Early Permian Granitic Magmatism in Middle Part of The Northern Margin of The North China Craton: Petrogenesis, Source, and Tectonic Setting

: The late Palaeozoic was an important period of tectonic evolution for the northern margin of the North China Craton (NCC). The source(s) and tectonic setting of early Permian granitoid rocks emplaced along the northern margin of the NCC are still unclear. These granitoids formed between ~295.4–276.1 Ma (uncertainties ranging from ±1.5 to ±7.8 Ma) according to zircon laser ablation inductively coupled mass spectrometry (LA-ICP-MS) and sensitive high-resolution ion microprobe (SHRIMP) U-Pb data. The Dadongou (DDG) pluton is an A1-type granite and the Dananfangzi (DNFZ) pluton is an A2-type granite. The Erdaowa (EDW), Lisicun (LSC), Wuhai (WH) and Gehuasitai (GHST) plutons are I-type granites. The Yuanbaoshan (YBS) dykes are diorite and syenodiorite. All the granitoids are enriched in large ion lithophile elements and light rare earth elements, depleted in high field strength elements and have negative εNd(t) and εHf(t) values. The A1-type granite was formed by the melting of the mafic crust. The A2-type granite was derived from partial melting of tonalite gneiss from the NCC crust and mantle materials. The EDW, LSC, WH and GHST granites mainly originated from partially melted granulite, with some mantle input. The YBS dykes are formed by the magma mixing of hot mantle melt and the relatively cold crustal magma. The northern margin of the NCC experienced anorogenic and collision tectonic stages, and the structural setting started to transform to post-collision at the later period of early Permian.


Introduction
The formation process and tectonic background of igneous rocks are important keys for understanding the geologic history of the continental margin. The northern margin of the North China Craton (NCC) is adjacent to the southern margin of the Xing Meng Orogenic Belt (XMOB) and within the eastern section of the Central Asian Orogenic Belt (CAOB), the largest Phanerozoic accretionary orogenic belt in the world. The CAOB formed by subduction-accretion during the Palaeozoic as a result of the closure of the Paleo-Asian Ocean (PAO) [1]. However, the timing of subduction, accretion and the PAO closure are uncertain. Previously proposed ages of the PAO closure include the late Silurian to Devonian [2], middle to late Devonian [3] and Permian to early Triassic [4][5][6]. Large volumes of Palaeozoic and Mesozoic magmatic rocks are distributed along the northern margin of the NCC. The evolution of the northern margin of the NCC was strongly influenced by the southward subduction and closure of the PAO from the early Palaeozoic to the Triassic [7,8]. Therefore, these magmatic rocks could provide insight into the timing of the PAO closure. Studies on early Permian magmatic rocks along the northern margin of the NCC have inferred that they were formed in different tectonic settings. Previous researches on granites, granodiorites and diorites have proposed the presence of an active continental margin related to the subduction of the PAO [4][5][6][9][10][11][12], while others [13] have suggested that the rocks resulted from intraplate extensional processes along lithospheric faults on the northern margin of the NCC. The authors of [14] analysed the early Permian A-type granites and emphasized that the A-type granites formed within a post-collisional tectonic setting.
To further discuss the sources, petrogenesis and tectonic setting in which these early Permian magmatic rocks formed, we collected granitoid rock samples from the central region of the northern margin of the NCC and obtained data on their petrography, major and trace element contents, Sr-Nd-Pb isotopes, laser ablation inductively coupled mass spectrometry (LA-ICP-MS) zircon U-Pb geochronology and Hf isotopes compositions. These new data, combined with existing data from previous studies, enable us to constrain the geologic evolution of the northern margin of the NCC and the XMOB.

