Whole-rock and zircon evidence for evolution of the Late Jurassic high Sr/Y Zhoujiapuzi granite, Liaodong Peninsula, North China Craton

. Middle-Late Jurassic high Sr/Y granitic intrusions are extensively exposed in the Liaodong Peninsula, in the eastern part of the North China Craton (NCC). However, the genesis of the high Sr/Y signature in these intrusions has not been studied in detail. In this study, we report results of zircon U-Pb dating, Hf isotopic analysis and zircon and whole-rock geochemical data for the Late Jurassic Zhoujiapuzi granite in the middle part of the Liaodong Peninsula. The Zhoujiapuzi 15 granite is high-K (calc-alkaline) and peraluminous in nature, with high SiO 2 (68.1–73.0 wt %) and Al 2 O 3 (14.5–16.8 wt %), low in TFe 2 O 3 (1.10–2.49 wt %) and MgO (0.10–0.44 wt %), and with high Sr/Y (19.9–102.0) and La N /Yb N (14.59–80.40). Morphological and chemical studies on zircon grains show that there are two stages of zircon growth, interpreted as magmatic evolution in two distinct stages. The early stage of zircons (ESZ) reflects a crystallization environment of low oxygen fugacity and high T Zr-Ti (Ti-in-zircon thermometer values: 669–792℃); the late stage of zircons (LSZ) formed with 20 high oxygen fugacity and lower T Zr-Ti (498–720℃). LA-ICP-MS U-Pb zircon dating yielded the formation ages of the ESZ and LSZ of ~162±1 Ma and ~158±1 Ma, respectively, with similar εHf(t) values in the range of -26.3– -22.8. Interpretation of the elemental and isotopic data suggests that the Zhoujiapuzi granite was a I-type granite derived from partial melting of basement in the region: ~2.17 Ga Liaoji granites. The high Sr/Y signature is most likely inherited from these source rocks. Based on the geochemical features and regional geological data, we propose that the Liaodong Peninsula in the Late Jurassic 25 was part of a mature continental arc, with extensive melting of thick crust above the Paleo-Pacific subduction zone. integrated with whole-rock geochemistry. We focus on the zircons, because of their potential to reveal the origins of the pluton (Belousova al., 2002; Wang et al., 2007; Breiter al., 2014; Zhao et al., 2014), and so provide a case study for the evolution of plutonic magma systems in general. Based on observations of the CL images and chemical analysis, two 50 zircon growth stages can be distinguished. We first determine the crystallization environments of the two zircon growth stages, and then decipher the petrogenesis, source characteristics and origin of the high Sr/Y signature of the pluton as a whole. Integrated with previous studies, our study provides insights into the tectonic evolution of the Liaodong Peninsula in the Late Jurassic. and yield a Concordia upper intercept age of 2167 Ma. Both assimilation of country-rocks and incomplete that the adakitic signatures of these rocks are inherited from their source rocks. Similar results have been obtained by studying the late Jurassic Zhoujiapuzi granite in the Liaodong Peninsula in this study. Therefore, we suggest that melting of a high Sr/Y (and La/Yb) source is an important process for the generation of Yanshanian high Sr/Y rocks in the NCC. This kind of high Sr/Y


Zircon CL images, Raman spectra and REE elements
CL images of zircons from the Zhoujiapuzi granite are shown in Fig. 5. Zircons commonly have crystal sizes between 120 and 200 μm, and have length/width ratios of 2:1-4:1, with euhedral, stubby to elongate prisms. According to the CL images, there were two stages for zircon growth, and the zircons can be divided individually and collectively into: the early stage of zircon (ESZ) and the late stage of zircon (LSZ). The ESZ is characterized by bright CL intensity and widely-spaced 130 oscillatory zoning patterns. The LSZ is characterized by extremely low CL emission and narrowly-spaced oscillatory zoning patterns.
Six ESZ spots and six LSZ spots were analyzed for Raman spectra. The ESZ have antisymmetric stretching vibration

