Petrogenesis and Tectonic Implications of Late Carboniferous Intrusions in the Tuwu-Yandong Porphyry Cu Belt (NW China): Constraints from Geochronology, Geochemistry and Sr–Nd–Hf Isotopes

:

In this contribution, we present new zircon LA-ICP-MS U-Pb dating and trace element geochemistry, whole-rock geochemical data, and Sr-Nd isotope analyses, as well as in situ zircon Hf isotopic compositions of the Late Carboniferous intrusions (granodiorite porphyry, gabbro and granite porphyry), combined with the available data of the Carboniferous magmatic rocks, to better constrain the tectonic settings of the magmatism and mineralization of the Tuwu-Yandong porphyry copper metallogenic belt.  [4]). (b) Tectonic framework and distribution of deposits in the eastern Tianshan (modified from [28]).
In this contribution, we present new zircon LA-ICP-MS U-Pb dating and trace element geochemistry, whole-rock geochemical data, and Sr-Nd isotope analyses, as well as in situ zircon Hf isotopic compositions of the Late Carboniferous intrusions (granodiorite porphyry, gabbro and granite porphyry), combined with the available data of the Carboniferous magmatic rocks, to better constrain the tectonic settings of the magmatism and mineralization of the Tuwu-Yandong porphyry copper metallogenic belt.

Geological Setting
The eastern Tianshan orogenic belt is a typical complex collage of island arc assemblages, remnants of oceanic crust, accretionary wedges, and continental fragments [29]. Generally, it may be divided from north to south into the Haerlike belt, Jueluotage belt, and central Tianshan block, separated by the regional-scale Qincheng and Aqikekuduke faults, respectively ( Figure 1b) [4,8]. The Haerlike belt contains Ordovician-Carboniferous volcanic rocks, granites, and Late Paleozoic mafic-ultramafic complexes but only hosts a few porphyry Cu and Au prospects [4]. The central Tianshan block is composed mainly of a Precambrian crystalline basement and hosts several volcanic Fe deposits (e.g., Tianhu and Weiya) and the giant Caixiashan Pb-Zn deposit [30]. The Jueluotage belt is the most important Cu, Fe, and Au metallogenetic unit in eastern Tianshan [31] and can be further divided into three sub-tectonic domains by the Kanggur and Yamansu faults, namely the  [4]). (b) Tectonic framework and distribution of deposits in the eastern Tianshan (modified from [28]).

Sampling
Large volumes of Carboniferous granitoid intrusions with minor mafic rocks (e.g., gabbro) are extensively distributed in the belt ( Figure 2). One mafic intrusion (gabbro) and two felsic intrusions (granodiorite porphyry and granite porphyry) were selected for further geochronological and geochemical analyses. These intrusions are undeformed and

Analytical Methods
The ten least altered samples were selected for whole-rock major and trace element as well as Sr-Nd isotopic analyses. Three samples were chosen for LA-ICP-MS zircon U-Pb isotopic dating, trace element geochemistry, and in situ Hf isotopic analyses.
The zircon grains were separated by routine physical elutriation, heavy liquid, and magnetic techniques and carefully hand-picked under a stereoscopic microscope. Subsequently, they were mounted on epoxy and polished to expose the crystal cross-sections. The documentation of the internal structures and selection of potential target sites for the U-Pb dating of all the mounted zircons were based on transmitted and reflected light photomicrographs, as well as cathodoluminescence (CL) images. Zircon U-Pb dating and trace element analyses were simultaneously conducted using an Agilent 7500 a inductively coupled plasma mass spectrometer (ICP-MS) coupled with a GeoLas 2005 at the Tianjin Institute of Geology and Mineral Resources. The analytical procedures were described by [40]. Laser ablation was operated at a constant energy of 60 mJ, with a repetition rate of 4 Hz and a spot diameter of 32 µm. NIST SRM 610 and zircons 91500, GJ-1, were used as external standards. Zircon 91500 was analyzed twice for every six analyses to calibrate the isotope fractionation. NIST SRM 610 was analyzed once every eight analyses to correct the instrumental drift and mass discrimination of the trace element analysis. Errors in individual analyses were cited at the 1σ level, and the weighted mean 206 Pb/ 238 U ages were quoted at the 95% confidence level. The adjustment of background and ablation signals, time drift correction, and quantitative calibration were performed using ICPMSDataCal software [40]. Concordia diagrams and weighted mean calculations were determined using Isoplot 3.71 [41]. Zircon Ce anomalies were calculated using the method based on the lattice strain model [42].
In situ Hf isotope analyses were undertaken on the adjacent spots used for the LA-ICP-MS zircon U-Pb dating in order to match the Hf isotope data with the U-Pb ages using a Neptune MC-ICP-MS and New Wave UP 213 ultraviolet LA-MC-ICP-MS at the National Research Center for Geoanalysis, Beijing, China. During the analyses, helium was used as the carrier gas. Based on the zircon size, the stationary beam spot size was set to either 55 or 40 µm. GJ1 international standard zircon samples were used as a reference. The weighted average of the 176 Hf/ 177 Hf of the GJ1 zircon samples was 0.282015 ± 31 (2 SD, n = 10), which is consistent with the values reported by [43]. Detailed operating conditions for the laser ablation system, the MC-ICP-MS instrument, and the analytical method are given by [44].
