Petrogenesis of the Cretaceous Intraplate Mafic Intrusions in the Eastern Tianshan Orogen, NW China

In this study, we conducted zircon U-Pb dating, and whole-rock geochemical and Sr-Nd isotope analyses on the Late Mesozoic dolerite dykes in the Bailingshan Fe deposit (Eastern Tianshan Orogen, NW China) to unravel their petrogenesis and regional tectonic significance. Zircon U-Pb dating on the dolerite yielded an Early Cretaceous age of 129.7 ± 1.4 Ma. The dolerite is calc-alkaline sodic (Na2O/K2O = 4.71 to 6.80), and enriched in LILEs (Rb, K, Sr, and Pb) but depleted in HFSEs (Nb, Ta, and Ti). The intermediate Nb/U (16.7 to 18.5) and Ce/Pb (6.33 to 6.90) values, and the presence of xenocrystic zircons in these dolerite dykes suggest crustal assimilation during the magma evolution. Petrological modeling suggests fractionation of olivine, pyroxene, garnet, and spinel. All the dolerite samples have low initial 87Sr/86Sr (0.7041 to 0.7043) and positive εNd(t) (+ 4.6 to + 5.1) values, indicative of a depleted asthenospheric mantle source. Partial melting modeling suggests that the melting has occurred in the spinel-garnet stability field. Integrating the data from ore deposit geology, geochronology, geochemistry and Sr-Nd isotopes, we proposed that the Late Cretaceous Eastern Tianshan mafic magmatism was developed in an intraplate extension setting.

In this contribution, we describe the newly-discovered Cretaceous dolerite dykes near the Bailingshan deposit in the Eastern Tianshan, and present their whole-rock geochemical and isotopic compositions. These dolerite dykes represent the youngest magmatic rocks in the Eastern Tianshan, and our data provide new petrogenetic insight and improve our understanding in the Late Mesozoic tectonic evolution of the region.

Sampling
Dolerite dyke samples in this study were collected near the Bailingshan Fe deposit (41 • 48 07 N, 91 • 11 40 E) in the Aqishan-Yamansu belt, with a sampling interval of about 2 -5 m from one dyke. In the field, the dolerite dyke (70 -80 m long, 2 -3 m wide) was observed to have intruded the Late Carboniferous volcaniclastic rocks (Figure 2a), suggesting a younger magmatic event. The dolerite samples are dark green, and medium-to fine-grained porphyritic (Figure 2b). Euhedral plagioclase is the dominant phenocryst phase and commonly 1 to 3 mm long (Figure 2b). The doleritic-/interstitial-textured groundmass is composed mainly of subhedral plagioclase, clinopyroxene and trace Fe-Ti oxides (Figure 2c). Minor plagioclase and clinopyroxene grains in the samples are partially altered to epidote and amphibole, respectively. All the sample preparation and laboratory analyses were performed at the Wuhan SampleSolution Analytical Technology Co., Ltd. (WSATCL), China.

Zircon U-Pb Geochronology
After separated with the conventional density and magnetic separation techniques, zircon grains were hand-picked under a binocular microscope. The zircon internal structure was studied via cathodoluminescence (CL) imaging using an Analytical Scanning Electron Microscope (JSM − IT100).
Laser Ablation-Inductively Coupled Plasma-Mass Spectrometry (LA-ICP-MS) zircon U-Pb dating and trace element analysis for the dolerite samples were performed at the WSATCL. The laser ablation used 5 Hz frequency and 24 µm spot size. Each analysis comprises a background acquisition of approximately 20 -30s, followed by 50s of sample data acquisition. The zircon 91500 and glass NIST610 were used as the external standard, and the Plešovice and GJ-1 zircons as the internal standard. Quantitative calibration for trace element analyses and U-Pb dating were conducted using ICPMSDataCal (Liu et al., 2008). Calculation of weighted mean ages and concordia diagram construction were performed using Isoplot/Ex 3.0 (Ludwig, 2003).

Whole-Rock Major and Trace Element Analyses
All the samples were first powdered to less than 200-mesh, then were placed in an oven at 105 • C for drying of 12 h to determine LOI. The major element contents were measured by X-ray fluorescence (XRF) spectrometry (1 g powder for each sample), and the analytical precision is better than 1%. For trace elements (including REEs), 50 mg powder for each sample was dissolved in a mixture of 100ml HNO 3 and 100 ml HF. This solution was then analyzed with an Agilent 7700e ICP-MS, and the analytical precision is better than ± 5%.