135
Twenty ESZ spots and eighteen LSZ spots were analyzed for trace and rare earth elements. The ESZ spots have lower U content (28-677 ppm) than that of the LSZ spots (U=641-3842 ppm). In the chondrite-normalized REE element diagram ( Fig. 6a, b), both the ESZ and LSZ characterized by HREE enrichment relative to LREE with positive Ce anomalies and negative Eu anomalies. The ESZ spots have ΣREE of 49-1115 ppm (average 390 ppm), ΣLREE of 3-72 ppm (average 14 ppm) and ΣHREE of 46-1100 ppm (average 377 ppm), whereas the LSZ spots have ΣREE of 327-1632 ppm (average 895 140 ppm), ΣLREE of 2-14 ppm (average 6 ppm) and ΣHREE of 325-1627 ppm (average 889 ppm). Hence, the REE content of the ESZ is significantly lower than that of the LSZ, and the difference between the two is mainly in HREE content. The ESZ spots have δEu of 0.07-0.60 (average 0.28) and δCe of 1. 26-166.54 (average 45.49). The LSZ spots have δEu of 0.08-0.24 (average 0.13) and δCe of 4.23-271.90 (average 94.37). These results indicate that the ESZ have a weaker negative Eu anomaly and a weaker positive Ce anomaly than those of the LSZ. 145 Some zircons have inherited cores, which have corroded and rounded shapes, such as the 1# in XY-001 and 6# and 41# in XY-008 (Fig. 5). These inherited zircons have oscillatory zoning in CL images. The inherited zircon spots have ΣREE of 602-1517 ppm, and show depletion of LREE, enrichment of HREE, a positive Ce anomaly (δCe of 1.52-216.08) and a negative Eu anomaly (δEu of 0.07-0.13) (Fig. 6c).

Zircon U-Pb and Hf isotope composition
150 Seventy-seven spots were analysed for U-Pb isotope composition from samples XY-001 and XY-008. In the U-Pb Concordia diagram (Fig. 7a, d), both the ESZ and LSZ spots are clustered on the Concordia curve. Overall, the ages of LSZ are generally older than those of ESZ (Fig. 8). On a single zircon, the 206 Pb/ 238 U age of the ESZ is older than that of the LSZ (Fig. 5 distinctly older ages ( 207 Pb/ 206 Pb ages ranging from 2494 to 2132 Ma) were obtained on inherited cores. Their ages are discordant, suggesting that these inherited cores were variably influenced by lead loss. Among these, 9 spots define a discordia line with an upper intercept age of 2167 ± 12 Ma (MSWD=0.59) (Fig. 7g), which is concordant with the weighted 160 mean 207 Pb/ 206 Pb ages of 2167 ± 12 Ma (1σ, MSWD = 0.54; Fig. 7h).
Twenty-four zircons were analyzed for Lu-Hf isotope composition. The variation in Hf isotopic data is limited, between 9 spots from ESZ and 8 spots from LSZ. 17 spots exhibit a range of 176 Hf/ 177 Hf ratios from 0.281921 to 0.282030, which converts to εHf(t) values between -26.3 to -22.8 (Fig. 9) Ma by using the upper intercept age.

Significance of the two stages of zircon
Generally, zircon with high U content can easily break down into the metamict state because of the radiation damage to the 170 lattice caused by α-particles originating from the decay of uranium (Mezger and Krogstad, 1997). The physical and structural changes often lead to the loss of Pb and addition of trace elements such as LREE. In this study, the LSZ are characterized by high U content. Hence, the metamictization degree of the zircons must be taken into consideration. Data from LSZ spots plot on the Concordia curve, indicating no obvious Pb loss. The internal structure of LSZ is relatively intact, with obvious oscillatory zoning, and few cracks, implying that the physical and structural of the LSZ remained unchanged. Nasdala et al. (1998) suggested that the metamictization of zircon can be well characterized by Raman spectroscopy. The half-width of the v3(SiO4) Raman band (b) of 10 cm -1 and 20 cm -1 are proposed to approximately distinguish well-crystallized, intermediate and metamict zircons (Nasdala et al., 1998). The LSZ have b values of 5.4-9.2, characterizing them as well-crystallized. Therefore, the above features indicate that the LSZ are not metamict. Consequently, it can be concluded that the U-Pb isotope and trace element systematics of the LSZ have not been changed by metamictization.