Whole-rock major and trace elements analyses were performed at the National Research Center for Geoanalysis, Beijing, China. The samples were chipped and powdered to approximately 200 mesh. The major elements were determined using a Philips PW 2404 X-ray fluorescence (XRF) spectrometer with a rhodium X-ray source. The testing precision was better than 1%. The sample powders for the trace element analyses were accurately weighed (25 mg) and placed into Savillex Teflon beakers within high-pressure bomb and were then digested using HF + HNO 3 + HClO 4 acid to ensure the complete dissolution of the refractory minerals. The trace elements, including rare earth elements, were determined using an Element-I plasma mass spectrometer (Finnigan-MAT Ltd. German), and the national geological standard reference samples GSR-3 and GSR-15 were used for the purpose of analytical quality control. The analytical precision for the trace elements was better than 5%, and the analytical procedures were described by [45].
Whole-rock Sr-Nd isotopic analyses were performed using a Micromass Isoprobe multi-collector ICP-MS at the National Research Center for Geoanalysis, Beijing, China, using analytical procedures described by [44]. The Sr and REE were separated using cation columns, and the Nd fractions were further separated using HDEHP-coated Kef columns. The measured 87

Zircon U-Pb Dating and Trace Element Geochemistry
All the analyzed zircons are colorless, euhedral, and prismatic, with an aspect ratio of 2:1 to 4:1. Most of them show typical magmatic oscillatory zoning in the CL images ( Figure 7). The zircon LA-ICP-MS U-Pb dating data are listed in Table 3 and graphically illustrated in Figure 8a-c, and the zircon trace element data are summarized in Table 4. The Th/U ratios of the analyzed zircons range from 0.16 to 1.4 (Figure 8d), which are higher than those of the metamorphic zircons (typically <0.1) but consistent with those of magmatic zircons [53]. The REE patterns of the analyzed zircons are characterized by HREE enrichment with positive Ce and negative Eu anomalies (Figure 9), consistent with those of magmatic zircon from igneous rocks [53,54]. Therefore, the zircon U-Pb dating results are interpreted to provide the age of magma crystallization.     x FOR PEER REVIEW 12 of 33 Ti-in-zircon thermometer and zircon Ce 4+ /Ce 3+ ratios are used to estimate the temperatures and oxidation states of silicate magmas [42,55]. The calculated Ti-in-zircon temperatures range from 642 to 763 °C (avg. 688 °C, median = 673 °C, n = 15) in the granodiorite porphyry, 697 to 825 °C (avg. 742 °C, median = 740 °C, n = 15) in the gabbro, and 665 to 813 °C (avg. 728 °C, median = 711 °C, n = 15) in the granite porphyry ( Table 4). The zircon Ce 4+ /Ce 3+ ratios of the granodiorite porphyry, gabbro, and granite porphyry vary from 55 to 254 (avg. 129, median = 112, n = 15), 23 to 229 (avg. 74, median = 40, n = 15) and 29 to 308 (avg. 100, median = 91, n = 15), respectively (Table 4).   [51]. The data for the diorite, plagiogranite porphyry, quartz albite porphyry, and quartz porphyry of the Tuwu-Yandong deposits are from [10].
Sixteen zircon grains from the granodiorite porphyry (YD-42) were analyzed and, excluding two discordant analyses (No. 02 and 04), the remaining fourteen data yielded 206 Pb/ 238 U ages ranging from 315 to 332 Ma, which form a coherent group and give a weighted mean age of 321.8 ± 3.1 Ma (MSWD = 1.6) (Figure 8a). The younger age (303 Ma) of analysis No. 02 is probably attributed to the effect of post-magmatic hydrothermal events, and the older age (359 Ma) of analysis No. 04 may indicate that the zircon grain is inherited. Among the sixteen analyses of the zircon grains from the gabbro (KBDB-5), fifteen concordant analyses yielded 206 Pb/ 238 U ages of 307-325 Ma, with a weighted mean age of 313.5 ± 1.2 Ma (MSWD = 0.54) (Figure 8b). One discordant analysis (No. 10) yielded an apparent 206 Pb/ 238 U age of 339 ± 7 Ma, which was interpreted as the age of the inherited zircon. Three of the sixteen zircon grains analyzed from the granite porphyry (LL-6) were discordant (No. 10, 13 and 15), possibly suggesting partial Pb loss. Zircon grain No. 03 and 06 yielded 206 Pb/ 238 U ages of 322 ± 5 Ma and 325 ± 5 Ma, respectively, which are similar to the previously determined U-Pb age of quartz albite porphyry (318.6 ± 3.0 Ma, [13]). The other eleven analyses fell on the concordia and yielded 206 Pb/ 238 U ages of 301 to 315 Ma, with a weighted mean age of 309.8 ± 2.5 Ma (MSWD = 0.82) (Figure 8c).

Whole-Rock Sr-Nd Isotopic Compositions
The whole-rock Rb-Sr and Sm-Nd isotope compositions are summarized in Table 6 and shown in Figure 10c (Table 6). Among these intrusive rocks, the gabbro has the highest εNd(t) values, and the granite porphyry has the lowest εNd(t) values (Figure 10d) but the oldest T DM .

