1 Introduction

The Qilian Orogenic Belt (QOB), located along the northeastern margin of the Tibetan Plateau, is a suture zone that has recorded an evolutionary history from continental breakup, oceanic subduction, and continental collision from the Proterozoic to the Paleozoic (Yin and Harrison 2000; Song et al. 2013; Liu et al. 2018; Wang et al. 2020). The North Qilian Orogenic Belt (NQOB) is a typical Early Paleozoic orogenic belt in the QOB, and studies have demonstrated that long-lived magmatism occurred during multiple episodes of orogenic processes in the Paleozoic (Xiao et al. 2009; Song et al. 2013, 2014; Zuza et al. 2017). The Paleozoic ophiolite sequences, island arc and collision-related granitoids, high-pressure metamorphic rocks, Silurian flysch formations, and Devonian molasses record the evolution from the subduction and closure of the North Qilian Ocean to the collision between the Qilian-Qaidam block and Alxa block (Song et al. 2013). Although intensive scientific research (Song et al. 2006, 2013; Wu et al. 2010; Yu et al. 2015; Yuan and Yang 2015; Zeng et al. 2016; Zhang et al. 2017; Wang et al. 2019) regarding the geodynamic events of the NQOB has focused on interpreting Paleozoic orogenic processes, several important issues, such as subduction polarity, timing of transition from subduction to collision, and closure of the North Qilian Ocean, are still debated.

As an important part of the continental crust, granites have been intensively studied because they contain valuable information about crustal evolution, crust–mantle interactions, and mountain building (Lundstrom and Glazner 2016; Wu et al. 2007; Hopkinson et al. 2017; Wang et al. 2018). The petrogenesis, source, evolution, and tectonic setting of granites have long been important topics in research on the evolutionary history and tectonic attributes of orogenic belts (Chappell and White 1974; Pearce et al. 1984; Clemens 2003; Zhu et al. 2009; Xu et al. 2010; Chen et al. 2018; Li et al. 2022). Lithologically, Early Paleozoic granites are widespread in the NQOB and are of utmost importance in understanding the tectonic evolution (Tseng et al. 2009; Yu et al. 2015). In addition, the identification of different types of granites can be used to reconstruct key geodynamic processes, such as the initial timing of the transition from arc to collision during subduction (Wang et al. 2018; Zhao et al. 2022) and the transition from compression to extension during collision (Yu et al. 2015; Zeng et al. 2016; Zhang et al. 2017). In this contribution, we conducted integrated research on the zircon geochronology and whole-rock geochemistry of the Early Paleozoic Beidaban granites in the NQOB. Based on this work, combined with previously published geochronological and geochemical data on Early Paleozoic igneous rocks (ca. 468–414 Ma), the purpose of this study is to better constrain their sources and petrogenesis, to improve our understanding of their tectonic evolution history, and to provide new constraints on the Early Paleozoic transitional tectonic events of the NQOB.

2 Geological background

The QOB is located at the intersection of the Alxa block, Tarim block, North China block, and Qaidam block (Fig. 1). It has experienced multiple episodes of tectonic evolution, including continental breakup, ocean basin formation, oceanic subduction, arc, back-arc systems, and continental collision (Xiao et al. 2009; Zuza et al. 2017; Zhu et al. 2022). The QOB consists of three nearly parallel tectonic subunits trending NW‒SE from south to north: the South Qilian Orogenic Belt (SQOB), the Central Qilian Block, and the NQOB (Fig. 1) (Xia et al. 2016; Li et al. 2018; Gao et al. 2021). The SQOB is bounded by the Zongwulong Orogenic Belt, the Quanji Massif, and the North Qaidam ultrahigh-pressure metamorphic belt of the Qaidam Block. This belt was thought to constitute the final suture zone in the Qilian-Qaidam region (Song et al. 2009; Yang et al. 2018). The Central Qilian Block mainly consists of Precambrian high-grade granitic gneisses and low-grade metamorphic assemblages, overlain by Paleozoic sedimentary rocks and intruded by granitoids (Smith 2006; Huang et al. 2015; Fu et al. 2018).

