Petrogenesis and Tectonic Significance of the ~276 Ma Baixintan Ni-Cu Ore-Bearing Mafic-Ultramafic Intrusion in the Eastern Tianshan Orogenic Belt, NW China

: The Baixintan mafic-ultramafic intrusion in the Dananhu-Tousuquan arc of the Eastern Tianshan orogenic belt is composed of lherzolite, olivine gabbro, and gabbro. Olivine gabbros contain zircon grains with a U-Pb age of 276.8 ± 1.1 Ma, similar to the ages of other Early Permian Ni-Cu ore-bearing intrusions in the region. The alkaline-silica diagrams, AFM diagram, together with the Ni/Cu-Pd/Ir diagram, indicate that the parental magmas for the Baixintan intrusion were likely high-Mg tholeiitic basaltic in composition. The Cu/Pd ratios, the relatively depleted PGEs and the correlations between them demonstrate that the parental magmas had already experienced sulfide segregation. The lower CaO content in pyroxenites compared with the Duke Island Alas-kan-type intrusion and the composition of spinels imply that Baixintan is not an Alaskan-type intrusion. By comparing the Baixintan intrusion with other specific mafic-ultramafic intrusions, this paper considers that the mantle source of the Baixintan intrusion is metasomatized by subduction slab-derived fluids’ components, which gives rise to the negative anomalies of Nb, Ti, and Ta elements. Nb/Yb-Th/Yb, Nb/Yb-TiO 2 /Yb, and Th N -Nb N plots show that the Baixintan intrusion was emplaced in a back-arc spreading environment and may be related to a mantle plume.


Introduction
The Eastern Tianshan orogenic belt is located in the southern margin of the Paleo-Asian Ocean. It is a convergence region between the Siberian Craton and the Tarim Craton, which has experienced a very complex process [1]. A series of magmatic Ni-Cu sulfide deposits occur in the Eastern Tianshan region, which is an important Ni-Cu resource base in China. All of these deposits are hosted in mafic-ultramafic intrusions. Most of the intrusions are located along the Jueluotage Belt, such as the Tudun, Huangshan, and Tulaergen mafic-ultramafic intrusions. A few of them are located in the Middle Tianshan Terrane, including the Tianyu and Baishiquan intrusions. It was demonstrated that the ages of these mafic-ultramafic intrusions range from 269 Ma to 300 Ma [2][3][4][5][6][7][8][9][10], indicating that most of the intrusions formed in the Early Permian. The parental magma of these mafic-ultramafic intrusions is tholeiitic, derived from a subducted slab-metasomatized mantle source, which has experienced different degrees of crustal contamination and sulfide segregation [11][12][13][14][15][16][17].

Petrography and Mineralization
To the north of the Dacaotan Fault, the Baixintan intrusion occurs in the Middle-Upper Ordovician Qiaganbulake Group, which consists of basalt, andesite, dacite, vocanic breccia, and tuff ( Figure 2a). To the south of the Dacaotan Fault, the Lower Jurassic Badaowan group is exposed which consists of conglomerate, sandstone, and siltstone. The intrusion has a gourd shape, with an exposed area of 1.5km 2 , and a length of 2800 m. Monzonitic granite and granodiorite of Late Paleozoic age occur on the southern side of the Baixintan intrusion. The zircon U-Pb ages of granodiorites range from 358.14 Ma to 367. 85 Ma [64]. The Baixintan intrusion is well-differentiated from north to south. It consists of lherzolite, olivine gabbro, and gabbro. The contacts between the gabbro, olivine gabbro and lherzolite are generally gradational. The ore bodies are lentoid, and are mainly hosted within lherzolite and gabbro (Figure 2b,c). Disseminated and patchy sulfides are the most important types of mineralization, massive and semi-massive sulfides are rare and only present in the basal contacts of the intrusion with country rocks [10]. Lherzolite is located in the southwest of the intrusion. The lherzolite is composed of 60% olivine, 20% clinopyroxene, 15% orthopyroxene, and 3% plagioclase (Figure 3a,c).