Timing of the Tuwu-Yandong Belt
The precise dating of intrusive rocks may be used to constrain the timing and duration of magmatic events, which is crucially important for understanding the rock-forming process and geodynamic setting [21,62,63]. Based on our new LA-ICP-MS zircon U-Pb data, the granodiorite porphyry, gabbro, and granite porphyry from the Tuwu-Yandong belt were formed at 321.8 ± 3.1 Ma, 313.5 ± 1.2 Ma, and 309.8 ± 2.5 Ma, respectively. In recent decades, a large number of geochronological studies on metallogenic age, mineralized porphyry, and pre-and post-mineralization granitic intrusive rocks have been completed [5,10,13,14,19,20,22,[64][65][66][67]. These studies, together with this study, reveal that Carboniferous intrusive magmatism events have widely taken place in the Tuwu-Yandong belt, ranging from 348 Ma to 310 Ma (Table 7 and references therein). On the basis of these available geochronological data and their magmatic associations, three major intrusive episodes have been identified: (1) the earliest intermediate intrusive rocks (e.g., diorite or diorite porphyry) emplaced at ca. 348-338 Ma [5,10,19]; (2) the felsic intrusive rocks, some of which show adakitic features (e.g., plagiogranite porphyry or tonalite porphyry [5,22,23]; granodiorite porphyry (this study); and porphyritic granodiorite [14]), formed during ca. 335-315 Ma ( [5,10,13,14,[19][20][21][22]; this study); and 3) the latest mafic (gabbro) and felsic (e.g., K-feldspar granite, granite porphyry) intrusive rocks, showing an interrupted sequence of SiO 2 values that are similar to typical bimodal values (Figure 4a) (Table 7). Two periods (ca. 335-330 Ma and 323-315 Ma) of porphyry Cu mineralization have also been identified in the belt, according to the geochronological data on the Cu mineralization and ore-related intrusive magmatism [15]. In this study, the granodiorite porphyry may be genetically related to the second episode of porphyry Cu mineralization, while the gabbro and granite porphyry most likely formed after the Cu mineralization, which is consistent with the spatial relationship between these intrusive rocks and Cu deposits ( Figure 2).

Magma Oxidization State
It is widely accepted that highly oxidized magmas are favorable for porphyry Cu (Mo) mineralization (e.g., [3,73]). Oxidized magmas can extract more Cu (and Mo) from source rocks during melting and scavenge sulfides during ascent [74]. A high oxygen fugacity also prevents the sulfide precipitation, and metals (e.g., Cu, Mo) remain in the exsolved aqueous phase for the later porphyry Cu mineralization ( [75]).
Zircon (ZrSiO 4 ) is an exceptionally robust mineral that retains its primary chemical and isotopic compositions from the time of crystallization and provides chemical information on the parental magmas [3]. Unlike other REEs that have only +3 valency, Ce and Eu commonly have two oxidation states in terrestrial magmas, and zircon more preferentially incorporates the oxidized cations Ce 4+ (0.97 Å) and Eu 3+ (1.07 Å) into the Zr 4+ (0.84 Å) site of its structure than the reduced Ce 3+ (1.14 Å) and Eu 2+ (1.25 Å) [75]. Thus, high Ce 4+ /Ce 3+ and Eu/Eu * (also known as δEu) ratios usually reflect the high oxygen fugacity (ƒO 2 ) of the parental magmas, which were used to quantify the oxidized nature of the parental magmas associated with porphyry deposits in northern Chile [42], Tibet (Ce 4+ /Ce 3+ > 120 and δEu > 0.4 [76,77]), and Qinling [78] in China. Recent research has also demonstrated that porphyry Cu deposits of large (>4 Mt Cu) and intermediate (1.5-4 Mt Cu) sizes are associated with granitic intrusions with zircon Ce 4+ /Ce 3+ ratios of >120, whereas the ratios are 54-69 for the small porphyry deposits in the CAOB [3,79].