Fig. 1
figure 1

Schematic map showing major tectonic units of the Qilian Orogenic Belt. The simplified geological map of the Qilian Orogenic Belt was modified from Zhu et al. (2022) by showing main boundaries of tectonic units and distribution of Paleozoic and Mesozoic granitoids rather than other tectonic units such as volcanics, sedimentary strata, and mafic–ultramafic rocks

The NQOB is situated in the northern part of the QOB and bounded by the Alxa block to the north (Fig. 1). The NQOB consists of two Early Paleozoic ophiolite belts (southern ophiolite belt (550–497 Ma, representing the spreading of the Oceanic basin) and a northern ophiolite belt (490–448 Ma, representing the back-arc basin extension), arc-related volcanic and intrusive rocks, high pressure metamorphic rocks, and accretionary complexes (Pan et al. 2004; Smith and Yang 2006; Song et al. 2007, 2013; Zhang et al. 2007; Xiao et al. 2009; Lin et al. 2010; Xia et al. 2012; Cheng et al. 2016; Li et al. 2017), which represent a typical western Pacific-type trench-arc-basin system. This zone is regarded as a suture zone resulting from the subduction of the North Qilian Ocean between the Alxa terrane and the Central Qilian Block in the Early Paleozoic (Xia et al. 2003, 2012; Song et al. 2007, 2013; Zhang et al. 2007; Li et al. 2016; Peng et al. 2017; Zhu et al. 2022).

Early Paleozoic intrusions are widespread throughout the NQOB. The Beidaban granite mass in the study area is located in the central part of the NQOB (Fig. 1). Cambrian metamorphic basement rocks, Early Paleozoic intrusive rocks, and Paleozoic to Mesozoic sedimentary sequences developed in the study area (Fig. 2). The oldest Cambrian Dahuangshan Group consists of metamorphosed feldspathic quartz sandstone and meta-arkose interbedded with low-grade metamorphosed slate. Ordovician and Silurian strata are absent in the study area. Carboniferous sedimentary rocks consist of clastic deposits of shallow marine and marine terrigenous facies with carbon-rich shale, sandstone, and coal. The conformable overlying these sedimentary rocks are Permian strata, which are mainly distributed in the southwestern part of the study area. The Permian rocks probably represent coarse clastic sediments (quartz sandstone and feldspathic sandstone) deposited in braided rivers. Jurassic strata consist of clastic rocks interbedded with sandstone and coal layers. The faults in the study area trend toward the SN and EW. Intermediate-acidic intrusive rocks are widely exposed on the northern side of the study area and include diorite, granodiorite, granite, monzogranite, and syenogranite (Chen et al. 2019). The Beidaban granite, with an exposed area of approximately 285 km2, is located in Jinshan town, Wuwei City, Gansu Province (Fig. 2).

Fig. 2
figure 2

Geological map of Beidaban pluton in the NQOB with localities of representative samples (modified from 1:50 000 geological map of Beidaban and surrounding areas)

3 Occurrence and petrography of granite samples

The Beidaban granite samples collected in the field are penetratively deformed by the development of spaced cleavage (Fig. 3a). The granite samples are pale red or pink in color and feature fine- to medium-grained granitic textures. They are mainly composed of euhedral K-feldspar (35%–40%), euhedral plagioclase (15%–20%), subhedral to anhedral quartz (25%–30%), and minor amphibole (3%–5%) and biotite (2%–3%) (Fig. 3b). Accessory minerals include small amounts of Ti–Fe oxides, allanite, apatite, and zircon.

Fig. 3
figure 3

Field (ab) and thin-section (cf) photographs of the Beidaban granites. a and b showing the establishing shot and close shot of granite; cf showing the mineral assemblage of granite; Abbreviations: Pl–plagioclase; Kfs–K-feldspar; Qtz–quartz; Am–amphibole; Bt–biotite

The K-feldspars are characterized by large euhedral laths or subhedral crystals that are 500–2000 μm in diameter (Fig. 3c, d). Most of them are microcline with typical tartan twinning. The plagioclase grains commonly exhibit euhedral to subhedral shapes and polysynthetic twinning and Carlsbad twinning, and they are generally 200–2000 μm in diameter (Figs. 3c, d). The plagioclase grains with hydromicazation on the surface are likely andesines (An = 32–40) based on the maximum extinction angle of the plagioclase crystals. Myrmekitic intergrowth of oligoclase and wormy quartz is common (Fig. 3c). The monocrystalline quartz crystals are typically subhedral to euhedral crystals ranging from 200 to 1000 μm in diameter (Fig. 3e). Amphiboles exist as perfect euhedral crystals in the Beidaban granite (Fig. 3e, f). They are mainly hexagonal and rhomboid in shape, and some are replaced by chlorite and carbonate minerals. The biotites are predominantly associated with amphiboles and display subhedral to euhedral crystals (Fig. 3e). Most of the biotites are replaced by chlorite. Minor Ti–Fe oxides with grain sizes ranging from 100 to 300 μm are occasionally observed in the Beidaban granite (Fig. 3f).