Whole-Rock Sr-Nd Isotopes
Whole-rock Sr-Nd isotope analyses were conducted on a Neptune Plus Multi-Collector (MC)-ICP-MS, with the detailed analytical techniques as described by Wang et al. (2019). The 87 Rb/ 86 Sr and 147 Sm/ 144 Nd ratios were calculated using the Sr, Rb, Nd and Sm contents obtained by ICP-MS. All the measured Sr and Nd isotope ratios were  normalized with 86 Sr/ 88 Sr = 0.1194 and 146 Nd/ 144 Nd = 0.7219, respectively (Lin et al., 2016). Analyses of the standards NIST SRM 987 and JNdi-1 during the measurement period yielded 87 Sr/ 86 Sr = 0.710244 ± 0.000022 (2σ) and 143 Nd/ 144 Nd = 0.512118 ± 0.000015 (2σ), similar to the recommended values (Tanaka et al., 2000).

Zircon U-Pb Ages
One sample was dated by zircon U-Pb analysis, and the results are given in Table 1. The zircons analyzed have wide ranges of U and Th contents, and all have high Th/U ratios (> 0.4). The zircon grains are transparent and euhedral-subhedral prismatic, indicating a magmatic origin (Figure 3; Belousova et al., 2002;Li, 2009). The 23 zircons analyzed show a wide 206 Pb/ 238 U age range ( Figure 4A). The concordant zircons (n = 16) range from 136 to 126 Ma, yielding a weighted mean age of 129.7 ± 1.4 Ma (MSWD = 1.4; Figure 4B), which likely represents the dolerite crystallization age. Seven inherited/xenocrystic zircons were identified, including a Precambrian (1860 Ma), three Devonian-Carboniferous (365 to 356 Ma), and three Permian-Triassic (263 to 243 Ma) ones.

Whole-Rock Major and Trace Elements
Whole-rock major element and trace element contents of the dolerite dykes are listed in Table 2. As our samples are fresh or only slightly altered (e.g., weak epidote alteration in some plagioclase grains), the alteration effect on geochemistry is likely negligible. After normalized to 100 wt.% (anhydrous), the samples contain 49.4 to 50.1 wt.% SiO 2 , 19.8 to 20.6 wt.% Al 2 O 3 , and 6.54 to 6.82 wt.% MgO. They have medium Fe 2 O 3 T content (8.43 to 8.71 wt.%) with corresponding Mg# (100 * Mg/(Mg + Fe)) of 47 -49. The rocks are sodic (Na 2 O/K 2 O = 4.71 to 6.80), and fall inside the basalt field in the TAS diagram (Figure 5; Wilson, 1989).

Whole-Rock Sr-Nd Isotopes
Whole-rock Sr-Nd isotope compositions of the studied samples are presented in  (Figure 7), all samples plot in the ocean island basalt (OIB) field (Hart, 1985), similar to the Permian Huangshanxi gabbro in the Eastern Tianshan (Zhang et al., 2011;Deng et al., 2015).