SHRIMP Zircon Analyses
A fresh olivine gabbro sample (BXT-B18) from the Baixintan intrusion was selected for zircon U-Pb dating, and the sample location is shown in Figure 2a. Zircons were separated using conventional heavy liquid and magnetic techniques at the Langfang Regional Geological Survey Lab in the Hebei Province. Zircons were hand-picked by using a binocular microscope and were selected by choosing the clearest, crack and inclusion-free grains. All zircon grains were inspected by backscattered scanning electron (BSE) microscope (Carl Zeiss AG, Jena, Germany) and cathodoluminescene (CL) microscopy (Gatan lnc, Las Vegas, NV, USA), to investigate their internal microstructures and check the position of the analytical spot with respect to the microstructures. U-Pb dating was carried out using the SHRIMP II ion microprobe at the Beijing SHRIMP Center, CAGS (Australian Scientific Instruments Pty Ltd, Canberra, Australia). The analytical procedure for zircon was similar to that described by [66,67]. The intensity of the primary 2-O ion beam was 4-6 nA. The primary beam size was ~30 µm, and each analytical site was rasterized for 2-3 min before analysis. Five scans through the relevant mass stations were made for each analysis. The standard zircon sample M257 (U = 840ppm) was measured to calibrate U concentrations [68], and the standard zircon TEMORA (416.8Ma) was used for the isotopic fractionation correction [69]. A correction for common lead was made by measuring the 204 Pb amount. The common lead composition was calculated at 207 Pb/ 206 Pb measured ages, using the Stacey and Kramers model [70]. The data processing was carried out using the SQUID 1.03d and ISOPLOT 3.75 programs [71,72]. Uncertainties for individual analyses are quoted at the 1σ level, whereas errors for weighted mean ages are quoted at 95% confidence.

Whole-Rock Major and Trace Elements Analyses
The major elements and trace elements in whole rocks were analyzed at the Key Laboratory for the Study of Focused Magmatism and Giant Ore Deposit, MNR in Xi'an, China. Major elements were analyzed by using an Axio-type XRF instrument. After one hour of heating at 1000 °C, the loss-on-ignition (LOI) was determined by the weight loss of a powdered sample. The sample powders (0.7 g), were fused with 5.200 ± 0.001 g of lithium tetraborate (Li2B4O7), 0.400 g ± 0.001 g lithium fluoride (LiF) and 0.30 ± 0.01 g ammonium nitrate (NH4NO3). The mixtures were dissolved in 1 mL lithium bromide solution at 1150-1250 °C for 10-15 min. The precision of the major elements was better than 2%. The accuracy and reproducibility were monitored by the Chinese national standard GBW07104 (andesite), and the standard values for GBW07104 are given in [73]. The standard deviation of the standard is better than 1%.
The trace elements were determined by using an ELEMENT inductively coupled plasma mass spectrometer (ICP-MS) by Thermo Fisher. Fifty-milligram powders of samples were dissolved in 1 mL of HF that was mixed with 0.5 mL of HNO3 in a screw-cap capsule at 185 ± 5 °C for 24 h. The sample solutions were dissolved again in 0.5 mL HNO3 in the capsules. After the solutions were dried, the previous step was repeated. The solutions were dissolved in 5 mL HNO3 again in a capsule at 130 °C for 3 h. Finally, the solutions were diluted with H2O to 50 mL for trace element analysis. The Chinese national standard GBW07105 (basalt) was used to monitor accuracy and reproducibility, and the standard values for GBW07105 are from [73]. The standard deviation of the standard is better than 3%. The precision is better than 5%.

Whole-Rock Sr-Nd Isotopes Analyses
Strontium and Nd in whole-rock samples were separated by using standard ion-exchange techniques [74]. Strontium and Nd isotopic analyses were performed on a Thermo Fisher Scientific Multi-receiving inductively coupled plasma mass spectrometer (MC-ICP-MS) at the National Research Center for Geoanalysis in Beijing, China (Thermo Fisher Scientific lnc, Waltham, MA, USA). The main analysis process is as follows: 0.25 g whole-rock powders were digested in sealed Teflon bombs with a mixture of 0.5 mL HNO3 and 1.5 mL HF. The sealed bombs were kept in an oven at 190 °C for 48 h. The decomposed samples were then dried at 160 °C, followed by adding 3 mL 1:1 HNO3. The solution was sealed and dissolved again at 150 °C for 6 h. An appropriate amount of aliquot was centrifuged, and the obtained supernatant was dried, followed by adjusting the pH. The solution containing Sr and Nd was separated by using SR specific resin and LN specific resin, respectively. The 87