4 Analytical methods

Zircon grains for the laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) analyses were separated using conventional magnetic and density techniques and then selected under a binocular microscope. The hand-picked zircon grains from the granite samples were mounted in epoxy and then polished to expose cross sections. Zircons under reflected and transmitted light, and cathodoluminescence (CL) were imaged to examine the internal microstructures. Before the experiment, zircons with clear oscillatory zoning structures were selected. LA-ICP-MS zircon U–Pb dating was conducted at the Tianjin Institute of Geology and Mineral Resources, Chinese Geological Survey. The detailed analysis involved a 193 nm laser ablation system coupled with an Agilent 7500 ICP-MS. Helium was taken as the carrier to enhance the transport efficiency of the ablated materials. A 32 μm spot size was used with an energy density of 5 J/cm2 and a repetition rate of 5 Hz. The U, Th, and Pb concentrations were calibrated using 29Si as an internal standard, and NIST 610 as a reference standard. Zircon 91500 was used as the external standard for U–Pb dating. Time-dependent drifts of U-Th-Pb isotope ratios were corrected using linear interpolation for every ten analyses. Data reduction was performed using the ICPMSDataCal software (Liu et al. 2008). Only data with a concordance in the range of 90%–110% were considered for interpretation of zircon ages. Concordia diagrams were produced using ISOPLOT 3.0 software (Ludwig 2003).

The analyses of major and trace elements (including rare earth elements) of the granite samples were performed at the Tianjin Institute of Geology and Mineral Resources, Chinese Geological Survey. In the process of major element analysis, samples were first weighed into moderate amounts of boric acid and then melted into glass at high temperatures. Then, the oxide contents were measured via an XRF instrument and the analysis accuracy was better than 1%. The trace element analysis was completed via an inductively coupled plasma mass spectrometer (ICP-MS) with Finnigan MAT Element I equipment according to the procedure for DZ/T0223-2001 ICP-MS. The relative error of the analysis element was better than ± 5%, the temperature and humidity during the progression of the experiment were 20 °C and 30%, respectively.

Whole-rock Sr–Nd isotopic compositions were analyzed at the Tianjin Institute of Geology and Mineral Resources, Chinese Geological Survey, using a multi-collector VG 354 mass spectrometer in static mode. Approximately 100–150 mg of powder was decomposed in a mixture of HF–HClO4 in screw-top Teflon beakers, and Rb, Sr, Sm, and Nd were separated using cation exchange columns. The Sr and Nd isotopic fractionations were corrected to 6Sr/88Sr = 0.1194 and 146Nd/144Nd = 0.7219, respectively. During the analytical period, several measurements of the NIST NBS987 Sr reference standard and the JNdi-1 Nd reference standard yielded 87Sr/86Sr = 0.710232 ± 15 (2σ) and 143Nd/144Nd = 0.512117 ± 11 (2σ), respectively. Analytical precision is approximately 1% for 87Rb/86Sr and 0.5% for 147Sm/144Nd. Detailed sample preparation and analytical procedures for Sr and Nd isotopic analysis followed those of Tang et al. (2007) and He et al. (2007).

Whole-rock Pb was separated and purified using anion exchange in HCl–Br columns. The Pb isotopic ratios of granite samples were measured via MC-ICP-MS on a Nu Instruments system in the same laboratory as the Sr and Nd isotopic analysis. Pb isotopic fractionation was corrected to 205Tl/203Tl = 2.3875. During the analytical period, repeated analyses of international standard NBS981 yielded ratios of 206Pb/204Pb = 16.939 ± 0.013 (2σ), 207Pb/204Pb = 15.497 ± 0.011 (2σ), and 208Pb/204Pb = 36.712 ± 0.033 (2σ), respectively. Detailed sample preparation and analytical procedures for the Pb isotope measurements followed those of He et al. (2005).

