Geochronology and geochemistry of volcanic rocks from the Tanjianshan Group, NW China: Implications for the early Palaeozoic tectonic evolution of the North Qaidam Orogen

3.1 Zircon morphology, U–Pb dating, and trace elements

Zircon crystal separation was carried out using standard techniques, purified by hand-picking under a binocular microscope and mounted in epoxy resin and polished down to expose the grain centre. Cathodoluminescence (CL) images were taken for all zircons at the State Key Laboratory of Geological Processes and Mineral Resources (GPMR), China University of Geosciences (Wuhan). U–Pb dating and trace element analyses were conducted synchronously by Laser Ablation Inductively Coupled Plasma Mass Spectrometry (LA-ICPMS) at GPMR. Laser sampling was performed using an excimer laser ablation system of GeoLas 2005. An Agilent 7500a ICPMS instrument was used to acquire ion-signal intensities. Laser energy and frequency were 70 mJ and 8 Hz, respectively, with spot size of 32 μm. Detailed analytical conditions and procedures for LA-ICPMS zircon U–Pb dating are described in Liu, Gao, et al. (2010), Liu, Hu, et al. (2010), and for trace element analyses in Liu et al. (2008) and Chen et al. (2011). Helium was used as the carrier gas. Argon was used as the make-up gas and mixed with the carrier gas via a T-connector before entering the ICPMS. Nitrogen was added into the central gas flow (Ar + He) of the Ar plasma to lower detection limits and improve precision (Hu et al., 2008).

Each LA-ICPMS analysis incorporated a background acquisition of ~20–30 s (gas blank) followed by 50 s data acquisition from the sample. Zircon 91500 was used as the external standard for U–Pb dating and was analysed twice for every five samples. Time-dependent drifts of U–Th–Pb isotopic ratios were corrected using a linear interpolation (with time) for every five analyses according to variations of the 91500 standard. Preferred U–Th–Pb isotopic ratios for 91500 were taken from Wiedenbeck et al. (2004). In addition, 25 spots on zircon standard GJ-1 were analysed as unknowns simultaneously with the analysed samples. The obtained weighted mean ²⁰⁶Pb/²³⁸U age for GJ-1 is 596.7 ± 5.3 Ma (2σ, n = 25), which is consistent with the reported or recommended values (GJ-1: 599.8 ± 1.7 Ma (2σ), Jackson, Pearson, Griffin, & Belousova, 2004).

The USGS reference glasses BCR-2G and BIR-1G were analysed as external standards for trace element calibration (Liu et al., 2008). The preferred values of element concentrations for the USGS reference glasses are from the GeoReM database (http://georem.mpch-mainz.gwdg.de/). Every 10 sample analyses were followed by one analysis of NIST SRM 610 to correct the time-dependent drift of sensitivity and mass discrimination in the trace element results. Trace element compositions of zircons were calibrated against multiple reference materials (BCR-2G and BIR-1G) combined with an internal standardization (Liu, Gao, et al., 2010). Offline selection and integration of background and analytical signals, time-drift correction, and quantitative calibration for zircon U–Pb dating and trace elements compositions were performed by the ICPMS DataCal 8.3 software. The start and end times of signals of measured samples conform to those of the zircon standard 91500. Common Pb was corrected based on the method proposed by Andersen (2002). Concordia diagrams and weighted mean calculations were made using Isoplot 4.5 (Ludwig, 2003).