Geologic Setting
The NCC formed at ~3.8 Ga and was modified by several tectonic episodes from the Neoproterozoic to the early Mesozoic [15]. The basement of the NCC is composed of late Archean tonalite-trondhjemite-granodiorite suites (TTGs), ~2.6-2.5 Ga gneisses, mafic-ultramafic intrusive rocks and a few supracrustal rocks (including sedimentary and bimodal volcanic rocks), as well as early Proterozoic TTGs (2.4-2.5 Ga) [6], mafic granulites (1.8 Ga) [16] and syntectonic or post-tectonic granites [11]. The northern margin of the NCC is located adjacent to the eastern section of the CAOB (Figure 1), which is one of the largest Phanerozoic accretionary orogenic belts in the world [1,4].  [5].
The northern margin of the NCC was strongly influenced by the PAO tectonic system between the Carboniferous and the Permian [7]. It is generally thought that the PAO subducted southward beneath the NCC [4], causing the formation of magmatic rock belts along the northern margin of the NCC during the late Palaeozoic and the early Mesozoic [7,11,17,18]. Magmatic rocks that formed in the late Carboniferous and early Permian are mainly diorites, quartz diorites, granodiorites and granites, as well as occasional tonalites and gabbros [18]. The granitoids are mostly I-type granites with small amounts of A-type granites [11].
Permian magmatic rocks are widely distributed in Inner Mongolia [17,, forming an E-W trending belt that runs parallel to the northern margin of the NCC (Figure 2). The rock assemblages include Figure 2. Distribution of early Permian granitoids in the middle part of the northern margin of the NCC (modified from [40,41]). diorite, quartz diorite and granite, followed by gabbro. Only a small amount of Permian volcanic rocks was emplaced in the northern margin of the NCC. Late Permian magmatic rocks are widely distributed on both sides of the Baiyun Obo-Chifeng-Fuxin Fault, which is considered the boundary between the northern margin of the NCC and the XMOB [4]. Early Permian ultramafic-mafic rocks occur on the northern margin of the NCC. Their distributions are controlled by the Jining-Chengde and Fengning-Longhua-Jianping faults, which are two long-term active lithospheric fault zones in the northern margin of the NCC. These ultramafic-mafic rocks intruded into lower carboniferous rocks and were penetrated by synchronous quartz diorites and granitoids.
We collected 32 whole rocks and 9 zircon samples from 8 early Permian intrusive granitoid rocks and dykes in the central region of the northern margin of the NCC ( Figure  2 and Figure 3a,b). The sampling transect was approximately 80 km along the northern margin of the NCC and 100 km in the vertical direction. The samples were from Dananfangzi (DNFZ), Dadonggou (DDG), Erdaowa (EDW), Lisicun (LSC), Gehuasitai (GHST), Wuhai (WH) and Yuanbaoshan (YBS) intrusive rocks and dykes. The early Permian intrusive rocks and dykes coexist with the early Palaeozoic magmatic rocks.

Whole-Rock Analysis
Of the rock samples, 32 that did not contain veins or evidence of alteration were crushed and powdered using a pollution-free agate ball mill. Then, they were passed through a 200-mesh sieve. Major and trace elements were measured using X-ray fluorescence (XRF) on a ZSX100e instrument (Rigaku Corporation, Japan), following the method of [42], and inductively coupled plasma mass spectrometry (ICP-MS) on an X SERIES II instrument (Thermo Electron Corporation, US), respectively, at the Henan Rock and Mineral Testing Centre, China. FeO was determined using the potassium dichromate volumetric method. A set of standards was analysed to monitor instrument precision during XRF analyses and yielded precision better than 3%. The GSR21 and GSR22 [43] standards were analysed during all ICP-MS analyses and yielded precision typically better than 10%.
Whole Sr-Nd-Pb isotopic compositions of fifteen powdered rock samples were measured using isotope dilution thermal ionization mass spectrometry (ID-TIMS) at the Beijing Research Institute of Uranium Geology, China. The Sm-Nd isotopes used 146 Nd/ 144 Nd = 0.7219 and the Rb-Sr isotopes used 88 [44,45] are listed in Table S1 and Table S2, respectively.

Zircon Analyses
Zircons were separated from 9 rock samples using standard methods (e.g., crushing, panning, magnetically separating and processing through heavy liquids) at the Hebei Institute of Regional Geology and Mineral Resources Survey, China. Approximately 100 zircon grains were handpicked from each sample using a binocular microscope. The grains were mounted on an epoxy resin disc along with several grains of the TEMORA [46] zircon standard, ground down to expose their interiors and polished. The zircon grains were then imaged using cathodoluminescence (CL) at the Institute of Geology and Geophysics, Chinese Academy of Sciences. The zircons were mostly euhedral stubby to long prismatic grains, but some grains had slightly or variably rounded terminations (Figure 5). The photomicrographs showed excellent crystal morphology under CL, with variable textures from shared oscillatory zoning to sector zoning, as was the case for magmatic zircons. The different zircon shapes and internal structures reflected variable transport and relatively heterogeneous source rocks. Magmatic zoning was characterized by moderate Th/U, which was ubiquitous under CL. Some of the grains were dark (high-U) or had dark rims and light-grey (low-U) cores. The U content of the analysed zircons was highly variable, ranging from 32.98 ppm to 3608.6 ppm (Table S3).