5 Results

5.1 Zircon U–Pb geochronology

Zircon grains from the Beidaban granite samples are euhedral and display similar crystal forms with long axes varying from 80 to 150 μm and length/width ratios ranging from 1 to 2 (Fig. 4). Cathodoluminescence (CL) images reveal that all zircon grains exhibit homogeneous oscillatory growth zoning, which is interpreted as typical magmatic features. Ten zircons from the granite samples were analyzed, and the individual zircon U–Pb dating results, U and Th contents, and isotopic ratios are listed in Table 1. The zircons have U contents ranging from 198 to 1180 ppm (632 ppm on average) and Th/U ratios ranging from 0.5 to 0.9, indicating magmatic zircons (Belousova et al. 2002). The ages shown in Fig. 5 represent the weighted averages of the concordant and clustered 206Pb/238U ages of individual zircons. The results show that the data for the granite (K2-ZW8) plot on or near the Concordia line (Fig. 5a) and that the 206Pb/238U ages are distributed in the range from 447 to 501 Ma; the concordant U–Pb age is 468 ± 10 Ma (Fig. 5a) with a weighted mean age of 467 ± 13 Ma (n = 10, MSWD = 2.7) (Fig. 5b). The rare earth element (REE) results of zircons are shown in Table 2. The REE patterns show that zircons have typical igneous zircon features that are characterized by high HREE contents, conspicuous positive Ce anomalies, and negative Eu anomalies (Fig. 6).

Fig. 4
figure 4

Representative cathodoluminescence (CL) images of measured zircons from the Beidaban granites

Table 1 Zircon U–Pb isotopic data from the Beidaban granites
Fig. 5
figure 5

Zircon U–Pb concordia age and mean age diagrams for the Beidaban granites

Table 2 Rare earth element data of zircons from the Beidaban granites
Fig. 6
figure 6

Chondrite-normalized REE patterns for zircons from the Beidaban granites (normalizing data from Sun and McDonough 1989)

5.2 Whole-rock geochemistry

The major and trace element compositions of the granite samples in this study are listed in Table 3. Notably, one granite sample has a high K2O content, which is probably caused by alteration because of its high loss on ignition (LOI). The contents of SiO2, Al2O3, P2O5, and total alkalis (K2O + Na2O) range from 63.0wt% to 78.0wt%, 13.0wt% to 16.6wt%, 0.1wt% to 0.4wt%, and 7.1wt% to 9.2wt%, respectively, which indicate that the Beidaban granite is calc-alkaline, shoshonitic, and metaluminous (A/CNK = 0.7–0.9) (Fig. 7). The Lumanshan S/I-type granites (450 Ma, Zhao et al. 2022) exhibit geochemical characteristics similar to those of the Beidaban granites (Fig. 7). However, the whole-rock geochemical compositions of the postcollisional intrusions (alkali-feldspar granite, Zhang et al. 2017; quartz diorite, Fu et al. 2018; monzogranite, Wang et al. 2018) in the NQOB demonstrate that they are mostly subalkalic and medium- to high-K alkaline rocks with variable A/CNK ratios (Fig. 7).

Table 3 Major element (wt.%) and trace element (ug/g) compositions of the Beidaban granites
Fig. 7
figure 7

K2O + Na2O vs. SiO2 (a, modified from Middlemost 1994), K2O vs. SiO2 (b, modified from Peccerillo and Taylor 1976), and A/NK vs. A/CNK (c, modified from Maniar and Piccoli 1989) classification diagrams of the Beidaban granites. Legends of cited data from the NQOB in Figs. 9, 10, 11 and 12 are the same as in this figure

The total REE (ΣREE) content in the Beidaban granite ranges from 215 to 530.0 μg/g, with an average of 325.1 μg/g. The light REE (LREE) contents are relatively high, ranging from 202.2 to 495.1 μg/g (302.3 on average), whereas the heavy REE (HREE) contents are relatively low, ranging from 12.4 to 34.5 μg/g (22.8 on average). The Beidaban granite shows negative europium anomalies (δEu = 0.5–0.8) in the chondrite-normalized pattern (Fig. 8a), with (La/Yb)N values ranging from 13.0 to 25.7. Except for 414 Ma A-type granites, most of the Paleozoic intrusions in the NQOB are characterized by small negative Eu anomalies in the REE distribution patterns (Fig. 8a–c). LREEs are generally enriched relative to HREEs, but the degree of enrichment varies among different Paleozoic intrusions. In the primitive mantle-normalized trace element diagram (Fig. 8d–f), the Beidaban and Lumanshan granites are enriched in large ion lithophile elements (LILEs; e.g., Rb, Th, Ba, K, and Sm) and depleted in high field strength elements (HFSEs; e.g., Nb, Ta, Ti, and Sr). The contents of Rb, Ba, Th, Sr, and P in the post-collisional intrusions (alkali-feldspar granite, quartz diorite, and monzogranite) are lower than those in the granites (Fig. 8f).

