequent delamination. During the period, widespread extensional tectonics occurred in the shallow crust of the NCC. The Liaonan–Wanfu MCC and rocks of different types and properties were generated (e.g. I-/A-type granite, adakites rock; Yang et al. 2004; Wu et al. 2005 b; this study). MCC – metamorphic core complexes.

6. Conclusions

1. (1) The Pulandian monzodiorite, granodiorite and porphyritic monzogranite are adakitic rocks that were emplaced at 127–120 Ma (i.e. Early Cretaceous).

2. (2) The Pulandian adakitic rocks were formed by partial melting of thickened eclogitic lower crust, and in an extensional setting related to slab rollback of the Palaeo-Pacific Ocean.

3. (3) As a result of the rapid slab rollback of the Palaeo-Pacific Ocean during c. 130–120 Ma, the asthenosphere upwelled and modified the thickened lithospheric mantle, which lost its stability, eventually resulting in the lithospheric delamination and thinning of the NCC.