Whole-Rock Ni, Cu, S and Platinum-Group Elements Analyses
Whole-rock Ni and Cu contents were determined using the Varian ICP735-ES inductively coupled plasma emission spectrometer and the Perkin Elmer Elan 9000 inductively coupled plasma mass spectrometer (ICP-MS) at the ALS Chemex (Guangzhou, China) Co. Ltd., respectively. The analytical precisions are ~7% and ~4% for Ni and Cu, respectively. Whole-rock S contents were measured by using a gravimetric method and IR absorption with the detection limits ~0.01%. The analytical precisions are ~4% of amount present for S. The values of standards GXR-1 and GXR-3 are from [75] and given in Table A4.
The concentrations of platinum-group elements (PGE) were determined by nickel sulfide fire assay-inductively coupled plasma mass spectrometry at the National Research Center of Geoanalysis in Beijing, China (PerkinElmer, Waltham, MA, USA). The precision and accuracy were monitored via analyses of the standards GBW07290 and GBW07291, and the recommended values and analytical values are from [73]. The analytical procedure was similar to that described by [76,77].

SHRIMP Analyses
The zircons separated from olivine gabbro are subhedral with a length of 60-140 µm and an elongation (length-to-width) ratios of 1-1.5, respectively. The grains are mostly homogeneous un-zoned and well defined with sharp tips (Figure 4a), and are similar with the igneous zircons described by [78]. The Th contents of the selected grains range from 544 ppm to2709 ppm, and those of U range from 294 ppm to 1455 ppm. The relatively high Th/U ratios (0.95-2.38) of the zircons indicate their magmatic origin [79]. Among the zircon grains from sample BXT-B18 (Figure 4a), twenty-one grains (No.  give concordant data within the error (Figure 4b), with 206 Pb/ 238 U age ranging from 272.4 ± 1.9 to 280.3 ± 2.3 Ma. (Table A1). The 21 analyses yield a mean 206 Pb/ 238 U age of 276.8 ± 1.1 Ma (95% confidence level, 2σ, MSWD = 1.8, probability = 0.020, see Supplementary Materials, Table S1. SHRIMP zircon U-Pb age data of sample BXT-B18 from the Baixintan intrusion) (Figure 4b).

Ni, Cu, S and PGE Concentration
The analytical results for Ni, Cu, S, and PGE are listed in Table A4. The PGE contents of rocks from the Baixintan intrusion are relatively low: the contents of ΣPGE range from 0.65 ppb to 4.85 ppb, with an average of 2.50 ppb, which is obviously depleted relative to the primitive mantle (ΣPGE = 23.5 ppb). The ΣPGEs of mafic rocks (gabbros) range from 0.65 ppb to 0.81 ppb, with an average of 0.71 ppb; the ΣPGEs of ultramafic rocks (lherzolites) range from 1.67 ppb to 4.85 ppb, with an average of 3.40 ppb. The mantle-normalized PGE patterns for the rocks from the Baixintan intrusion have an overall trough shape (Figure 7), showing that the rocks are depleted in PGEs relative to Ni and Cu, which are also characterized by depletions in Os, Ir, and Ru relative to Rh, Pt, and Pd. The Ni/Cu and Pd/Ir ratios of the Baixintan intrusions are 1.91-15.01 and 6.56-11.17, respectively. The Cu/Pd ratios range from 0.63 × 10 5 to 4.37 × 10 5 .