180
Both the ESZ and LSZ have oscillatory zoning patterns, and their chondrite-normalized REE patterns are characterized by steeply positive slopes from the LREE to HREE with strong negative Eu anomalies and pronounced positive Ce anomalies. The above characteristics are consistent with those of igneous zircon (Hoskin and Schaltegger, 2003). Although hydrothermal zircon can also have oscillatory zoning patterns similar to magmatic zircons, there are obvious differences in trace elements between the magmatic and hydrothermal zircon (Hoskin et al., 2005). In the discrimination diagram ( Fig. 10), 185 both the spots of ESZ and LSZ fall in or near the magmatic field, which is obviously different from hydrothermal zircon.
Hence, the above characteristics indicate that both the ESZ and LSZ from the Zhoujiapuzi granite have a magmatic origin.
The ESZ was overgrown continuously by the LSZ. In addition, the contact between the ESZ and LSZ is euhedral. Such core-mantle overgrowth relationships indicate that both the ESZ and LSZ formed in two stages of the same magmatism (i.e. the ESZ are not xenocrysts or inherited zircons). Furthermore, the similar Hf isotopic data of the ESZ and LSZ is also 190 consistent with this interpretation. The obvious difference in internal structure and trace element composition between the ESZ and LSZ could be due to significant and abrupt changes in their crystallization environments (Wang et al., 2007). Watson and Harrison (2005) found that the Ti content of zircon has a strong dependence on temperature (T), and obtain a Ti-in-zircon thermometer (TZr-Ti). Since then, Ferry and Watson (2007) suggested that the solubility of Ti in zircon depends not only on T and activity of TiO2 (aTiO2) but also on the activity of SiO2 (aSiO2), and revised the TZr-Ti. We use the TZr-Ti 195 from Ferry and Watson (2007) and the recommended values (aSiO2=1, aTiO2 = 0.6) for the activity of SiO2 and TiO2, due to the presence of ilmenite and quartz and lack of sphene and rutile in the Zhoujiapuzi granite. The TZr-Ti from the ESZ and LSZ are 669-792 ℃ (average 734 ℃) and 498-720℃ (average 621 ℃), respectively, i.e. the ESZ formed at higher temperatures than the LSZ. The U content shows a significant negative correlation with T ( Fig. 11a).
Cerium exists in magmas as both Ce 3+ and Ce 4+ . Because the 0.84-Å radius of the Zr 4+ ion is more closely matched by 200 the Ce 4 + (0.97-Å radius) than the Ce 3 + (1.143-Å radius) (all ionic radii are from Shannon, 1976), Ce 4 + is more compatible in the zircon structure than the Ce 3 + . Hence, zircon Ce 4 + /Ce 3 + ratio is a useful tool for evaluating the oxygen fugacity condition of crystallization environment (Ballard et al., 2002). In this study, the Ce 4 + /Ce 3 + ratio of the ESZ and LSZ are 6.30-153.36 (average 32.51) and 21.81-5773.06 (average 787.39), respectively (the calculation method after Ballard et al., 2002). As shown in the Ce 4 + /Ce 3 + -U diagram (Fig. 11b), U has a significant positive correlation with Ce 4 + /Ce 3 + . This result implies 205 that the LSZ formed in a higher oxygen fugacity environment than the ESZ.
The Zr/Hf ratio in zircon has a negative correlation with the degree of fractionation in the parent melt (Breiter et al., 2014). The Zr/Hf ratios of the ESZ (39-56) are obviously higher than those of the LSZ (21-40) (Fig. 11c). Hence, the above features reflect that the LSZ crystallized from a later and more evolved magma. The lowest Zr/Hf value of the LSZ is less than 25, which indicates that the Zhoujiapuzi granite finally evolved into a highly fractionated granite due to crystallization 210 differentiation. It is consistent that all the samples of the Zhoujiapuzi granite are characterized by relatively high SiO2 and total alkali contents, low TFe2O3, MgO and CaO contents, and enrichment of Rb, Th and U (Xiao et al., 2014).
The absence of enclaves and disequilibrium textures in the Zhoujiapuzi granite and uniform εHf(t) values of the ESZ and LSZ do not support magma mixing and wall-rock assimilation. Consequently, the abrupt change between the crystallization environment of the ESZ and LSZ is not due to the magma mixing or contamination during magma evolution.