Fig. 8
figure 8

Chondrite-normalized rare-earth element patterns (a, b, c) and primitive mantle-normalized trace element patterns (d, e, f) for the Beidaban granites and other cited data in the NQOB. Chondrite values and primitive mantle values are from Sun and McDonough (1989)

5.3 Sr–Nd and Pb isotopic compositions

The Sr–Nd isotopic compositions of the granite samples are given in Table 4. The Rb, Sr, Sm, and Nd values range from 195 to 302 μg/g, 165 to 473 μg/g, 5.4 to 14.9 μg/g, and 36.7 to 91.3 μg/g, respectively. The measured 87Rb/86Sr, 87Sr/86Sr, 147Sm/144Nd, and 143Nd/144Nd values are 1.3280374–1.9620178, 0.717809–0.740767, 0.07867069–0.10503226, and 0.511751–0.512016, respectively. When the age of 468 Ma is used in the calculation, the initial (87Sr/86Sr)i, (143Nd/144Nd)i, and εNd(t) values vary from 0.70545 to 0.71082, 0.511478–0.511695, and − 10.9 to − 6.7, respectively. The calculated Nd isotope model ages (TDM) and two-stage model ages (TDM2) range from 1.49 to 1.80 Ga and from 1.74 to 2.08 Ga, respectively.

Table 4 Sr and Nd isotopic data and calculated values for the Beidaban granites

The Pb isotopic compositions are given in Table 5. The measured whole-rock 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb ratios are 19.89–21.93 (21.10 on average), 15.76–15.87 (15.82 on average), and 40.09–42.16 (40.83 on average), respectively. When the age of 468 Ma is used in the calculation, the initial (206Pb/204Pb), (207Pb/204Pb), and (208Pb/204Pb) ratios are 19.14–20.26 (19.81 on average), 15.71–15.77 (15.75 on average), and 37.70–38.26 (38.01 on average), respectively.

Table 5 Pb isotopic data and initial ratios for the Beidaban granites

6 Discussion

6.1 Petrogenesis and sources

The Beidaban granites and other Middle–Late Ordovician samples (> 450 Ma) in the NQOB are characterized by high contents of SiO2 and K2O and show metaluminous features (Fig. 7). In addition, euhedral primary hornblende, biotite, and Ti–Fe oxides frequently occur in the Beidaban granite (Fig. 3). These characteristics are indicative of I-type granite affinity (Miller 1985). The geochemical features of low A/CNK ratios (<1) (Fig. 7c), enrichment of LREEs and LILEs (Rb, Th, K), and depletion of HFSEs (Nb, Ta, Ti) (Fig. 8) also suggest that the Middle–Late Ordovician granites are likely arc I-type granitoids (Zhao et al. 2014). However, the Late Ordovician to Silurian granites (<440–438 Ma) are characterized by relatively low contents of SiO2 and K2O, high A/NK (Fig. 7), and low LREEs and Sr, P, and Ti (Fig. 8). In the FeOt/MgO vs. 10,000 ∗ Ga/Al and Y vs. 10,000 ∗ Ga/Al diagrams (Fig. 9a, b), nearly all the Early Paleozoic samples plot in the I-, and S-type fields except for several Late Silurian samples (414 Ma). These 414 Ma granites are A-type granitoids (Zhang et al. 2017) based on their major and trace element geochemical features (Figs. 7 and 8). The negative relationship between SiO2 and P2O5 (Fig. 9c) suggests that the Beidaban granites and other Middle-Late Ordovician samples are I-type granites but that several Late Ordovician to Silurian granites are likely S-type granites. Nevertheless, some Beidaban granites also feature geochemical characteristics of S-type granites, such as relatively high contents of Al2O3 and K2O. These characteristics are similar to those of the Lumanshan S/I-type granites (450 Ma, Zhao et al. 2022) in the NQOB. According to the ACF classification diagram (Fig. 9d), most of the samples from Beidaban granites and Middle-Late Ordovician granites plot in the S-type field, whereas most of the Silurian granite samples (<434 Ma) plot in the I-type field. Significantly, some of the Beidaban samples in this study straddle the boundary between S-type and I-type granitoids. Therefore, recent studies suggested that these types of granites do not belong to pure I-type granites because of their multiple melt sources (called transitional I/S-type granites) (Chappell et al. 2012; Gao et al. 2016; Wang et al. 2018; Zhao et al. 2022). Thus, based on the above discussions, we propose that the Beidaban granites probably represent transitional I/S-type granitoids.