Fractional Crystallization and Crustal Contamination
The dolerite samples have much lower Mg# (47 -49), Cr (144 -150 ppm) and Ni (112 -121 ppm) contents ( Table 2) than typical mantle-derived melts (Mg# = 71 -83, Cr > 1000 ppm and Ni > 400 ppm; Wilson, 1989;Wang et al., 2019), suggesting that their parental magma may have experienced fractional crystallization. Rayleigh fractional crystallization model calculations were conducted to determine the fractionation phase (Wilson, 1989), which shows that all samples plot between the olivine and pyroxene evolution trends in the Cr-Ni diagram ( Figure 8A). This indicates significant fractionation of olivine and pyroxene. Fractionation of Tibearing minerals (e.g., spinel) in the dolerite is evidenced by the negative Nb, Ta and Ti anomalies and the Nb-Ta fractional crystallization model diagram ( Figure 8B; Saunders et al., 1992;Hawkesworth et al., 1993). The lack of negative Sr and Eu anomalies in the samples suggests limited plagioclase fractionation (Figure 6). Mantle-derived magmas commonly assimilate crustal components during their ascent, which altered their geochemical characteristics (e.g., LREE and LILE enrichments and HFSE depletions; Wilson, 1989;Rudnick and Gao, 2014), as found also in the dolerite dykes (Figure 6). This clearly suggests certain degrees of crustal contamination for the doleritic magma. Element pairs (e.g., Nb and U, Ce, and Pb) with similar bulk-solid/melt partition coefficients cannot be significantly    (Zhang et al., 2011); DM (Salters and Stracke, 2004). The OIB and MORB fields are after Hart (1985) and Wilson (1989), respectively. segregated through partial melting or fractional crystallization, and their ratios remain roughly constant and reflect those of the magma source (Hofmann, 1997). The dolerite samples have Nb/U (16.7 to 18.5) and Ce/Pb (6.33 to 6.90) values intermediate between the mantle array (OIB/MORB) and continental crust, also suggesting crustal assimilation (Figure 9). Crustal contamination is also evidenced by the presence of xenocrystic zircons in the samples (Figure 4A). To summary, both fractional crystallization and crustal contamination occurred during the Cretaceous magma emplacement at/around Bailingshan.