SHRIMP Zircon Analyses
Sample NMH9-5 from the DDZ was dated using a sensitive high-resolution ion microprobe (SHRIMP) with an accelerating voltage of ~10 kV, an ion beam current of ~7.5 nA and a beam diameter of 25-30 μm at the Beijing SHRIMP Centre, Institute of Geology, Chinese Academy of Geological Sciences. Data obtained from the SHRIMP analyses were processed and analysed using the PRAWN software package from the Australian National University [47].

LA-ICP-MS Zircon Analyses
The other zircon samples were dated using laser ablation (LA) single collector ICP-MS with an acceleration voltage of ~15 kV, an ion beam current of ~4 n, and a beam diameter of ~32 μm at the China University of Geoscience, Wuhan, China. The zircon standards 91500 (primary standard ~1065 Ma) [48] and GJ-1 (~608 Ma) [49] were also analysed at regular intervals throughout each run to assess the reproducibility and accuracy of the measured isotopic ratios and to correct for U/Pb fractionation. The weighted mean ages of the 91500 and GJ-1 zircon standards were 1061.8 ± 2.2 Ma (n = 74, MSWD = 0.49) and 598.6 ± 4.4 Ma (n = 35, MSWD = 7.8), respectively (all errors are 2σ). The analytical procedure for U-Pb dating has been described in detail previously [50]. The measured 204 Pb was used to correct for common Pb. Data obtained from the LA-ICP-MS analyses were processed using the ICPMSDataCal software package [50].
Zircon Hf isotopic analyses were performed in situ on the same samples used for U-Pb dating using a Newwave UP213 laser-ablation microprobe (Elemental Scientific, US) attached to a Finnigan Neptune multi-collector ICP-MS (Thermo Fisher Scientific, Germany), following the analytical procedures of [51], at the Institute of Mineral Resources, Chinese Academy of Geological Sciences, Beijing, China. A stationary spot with a beam diameter of 55 μm was used. The ablated sample was transported, using He as the carrier gas, from the laser-ablation cell to the ICP-MS torch via a mixing chamber filled with He. The GJ-1 zircon standard was also analysed, and yielded 176 Hf/ 177 Hf ratios indistinguishable from the weighted mean 176 Hf /177 Hf ratio of 0.282006±14 (2σ). Care was taken to ablate spots immediately adjacent to those used for U-Pb determination within the same CL imaged zone. LA-ICP-MS zircon U-Pb ages and Lu-Hf isotopic data [52][53][54][55] are listed in Table S3 and Table S4, respectively. All reported errors include only the analytical error.

Zircon U-Pb Geochronology
We report the zircon 206 Pb/ 238 U age as the best ages for grains younger than 1000 Ma and the 207 Pb/ 206 Pb age for grains older than 1000 Ma [56]. Figure 5 and Figure 6 show representative CL images of the zircon samples, zircon U-Pb Concordia diagrams and weighted mean 206 Pb/ 238 U ages. The majority of the samples were within 10% concordance and most of the ages were well grouped ( Figure 6).

Alkali-Feldspar Granites
We found that 6 zircons from sample HB25-2 from the DNFZ provided concordant results, with a mean 206 Pb/ 238 U age of 284.7 ± 7.8 Ma (Figure 6a), whereas 12 concordant zircon grains from sample NMH9-5 from the DDG yielded a mean age of 295.4 ± 4.7 Ma (Figure 6b). No inherited cores were involved in the calculation ( Figure 5).