In this study, the zircon Ce 4+ /Ce 3+ ratios of the granodiorite porphyry (avg. 129, median = 112, n = 15) are higher than those of the gabbro (avg. 74, median = 40, n = 15) and granite porphyry (avg. 100, median = 91, n = 15), both of which are consistent with those of the granitic intrusions associated with large and medium porphyry Cu deposits in the CAOB (Figure 11a,c; [10]). The samples of the granodiorite porphyry (and plagiogranite porphyry [10]) are mainly plotted between the fayalite-magnetite-quartz (FMQ) buffer curve and the magnetite-hematite (MH) buffer curve in the temperature (T) vs. logƒO 2 diagram (Figure 11d; [74,80]), further indicating the high ƒO 2 of the parental magmas. However, the low zircon Eu/Eu * ratios (<0.4) of the granodiorite porphyry (and plagiogranite porphyry [10]) suggest that the Eu/Eu * ratios of the zircon grains may be affected by another factor in addition to the oxidation conditions of the parental magmas. Since Eu 2+ is preferentially incorporated into the Ca 2+ site of plagioclase, the crystallization of plagioclase lowers the Eu in the residual melt and results in low Eu/Eu * in any latecrystallizing phases after plagioclase [3]. Indeed, the presence of plagioclase phenocrysts in the granodiorite porphyry (Figure 3b,c) suggests that the plagioclase crystallized early and preferentially removed Eu 2+ from the melt to cause low Eu/Eu * ratios in the zircon grains. Therefore, the low Eu anomalies in the zircon is not in conflict with the high Ce 4+ /Ce 3+ in the zircon grains [3]. In addition, the lower magma temperature of the granodiorite porphyry (avg. 688 • C, median = 673 • C, n = 15; determined by the Ti-inzircon thermometry) compared to that of the gabbro (avg. 742 • C, median = 740 • C, n = 15) and granite porphyry (avg. 728 • C, median = 711 • C, n = 15) suggests that the parental magmas of the granodiorite porphyry may be water-rich and based on a lower water fugacity (ƒH 2 O) reflecting the higher magma temperature [79]. Thus, the granodiorite porphyry (321.8 ± 3.1 Ma) was likely derived from a more oxidized and hydrous magma source than the gabbro and granite porphyry, implying its Cu fertility and capacity to form medium-large porphyry Cu deposits [3]. The copper mineralization potential is further supported by the molybdenite Re-Os age of 323-317 Ma in the Tuwu-Yandong porphyry Cu belt [15,68].  [58,74]. Data on the ore-related/barren intrusions in northern Chile are from [37]. Data on the medium-large and small porphyry deposits in the CAOB are from [3]. The data on the diorite, plagiogranite porphyry, quartz albite porphyry, and quartz porphyry of the Tuwu-Yandong deposits are from [10].
The granodiorite porphyry samples are strongly enriched in LREE relative to HREE ([La/Yb]N = 6.8-9.5) (Figure 5a) and show clear depletion in Nb, Ta, and Ti and positive  [58,74]. Data on the ore-related/barren intrusions in northern Chile are from [37]. Data on the medium-large and small porphyry deposits in the CAOB are from [3]. The data on the diorite, plagiogranite porphyry, quartz albite porphyry, and quartz porphyry of the Tuwu-Yandong deposits are from [10].
The granodiorite porphyry samples are strongly enriched in LREE relative to HREE ([La/Yb] N = 6.8-9.5) (Figure 5a) and show clear depletion in Nb, Ta, and Ti and positive Ba, U, K, and Sr anomalies (Figure 5b) similar to those of most modern subduction-related magmatic rocks [88,89]. The subduction-unrelated magmas are mostly plotted within yhr MORB-OIB array in the Th/Yb versus Nb/Yb diagram (Figure 12a), whereas those of the subduction zone show a significant shift away from the mantle array. The granodiorite porphyry samples possess elevated Th/Yb ratios, indicating their subduction-related enrichment. Their high Ba (188-419 ppm) and Ba/Th (mainly >170) values and low Th (<1.5 ppm) and Th/Nb (<0.6) ratios (Figure 12c) further indicate that the parental magmas of the granodiorite porphyry may be metasomatized by slab-derived fluids [84]. In the Mg # versus SiO 2 diagram (Figure 6b), the granodiorite porphyry samples plot within the field of the subducted slab-derived adakites. Furthermore, the La/Sm ratios of the granodiorite porphyry show a positive correlation with the La contents, implying that the partial melting process was dominant in the petrogenesis (Figure 12d). These geochemical and isotopic signatures indicate that the granodiorite porphyry was most likely derived from the partial melting of a subducting oceanic crust rather than the partial melting of a delaminated lower crust or thickened mafic lower crust or the AFC process of basaltic magmas. porphyry show a positive correlation with the La contents, implying that the partial melting process was dominant in the petrogenesis (Figure 12d). These geochemical and isotopic signatures indicate that the granodiorite porphyry was most likely derived from the partial melting of a subducting oceanic crust rather than the partial melting of a delaminated lower crust or thickened mafic lower crust or the AFC process of basaltic magmas. Melts derived from the crust are characterized by Mg # values of less than 40 regardless of the degree of melting, whereas those generated from the mantle exhibit high Mg # values (greater than 40) [13,22]. In general, the reaction of slab-derived melts with overlying peridotite in the mantle wedge can result in the high Mg # values [5,52]. Therefore, we speculate that the formation of the granodiorite porphyry may involve the addition of mantle-derived components. The granodiorite porphyry samples have low ( 87 Sr/ 86 Sr)i ratios and positive εNd(t), which, in part, overlap with the mantle array in the εNd(t) versus ( 87 Sr/ 86 Sr)i diagram (Figure 10c), and positive εHf(t) values, which fall between the depleted mantle (DM) and the 0.88 Ga crustal evolution line, as well as young Hf crustal (single-stage) model ages (385-504 Ma), further reflecting the interaction between melts generated from a subducted oceanic slab and mantle melts. The petrogenesis of the granodiorite porphyry (ca. 322 Ma) is similar to that of adakitic plagiogranite porphyry (ca. 335-332 Ma, [5,22,23]) in the Tuwu-Yandong belt. Melts derived from the crust are characterized by Mg # values of less than 40 regardless of the degree of melting, whereas those generated from the mantle exhibit high Mg # values (greater than 40) [13,22]. In general, the reaction of slab-derived melts with overlying peridotite in the mantle wedge can result in the high Mg # values [5,52]. Therefore, we speculate that the formation of the granodiorite porphyry may involve the addition of mantle-derived components. The granodiorite porphyry samples have low ( 87 Sr/ 86 Sr) i ratios and positive εNd(t), which, in part, overlap with the mantle array in the εNd(t) versus ( 87 Sr/ 86 Sr) i diagram (Figure 10c), and positive εHf(t) values, which fall between the depleted mantle (DM) and the 0.88 Ga crustal evolution line, as well as young Hf crustal (single-stage) model ages (385-504 Ma), further reflecting the interaction between melts generated from a subducted oceanic slab and mantle melts. The petrogenesis of the granodiorite porphyry (ca. 322 Ma) is similar to that of adakitic plagiogranite porphyry (ca. 335-332 Ma, [5,22,23]) in the Tuwu-Yandong belt.