Parental Magma and Sulfide Segregation
As shown in Figure 8, the MgO and FeO contents show negative relationships with the SiO2 content (Figure 8a,b), which is in agreement with the fractionation of olivine and chromite; the CaO and Al2O3 contents show positive relationships with the SiO2 content ( Figure 8c,d), which is consistent with the fractionation of clinopyroxene and plagioclase. The TiO2 and the total alkali (K2O + Na2O) contents are positively related to the SiO2 content (Figure 8e,f). In the alkaline-silica diagram, all the samples plot in the subalkaline field (Figure 8f). In the AFM (Na2O + K2O-FeO-MgO) diagram, the ultramafic rocks show a tholeiitic trend, and the mafic rocks are closer to the calc-alkaline trend ( Figure 9).  In a plot of Ni/Cu (1.91-15.01) and Pd/Ir (6.56-11.17) ratios, the majority of the mafic-ultramafic rocks from the Baixintan intrusion fall into the area of layered intrusions, close to the high-Mg basalts region (Figure 10), indicating that the parental magmas for the Baixintan intrusion were likely high-Mg basalt in composition. The rocks from the Baixintan intrusion have relatively consistent Pd/Ir ratios but variable Ni/Cu ratios. On average, the ultramafic rocks seem to have higher Ni/Cu ratios than the mafic rocks, and are consistent with the "olivine removal" trend ( Figure 10). Moreover, Feng et al. [10] found that the parental magma for the Baixintan intrusion was depleted in PGE, and the Ir and Pd contents in the parental magma were estimated at ~0.0022 ppb and ~0.18 ppb, respectively. Figure 10. Ni/Cu-Pd/Ir diagram of the rocks from the Baixintan intrusion (after [93]). Data are from this paper and [94].
The host magma becoming saturated in sulfide and segregating immiscible sulfide is one of the key aspects in the genesis of a magmatic sulfide ore deposit [95]. Because the elemental compatibilities of PGEs in rock-forming minerals are quite different, we can use the correlations of the PGEs and the Cu/Pd ratios to reveal the sulfide segregation process of the parental magma [96].
The estimated bulk solid-liquid partition coefficients for PGE, Ni, and Cu in a very differentiated komatiitic basalt are Ir 6.6, Ru 4.5, Pt 0.53, Pd 0.09, Ni 6.2, Cu, 0.01 [97]. Iridium is compatible in clinopyroxene, but incompatible in olivine; Rh is compatible in olivine, but incompatible in both clinopyroxene and orthopyroxene; Pt is compatible in orthopyroxene, but incompatible in olivine and clinopyroxene; Pd and Cu are incompatible elements in all of the minerals mentioned above [98][99][100]. Hence, if no sulfide saturation occurs, the PGEs will become progressively fractionated [93], the fractional crystallization of the mafic minerals can give rise to the decrease of the Ni, Ir, and Rh contents, but increases the Pt, Pd, and Cu contents in the residual magma. Consequently, the Pd/Ir ratios of the residual magma increase significantly, while the Cu/Pd ratios remain stable [96]. Once sulfide saturation occurs, most of the PGEs will tend to enter the sulfide [93], which causes a strong loss of PGEs in the parental magma, while the contents of Ni and Cu decrease slightly. Consequently, the Pd/Ir ratios in the residual magma remain stable, while the Cu/Pd ratios increase significantly [96].
With the decrease of Ir, the contents of Rh, Pt, and Pd in the rocks from the Baixintan intrusion decrease rapidly (Figure 11a-c). A positive correlation between Pt and Pd can be seen in the rocks (Figure 11d). The Cu/Pd ratios of rocks from the Baixintan intrusion range from 0.63 × 10 5 to 4.37 × 10 5 (Table A4), which are distinctly higher than that of the primitive mantle (10 3 -10 4 , [95]). The Pd/Ir ratios of the samples do not vary a lot (Figure 10), ranging from 6.56 to 11.17 (Table A4). The positive correlations between PGEs and the higher Cu/Pd ratios of samples from the Baixintan intrusion indicate that the parental magma has experienced a sulfide segregation process. Ruan et al. [101] considered that the assimilation of crustal Si and S components played important roles on sulfide segregation in the parental magma of the Baixintan intrusion.