3.2 Zircon Hf isotopes

In situ analysis of zircon Lu–Hf isotopes was carried out by laser ablation multicollector inductively coupled plasma mass spectrometery (LA-MC-ICPMS) at GPMR. The analyses were conducted with a spot size of 44 μm, a 10-Hz repetition rate, and a laser power of 100 mJ/pulse. The ablation spots for the Hf isotope analyses were situated close to the U–Pb age analysis positions on each grain. Interference of ¹⁷⁶Lu on ¹⁷⁶Hf was corrected by measuring the intensity of the interference-free ¹⁷⁵Lu, using the recommended ¹⁷⁶Lu/¹⁷⁵Lu ratio of 0.02669 (De Biévre & Taylor, 1993) to calculate ¹⁷⁶Lu/¹⁷⁷Hf. Similarly, the isobaric interference of ¹⁷⁶Yb on ¹⁷⁶Hf was corrected using a recommended ¹⁷⁶Yb/¹⁷²Yb ratio of 0.5886 (Chu et al., 2002) to calculate ¹⁷⁶Hf/¹⁷⁷Hf ratios. Zircon 91500 was used as the reference standard (Wiedenbeck et al., 2004). Details of the operating conditions and data reduction were described by Hu et al. (2012). A decay constant of 1.867 × 10⁻¹¹ year⁻¹ was adopted for ¹⁷⁶Lu (Soderlund, Patchett, Vervoort, & Isachsen, 2004). Initial ¹⁷⁶Hf/¹⁷⁷Hf ratio, denoted as (¹⁷⁶Hf/¹⁷⁷Hf)_i, was calculated relative to the chondritic reservoir with a ¹⁷⁶Hf/¹⁷⁷Hf ratio of 0.282785 and ¹⁷⁶Lu/¹⁷⁷Hf of 0.0336 (Bouvier, Vervoort, & Patchett, 2008). Hf model ages (TDM) were calculated relative to the depleted mantle with a present-day ¹⁷⁶Hf/¹⁷⁷Hf ratio of 0.28325 and ¹⁷⁶Lu/¹⁷⁷Hf of 0.0384 (Vervoort & Blichert-Toft, 1999). Crustal Hf model ages (T_DM^C) were calculated by assuming a mean ¹⁷⁶Lu^/177Hf value of 0.015 for the average continental crust (Griffin et al., 2002).

3.3 Whole-rock major and trace elements

Whole-rock major-element compositions were determined at GPMR using conventional X-ray fluorescence (XRF) techniques, and loss on ignition (LOI) was measured for each sample. The analytical uncertainty is better than 5%. Whole-rock trace-element compositions were analysed by fusion ICPMS using an Agilent 7500 system, as described by Liu et al. (2008). The analytical precision is better than 5% for elements with concentrations >10 ppm, and better than 10% for those <10 ppm.

3.4 Whole-rock Nd–Sr isotopes

Neodymium and strontium isotopic ratios were analysed using a Finnigan MAT-261 multicollector isotope-ratio mass spectrometer (MC-IRMS) at GPMR. Sample powders were digested in custom-made Teflon bombs using a mixture of double-distilled HNO₃ and HF acids at 190°C for 48 hr. Nd and Sr were sequentially separated and purified in a clean laboratory using ion exchange columns of Dowex AG50WX12 cation resin and Eichrom Ln-Spec resin, respectively. Sr and Nd isotopic ratios was corrected to ⁸⁶Sr/⁸⁸Sr = 0.119400 and ¹⁴⁶Nd/¹⁴⁴Nd = 0.721900, respectively. Repeated measurements of the NBS987 and La Jolla standards yielded ⁸⁷Sr/⁸⁶Sr = 0.710254 ± 4 (2σ) and ¹⁴³Nd/¹⁴⁴Nd = 0.511854 ± 8 (2σ), respectively. Total procedural Sr and Nd blanks were < 1 ng and < 50 pg, respectively. Ratios of ¹⁴⁷Sm/¹⁴⁴Nd and ⁸⁷Rb/⁸⁶Sr were calculated using Sm, Nd, Rb, and Sr concentrations determined by ICPMS, with relative uncertainties of ~0.3% and ~1%, respectively, based on the ICPMS results for the standards cited above. Detailed analytical procedures for Sr and Nd isotopic analyses are given in Zhang et al. (2002) and Ling et al. (2009). Sm–Nd isotopic parameters were calculated following the procedures described in Möller, Mezger, and Schenk (1998).