The High-Al Gabbro
The gabbro samples are characterized by low SiO 2 contents (49.08-52.02 wt.%) and high MgO (5.62-7.00 wt.%), Fe2O3 T (9.10-9.74 wt.%), V (209-281 ppm), Cr (107-172 ppm), and Ni  contents and Mg # values (53.3-59.9), suggesting that they are unlikely to have originated from the lower crust [92] or the mantle-derived primary magma (usually with MgO contents > 15 wt.%, Mg # > 65, Cr > 2000 ppm, and Ni > 500 ppm) [93] but rather from an evolved magma. They have low ( 87 Sr/ 86 Sr) i and high εNd(t) (6.63-7.23) and εHf(t) (11.6-15.9) values, as well as young Nd (T DM = 549-646 Ma) and Hf (T DM1 = 310-493Ma, T DM2 = 310-591 Ma) model ages, reflecting a depleted mantle source, which is supported by the ( 87 Sr/ 86 Sr) i vs. εNd(t) diagram (Figure 10c). These geochemical and isotopic signatures indicate that the gabbro was most likely derived from the partial melting of the mantle peridotite. The ratios of REE are useful criteria for constraining the composition of the mantle source and degree of partial melting [93]. Experimental studies have shown that the partition coefficients of REE are different for garnet-and spinel-facies peridotites. The HREE is commonly preferentially retained by garnet, while spinel preferentially incorporates the MREE [94]. The gabbro samples exhibit moderate [La/Yb] N (2.7-3.0) and low [Dy/Yb] N (ca. 1.1) ratios with fairly flat REE patterns, probably indicating their formation at the depth of the spinel stability field.
The gabbro samples exhibit an enrichment in LILEs (e.g., Ba, U, K, and Sr) and depletion in HFSEs (Nb, Ta, and Ti), which may be ascribed to the partial melting of a depleted mantle wedge with the addition of slab-derived components (fluids or melts) [95], which is further supported by the elevated Th/Yb ratios (Figure 12a,b) [90,91]. If the depleted mantle wedge peridotite is modified by a low level of slab-derived melts, it will produce high-Nb or Nb-enriched basaltic rocks [94], which is inconsistent with the studied gabbro, which has low Nb (2.08-3.23) abundances. The gabbro samples have high and variable Ba/Th (337-410) but low and constant Th/Nb (0.24-0.47) ratios, strongly suggesting that the subduction components were dominated by slab-derived hydrous fluids instead of sediments (Figure 12c). In addition, the gabbro samples possess a narrow La/Sm ratio range with variable La contents (Figure 12d), suggesting that the magma that formed the gabbro underwent significant fractional crystallization. The positive correlations of CaO, Cr, and Ni with Mg # and negative correlation of MgO with SiO 2 ( Table 2) are consistent with the fractionation of olivine, pyroxene, or amphibole. The positive Eu and Sr anomalies (Figure 5a,b) and high Al 2 O 3 contents (Table 2) argue against plagioclase fractionation. Therefore, the gabbro was likely derived from the partial melting of a depleted mantle wedge hydrated by slab-released fluids, and the parental magma underwent the crystal fractionation of Al-poor phases such as olivine, pyroxene, or amphibole, resembling highalumina basalt that was formed by the fractional crystallization of mantle-derived hydrous magma [25].  [96]). They possess calculated Ti-in-zircon temperatures [57] in the range of 665-813 • C (avg. 728 • C, median = 711 • C, n = 16), which contrast the high-temperature formation conditions of A-type granites (>800 • C, [97]). Moreover, A-type granites generally contain some special alkali mafic minerals, such as arfvedsonite, sodium pyroxene, riebeckite, and late-crystallizing biotite and amphibole [98]. However, these mineral assemblages were not identified in our petrographical observations ( Figure 3). Therefore, the petrological and geochemical features rule out an affinity with A-type granites. The granite porphyry samples show a negative correlation between the P 2 O 5 and SiO 2 contents (Figure 13c) and positive correlation between the Y and Rb values (Figure 13d), which are typical I-type granite evolution trends [99]. Furthermore, the absence of aluminous minerals (e.g., muscovite, tourmaline, and garnet [100]), low A/CNK ratios of 1.01-1.09 (Figure 4d), and high Na 2 O contents of 5.08-5.25 wt.% indicate that the granite porphyry is I-type rather than S-type, which commonly contains Al-rich minerals with high A/CNK values (>1.1) and low Na 2 O contents [101]. Therefore, we classify the granite porphyry as the non-fractionated I-type rather than the S-or A-type. (a) Na2O + K2O vs. 10,000 Ga/Al diagram [96]; (b) FeO T /MgO vs. Zr + Nb + Ce + Y diagram [96]; (c) P2O5 vs. SiO2 diagram; (d) Y vs. Rb diagram [99]. FG, fractionated felsic granites; OGT, unfractionated M-, I-, and S-type granites. The data from previous studies can be found in Supplementary  Table S1.