Tectonic Settings and Magma Source
In this part, this paper will estimate the tectonic setting and the magma source of the Baixintan intrusion by comparing it with other two specific mafic-ultramafic intrusions, such as the Limahe and Duke Island intrusions, which are generated in different acknowledged tectonic settings. The Limahe intrusion is located in the Emeishan Large Igneous Province (ELIP), SW China. The ELIP is composed of huge volumes of Emeishan flood basalts, numerous mafic-ultramafic intrusions, granites, and syenites [102]. The genetic relationship between the mafic-ultramafic intrusions emplaced in the ELIP and the mantle plume has been discussed in previous studies [87,[103][104][105][106][107][108][109][110][111]. The Duke Island intrusion, located in Southeastern Alaska, is generally regarded as a zoned Alaskan-type intrusion [112][113][114][115].
Some researchers held the view that the Ni-Cu ore-bearing mafic-ultramafic intrusions (or some of them) located in the Eastern Tianshan orogenic belt are Alaskan-type intrusions which are formed in the island arc or active continental margin because of their zoned bodies and some arc-like geochemical characteristics, such as relatively high large ion lithophile elements (LILE) and low high field strength elements (HFSE), negative anomalies of Nb and Ta [5,[34][35][36]. Most of the classic Alaskan-type intrusions show the following characteristic features [113]: 1. rock types that are mostly dunite, wehrlite, olivine clinopyroxenite, magnetite-rich clinopyroxenite, and gabbro; 2. crude concentric zoning in some of the larger bodies with dunite in the central-most parts and gabbro in the outer-most parts; 3. principal mineral associations in the ultramafic rocks consisting of olivine, diopside, magnetite, and hornblende, orthopyroxene and plagioclase are characteristically absent. The mineral assemblage of Alaskan-type intrusions means that the rocks ought to be depleted in SiO2 and enriched in CaO, especially for the pyroxenites [116]. In the CaO-SiO2 discrimination diagram (Figure 12), samples from the Duke Island intrusion are invariably plotted into the Alaskan field. However, all the samples from the Baixintan intrusion, as well as the Limahe intrusion are plotted into the layered intrusion field, showing lower CaO contents. As we can see in Figure 13, the spinel compositions of the Baixintan and Limahe intrusions are quite different from that of Alaskan-type zoned ultramafic intrusions. Samples from the Limahe intrusion falls into the area of "Subvolcanic intrusions related to flood basalts", samples from the Baixintan intrusion are plotted into the "Layered intrusions", and most of them are precisely distributed along the right margin of "subvolcanic intrusions related to flood basalts".
Based on the evidences from the compositions of pyroxenite and spinel, we consider that the Baixintan intrusion is not an Alaskan-type intrusion.

Figure 12.
SiO2-CaO diagram of pyroxenites from the Baixintan, Limahe, and Duke Island intrusions (after [116]. 1 = Alaskan field, 2 = Ophiolitic field, and 3 = Layered intrusion field. Data source of Baixintan: [84]; data sources of Duke Island: [114,115]; data sources of Limahe: [80]. It was demonstrated that contamination by continental crust or lithosphere, or a metasomatized source by the subduction-related fluids can give rise to negative Nb, Ta, and Ti anomalies [121]. By contrast, uncontaminated plume-generated basaltic rocks will normally have flat REE patterns or LREE-enriched patterns and lack negative Nb, Ta, and Ti anomalies [122][123][124]. Compared with the plume-derived Limahe intrusion, samples from the Baixintan intrusion have strongly negative Nb, weakly negative Ta and Ti anomalies ( Figure 5a) and relatively flat REE patterns (Figure 5b). Feng et al. [10] stated that the Baixintan mafic-ultramafic intrusive rocks are characterized by moderate light REE enrichments relative to heavy REE and pronounced negative Nb-Ta anomalies. The prominent negative Nb-Ta anomalies in the Baixintan intrusion were also found by the authors of [84,125]. However, most samples from the Baixintan intrusion are plotted into the ocean island basalt (OIB) area in the 87 Sr /86 Sr(t)-εNd(t) diagram ( Figure 6), implying that they are derived from a more enriched mantle source compared to MORB, which may be explained by mantle plume material. The lower 87 Sr/ 86 Sr(t) ratios and higher εNd(t) values compared with the Limahe intrusion also suggest the magma of the Baixintan intrusion may have experienced a lower degree of crustal contamination (Figure 6). In consideration of a low degree of crustal contamination suggested by the Sr and Nd isotopes, this paper tends to attribute the negative Nb, Ta, and Ti anomalies of the Baixintan intrusion to its metasomatized mantle source by subduction-related fluids. This is also supported in the Ba/La-Th/Yb diagram ( Figure 14). Rocks produced in oceanic subduction systems can be divided into two trends [126]. The high Ba/La trend is thought to be related to slab-derived fluids, and the high Th/Yb trend is attributed to either melting of subducted sediment or bulk assimilation of sedimentary material, intercalated within the volcanic pile, during magma ascent through the arc crust [126]. In the Ba/La-Th/Yb diagram (Figure 14), the Limahe intrusion displays a high Th/Yb ratio trend, while the Baixintan, similar to the Duke Island intrusion, follows a high Ba/La ratio trend. These features further indicate that: 1. the assimilation and contamination of crustal materials played a key role during the formation of the Limahe intrusion, which is consistent with the information reflected by the Sr and Nd isotopic characteristics ( Figure 13); 2. the Baixintan intrusion may derive from a mantle source that is metasomatized by slab-derived fluids' components.