215
Therefore, we propose that the ESZ was formed in a relatively deep magma chamber, which had low oxygen fugacity, low Zr saturation and high temperature. The low Th, U and REE, and widely-spaced oscillatory zoning patterns indicate a low growth rate of zircon (Hoskin and Schaltegger, 2003;Wang et al., 2011). In contrast, the LSZ was formed during the ascent and/or at the emplacement location of the magma. At this stage, the oxygen fugacity significantly increased, the temperature decreased, and Zr saturation increased due to the crystallization differentiation. In this environment, the crystallization rate 220 of zircon significantly increased, forming the zircons with a higher content of Th, U and REE elements, low CL emission and narrowly-spaced oscillatory zoning patterns. As shown in the U-206 Pb/ 238 U age diagram (Fig. 11d) phenomenon indicates that the cooling rate of magma in the deeper magma chamber was relatively slow, thus the zircon crystallized continuously over a relatively long period. The magma finally solidified in a relatively short period, so the zircon crystallized rapidly.
Zircon U-Pb dating is the most commonly used method in magmatic rock dating. A weighted mean age or upper intercept age is usually obtained to represent the emplacement time of a magmatic rock. However, this study shows that zircons in magmatic rocks record two different magmatic evolution stages: in the deeper magma chamber and in the ascent passage or/and emplacement site. Previous studies, such as Wang et al. (2007), Zhao et al. (2014) and Chen et al. (2020), 230 also show that zircons can crystallize continually or intermittently in a single phase of magmatism, showing several growth zones of clearly different internal structure and distinct time difference. Some scholars even regard that the age difference of different stages can be more than dozens of Ma (Wang et al., 2007). Therefore, if the zircon ages in the same magmatic rock have a large range of variation, this could be caused by the zircons recording different stages in magmatic evolution, related to different levels of magma within the crust and/or different temperature regimes. It is notable that the bulk petrology and 235 geochemistry of the host pluton does not record and reveal this two-stage magmatic evolution, which can only be detected in the zircon analysis.

Genetic type: I-type affinity
The samples of the Zhoujiapuzi granite have A/KNC < 1.1, relatively high Na2O (3.96-4.65 wt.%) and lack peraluminous minerals (e.g. cordierite, andalusite, muscovite and garnet), which are clearly different from S-type granites (Chappell and 240 White, 1992). With the rise of the degree of crystallization, P2O5 contents (generally>0.1 wt.%) increase in S-type granites, accompanied by an increase/immutability in SiO2 (Wolf and London, 1994). However, the Zhoujiapuzi granite samples have low P2O5 contents (0.02 -0.08 wt.%), and decrease with increasing SiO2 (Fig. 12a), which is also inconsistent with the characteristics of S-type granite (Chappell and White, 1992). Additionally, Rb has a negative correlation with Y ( Fig. 12b), which has been considered as an indicator of I-type granite rather than S-type granite (Jiang et al., 2018). Thus, the 245 Zhoujiapuzi granite is not an S-type granite.
Although the samples of the Zhoujiapuzi granite have high 10000Ga / Al ratios of (2.74-3.13) akin to A-type granites, the low Zr (113-242 ppm) and Zr+Nb+Ce+Y (152.0-382.6 ppm) contents are distinct from the diagnostic features of A-type granites (Whalen et al., 1987). This observation is consistent with the Zhoujiapuzi granite lacking alkaline mafic minerals (Zhang et al., 2017). Wu et al. (2017) suggested that a high formation temperature is one of the most important 250 characteristics of A-type granite. Zircon saturation thermometry (TZrn) and Ti-in-zircon thermometer (TZr-Ti) are two methods for estimating magma temperatures. As discussed before, TZr-Ti is negatively correlated with the U content, which indicates that the crystallization temperature of zircon gradually decreases with magma. Hence, the initial magma temperature should be greater than or equal to the highest Ti temperature (792 ℃). Zircon saturation thermometry was introduced by Watson and Harrison (1983)  magma temperature should be lower than or equal to the lowest TZrn (803 ℃) (Miller et al., 2003). Therefore, the initial magma temperature of the Zhoujiapuzi granite was at a range of 792-803 ℃, which is significantly lower than that of the typically A-type granite (>900 ℃, Skjerlie and Johnston, 1992;Douce, 1997). Furthermore, the Zhoujiapuzi granite samples show similar trends to the Late Jurassic granitoids in the Liaodong peninsula, which are typical I-type granites (Fig. 12).