Prior studies uncovered that I-type granitoids may be formed by three petrogenetic scenarios, including (1) a complete process of fractional crystallization from primary mafic magmas [102]; (2) the mixing of crustal-derived materials with mantle-derived magmas [103]; and (3) the partial melting of intermediate to mafic metaigneous rocks without sedimentary contamination [104]. The non-fractionated I-type granite porphyry samples contain lower MgO (<0.9 wt.%), Cr (5.6-14.3 ppm), and Ni (1.6-4.7 ppm) contents but higher SiO2 (>71 wt.%) contents compared with the magmas derived from the direct partial melting of the mantle, which generally possess high Mg # values and exhibit mafic to intermediate compositions [105]. The absence of mafic microgranitoid enclaves in the rocks and consistent Sr-Nd isotopic compositions (Figure 10c) indicate that the mixing of mafic and silicic melts is unlikely to have occurred. The La/Sm ratios of the granite porphyry samples are positively correlated with the La contents, implying that the granite porphyry was likely formed by partial melting (Figure 12e). The Mg # (38.7-41.0) of the granite porphyry samples resemble those of the experimental melts from the metabasalts and eclogites (Figure 6b), suggesting that the magma may have been formed by the partial melting of a mafic crustal source. The Th/La ratios (0.23-0.29) of the granite porphyry samples are close to the average Th/La ratio of the crust (~0.3 [106]) but higher than that of the mantle (~0.12 [51,107]), further indicating a crustal origin. The granite porphyry samples have high εHf(t) (10.3-13.0; Figure 10a,b) and εNd(t) values (4.78-5.21) (Figure 10c,d) and young Hf (TDM1 = 429-540Ma, TDM2 = 494-663 Ma; Table 5) and Nd (TDM1 = 619-688 Ma; Table 6) model ages. These geochemical and isotopic signatures demonstrate that the granite porphyry was likely derived from the juvenile lower crust. Their U-shaped REE patterns (low [Dy/Yb]N = 0.83-0.94) and relatively high Y and Yb contents (Figure 5a) indicate that amphibole (rather than garnet) acts as a residual phase during crustal melting [25].  [96]; (b) FeO T /MgO vs. Zr + Nb + Ce + Y diagram [96]; (c) P 2 O 5 vs. SiO 2 diagram; (d) Y vs. Rb diagram [99]. FG, fractionated felsic granites; OGT, unfractionated M-, I-, and S-type granites. The data from previous studies can be found in Supplementary Table S1.
Prior studies uncovered that I-type granitoids may be formed by three petrogenetic scenarios, including (1) a complete process of fractional crystallization from primary mafic magmas [102]; (2) the mixing of crustal-derived materials with mantle-derived magmas [103]; and (3) the partial melting of intermediate to mafic metaigneous rocks without sedimentary contamination [104]. The non-fractionated I-type granite porphyry samples contain lower MgO (<0.9 wt.%), Cr (5.6-14.3 ppm), and Ni (1.6-4.7 ppm) contents but higher SiO 2 (>71 wt.%) contents compared with the magmas derived from the direct partial melting of the mantle, which generally possess high Mg # values and exhibit mafic to intermediate compositions [105]. The absence of mafic microgranitoid enclaves in the rocks and consistent Sr-Nd isotopic compositions (Figure 10c) indicate that the mixing of mafic and silicic melts is unlikely to have occurred. The La/Sm ratios of the granite porphyry samples are positively correlated with the La contents, implying that the granite porphyry was likely formed by partial melting (Figure 12e). The Mg # (38.7-41.0) of the granite porphyry samples resemble those of the experimental melts from the metabasalts and eclogites (Figure 6b), suggesting that the magma may have been formed by the partial melting of a mafic crustal source. The Th/La ratios (0.23-0.29) of the granite porphyry samples are close to the average Th/La ratio of the crust (~0.3 [106]) but higher than that of the mantle (~0.12 [51,107]), further indicating a crustal origin. The granite porphyry samples have high εHf(t) (10.3-13.0; Figure 10a,b) and εNd(t) values (4.78-5.21) (Figure 10c,d) and young Hf (T DM1 = 429-540Ma, T DM2 = 494-663 Ma; Table 5) and Nd (T DM1 = 619-688 Ma;  Table 6) model ages. These geochemical and isotopic signatures demonstrate that the granite porphyry was likely derived from the juvenile lower crust. Their U-shaped REE patterns (low [Dy/Yb] N = 0.83-0.94) and relatively high Y and Yb contents (Figure 5a) indicate that amphibole (rather than garnet) acts as a residual phase during crustal melting [25].