Figure 14.
Ba/La-Th/Yb diagram (after [126]). Data sources of Baixintan: this paper, [84]; data sources of Limahe: [80,81]; data source of Duke Island: [115]; data sources of the Tarim flood basalts field (51 samples): [88,90,127,128]. Furthermore, previous studies have shown that subduction slab-derived fluids have made a contribution to the source of intrusions located in the Eastern Tianshan orogenic belt [11][12][13][14][15][16][17], but the contribution of subduction slab-derived fluids to the source of the Pobei and Hongshishan intrusions in the Beishan Rift was excluded [27,49,50,129,130]. It means that the scope of impacts on the intrusions by the subduction event is limited in the region. The impact of subduction seems to be weakened gradually from the Eastern Tianshan orogenic belt to the Beishan Rift.

A Proposed Plume-Related Back-arc Spreading Tectonic Setting for the Baixintan Intrusion
The NbN-ThN discrimination diagram [131] displays possible tectonic settings for the Baixintan and the other two different types of mafic-ultramafic intrusions discussed previously (Figure 15a,b). In Figure 15a, most of the samples plot into the low-Ti island arc tholeiite (IAT) field, showing similar geochemical features with island arc tholeiites. Based on the discussion before, we consider that these arc-like features are caused by its metasomatized mantle source. In Figure 15b, nearly all of the samples from the Baixintan intrusion are limited to the Back-arc A field, implying a back-arc setting of Baixintan and an input of subduction components. In contrast, a large proportion of samples from the Limahe intrusion fall into the alkaline OIB and the overlap between OIB and P-MORB field (Figure 15a), which is consistent with its plume source. Moreover, there seems to be a trend of increasing the OIB-type component from the Baixintan to the Limahe intrusion (Figure 15b). Samples from the Duke Island intrusion fall into supra-subduction zone (Figure 15a), which is indicative of a nascent forearc convergent setting (Figure 15b). The Early Permian Tarim large igneous province (TLIP) in North West China consists of two magmatic phases [132]: the ~290 Ma magmatic phase is characterized by bimodal volcanic rocks that consist of rhyolites and basalts, which occur within the Tarim Craton; the ~280 Ma magmatic phase is mainly composed of intrusive rocks and mafic dikes, which mainly appear on the northern border of the Tarim Craton and inside of the CAOB [25]. Recently, several zircon U-Pb ages of the Baixintan intrusion have been published. Zircon U-Pb LA-ICP-MS age of plagioclase-bearing lherzolite in the Baixintan intrusion is 277.9 ± 2.6 Ma [125], zircon U-Pb LA-ICP-MS age of olivine gabbro is 287.3 ± 3.1 Ma [84], and a U-Pb age of olivine norite (olivine gabbro in this paper) is 286.0 ± 1.6 Ma [10]. In this study, zircons separated from olivine gabbro yielded a SHRIMP U-Pb age of 276.8 Ma (Figure 4b), which is consistent with [125], but ~10Ma younger than the age reported by the authors of [10,84]. Even so, the ages obtained indicate that the Baixintan intrusion located in the Eastern Tianshan belt formed in the Early Permian, which is consistent with the later magmatic phase of the TLIP. Other ore-bearing intrusions in the Eastern Tianshan belt and the Beishan Rift formed almost simultaneously (Figure 1b, Table 1). Furthermore, the mafic-ultramafic intrusions are normally associated with A-type granites spatially and temporally, suggesting that these magmatic events occurred during an extensional regime, possibly related to a mantle plume event [25]. In the Nb/Yb-Th/Yb discrimination diagram (Figure 16a), the oceanic basalts (intra-plate islands, plume-distal ocean ridges and oceanic plateau) plot predominantly within the MORB-OIB array (shaded), while crustally contaminated basalts and alkalic basalts containing a large recycled component mainly plot above the MORB-OIB array, or on a vector at a steep angle to the array [140]. As Figure 16a shows, both the Limahe and Baixintan intrusions plot above the MORB-OIB array, indicating crustal contamination and a metasomatized mantle source of them, respectively. However, unlike the Limahe intrusion showing an affinity with the OIB components, the samples of Baixintan have lower Th/Yb ratio, which are closer to the E-MORB components. There is a similar case in the Nb/Yb -TiO2/Yb diagram (Figure 16b), where the samples from the Limahe intrusion mostly plot in the OIB array due to its mantle plume source, while more than half of the samples from the Baixintan intrusion plot in the E-MORB field inside of the MORB array. Figure 16. Nb/Yb-Th/Yb diagram (a) and Nb/Yb-TiO2/Yb diagram (b) for the Baixintan, Limahe, and Duke Island intrusions (after [140]). Data sources are the same as in Figure 14.
It is generally believed that N-MORB and OIB are two separate end members, while E-MORB is mixed by N-MORB and OIB in different degrees [141][142][143], which is considered to be the result of plume and Mid-Ocean Ridge interaction [144][145][146]. It is suggested that E-MORB formed at plume-proximal ridge settings [140]. Similar geochemical characteristics of the Baixintan intrusion to those of E-MORB imply that it may be generated from a plume-related setting.
In consideration of geochronology and geochemical characteristics of the Baixintan intrusion, we suggest that the Baixintan intrusion may be generated from a back-arc spreading environment and related to a mantle plume.