Fig. 9
figure 9

Whole-rock major and trace element discrimination diagrams for the Beidaban granites. a FeOt/MgO vs. 10 000× Ga/Al; b Y vs. 10,000× Ga/Al; c SiO2 vs. P2O5; d A-C-F. a and b are modified from Whalen et al. (1987); d is modified from Lameyre and Bowden. (1982)

The nearly equal Sr and Nd isotopic values (Table 4) of the Beidaban granites suggest that they were cogenetic. Generally, crustal melts have negative εNd(t) values and high initial (87Sr/86Sr)i values, whereas mantle magmas have positive εNd(t) values along the mantle array (Kinny and Maas 2003). The high initial (87Sr/86Sr)i values (0.70545–0.71082) and negative εNd(t) values (-10.9 to -6.7) strongly indicate that continental crust was involved in the genesis of the Beidaban granite. In the εNd(t) vs. t(Ga) diagram (Fig. 10a), the Beidaban granites plot in Paleo-Mesoproterozoic crustal field, indicating that they were generated by melting of the Paleo-Mesoproterozoic crustal basement and were probably caused by magma underplating (Huang et al. 2015; Zhu et al. 2022). As illustrated in Fig. 10b, although the Beidaban granites overlap the field of Paleozoic granitoids in Qilian, they plot in the lower left field of I-type granitoids. This means that the magma sources for the Beidaban granites were different from those for the I-type granitoids in Qilian and were more likely derived from Paleo-Mesoproterozoic continental crustal material (Chu et al. 2006; Zhou et al. 2017; Yang et al. 2018). This explanation is supported by the two-stage Nd model ages (TDM2) of 1.74–2.08 Ga. Similar cases are observed for the Lumanshan granites (450 Ma, Zhao et al. 2022) in the NQOB (Fig. 10b). However, the high εNd(t) values and low initial (87Sr/86Sr) values (Fig. 10a, b) suggest that the 446 Ma granites were generated by partial melting of thickened crust (Yu et al. 2015). Compared to those of the Beidaban granites, the 440–438 Ma, 434 Ma, and 433–431 Ma granitoids in the NQOB have higher εNd(t) values and lower initial (87Sr/86Sr)i values (Fig. 10b), which are indicative of mantle isotopic signatures or material evidence of melting ocean crust (Zhang et al. 2006; Chen et al. 2018; Wang et al. 2018). The 414 Ma granites are possibly derived from partial melting of felsic crustal material, which was caused by lithospheric delamination after the collision (Zhang et al. 2017). The Pb isotope values, which plot in the field of continental crust and above the Northern Hemisphere Reference Line (NHRL) in the (206Pb/204Pb) vs. (207Pb/204Pb) diagram (Fig. 10c), suggest continental crust source for the Beidaban granite. In addition, geochemical features such as high Th/Ta (17.43–30.12) and Rb/Nb (6.01–15.49), which are obviously greater than those of the continental crust (Rudnick and Gao 2003), are also consistent with a recycled crustal component (Taylor and Mclennan 1985).