Granites
Six analysed zircon grain spots from sample HB75-2 from the EDW were calculated. Analysis spots were located on zircon rims with oscillatory zoning, which provided concordant results with a mean 206 Pb/ 238 U age of 278.1 ± 5.3 Ma (Figure 6c). Three dark (high-U) inherited cores yielded much older ages of 902 ± 17 Ma, 1028 ± 79 Ma, and 2420 ± 38 Ma, and may have been associated with the Proterozoic magmatic thermal event (2.4-1.8 Ga) [16,57] on the northern margin of the NCC. The inherited ages suggest that the continental basement was involved in later magmatic processes.
Eight well-zoned magmatic zircon grains from sample NMH15-1 from the LSC provided concordant data, with a weighted mean 206 Pb /238 U age of 281.9 ± 1.5 Ma (Figure 6d). Ten zoned zircon grains from sample HB53-2 from the GHST provided concordant results, with a mean 206 Pb/ 238 U age of 276.1 ± 3.2 Ma (Figure 6e). Eight zircons analysed from sample HB44-2 from the WH yielded a weighted mean 206 Pb/ 238 U age of 279.8 ± 4.3 Ma ( Figure  6f).

Diorite and Syenodiorite
Ten zircon grains from sample HB179-2 from the YBS syneodiorite dyke yielded a weighted mean age of 284.4 ± 5.2 Ma (Figure 6g). Eleven zircon spots from sample HB179-4 from the YBS diorite dyke provided concordant results, with a mean 206 Pb/ 238 U age of 282.2 ± 4.3 Ma (Figure 6h).

Sr-Nd-Pb Isotopic Compositions
Whole-rock Sr-Nd-Pb isotopic compositions and calculated initial Sr-Nd-Pb isotopic compositions based on the zircon U-Pb ages are presented in Table S2.
Alkali-feldspar granites had a wide range of ( 87

Zircon Hf Isotopic Compositions
Some concordant zircons dated using LA-ICPMS were also analysed for their Hf isotopic compositions. The zircon Lu-Hf isotopic data, depleted mantle extraction (TDM) and crustal model ages (TDM c ) of these samples are listed in Table S4 and plotted in Figure 10.
Alkali   Both A-type granites and highly differentiated granites could exhibit these geochemical characteristics. A-type granites have high TFe2O3 (>1%), while TFe2O3 is usually low (<1 %) in highly fractionated granites [64]. The DNFZ and DDG plutons also have relatively high TFe2O3 (1.87% on average) content. Furthermore, the samples from the DNFZ and DDG contain no elbaite, lepidolite, lithium muscovite or beryl, which are lithogenous mineral markers of highly fractionated granites [65]. The zircons in highly fractionated granite are usually anhedral because of their late crystallization [65,66]. The zircon grains of samples are also mostly euhedral. Therefore, the DNFZ and DDG plutons are interpreted as Atype granites rather than highly fractionated granites.

Rock Classification
The source area of A1-type granite is similar to that of ocean island basalts, while the source area of A2-type granite is derived from the continental crust or curst formed by underplating [67]. The Y/Nb ratio can be used to distinguish two subtypes of A-type granites because the Y/Nb ratio is mainly controlled by the source rock types, and the influence of crystallization differentiation of pyroxene, amphibole and opaque minerals on the Y/Nb ratios is relatively weak [67]. A1-type granites are rich in Nb, while A2-type granites are rich in Y [67]. The DDG pluton is high in Nb while it is low in Y, displaying an A1type feature. The DNFZ pluton has high values of Y and can be classified as A2-type granite. The DDG pluton shows more distinctive A-type features than the DNFZ pluton, is peralkaline and has low contents of Al2O3 and CaO. The DDG pluton appears to be a peralkaline A1-type granite. The DNFZ pluton contains perthite without alkaline melaminerals, such as arfvedsonite or riebeckite, and is metaluminous-peraluminous. The DNFZ pluton appears to be an aluminous A2-type granite.

Granites
The granites have higher SiO2 contents and large A/CNK ranges, are generally weakly peraluminous and contain small amounts of normative corundum (typically < 1%). The majority of the samples belong to the high-K calcium alkaline series and display Itype granite characteristics in a K2O-Na2O diagram [62]. The granites contain no alkaline melaminerals and have low ratios of 1000Ga/Al. The samples contain biotite and iron oxides rather than muscovite, cordierite and other strong aluminium diagnostic minerals. When the SiO2 content of granite is above 75%, the P2O5 content of I-type granites is less than 0.05%, whereas the P2O5 content of S-type granites is greater than 0.1% [68]. The majority of our samples had P2O5 contents less than 0.1% (mostly < 0.05%), and the P2O5 was content negatively correlated with the SiO2 content. Thus, our samples were interpreted to be I-type granites.