Conceptional Genetic Model
Based on the nature of the magma source and the tectonic setting of the Baixintan intrusion reflected by geochemical characteristics, this paper proposes a geological evolutionary process for the Eastern Tianshan region: before the Early Permian, the subduction of the Paleo-Asian Ocean had altered the lithospheric mantle beneath the Eastern Tianshan region, making it contain some subduction slab-derived fluids' components.
However, the subduction event did not impact the sources of intrusions located in the Beishan Rift on the south (Figure 17a). By the Early Permian, the Paleo-Asian Ocean had closed, and the Eastern Tianshan region was then in a back-arc spreading setting. By ~280 Ma, the mantle plume upwelled, and the magma derived from the mantle plume was influenced by lithospheric mantle containing subduction components. At last, the magma intruded into the crust, and ore-bearing mafic-ultramafic intrusions with the arc-like geochemical characteristics formed in the Eastern Tianshan orogenic belt ( Figure  17b).

Conclusions
The ~276 Ma Baixintan Ni-Cu ore-bearing mafic-ultramafic intrusion was emplaced in a back-arc spreading environment, and may be related to a mantle plume. The parental magmas for the Baixintan intrusion were likely high-Mg tholeiitic basaltic in composition and had already experienced sulfide segregation. The magma derived from the mantle plume was influenced by lithospheric mantle containing subduction slab-derived fluids' components, which give rise to the negative anomalies of Nb, Ta, and Ti elements in the Baixintan intrusion.

Supplementary
Materials: The following are available online at www.mdpi.com/2075-163X/11/4/348/s1, Table S1. SHRIMP zircon U-Pb age data of sample BXT-B18 from the Baixintan intrusion.  Data Availability Statement: .Data generated in this study is available in the article.

Acknowledgments:
We are very grateful to the Editor and the reviewers for their great help in improving this manuscript and constructive opinions.

Conflicts of Interest:
The authors declare no conflicts of interest.