260
Therefore, we conclude that the Zhoujiapuzi granite is a highly fractionated I-type granite.

Petrogenesis of the high Sr/Y granite
The samples of the Zhoujiapuzi granite have high Sr/Y and (La/Yb)N ratios and low Y and Yb contents (Fig. 13a)

Model A: Partial melting of subducting oceanic crust
The partial melting of the young, hot and hydrated subducted oceanic slab in the garnet stability field is the classical 275 formation model of adakite (high Sr/Y rock) (Defant and Drummond, 1990). Studies have shown that the rock with this genetic model generally has the characteristics of high mantle components (such as MgO, CaO and Cr) because of the involvement of mantle magma (Wang et al., 2018). However, this phenomenon was not seen in the Zhoujiapuzi granite. In addition, the Zhoujiapuzi granite has high K2O/Na2O ratios (0.92-1.22, average 1.13), which is inconsistent with the slabderived adakites (K2O/Na2O=~0.4, Martin et al., 2005). Moreover, the low εHf(t) values (-26.3 to -22.8) of the Zhoujiapuzi 280 granite are also inconsistent with the magmas derived from the partial melting of oceanic crust, which generally have depleted isotopic character (Zhan et al., 2020). In summary, the Zhoujiapuzi granite is difficult to explain by Model A.

Model B: Delaminated lower continental crust (LCC)
High-density, garnet-bearing mafic lower crust delaminating or foundering into the asthenosphere mantle and subsequent interaction with mantle peridotite could produce high Sr/Y magmas (Kay and Kay 1993). Because the melt formed by partial 285 melting of the delaminated lower crust would interact with mantle peridotite during magma ascent, the high Sr/Y magmas  , 1978;King et al., 2015), rocks formed by partial melting of the delaminated lower crust should possess a high-temperature fingerprint. As mentioned before, the initial 290 magma temperature of the Zhoujiapuzi granite was~800 ℃, which is markedly lower than the temperature of the asthenosphere. Therefore, the petrogenetic model of delaminated lower continental crust (Model B) is also inconsistent with the Zhoujiapuzi granite.

Model C: Differentiation of basaltic arc magma
Low-pressure fractional crystallization (involving olivine + clinopyroxene + plagioclase + amphibole+ titanomagnetite) or 295 high-pressure fractional crystallization (involving garnet) from basaltic magmas have been proposed as two ways to generate adakitic characteristics (Castillo et al., 1999;Macpherson et al., 2006). As shown in the La/Sm versus La diagram (Fig. 14a), increasing La content with constant La/Sm shows that fractional crystallization rather than partial melting is the main factor controlling the composition of the Zhoujiapuzi granite. Hence, the samples of Zhoujiapuzi granite displayed variable Eu and Sr contents, implying that the plagioclase is likely a fractional phase. Separation of titanomagnetite could explain the 300 positive in TFe2O3 with increasing TiO2 content (Fig. 14b), consistent with the occurrences of magnetite in some studied rocks. This possible mineral assemblage of fractional crystallization is also reflected by the chemical variations in the Sr/Y-Y diagram (Fig. 13b).
However, fractionation of olivine and clinopyroxene is inconsistent with the depletion of HREE (e.g., Yb) given their low distribution coefficients for these elements in mafic-intermediate magmas (Dunn and Sen, 1994). Moreover, the 305 crystallization of amphibole would result in a negative correlation between the (Dy/Yb)N and (La/Yb) N (Davidson et al., 2007), which is not seen in the Zhoujiapuzi granite (Fig. 14c). No positive correlations between Dy/Yb and Sr/Y ratios and SiO2 contents (Figure 14d, e) suggest that fractional crystallization of garnet was not a significant process for the Zhoujiapuzi granite (Macpherson et al., 2006). Therefore, both the low-pressure and high-pressure crystallization fractionation of basaltic melts could be ruled out for the negligible fractional crystallization of olivine, clinopyroxene, garnet 310 and amphibole.
In addition, crystal fractionation of basaltic melts can only form minor volumes of granitic melts, the ratio of the two is about 9:1 (Zeng et al., 2016). However, in the Liaodong Peninsula, most of the Middle-Late Jurassic magmatic rocks are acidic, mafic-ultramafic rocks are only reported in the Huaziyu area (lamprophyre dikes, Jiang et al., 2005), and cover a much smaller area than the Zhoujiapuzi granite. For these reasons, it is highly improbable that Zhoujiapuzi granite was 315 derived by differentiation of basaltic magma (Model C).