4.1 U–Pb geochronology

The LA-ICPMS analytical results of zircon U–Pb dating on the volcanic rocks are shown in Table 2. A total of 132 zircon spots on four samples from Formation d-1, 110 spots on four samples from Formation d-4, and 111 spots on five samples from Formation d-3 were analysed for U–Pb isotopic ratios and trace-element concentrations. After excluding ages ≥800 Ma (considered to be inherited zircons) and discordant data (concordances <90%), 102 concordant ages were obtained, with ²⁰⁶Pb/²³⁸U ages ranging from 402 to 493 Ma (Table 2). Based on evaluation of U–Pb dates and CL images, the age results can be subdivided into two groups: Group 1 (≥ 440 Ma) and Group 2 (< 440 Ma). In CL images, most Group 1 zircons from Formations d-1 and d-4 possess stubby prismatic to equant forms, with grain sizes ranging from 50 to 150 μm and length/width ratios ranging from 1:1 to 2:1 (Figure 4a,b). They exhibit obvious oscillatory zoning and have Th/U ratios mostly >0.3, indicating a magmatic origin (Corfu, Hanchar, Hoskin, & Kinny, 2003; Hoskin & Schaltegger, 2003). In contrast, most Group 2 zircons from Formations d-1 and d-4 show irregular zoning or dark hues in CL images, indicating fluid alteration (Li, Sun, et al., 2017; Li et al., 2018; Li, Watanabe, & Yonezu, 2014; Sun, Li, Evans, Yang, & Wu, 2017). In addition, some Group 2 zircons (e.g., zjg-24-2-15; Figure 4a) exhibit core-rim textures reflecting crystallization of the core form a magma and growth of the rim from an orogenic fluid. On the other hand, Group 1 and Group 2 zircons from Formation d-3 exhibit no pronounced differences in CL images (Figure 4c). Most Formation d-3 zircons show magmatic zoning but some display slightly dark hues, suggesting weak fluid modification.

Group 1 and Group 2 zircons from Formation d-1 yielded weighted mean ²⁰⁶Pb/²³⁸U ages of 454.3 ± 6.4 Ma (MSWD = 3.5, Figure 5a) and 424.5 ± 6.9 Ma (MSWD = 0.68, Figure 5b), respectively. Zircons from Formation d-4 exhibit similar ages, with Group 1 zircons yielding a weighted mean U–Pb age of 458.1 ± 3.3 Ma (MSWD = 1.5, Figure 5c), and Group 2 zircons a weighted mean age of 428.2 ± 6.3 Ma (MSWD = 1.3, Figure 5d). These age results indicate that Formations d-1 and d-4 formed simultaneously at 460–440 Ma and experienced fluid alteration at 430–420 Ma. Contamination of the Group 1 zircons in Formations d-1 and d-4 during magma ascent is unlikely because the precursor magmatic activities, as recorded in Formations a-1/b and a-2, date to 520–460 Ma (Liang et al., 2014; Zhu et al., 2012) and thus are older than 460 Ma. However, Formation d-3 yielded a different zircon age distribution than Formations d-1 and d-4. In Formation d-3, Group 1 zircons, which are regarded as zircons that crystallized early or were captured from Formations d-1 and d-4, have a weighted mean ²⁰⁶Pb/²³⁸U age of 443 ± 7 Ma (MSWD = 0.68, Figure 5e). On the other hand, Group 2 zircons from Formation d-3 yield a weighted mean age of 430.8 ± 5.1 Ma (MSWD = 0.69, Figure 5f). These results suggest that the magma of Formation d-3 erupted during 440–430 Ma, in the process of which it captured a few zircons from the older Formations d-1 and d-4.

4.2 Zircon trace-element compositions

The LA-ICPMS analytical results of zircon trace elements are shown in Table 3. Corresponding to their age variations, Group 1 and Group 2 zircons have distinct trace element concentrations. Compared with Group 1 zircons, Group 2 zircons from Formations d-1 and d-4 are more enriched in Hf (mostly >10,000 ppm) and light rare earth elements, with higher LREE/HREE ratios (mostly >0.1) but lower Eu/Eu* and Ce/Ce* values, yielding flatter patterns in a chondrite-normalized distribution (Figure 6a). On the other hand, Group 1 and Group 2 zircons of Formation d-3 have relatively similar trace element concentrations, although the Group 2 zircons exhibit lower LREE concentrations and more pronounced Eu and Ce anomalies (Eu/Eu* = 0.15–0.65, Ce/Ce* = 1.64–254) than the Group 1 zircons (Figure 6b).