Fig. 10
figure 10

Sr, Nd, and Pb isotopic composition and ratios for the Beidaban granites and other cited data in the NQOB. a εNd(t) vs. t (Ga) diagram. b εNd(t) vs. (87Sr/86Sr)i diagram (modified from Zindler and Hart 1986). Fields for Paleozoic granitoids and I-type granitoids in Qilian are based on Zhu et al. (2022) and Zhang et al. (2017). c (206Pb/204Pb) vs. (207Pb/204Pb) diagram. Data of EM I, EM II, BSE, MORB, and NHRL are from Zindler and Hart (1986); continental crust data from Zartman and Doe (1981)

Studies reveal that dehydration melting of metapelites and metagreywackes yields higher Al2O3/(TFeO + MgO + TiO2) and lower CaO + TFeO + MgO + TiO2 values compared to that of metabasaltic rocks (Kaygusuz et al. 2008). Melting experiments also reveal that melting products of mafic lower crust yield lower K2O/Na2O, Rb/Ba, and Rb/Sr ratios and Al2O3/(MgO + TFeO) contents or higher Al2O3 + TFeO + MgO + TiO2 contents compared to those of metasedimentary rocks (Rushmer 1991; Patiño Douce and Beard 1996; Kaygusuz et al. 2008). Thus, the Beidaban granites and the 450 Ma granites with relatively high K2O/Na2O, Rb/Ba, and Rb/Sr ratios and low Al2O3 + TFeO + MgO + TiO2 contents are thought to be mainly derived from metasedimentary rocks, which is consistent with the plot of the Rb/Sr vs. Rb/Ba diagram (Fig. 11a). However, scholars have proposed that the CaO/Na2O ratios of felsic rocks derived from partial melting of metagraywackes and igneous sources range from 0.3 to 1.5 (Jung and Pfänder 2007; Sylvester 1998). The high CaO/Na2O ratios (0.49–1.08) of the Beidaban granites and the 450 Ma granites are supposed to indicate the partial melting of metagraywackes and igneous sources. Furthermore, the CaO/Na2O vs. CaO/TFeO diagram suggests metagraywacke and granitoid sources for the Beidaban granites (Fig. 11b). Robust persuasiveness cannot be founded upon a single discriminant diagram. In the discriminant diagrams of Al2O3 + TFeO + MgO + TiO2 vs. Al2O3/(TFeO + MgO + TiO2) and molar CaO/(MgO + TFeO) vs. molar Al2O3/(MgO + TFeO), the Beidaban granites and the 450 Ma granites plot not only in the field of partial melting of metagraywacke sources but also in amphibolites or metabasaltic to metatonalitic sources (Fig. 11c, d). Therefore, it is reasonable to confirm that the transitional I/S-type Beidaban granites and the 450 Ma granites were derived from heterogeneous magma sources produced by partial melting of metagreywackes and igneous rocks (Sylvester 1998; Altherr and Siebel 2002; Kaygusuz et al. 2008).

Fig. 11
figure 11

Major and trace element discrimination diagrams for the Beidaban granites. a Rb/Sr vs. Rb/Ba (modified from Sylvester 1998). b CaO/Na2O vs. CaO/TFeO (modified from Yang et al. 2016). c Al2O3 + TFeO + MgO + TiO2 vs. Al2O3/(TFeO + MgO + TiO2) (modified from Kaygusuz et al. 2008). d molar CaO/(MgO + TFeO) vs. molar Al2O3/(MgO + TFeO) (modified from Altherr et al. 2000)

6.2 Tectonic implications

Zircon geochronology studies of intermediate-acidic igneous rocks in the NQOB imply that these igneous rocks span a long history of tectonic evolution between 853 and 211 Ma (Zhu et al. 2022 and references therein). Paleozoic igneous rocks, especially Early Paleozoic granitoids, are distributed in the NQOB, and geochronological and geochemical studies on these granitoids indicate that continental breakup (680–520 Ma) and development of the Paleo Qilian Ocean (520–495 Ma) occurred in the NQOB (Xia et al. 1999; Xiao et al. 2009; Zhang et al. 2012). Although controversies regarding the subduction polarity of the North Qilian Ocean exist (Zhang et al. 1997, 2012; Xia et al. 2003), north-dipping subduction along the northern margin of the QOB during the early Paleozoic has been accepted by most scholars (Song et al. 2013; Xia et al. 2016; Zhu et al. 2022). However, despite previous studies on Paleozoic granites in the NQOB (Xia et al. 2012; Zhang et al. 2017; Yang et al. 2018; Wang et al. 2018; Fu et al. 2018; Zhao et al. 2022), the important first-order problems related to the transitional tectonic setting (from arc to initial collision) or the initial closure timing of the North Qilian Ocean have been poorly constrained, which limits the ability to reconstruct the tectonic mechanism of magma generation in the NQOB. Therefore, the Middle Ordovician to Silurian granitoids with ages ranging from 468 to 414 Ma that occurred in the NQOB are summarized for comparison.