Model D: Magma mixing between mantle-derived mafic and crust-derived silicic magmas
The Zhoujiapuzi granite has high K2O/Na2O ratio (>1) and A/CNK value (>1), together with the absence of mafic https://doi.org/10.5194/se-2021-129 Preprint. Discussion started: 17 November 2021 c Author(s) 2021. CC BY 4.0 License. microgranular enclaves (MMEs), felsic xenocrysts and melting texture of plagioclase, implying that the mantle-derived magma is unlikely to have played an important role in the genesis of the Zhoujiapuzi granite (Castro et al., 1991). In addition, 320 the Zhoujiapuzi granite is characterized by the development of biotite, but lacks amphibole and pyroxene. These features, coupled with the high A/CNK value, are consistent with an origin as a crust-derived granitoid, but obviously different from the granitoids formed by crust-mantle-derived magma mixing (Barbarin, 1990). Moreover, granites formed by magma mixing generally have variations in εHf(t) values, high MgO, TFe2O3, CaO and Cr contents and low SiO2 content (Ma et al., 2013;Wang et al., 2018). These features are obviously inconsistent with the Zhoujiapuzi granite in this study. Hence, magma 325 mixing of mantle-derived and crust-derived magmas (Model D) is also unlikely to have produced the Zhoujiapuzi granite.

Model E: Partial melting of thickened basaltic LCC
Experimental studies have shown that the partial melt of basaltic LCC in the garnet stabilization zone (> 40 km, i.e.~1.2 GPa) can produce magma with a high Sr/Y ratio (Rapp et al., 2003 and references therein). In these scenarios, high Sr/Y and overall adakitic affinity are caused by leaving garnet as residual phases (e.g. Gao et al., 2004). Based on geochemical data 330 for the Zhoujiapuzi granites, partial melting of thickened basaltic LCC is also unlikely to account for the high Sr/Y Zhoujiapuzi granite (Model E). This conclusion is based on the following observations: (1) This ratio of (Gd/Yb)N is the most important feature to judge whether garnet is involvement in magma genesis (Ma et al., 2012). If the (Gd/Yb)N ratio of the source is similar to the average value of the LCC (1.71, Rudnick and Gao, 2003), partial melting of these crustal materials controlled by garnet at high pressure can produce melt with (Gd/Yb)N of 5.8 (Huang 335 and He, 2010). In contrast, the (Gd/Yb)N values (1.22-5.06, average 2.69) of the Zhoujiapuzi granite are relatively low. (2) Melt in equilibrium with a garnet-rich residue will result in positive correlations between the Dy/Yb and Sr/Y ratios and SiO2 contents. However, these phenomena were not seen in the Zhoujiapuzi granite (Fig.14 d,  (4) As shown in the discrimination diagrams of granite sources (Fig. 14f, g), all samples fall in the range of metagreywacke-340 derived melts. Therefore, the Zhoujiapuzi granite was considered to have been derived from crustal anatexis of metagraywacke (or intermediate-acid igneous rock with similar mineral composition), rather than basaltic lower crust.