Alkali-Feldspar Granites
A-type granite magmas can be produced by 1) the melting of crustal igneous rocks [69], including tonalities, granodiorites [70] and tonalitic gneisses [71]; 2) mantle-derived alkaline basaltic magma [67]; or 3) the mixing of mantle-derived alkaline basaltic magma and crustal anatectic magma [72]. It is unlikely that the fractional crystallization of mantlederived alkaline basaltic magma formed DNFZ and DDG pluton. First, A-type granite formed by the crystallization differentiation of mantle-derived basic magma is usually accompanied by large-area contemporaneous basic-ultrabasic rocks [73]. However, only the basic dykes were exposed in the study area without the large-scale exposure of contemporaneous basic rocks. The differentiation and assimilation of mantle basic magma usually produce a series of rocks which change continuously from basic to intermediate and acidic [74]. There were no intermediate rocks accompanying the A-type granite plutons in study area. The DNFZ and DDG plutons had high Pb content, large negative Eu and Sr anomalies and strong depletions in high field strength elements (HFSEs) such as Nb and Ti. These characteristics indicate that plagioclase remained in the magma source and the granites formed at low pressure [75], which does not correlate with the results of differentiation and evolution of a mantle-derived basaltic magma [76].
The DDG A1-type granite was characterized by high SiO2, K2O and Na2O, while it was low in CaO, Mg # , Cr and Ni, showing major element characteristics reflecting middlelower crustal partial melting [77]. The Sr and Nd isotopic compositions were similar to those of the lower crust in the northern margin of the NCC (Figure 12). Relatively low εNd(t) values and older crustal model ages for the zircon Hf data show that the magma dominantly originated from ancient crustal intermediate rocks. The DDG pluton had relatively high contents of Nb and Ta, suggesting that the melting of mafic crust was involved in the formation.
The DNFZ A2-Type granite was high in SiO2, K2O, Na2O, Al2O3 and Pb. The DNFZ pluton had low contents of Eu, Ni, Nb and Ta, showing the features of felsic crust source. In the study area, exposed tonalite gneisses formed at 2.5-2.4 Ga [16]. Therefore, we consider the tonalite gneisses present in the lower NCC crust as a main source of the observed A2-type granite. The DNFZ granite had more significant mantle feature than DDG granite through Hf isotopic composition. The zircon Hf data of the DNFZ pluton show a mixture source of mantle and crust ( Figure 10). The DNFZ granite, as the later pluton, had more mantle input, implying that the mantle activity became more intense later. The mantle magma provided heat that led to the crust melting and mantle material input during the formation of A2-type granites.

Granites
The granites had low Sr and Yb contents (< 2 ppm), suggesting that they formed at medium to low pressures [70]. The partial melting of pre-existing crustal igneous rocks played a key role in the formation of I-type granitic magma [79]. High-K calc-alkaline mafic-intermediate rocks such as pyroxene granulites can evolve into a magma source for granite [80]. Petrologic data [80] have shown that the Neoarchean and Paleoproterozoic continental crust of the NCC is mainly composed of TTG and granulites metamorphosed from high-K calc-alkaline mafic-intermediate rocks. However, high-K calc-alkaline I-type granites can only evolve from the partial melting of calc-alkaline to high-K calc-alkaline mafic to intermediate metamorphic rocks [81]. The granites were likely formed by the partial melting of granulites metamorphosed from high-K calc-alkaline mafic-intermediate rocks in the NCC. A slight negative Eu anomaly and depletion in HFSEs such as Ti, P, Nb, and Ta indicate that residual plagioclase, rutile, titanite, apatite, and amphibole remained in the magma source. The strong depletions in Nb and Ta also show a crustal origin or magma that was contaminated by crustal material [82].
The ratios of K/Rb, Zr/Hf, Nb/Ta and Y/Ho are invariant in ordinary magma systems [83], while they are largely reduced in highly fractionated granites [66,84,85]. In this study, the average elemental ratios were 52.77 (Zr/Hf), 11.09 (Nb/Ta) and 29.29 (Y/Ho), which is not similar to the elemental ratios of highly fractionated granites. The ratios of Zr/Hf (> 26) and Nb/Ta (< 5), along with the low Be content, support that the granites mainly formed due to initial partial melting without strong fractional crystallization [84,86].
The Nd model ages and zircon εHf(t) values of most of the samples are consistent with the ancient lower NCC crust. The zircon εHf(t) values of granites were distributed between the lower crust and depleted mantle lines (Figure 10), implying that the crust was the main origin and that the mantle played a small part. The granites had a few ancient inherited zircons, which indicate the old crustal source. The samples had Hf crust model ages peaking at 2.2-1.8 Ga, corresponding to a period of NCC continental crustal accretion [57,87,88]. High SiO2 and low concentrations of MgO, Ni and V indicate little or no mantle materials in the source. Thus, the granites in this study mainly originated from partial melting of the ancient mafic crust with a small added component of mantle material.