Tectonic Implications
The eastern Tianshan orogenic belt occupies the middle part of the CAOB, constituting an important Cu-Mo-Au-Ni-Fe-Ag metallogenic province in China [4,14]. Previous studies have revealed that the eastern Tianshan orogenic belt underwent a long and complex tectonic evolution during the Paleozoic to Mesozoic, including subduction and accretion followed by the collision of the Siberian and Tarim Cratons, and post-collision extension [25,108]. Recently, a large number of Late Ordovician-Late Carboniferous magmatic rocks with arc affinity have been reported in the Dananhu island arc belt, such as Yudai diorite porphyry (452.7 ± 2.8 Ma [7]), Sanchakou-Yuhai diorites and granodiorites (444-430 Ma and 325-318 Ma, respectively, [8,9,79,109,110]), and Tuwu-Yandong intrusive rocks (348-315 Ma; Table 7), which are interpreted to be related to the northward subduction of the ancient Tianshan Ocean (e.g., Kangguer Ocean). The studied granodiorite porphyry (322 Ma) and granite porphyry (310 Ma) have relatively low Y, Yb, Ta, Nb, and Rb contents which are similar to those of typical oceanic volcanic arc granites (Figure 14a-c). The studied gabbro samples are plotted in the field of the island arc basalt in the Th-Ta-Hf/3 diagram (Figure 14d [111]), and on the Th/Yb vs. Nb/Yb diagram and the Ta/Yb-Th/Yb diagram (Figure 12a,b), all the intrusive rock samples fall into the "Oceanic Arcs" field. Consequently, these Late Carboniferous intrusions in the Tuwu-Yandong belt most likely formed in an island arc setting. Nevertheless, the subduction process of the ancient oceanic basin during the Late Carboniferous remains controversial, with the proposed processes including slab rollback, flat subduction, and ridge subduction [10,25,39,105,112].

Tectonic Implications
The eastern Tianshan orogenic belt occupies the middle part of the CAOB, constituting an important Cu-Mo-Au-Ni-Fe-Ag metallogenic province in China [4,14]. Previous studies have revealed that the eastern Tianshan orogenic belt underwent a long and complex tectonic evolution during the Paleozoic to Mesozoic, including subduction and accretion followed by the collision of the Siberian and Tarim Cratons, and post-collision extension [25,108]. Recently, a large number of Late Ordovician-Late Carboniferous magmatic rocks with arc affinity have been reported in the Dananhu island arc belt, such as Yudai diorite porphyry (452.7 ± 2.8 Ma [7]), Sanchakou-Yuhai diorites and granodiorites (444-430 Ma and 325-318 Ma, respectively, [8,9,79,109,110]), and Tuwu-Yandong intrusive rocks (348-315 Ma; Table 7), which are interpreted to be related to the northward subduction of the ancient Tianshan Ocean (e.g., Kangguer Ocean). The studied granodiorite porphyry (322 Ma) and granite porphyry (310 Ma) have relatively low Y, Yb, Ta, Nb, and Rb contents which are similar to those of typical oceanic volcanic arc granites ( Figure  14a-c). The studied gabbro samples are plotted in the field of the island arc basalt in the Th-Ta-Hf/3 diagram (Figure 14d [111]), and on the Th/Yb vs. Nb/Yb diagram and the Ta/Yb-Th/Yb diagram (Figure 12a,b), all the intrusive rock samples fall into the "Oceanic Arcs" field. Consequently, these Late Carboniferous intrusions in the Tuwu-Yandong belt most likely formed in an island arc setting. Nevertheless, the subduction process of the ancient oceanic basin during the Late Carboniferous remains controversial, with the proposed processes including slab rollback, flat subduction, and ridge subduction [10,25,39,105,112].  [113], (b) Ta vs. Yb diagram [113], and (c) Rb/30-Hf-3×Ta diagram [114] for the granitic rocks. WPG, within-plate granites; VAG, volcanic arc granites; Syn-COLG, syn-collision granites; Post-COLG, post-collision granites; ORG, ocean ridge granites. (d) Hf/3-Th-Ta diagram [111] for the gabbro. IAB, island arc basalt; N-MORB, normal-type mid-ocean ridge basalt; E-MORB, enriched-type mid-ocean ridge basalt; WPA, within-plate alkalic; WPT, within-plate tholeiite.