4.3 Whole-rock major- and trace-element compositions

The major- and trace-element concentrations of 11 mafic volcanic rock samples from Formations d-1, d-4, and d-3 are shown in Table 4. The rocks have varied chemical compositions in whole-rock geochemistry, with contents of SiO₂ = 45.0–53.7%, TiO₂ = 0.93–2.25%, Al₂O₃ = 11.6–19.8%, MgO = 4.03–8.86%, FeO_T = 6.07–13.5%, CaO = 5.55–11.8%, Na₂O = 1.77–5.96%, K₂O = 0.12–1.10%, and P₂O₅ = 0.07–0.45%. Formations d-1 and d-4 have similar geochemical signatures, which differ from those of Formation d-3. Specifically, Formation d-3 has higher average concentrations of CaO (10.3%), FeO_T (12.6%), and MgO (7.82%) but lower average concentrations of Al₂O₃ (12.7%), K₂O (0.18%), Na₂O (2.88%), and P₂O₅ (0.08%), thus exhibiting more mafic characteristics. After normalization of total major-element oxides to 100%, these rock samples plot mainly in the basaltic andesite, basaltic trachy-andesite, and subalkaline basalt fields (Figure 7a,b) and exhibit magmatic affinities transitional between the calc-alkaline and low-potassium tholeiitic series (Figure 7c,d), although with large variations among samples.

Similar to the major elements, the trace-element characteristics of Formation d-3 are different from those of Formations d-1 and d-4 (Table 4). Formation d-3 exhibits the lowest average values of total rare earth elements (ΣREE, 38.6 ppm), LREE/HREE (1.7), and (La/Yb)_N (1.1), and is characterized by relatively flat REE patterns (Figure 8a). In contrast, Formations d-1 and d-4 are characterized by the pronounced differentiation of LREE and HREE with enrichment of LREE, and by LREE/HREE and (La/Yb)_N values ranging from 2.8 to 5.3 and from 2.3 to 6.3, respectively. In addition, the Eu and Ce anomalies are too weak to be observed, with average Eu/Eu* and Ce/Ce* values ranging from 0.98 to 1.05 for all the formations. On an N-MORB-normalized distribution, all samples display enrichment in LILE (e.g., Rb, Ba, K, Sr, and Th; Figure 8b), but Formation d-3 has lower concentrations of LILE than Formations d-1 and d-4.

4.4 Sr–Nd isotopes

The Sr and Nd isotope compositions of the volcanic rock samples are shown in Table 5. These samples show ⁸⁷Sr/⁸⁶Sr ratios of 0.705620–0.711084 and ¹⁴³Nd/¹⁴⁴Nd ratios of 0.512297–0.512973. Compared with Formations d-1 and d-4 (average ⁸⁷Sr/⁸⁶Sr = 0.706851; ¹⁴³Nd/¹⁴⁴Nd = 0.512564), Formation d-3 has lower average ⁸⁷Sr/⁸⁶Sr ratios (0.705571) but higher average ¹⁴³Nd/¹⁴⁴Nd ratios (0.512848). Moreover, Formation d-3 is characterized by exclusively positive εNd (t) values (1.3–5.0), whereas for Formations d-1 and d-4 half of the εNd (t) values are negative (−2.7 to 7.0). The calculated two-stage model ages (T_2DM) for Formations d-1 and d-4 vary widely from 618 to 1410 Ma, but they are less variable and younger on average for Formation d-3 (764–1065 Ma).

4.5 Zircon Hf isotopes

The zircon Hf isotopic results of the volcanic rocks (Formations d-1 and d-4) are shown in Table 6. A total of 16 spots of zircon grains with ages ranging from 452 to 494 Ma were chosen for Hf isotopic analysis. These zircons possess oscillatory zoning characteristics in CL images, typical of a magmatic origin. All of these zircons have positive εHf (t) values (7.5–16.1), with Hf model ages (T_DM) and average crustal Hf model ages (T_DM^C) ranging from 443 to 778 Ma and from 428 to 970 Ma, respectively. In comparison, zircons from Sample XTS-103-2 (Formation d-4) have higher average ¹⁷⁶Hf/¹⁷⁷Hf (0.282892) values and lower ¹⁷⁶Yb/¹⁷⁷Hf ratios (0.023826). This results in relatively lower average initial ¹⁷⁶Hf/¹⁷⁷Hf (0.282892) and εHf (t) (9.1) values but higher T_DM (717 Ma) and T_DM^C (870 Ma) ages for Sample XTS-71-2 (Formation d-1).