As mentioned above, the Beidaban granites yield a zircon U–Pb age of 468 ± 10 Ma and contain K-feldspar, amphibole, biotite, and Ti–Fe oxides with A/CNK values less than 1, which illustrates that they are K-rich porphyritic calc-alkaline granitoids (KCGs). This type of granitoid is indicative of an initial collisional setting (Barbarin 1999). However, the transitional I/S-type Beidaban granites and the Middle Ordovician granites (> 446 Ma) also show geochemical characteristics similar to those of adakitic island-arc-type granitoids (Fig. 12a). Based on these findings, we suggest that the transition from subduction to collision environment in the NQOB probably did not occur earlier than 468 Ma. Previous studies proposed that the North Qilian Ocean closed before 450 Ma (Zhao et al. 2022). In the Yb + Ta vs. Rb and Y vs. Nb diagrams, nearly all the Beidaban granites and Middle-Late Ordovician granites (> 446 Ma) plot in the fields of VAG and Syn-COLG (Fig. 12b, c), indicating that they likely represent the products of final arc and incipient syncollisional magmatism (Sylvester 1998). These geochronological and geochemical features provide significant information about tectonic processes after the closure of the North Qilian Ocean (Yu et al. 2015). As the North Qilian Ocean closed, the island arc crust was considerably shortened and thickened, and the syn-collisional magmatism in the NQOB contributed to continental crust growth (Zhang et al. 2006; Chen et al. 2018; Fu et al. 2018). Several recent studies proposed that the NQOB featured a collisional setting at ca. 442–422 Ma (Wu et al. 2010; Song et al. 2013; Zhang et al. 2017; Li et al. 2017; Wang et al. 2018; Zhao et al. 2022), as evidenced by the Qingshan monzogranite (440–438 Ma), Laohushan quartz diorites (426 Ma), and Shengrongsi granites (422 Ma). These granites are distributed mainly in the conjunction area between the NQOB and the Alxa block and have geological and geochemical features related to the compressional environment generated by the collision between the Qilian-Qaidam block and the Alxa block (Fu et al. 2018). Comparatively, the 414 Ma A-type granites are different from the island-arc-type and syn-collisional granites in the NQOB (Fig. 12). These A-type magmatic suites were post-collisional granites and were probably generated by lithospheric delamination after the collision (Zhang et al. 2017). Therefore, based on our work and previous studies, we propose that the Beidaban granites record the tectonic transition setting from subduction in the Qilian Ocean to initiation of collision and that the final closure of the Qilian Ocean can be constrained to the Middle–Late Ordovician (ca. 468–450 Ma).

Fig. 12
figure 12

Discrimination diagrams for the Beidaban granites. a Y vs. Sr/Y (modified from Drummond and Defant 1990). b Yb + Ta vs. Rb. c Y vs. Nb. d Y + Nb vs. Rb. Normalization values of b, c, and d are from Pearce et al. (1984). ORG: ocean ridge granites; VAG: volcanic arc granites; WPG: within plate granites; COLG: collision granites

7 Conclusions

  1. (1)

    Zircon U–Pb ages show that the Beidaban granite were emplaced at 468 ± 10 Ma, indicating Middle Ordovician magmatism in the central part of the North Qilian Orogenic Belt.

  2. (2)

    The Beidaban granites contain amphibole, biotite, and Ti–Fe oxides and are K-rich porphyritic calc-alkaline granitoids. The geochemical characteristics indicate that the Beidaban granites represent transitional I/S-type granitoids.

  3. (3)

    The high initial (87Sr/86Sr)i values (0.70545–0.71082), clearly negative εNd(t) values (− 10.9 to − 6.7), and high initial Pb isotopic compositions (206Pb/204Pb = 19.14–20.26; 207Pb/204Pb = 15.71–15.77) suggest that the Beidaban granites originated from recycled crustal components and were probably derived from partial melting of a metasedimentary source with the involvement of igneous rocks.

  4. (4)

    The timing of the tectonic transition in the NQOB from an arc to a collisional setting and the final closure of the Qilian Ocean can be constrained to the Middle-Late Ordovician (ca. 468–450 Ma).