Model F: Melting of a high Sr/Y (and La/Yb) source
Studies have shown that when a source rock has a high Sr/Y ratio, the high Sr/Y signature of the derived magma can inherit from their source, regardless of pressure (Kamei et al., 2009;Moyen, 2009;Ma et al., 2015). The Zhoujiapuzi granite has 345 similar K2O/Na2O ratios and Al2O3 contents to the Tsutsugatake intrusion (Fig. 14h), which is explained by partial melting of arc-type tonalite or adakitic granodiorite (Kamei et al., 2009). We suggest that partial melting of high Sr/Y Liaoji granite was most probably the origin of the high Sr/Y Zhoujiapuzi granite, as discussed below (Model F).
Among the inherited zircons from Zhoujiapuzi granite, the 207 Pb / 206 Pb ages of all the spots are between 2132 and 2200 Ma, except one, and yield a Concordia upper intercept age of 2167 Ma. Both assimilation of country-rocks and incomplete melting of source rocks can explain the genesis of inherited zircon in granite. Due to the similar TDM2 of syn-magmatic zircons (ESZ and LSZ) and the~2.17 Ga inherited zircon, the~2.17 Ga inherited zircons most likely come from the source of the Zhoujiapuzi granite. In the study area, meta-sedimentary rocks and meta-volcanic rocks of the South Liaohe Group, 2.17 Ga mafic rocks, as well as the Liaoji granites, have~2.17 Ga zircon. In spite of an age peak of~2.17 Ga in detrital zircon age spectra of the metasediments from the South Liaohe Group, melting of a sediment-dominated source is unlikely to 355 have occurred, as it would have also introduced other age peaks such as~2.03 Ga and~2.09 Ga (Li et al., 2015). In addition, given the I-type characteristics of the Zhoujipuzi granite, derivation from an igneous precursor is more plausible rather than a metasedimentary origin (Chappell and White, 1992). Therefore, these~2.17 Ga zircon from Zhoujiapuzi granite is unlikely to come from the South Liaohe Group. As shown in the host rock discrimination diagrams (Fig. 15, introduced by Belousova et al., 2002), all the~2.17 Ga inherited zircons from Zhoujiapuzi granite fall into the granitoid area (Fig. 15), precluding that 360 these~2.17 Ga zircon come from the~2.17 Ga mafic rocks. In addition, the~2.17 Ga inherited zircons from Zhoujiapuzi granite and the zircons from the Liaoji granites lie in a similar area in the εHf(t)-age (Ma) diagram (Fig. 9). Hence, the~2.17 Ga inherited zircon most likely come from the Liaoji granites.
Some of the Liaoji granites, such as the Muniuhe granite (comprising granodiorite and syenogranite with no distinct boundary between the two), have adakitic signatures, and similar REE and trace element patterns as the Zhoujiapuzi granite 365 (Fig. 4). Based on a model of modal batch melting (Shaw, 1970) using the experiments of Conrad et al. (1988), the high Sr/Y characteristic of the Zhoujiapuzi granite can be explained by partial melting of Muniuhe granitic pluton leaving amphibole as the main residue (Fig. 13b). In this model, we choose the XY-005 sample to approximately represent a primitive melt composition because the fractional crystallization of plagioclase decreased of Sr content and Sr/Y ratio (Fig. 13a). To find the best matching experimental melts, we have compared the major elements of the XY-005 sample with that of 370 experimental melts, coupled with the initial magma temperature of~800 ℃ and the characteristics of no garnet residue discussed above. Results are shown in Fig. 13b. The Sr and Y compositions of the starting material used in these experiments resemble those of the average composition of the Muniuhe granitic pluton (Sr=475 ppm, Y=9.77 ppm), if the residue contains a large volume of amphibole (>90 %). However, if more plagioclase is retained in the residue (e.g. 18.3 %), a source region with a higher Sr content is required. Therefore, a similar high Sr/Y Liaoji granite to the Muniuhe granitic 375 pluton can produce the high Sr/Y signatures of the Zhoujiapuzi granite.
A large number of Yanshanian adakites (or high Sr/Y rocks) are developed in the NCC, which are generally considered to be derived from the thickened basaltic LCC (e.g. Gao et al., 2004;Wu et al., 2005;Ma et al., 2013). Zhang et al. (2001Zhang et al. ( , 2003 suggested that these so-called "C-type adakites" indicated a large-scale crustal thickening event, so it was speculated that a Mesozoic plateau once existed in the eastern China. However, according to the studies on the Triassic and Jurassic 380 adakitic rocks near the Pingquan area, the northern part of the NCC, Ma et al. (2012Ma et al. ( , 2015 suggested that the adakitic signatures of these rocks are inherited from their source rocks. Similar results have been obtained by studying the late Jurassic Zhoujiapuzi granite in the Liaodong Peninsula in this study. Therefore, we suggest that melting of a high Sr/Y (and La/Yb) source is an important process for the generation of Yanshanian high Sr/Y rocks in the NCC. This kind of high Sr/Y