5.1 Magma genesis and evolutionary processes of the volcanic rocks

Primary mafic magmas are generally sourced from the mantle without pronounced fractional crystallization, fluid interaction, or crustal contamination. Mafic rocks formed from such magmas are characterized by low contents of SiO₂ and incompatible elements (e.g., LILE and LREE; Jesus, Mateus, Munhá, & Tassinari, 2014; Li et al., 2016). The volcanic rocks in the Xitieshan area (especially from Formations d-1 and d-4) are enriched in LILE and LREE (Table 4), indicating that they were not the direct products of primary mafic magmas but, rather, resulted from a complex magma evolution.

The roles of fractional crystallization and post-crystallization alteration and metasomatism should be taken into account before inferring magma sources (Buchanan, Reimold, Koeberl, & Kruger, 2004; Kuritani, Kitagawa, & Nakamura, 2005). Major-element compositions are useful indicators of fractional crystallization and fluid interaction processes in volcanic rocks (Li, Watanabe, Xi, & Yonezu, 2013; Wu et al., 2017). For the Formation d volcanic rocks in this study, SiO₂ is poorly correlated to Al₂O₃, CaO, P₂O₅, and TiO₂ (Figure 9). This indicates that fractional crystallization did not play an important role in magma evolution. On the other hand, most major elements of Formation d volcanic rocks (especially Formations d-1 and d-4) show large concentration ranges among different samples (Table 4), possibly indicating varying degrees of post-crystallization alteration or metasomatism. Because orogeny-related fluids are widely existed in continental subduction and exhumation processes and can dissolve and mobilize substantial Si, Al, Ca, Na, and K (Gao, John, Klemd, & Xiong, 2007; Tian, Huang, Hui, & Xiao, 2015), the large variations of these elements in Formations d-1 and d-4 can be ascribed to alteration/metasomatism caused by orogeny-related fluids (Xia, Zheng, & Hu, 2010; Xiao et al., 2015). Hydrous melting commonly generates a magma from both the sediment and basalt layers of a subducting slab at relatively shallow depths, and the channelized fluid is likely to have migrated upward during the exhumation of the Xitieshan UHP rocks, as evidenced by formation of felsic veins (weighted mean zircon ²⁰⁶Pb/²³⁸U age of 420 ± 4 Ma) in the Xitieshan area (Liu et al., 2014).

Crustal contamination is an important evolutionary process during ascent and emplacement of magma. The enrichment of LILE and LREE in volcanic rocks of Formations d-1 and d-4 may indicate that crustal materials were dissolved into the magma, either in the source region or during magma ascent. Low Nb/La ratios (Nb/La <1), as shown by most volcanic rocks in the Xitieshan area (Table 4), are a reliable indicator of crustal contamination (Kieffer et al., 2004). Nb/U ratios generally do not change during partial melting of the mantle and thus can be used to trace magma evolution. Nb/U ratios in Formations d-1 and d-4 (5.9–19.6, average 12.5) are lower than in Formation d-3 (13.3–23.9, average 18.2) and closer to that of continental crust (9–12; Hofmann, 1988) than to that of the primitive mantle (~34; Sun & McDonough, 1989). This suggests that Formations d-1 and d-4 experienced greater crustal contamination than Formation d-3. The ratios of Nb/Ta in Formations d-1 and d-4 range from 8.7 to 14.9 with an average value of 12.9, significantly different from that of primitive mantle (17.4; Sun & McDonough, 1989). Ratios of elements having similar partitioning coefficients are not markedly influenced by fractional crystallization and degree of partial melting, and, thus, their correlation can be used to determine the degree of crustal contamination. As shown in Figure 10a,b, the linear correlations of Zr versus Th and Nb versus Th support crustal assimilation and contamination during magma formation or emplacement.

To further reveal tectonic settings of volcanic rocks, it is critical to determine the type of mantle source for intermediate to mafic magmas, that is, either from a depleted mantle or an enriched mantle. The N-MORB-associated depleted mantle is characterized by low Rb/Sr and ⁸⁷Sr/⁸⁶Sr, high Sm/Nd and ¹⁴³Nd/¹⁴⁴Nd, and positive Nd (t) values, whereas the E-MORB-related enriched mantle, which is thought to be generated through recycling of subducted crust, is relatively enriched in Si, Al, Ca, Na and K (Donnelly, Goldstein, Langmuir, & Spiegelman, 2004). The low ⁸⁷Sr/⁸⁶Sr (mean = 0.705571), high ¹⁴³Nd/¹⁴⁴Nd (mean = 0.512848), and positive Nd (t) (1.3–5.0) of Formation d-3 indicate a magma source with N-MORB characteristics.