Diorite and Syenodiorite
Acicular apatites were developed in YBS syenodiorite, showing the process of fast cooling. Acicular apatites imply that the hot basic magma was injected into relatively cold acidic magma and crystallized rapidly. In the TFeO/MgO plot (Figure 13a), the samples followed the magma mixing trend [89]. The YBS diorite and syenodiorite had high Mg # , MgO, FeO, Cr and Ni content. The Mg # (48.57-55.61) of the YBS rocks was higher than those of the experimental melt formed by partial melting of basalt and eclogite at different pressures [90], suggesting that both crustal melting and mantle-derived magma acted as the magmatic source. The high ( 87 Sr/ 86 Sr)t and lower εNd(t) and εHf(t) also illustrate a mixture of mantle-derived components and crustal materials. Thus, we state that the YBS dykes were formed by the magma mixing of mantle basic magma and crust magma. The low SiO2 and high Mg # content show that the mixed magma was dominated by mantle materials. The divergence of chemical composition between diorite and syenodiorite were caused by the different degrees of magma mixing.
The Al2O3 content in the YBS dykes was high and may have been caused by the melting of enriched mantle (EM) or plagioclase accumulation. However, the absence of an Eu anomaly in these samples indicates that plagioclase accumulation was not significant, and that the high Al content was likely from the mantle part of the magma source of the dykes. The YBS dykes were depleted in Nb, Ta and Ti, which was caused by lithospheric mantle metasomatism by a fluid or by crustal contamination. Furthermore, the YBS dykes had high Ba content, ranging from 832-1821 ppm, reflecting a large fluid-related enrichment, while the nearly constant Th/Nb ratio showed a limited melt influence [91]. The plots of Ba-Nb/Y and Ba/Y-Nb/Y (Figure 13b,c) suggest that the YBS dykes were partly the result of fluid-related partial melting. Aqueous fluids were responsible for the enrichment of LREE and REE in these rocks [91]. The subducted PAO slab likely supplied fluid for partial mantle melting and enrichment.