As discussed above, the partial melting of the Kangguer oceanic slab produced parental magmas of the plagiogranite porphyry (335-332 Ma, [5,22,23]), granodiorite porphyry (321.8 ± 3.1 Ma, this study), and, possibly, Chihu porphyritic granodiorite (314.5 ± 2.5 Ma, [14]). Since normal subduction zones have lower temperatures than the adjacent mantle, it is generally believed that the partial melting of the subducting slab cannot occur, but rather dehydration leads to the partial melting of the overlying mantle wedge [115], with the formation of arc-related calc-alkaline basaltic-andesitic-daciticrhyolitic igneous rocks [116]. Therefore, the existence of adakites derived from the partial melting of the subducting slab may indicate a special environment. Adakites are originally thought to be associated with the subduction of the young (≤25 Ma) and hot oceanic lithosphere [52]. Other studies revealed that adakitic rocks can be formed in various tectonic settings as long as a high geothermal gradient exists [105], such as the initial subduction of an old crust [117], ridge subduction [118], flat subduction [119], or post-collision [116]. Combined with regional sedimentation, magmatism, and tectonism, the subduction of the young and hot oceanic lithosphere, initial subduction of an old crust, and post-collision setting during the Early Late Carboniferous (335-315 Ma) seem unlikely. In addition to adakites, ridge subduction usually produces high-Mg andesites and Nb-enriched basalts (Nb >20 ppm) [120]. The reported basalts and the studied gabbro that are temporally and spatially close to the adakitic rocks in the Tuwu-Yandong belt have low Nb contents (almost all <10 ppm, and mostly <5 ppm; see Supplementary Table S1), ruling out ridge subduction. As an unusual mode of subduction, flat subduction, which occurs in ca. 10% of the world's convergent margins, can produce the temperature and pressure conditions necessary for the fusion of a moderately old oceanic crust [105]. The upper part and leading edge of the slab can melt during the early stages of flat subduction [119]. Hence, we speculate the formation of the adakites in the Tuwu-Yandong belt was most likely caused by the northward flat subduction of the Kangguer Ocean. This is further supported by the increasing Ce/Y ratios of the basic rocks and Ho/Yb ratios of the felsic rocks (from the latest part of the Early Carboniferous to the Late Carboniferous) in the Dananhu and Bogeda belt, indicating significant crustal thickening, which is likely associated with high-to low-angle subduction transition [25]. Furthermore, prolonged flat subduction both cools the lithospheres and impedes the partial melting of the subducting oceanic crust [119], which may be consistent with the gradually weakening adakitic features (e.g., the decreasing Sr/Y ratio and increasing Y content of the adakites) from Early Carboniferous to Late Carboniferous (335-315 Ma) in the Tuwu-Yandong belt (Figure 6a; [5,14] and references therein; this study).
As flat subduction continues, the gradually cooling subducting slab phases into the eclogite facies, with the gravity increasing, which leads the subducting slab to become increasingly unstable [121], possibly transforming into a higher-angle subduction. The rollback (or low-to high-angle subduction transition) of the subducted slab causes the strong upwelling of the asthenosphere, provoking strong asthenosphere-lithosphere interactions and the partial melting of juvenile lower crust [10]. Some magmatism and mineralization data indicate that slab rollback may have occurred in the Dananhu island arc in the Late Carboniferous (ca. 314 Ma), such as (1) the high-Al gabbro (313.5 ± 1.2 Ma) and nonfractionated I-type granite porphyry (309.8 ± 2.5 Ma) in the Tuwu-Yandong belt, of which the latter is considered to be derived from the partial melting of the juvenile lower crust under garnet-free amphibolite facies conditions (this study), and (2) the Haibaotan gabbro (315.5 ± 1.9 Ma), associated with magmatic Cu-Ni mineralization along the Dacaotan fault, which is thought to have formed in a subduction setting [122]. The inference of slab rollback is reinforced by the ca. 311 Ma K-feldspar granites (located approximately 30 km northeast of the Chihu deposit) showing variable Nb/La ratios (0.38-1.07) (Nb/La <0. 71 for the rocks formed in subduction settings, Nb/La >0.71 for the rocks in a lithospheric extension or mantle plume environments [95,123]), indicating that they were formed by subduction-related materials with a significant addition of intraplate components [61]. No younger magmatism after 310 Ma has been identified in the Tuwu-Yandong belt, which may indicate a "quiet period" before the final closure of the ancient Tianshan Ocean along the Kangguer Fault in the belt [10].

Conclusions
(1) New LA-ICP-MS zircon U-Pb geochronology indicates that the granodiorite porphyry, gabbro, and granite porphyry were emplaced at 321.8 ± 3.1 Ma, 313.5 ± 1.2 Ma and 309.8 ± 2.5 Ma, respectively. (2) The zircon trace elements of the Carboniferous intrusions in the Tuwu-Yandong belt imply that the granodiorite porphyry is likely derived from a more oxidized and hydrous magma source than that of the gabbro and granite porphyry, which may favor the formation of porphyry Cu deposits. (3) The adakitic granodiorite porphyry is derived from the partial melting of the subducted oceanic slab, with subsequent interactions with mantle peridotite. The high-Al gabbro is derived from the partial melting of a depleted mantle wedge hydrated by slab-released fluids, while the non-fractionated I-type granite porphyry is derived from the partial melting of the juvenile lower crust. Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/min12121573/s1, Tables S1-S3. Table S1: Published whole-rock major (in wt%) and trace element (in ppm) compositions for Carboniferous magmatic rocks in the Tuwu-Yandong belt (and adjacent areas in the middle section of the Dananhu island arc). Table S2: Published in-situ zircon Lu-Hf isotope compositions for Carboniferous magmatic rocks in the Tuwu-Yandong belt (and adjacent areas in the middle section of the Dananhu island arc).