Some trace element ratios are useful in characterizing magma sources because they do not change appreciably as a function of magma evolution or post-crystallization processes (Li, Watanabe, Xi, & Yonezu, 2013; Wu et al., 2017). Immobile elements, such as Nb, Yb, and Zr, can be employed to discriminate the nature of the magma source. The Zr/Hf ratios of the volcanic rocks in the Xitieshan area range from 34.2 to 43.9 with an average value of 39.5, which is close to that of the primitive mantle (36.7, Sun & McDonough, 1989) and implies a mantle origin for the magma. Nb/Yb ratios, which can be a robust indicator of mantle fertility (Pearce & Stern, 2006; Xu et al., 2014) range from 2.8 to 8.9 with an average value of 6.0 in Formations d-1 and d-4, which is close to that of E-MORB (3.5, Sun & McDonough, 1989). In contrast, Nb/Yb ratios in Formation d-3 are much lower (range = 1.2–1.3; average 1.27), which is similar to that of N-MORB (0.76, Sun & McDonough, 1989). These observations suggest that the mantle source of Formations d-1 and d-4 was enriched mantle, whereas that of Formation d-3 was normal mantle.

LREE enrichment can also be an indicator of enriched mantle magma sources. Strong LREE enrichment in Formations d-1 and d-4 (LREE/HREE = 2.8–5.3) suggests the influence of enriched mantle components (Zhou et al., 2015). The mantle association of the volcanic rocks can be further examined using Zr/Nb versus Y/Nb and Zr versus Y diagrams (Figure 10c,d). The magmas of Formations d-1 and d-4 had enriched mantle sources but probably experienced different degrees of assimilation and contamination of crustal materials during their ascent through the lithosphere. In contrast, the magma of Formation d-3 was probably generated from a normal mantle source and experienced little subsequent modification.

Whole-rock Sr–Nd and zircon Hf isotopic compositions can yield information on the origin and evolutionary processes of mafic magmas (Xia, Xu, Zhao, & Liu, 2015; Zhu, Zhong, Li, Bai, & Yang, 2016). In a ¹⁴³Nd/¹⁴⁴Nd versus ⁸⁷Sr/⁸⁶Sr crossplot (Figure 11a), Formations d-1 and d-4 exhibit only slightly elevated ⁸⁷Sr/⁸⁶Sr ratios, but their ¹⁴³Nd/¹⁴⁴Nd values do not preclude contamination by an older crustal component as evidenced by the frequency of negative εNd (t) values and relatively old T_2DM ages. High and variable ⁸⁷Rb/⁸⁶Sr and ⁸⁷Sr/⁸⁶Sr ratios in Formation d-1 and d-4 can be explained by the disturbance of the Rb–Sr system through crustal contamination or late-stage metamorphism (Uno et al., 2014). In contrast, all samples from Formation d-3 have positive εNd (t) values (Figure 11b), indicating that they likely formed from primitive mantle and underwent less contamination by older crustal components. In an εHf (t) versus age diagram (Figure 11c), Formations d-1 and d-4 have positive εHf (t) values greater than 5 and plot between the fields of depleted mantle (DM) and chondritic uniform reservoir (CHUR), but they still show arc basalt affiliation and crustal contamination feature when compared with their εNd (t) values (Figure 11d).