Tectonic implications
A large number of Early Jurassic arc-like igneous rocks occur in the northeast part of NCC-Korean Peninsula-Hida belt, which belong to the middle-high K calc-alkaline series and are characterized by the enrichment in LILE and depletions in HFSE (Wu et al., 2007;Tang et al., 2018 and references therein). In addition, the Early Jurassic accretionary complexes in the eastern margin of the Eurasian continent and the Japan islands, such as the Heilongjiang complex, the Khabarovsk 390 complex and the Mino-Tamba complex, are considered to be related to subduction (Wu et al., 2007;Tang et al., 2018 and references therein). It is generally accepted that the Paleo-Pacific slab subducted westwards in the Early Jurassic (Tang et al., 2018;Zhu and Xu, 2018).
In the middle-late Jurassic, I-type granites are dominant in the Liaodong Peninsula, such as the Zhoujiapuzi granite (this study), Heigou pluton, Gaoliduntai pluton (Wu et al., 2005a), Waling granite (Yang et al., 2015b) and Sanguliu granite 395 . There are not A-type granites, and mantle derived magmatism is extremely rare. These granites were formed by partial melting of crustal materials without obvious contribution of mantle derived magma (Wu et al., 2005a;Yang et al., 2015bYang et al., , 2018Xue et al., 2020). In addition, WNW-ESE compression during 157-143 Ma was widespread in the Liaodong Peninsula (Yang et al., 2004;. It not only mylonitized the granite plutons in middle-lower crust levels, but also intensely deformed the thick sedimentary cover in the upper crust (Qiu et al., 2018;Ren et al., 2020). Low 400 angle subduction will lead to the separation of lithospheric mantle and asthenospheric mantle, which will lead to tectonic compression, a cold orogeny and a lack of mafic magmatism . Hence, the Late Jurassic magmatism in the Liaodong peninsula is most likely to be related to the thinning of the NCC mantle lithosphere, itself triggered by the subduction of the Paleo-Pacific plate . A likely setting is a mature continental arc, with crust previously thickened by compressional tectonics, related to both the oceanic subduction and the earlier Mesozoic collisions at the north 405 and south margins of the NCC. There is a potential resemblance to the modern arc of the Central Andes (Allmendinger et al., 1997), where crustal thickening and plateau growth has developed over the Cenozoic (Scott et al., 2018), and melting of older basement has taken place during subduction of the Nazca plate (Miller and Harris, 1989).

Conclusion
(1) LA-ICP-MS zircon U-Pb dating indicates that the Zhoujiapuzi granite in the Liaodong Peninsula formed during the 410 Late Jurassic (158-162 Ma).
(2) Zircon growth in Zhoujiapuzi granite can be divided into two distinct stages, the ESZ and LSZ. The ESZ was formed in a deeper, hotter, magma chamber, which had low oxygen fugacity and high temperature. Whereas, the LSZ formed from later, more evolved, magma. Oxygen fugacity significantly increased and the temperature decreased at this https://doi.org/10.5194/se-2021-129 Preprint.