Magma Source
Mafic magmas can be derived from the lithospheric or asthenospheric mantle (Wilson, 1989;Shellnutt, 2014). The lithospheric mantle is generally cooler and isotopically moreenriched due to its interactions with subduction-derived melts and/or fluids and its long isolation from the mantle convection underneath. In contrast, asthenospheric mantle is commonly hotter and isotopically more-depleted (Saunders et al., 1992;Hofmann, 1997). Our dolerite samples have low initial 87 Sr/ 86 Sr (0.7041 to 0.7043) and positive ε Nd (t) (+ 4.6 to + 5.1) values, plotting near the MORB field and overlap with typical OIB (Figure 7), which suggests an asthenospheric mantle source. This conclusion agrees with published work on the Huangshanxi gabbro, which also pointed to a depleted asthenospheric mantle source beneath the Eastern Tianshan Orogen (Zhang et al., 2011;Deng et al., 2015). As above-discussed, the slight LREE enrichments and distinct negative Nb, Ta and Ti anomalies of the dolerite samples were likely caused by fractional crystallization and crustal assimilation during the magma emplacement.
Since garnet has high partition coefficients for Yb (D garnet/melt = 6.6) relative to Sm (D garnet/melt = 0.25), low-degree partial melting of mantle lherzolite (with garnet residue) would strongly increase the Sm/Yb ratios (Green, 2006;Jung et al., 2006). In contrast, partial melting of mantle spinel lherzolite does not markedly change the Sm/Yb ratios, since spinel has similar partition coefficients for Sm and Yb (McKenzie and O'Nions, 1991;Kelemen et al., 1993). Therefore, mafic magma source can be effectively constrained by fractionation of Sm and Yb (Ellam, 1992;Aldanmaz et al., 2000;Zhang et al., 2015). Our dolerite dykes have slightly elevated Sm/Yb ratios (2.01 to 2.14), and all samples plot near the spinel-garnet lherzolite model curve in the partial melting model diagram (Figure 10). This suggests that the parental magma was likely generated by low-degree partial melting of a spinel-garnet lherzolite mantle source at 70 -80 km depth, which is where the spinel-to-garnet transition is located (McKenzie and O'Nions, 1991).
Slab break-off usually occurs during incipient continentcontinent collision (Davies and von Blanckenburg, 1995). Under this setting, the hot asthenospheric mantle would rise through the slab window, and partially melted the overlying metasomatized lithosphere (Rogers et al., 2002;Bonin, 2004;Ferrari, 2004). Therefore, the resulting magmatism would be focused along linear trends (Davies and von Blanckenburg, 1995). In/around our study area, no coeval linear magmatic trends are observed or ever reported, inconsistent with a slab break-off model. Considering the regional tectonic evolution, it is widely considered that the Eastern Tianshan was in a post-collisional setting since the Triassic (Wu et al., 2006;Zhang et al., 2015;Zhao et al., 2019), after the Kanggur Ocean closure and the subsequent Junggar-Central Tianshan collision (Xiao et al., 2008;Wang et al., 2014;Zhang et al., 2016;Zhao et al., 2019). Hence, we suggested that the Cretaceous dolerite was unlikely produced by slab breakoff. The dolerite samples display obvious negative Nb, Ta and Ti anomalies, significantly different from typical OIBlike, mantle plume-related Tarim flood basalts (Zhou et al., 2009). The small volume of Cretaceous Eastern Tianshan mafic rocks is also inconsistent with a large igneous province (LIP) origin (Campbell and Griffiths, 1990;Bryan and Ernst, 2008). Furthermore, no geophysical evidence is available to show the presence of mantle plume beneath the Eastern Tianshan. Hence, we considered that the dolerite was unlikely to be mantle plume-related. The continental rift model involves narrow and long tectonic depressions in the lithosphere, leading to partial melting of the upwelling asthenospheric mantle (Corti, 2009;Thybo and Nielsen, 2009).  Hofmann (1997), and data for primitive mantle (Sun and McDonough, 1989) and average continental crust (Rudnick and Gao, 2014) are also shown for comparison.
Key features of this model include an elongated topographic trough and Moho shallowing (Thybo and Nielsen, 2009). However, graben structures are absent in the Eastern Tianshan, suggesting that a continental rift scenario was unlikely either.
Decompression through lithospheric delamination, accompanied by crustal extension, is another possible mechanism to explain the melting of upwelling asthenosphere (Marotta et al., 1998;McGee and Smith, 2016). Jull and Kelemen (2001) reported that the subduction-modified lower lithospheric mantle is denser than the normal upper lithospheric mantle. This makes the lower lithospheric mantle to sink into the underlying asthenosphere, and the latter rises up to take its place (Kay and Kay, 1993;Xu et al., 2002). In the Eastern Tianshan, many recent works documented that Carboniferous arc-related igneous rocks were extensively developed due to the subduction of Kanggur ocean basin (Xiao et al., 2008;Wang et al., 2014;Luo et al., 2016;Zhang et al., 2016;Zhao et al., 2019). This process has likely added oceanic materials into the deep lithospheric mantle beneath the Eastern Tianshan. After that, the high-density lithospheric root was probably removed, resulting in asthenospheric mantle upwelling and the Bailingshan dolerite dyke emplacement. This suggestion is supported by the intraplate tectonic classification in the Zr vs. Zr/Y diagram (Figure 11). Zhao et al. (2019) suggested that the Triassic (ca. 235 Ma) felsic magmatism was derived from partial melting of thickened juvenile lower crust, and Zhang et al. (2017) reported that the Duotoushan adakitic dacite porphyry was emplaced via delamination process at around 197 Ma. These FIGURE 11 | Zr vs. Zr/Y diagram for the dolerite dyke (modified after Pearce and Norry, 1979).
Frontiers in Earth Science | www.frontiersin.org observations indicate an Early Jurassic tectonic transition from post-collisional to intraplate setting in the Eastern Tianshan. Together with our new newly data, we proposed that the removal of the high-density lithospheric root may have continued to the Cretaceous.
To summarize, we have constructed a modified tectonic model to explain the Cretaceous intraplate magmatism in the Eastern Tianshan (Figure 12). In our model, the lower lithospheric mantle was modified by oceanic crustal input from the Late Paleozoic subduction, which increased its density. The density imbalance with the less-dense asthenospheric mantle beneath may have caused a small-scale lithospheric root removal. Consequently, the asthenospheric mantle upwelled and partially melted, and formed the parental melt of the Bailingshan dolerite dykes.

CONCLUSION
(1) Zircon U-Pb dating of the Bailingshan dolerite dykes from the Eastern Tianshan yielded an Early Cretaceous age (129.7 ± 1.4 Ma).
(2) Parental magma of the dolerite dykes was likely derived from low-degree partial melting of the asthenospheric mantle in the spinel-garnet stability field, and undergone fractionation and crustal assimilation during its ascent.
(3) Formation of the Cretaceous Eastern Tianshan dolerite dykes was likely in an intraplate extension setting, and related to the sinking of dense, subduction-modified lower lithospheric mantle.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author/s.

AUTHOR CONTRIBUTIONS
WZ and XD conceived this research. WZ wrotes the manuscript and prepares the figures. BT, LP, and XJ reviews and supervises the manuscript. The co-authors XD are involved in the discussion of the manuscript. All authors finally approved the manuscript and thus agreed to be accountable for this work.