Tectonic Implications
The A1-type granites, in general, represent an anorogenic magmatism of continental rift or intraplate environment [67]. In this study, the DDG A1-type granite suggests that an anorogenic environment was present. The DDG granite is plotted in the anorogenic field in the R1-R2 diagram and within the plate granite field in the Rb-(Y+Nb) diagram ( Figure 14). The tectonic setting of study area was likely to be anorogenic in the early period of the early Permian (295.4 Ma). It is generally believed that the A2-type granite formed in a post-orogenic environment or island arc environment and could be genetically related to subduction-related crustal melting or regional-scale mantle upwelling in the CAOB [92,93]. The DNFZ A2-type granite (284.7 Ma) showed the trend of post-collision or arc granite to syncollision granite (Figure. 14). The DNFZ pluton was likely formed at the transient structural relaxed stage of subduction before the final suture of the PAO.
The YBS dykes (284.4-282.2 Ma) formed though magma mixing and fluid-related partial melting (Figure. 13). The fluid evolved as the result of dehydration of the subducting slab following the subduction of the CAOB. The YBS dykes in the pre-orogenic and volcanic arc granite field are plotted in Figure 14. The collision had not happened but was about to start between 284.4-282.2 Ma.
The LSC, WH, EDW and GHST granites (281.9-276.1Ma) mainly had the features of syncollision granites (Figure 14). These granites showed the trend of volcanic arc granites to syncollision granites and had slight post-collision granite features at the late period of early Permian. We hypothesized that the collision of the Siberian plate and the North China plate in study area occurred during the middle-late period of early Permian (281.9-276.1Ma), while the remnant arc may have still existed in some areas. In the study area, granites (272-270 Ma) [32][33][34][35] that were formed slightly later than the LSC, WH, EDW and GHST plutons were formed in a post-collision background. These post-collision granites indicate that the closure of Paleo Asian Ocean was likely to take place before 272 Ma. The closure time of the Paleo Asian Ocean has not been determined yet. The unconformable contact relationships and different deformation modes of the late Devonian molasse formation and ophiolitic mélange show that the collision and suture took place during the Devonian at the latest [96]. Other researchers agree the PAO was closed during the late Permian and early Triassic according to the geochronology and geochemistry of the late Palaeozoic basic-ultrabasic rocks and granitoids [7,11,97]. The geochemistry of granitoids mainly represents the features of source rocks and the information of the tectonic environment [95]. We speculate that the collision of the Siberian plate and the North China plate had not yet happened at 295.4-282.2 Ma and was likely to take place between 281.9-276.1 Ma. The structural setting started to transform to post-collision at the later period of the early Permian. The Paleo Asian Ocean was closed before the middle Permian in the middle part of the northern margin of the NCC, and the study area was in the postcollision setting during the middle Permian [32,98]. The early Permian granitoids with negative whole-rock εNd(t) and zircon εHf(t) values [7,11,27,50] with crustal characteristics are widely distributed in the eastern CAOB, Southern Mongolia, and the western CAOB [91,99,100]. Numerous researches have proven that the late Paleozoic magmatism in the northern margin of the NCC is related to the southward subduction of the PAO plate [7,11]. Large-scale magmatism indicates the occurrence of regional scale crustal remelting. Therefore, we suggest that the dehydration of the subducted PAO slab induced the interaction between fluids and the lower crust, causing partial melting of the lower crust. Accumulation of subducted slabs at the base of the mantle could cause widespread mantle upwelling events. Subsequently, mantle upwelling aggravated the melting of crust.

Conclusions
The integrated geochronological, major and trace elemental, and Sr-Nd-Pb isotopic and zircon Hf isotopic compositions of the granitoids from the central region of the northern margin of the NCC allowed us to reach the following conclusions: 1) These granitoids can be classified as alkali-feldspar granites, granites, diorit, and syenodiorite. Geochronological analyses suggest that the alkali feldspar granites formed at 295.4-284.7 Ma, the granites formed at 281.9-276.1 Ma, and the diorite and syenodiorite formed at 284.4-282.2Ma. 2) The DDG pluton appeared to be A1-type granite, while the DNFZ pluton had a pronounced A2-type granite affinity. The A1-type granite was formed by the melting of the mafic crust. The A2-type granite was derived from partial melting of tonalite gneiss from the NCC crust and mantle materials. Granites mainly originated from partially melted granulite, with some mantle input. The YBS dykes were formed by the magma mixing of hot mantle melt and the relatively cold crustal magma. Our findings showed that the early Permian crustal evolution in northern margin of the NCC included a few juvenile crust inputs besides intracrustal circulation.
3) The subduction and slab breakoff of the PAO led to widespread mantle upwelling events, and this mantle upwelling initiated large-scale remelting of the lower crust. The northern margin of the NCC experienced anorogenic and collision tectonic stages, and the structural setting started to transform to post-collision at the later period of early Permian. The studies of basic rocks, xenoliths and minerals are needed to better constrain our findings.

Supplementary Materials:
The following are available online at www.mdpi.com/2075-163X/11/2/99/s1, Table S1: Whole-rock major (%) and trace-element (ppm) data of the Early Permian plutons in the middle segment of the north margin of the NCC, Table S2: Whole-rock Sr-Nd-Pb isotopic compositions of the Early Permian plutons in the middle segment of the north margin of the NCC, Table S3: Zircon U-Pb data of the Early Permian plutons in the middle segment of the north margin of the NCC, Table S4: Zircon Hf isotopic data of the Early Permian plutons in the middle segment of the north margin of the NCC.