5.2 Tectonic setting of the volcanic rocks

Major-element data show that the volcanic rocks in the Xitieshan area can be classified as transitional between the calc-alkaline and low-potassium tholeiite series (Figure 7c), showing continental arc affinities. The distinctly lower concentrations of FeO_T and MgO and higher concentrations of Al₂O₃, K₂O, Na₂O, and P₂O₅ (Figures 7 and 9) indicate that Formations d-1 and d-4 are continental volcanic-arc basalts, whereas the opposite concentration patterns in Formation d-3 suggest a mid-ocean ridge environment. These interpretations are supported by REE and trace element variations: Enrichment of LILE and LREE in Formations d-1 and d-4 are consistent with volcanic-arc rocks, and the relatively flat REE and trace element patterns for Formation d-3 are consistent with mid-ocean ridge basalts (Figure 8). On a TiO₂–Zr diagram, Formation d-3 clusters in the mid-ocean ridge basalt field, whereas Formations d-1 and d-4 are scattered in the volcanic-arc basalt and within-plate basalt fields (Figure 12a). A Th/Yb–Ta/Yb diagram reveals that Formation d-3 plots in the MORB field, whereas Formations d-1 and d-4 are mostly classified as transitional basalt and tholeiitic basalt (Figure 12b). Hf–Th–Nb and Nb–Zr–Y ternary diagrams show that Formation d-3 clusters in the N-MORB field, and that Formations d-1 and d-4 plot as intraplate tholeiitic basalt with E-MORB characteristics (Figure 12c,d). However, the enrichments of some LILE (such as Rb and Ba) in Formation d-3 are somewhat different from those of normal MORB, which is generally characterized by depletion of mobile incompatible trace elements such as LILE and LREE and radiogenic isotopes (Jiao, Wang, Lu, & Duan, 2017; Wu et al., 2017). The differences in trace element compositions of the studied volcanic rocks could reflect different amounts of crustal materials admixed with their mantle sources.

The tectonic evolutionary history of the study units can be inferred from their observed geochemical signatures. From Formations d-1 and d-4 to Formation d-3, the tectonic environment evolved from a transitional stage between continental arc and mid-ocean ridge to a back-arc mid-ocean ridge stage, with the dominant lithology progressing from enriched transitional basalt (E-MORB) to normal tholeiitic basalt (N-MORB; see Section 5.4 for a detailed description of the tectonic evolution of the study area and its relationship to volcanic activity).

5.3 Late-stage fluid modification of the Tanjianshan Group

During continental exhumation, zircons may undergo orogenic reworking by replacement recrystallization in the presence of various fluids (Liu et al., 2014). The fluid-modified Group 2 zircons from Formations d-1 and d-4 possess younger ages than regional peak of metamorphism and volcanism (460–440 Ma), suggesting a relationship of zircon alteration to continental exhumation. Low-temperature or supercritical fluids related to orogenic processes can transfer LREE, HREE, Th, U, and high-field-strength elements (HFSE) from accessory minerals to recrystallized zircons during the exhumation of continental and oceanic crust (Xia, Zheng, & Hu, 2010). Fluid flow within UHP slabs is known to operate vigorously during the exhumation of deeply subducted continental crust (e.g., Chen et al., 2007; Wu, Gao, et al., 2009; Xia, Zheng, & Zhou, 2008). Abrupt decompression during the initial exhumation may release significant amounts of fluids from UHP metamorphic rocks (Zheng, 2009). Melting of UHP metamorphic rocks may occur at elevated temperatures during the initial exhumation, resulting in small amounts of hydrous melt (Zheng, Fu, Gong, & Li, 2003). Consequently, both aqueous fluids and hydrous melts are active during the exhumation of UHP slabs, enabling the formation of orogenic fluids (Xia, Zheng, & Hu, 2010; Zheng, 2009).

Most Group 2 zircons of Formations d-1 and d-4 from the Tanjianshan Group show dark hues in CL images with unclear oscillatory zoning, and some of them show core-rim textures. These features suggest that the zircons have been affected by fluids (Sun, Li, Evans, Yang, & Wu, 2017). The zircons are enriched in Th, U, and LREE, with minor positive Ce anomalies. They plot in the orogenic fluid alteration field, which thus supports an interpretation of fluid alteration (Figure 13a–e). Considering the age differences between Group 1 and Group 2 zircons, it can be concluded that Formations d-1 and d-4 underwent orogenic fluid modification at 430–420 Ma, which is 20–30 Myr later than the main volcanic stage (460–440 Ma). In contrast, neither Group 1 nor Group 2 zircons from Formation d-3 suffered intense fluid modification, which can be ascribed to the younger eruption age (440–430 Ma) of this formation.

5.4 Tectonic evolution of North Qaidam Orogen