Geochronology and Sr-Nd-Pb-Hf isotopic geochemistry of middle-late Permian granitic and volcanic rocks within the eastern margin of the Khanka Massif: petrogenesis and implications for the tectonic nature

ABSTRACT The Khanka Massif forms the easternmost part of the Central Asian Orogenic Belt (CAOB) and represents an ideal location for examination of the timing and processes involved in the transition from the Palaeo-Asian to Palaeo-Pacific Ocean tectonic systems. However, it is obstructed because the late Palaeozoic tectonic attributes of the eastern margin of this massif remain unclear. In this paper, we provide zircon U-Pb ages and Hf isotopic, as well as the whole-rock Sr-Nd-Pb-Hf isotopic data for Permian granitic and volcanic rocks within the eastern margin of the Khanka Massif, with the aim of exploring the petrogenesis and their geodynamics significance. These data suggest that the Permian magmatism within the eastern margin of the Khanka Massif, composed of monzogranites and rhyolites, can be split into two stages; i.e. middle (ca. 264 to ca. 263 Ma) and late (ca. 252 Ma) Permian. The highly fractionated middle Permian monzogranites are formed in an N-S extensional setting and represent the magmatic response to the amalgamation of the Khanka and Jiamusi massifs. Middle-late Permian rhyolites with decoupled Nd-Hf isotopic compositions were derived from partial melting of metasedimentary crustal rocks. Their moderate-high zircon saturation temperatures (796–817°C) are thought to represent the involvement of ridge subduction of the Palaeo-Asian Ocean beneath the western margin of the Khanka Massif. Our findings further reinforce the existence of the ridge subduction of the Palaeo-Asian Ocean at the western margin of Khanka Massif in the Permian, while excluding the possibility of Permian Palaeo-Pacific oceanic plate subduction occurred beneath the eastern margin of the Khanka Massif in the Permian.


Introduction
The Central Asian Orogenic Belt (CAOB) is one of the largest Phanerozoic accretionary orogens in the world, and its structural overprinting and tectonic history have been the focus of a substantial amount of recent research (Windley et al. 2007;Kröner et al. 2014;Xiao and Santosh 2014;Wilde 2015). The eastern segment of the CAOB (Figure 1(a), as the typical area recording the accretionary processes, underwent the evolution not only of the Palaeo-Asian Ocean tectonism but also of the subsequent Palaeo-Pacific Ocean tectonism (Wilde 2015;Xiao et al. 2015;Liu et al. 2017;Tang et al. 2018;Wang et al. 2019). However, the transformation between these two tectonic systems and the processes involved in the tectonism remain controversial.
The Khanka Massif (hereinafter referred to as KM), as the easternmost massif in the eastern CAOB, is an ideal region to explore the transformation timing and mechanism of the Palaeo-Asian Ocean and Palaeo-Pacific Ocean tectonic systems (Li 2006;Xu et al. 2009Xu et al. , 2012Xu et al. , 2013. Previous researchers have focused largely on the western margin of the Khanka Massif (hereinafter referred to as WKM), and suggest that the region was in an active continental margin tectonic setting during the Permian in response to the eastward subduction of the Palaeo-Asian oceanic plate beneath the WKM (Cao et al. 2011;Chen et al. 2017;Ma et al. 2019;Xing et al. 2019). However, some scholars hold the different view that the so-called active continental margin was related to the westward subduction of the Palaeo-Pacific plate beneath the eastern margin of the Khanka Massif (hereinafter called EKM) . Recently, it was further identified that the eastward ridge subduction of the Palaeo-Asian oceanic plate beneath the WKM (Wang et al. 2021). This type of ridge subduction generally leads to the inward migration of arc-type magmatism in subduction zone settings, with normal arc-type magmas being erupted and emplaced at distances farther from the subduction zone (Thorkelson and Breitsprecher 2005;Huangfu et al. 2016). It is unconvincing to attribute the active continental margin in the western Khanka Massif to the westward subduction of the Palaeo-Pacific oceanic plate, meaning that understanding the tectonic history of the EKM is vital to supply further restrictions on tectonic evolution in this regard.
This paper shows new zircon U-Pb geochronological and geochemical data for middle-late Permian granitic and volcanic rocks from the EKM. These data provide new insights into whether Permian subduction actually took place along the EKM.

Geological background and sample descriptions
mainly in eastern Lesozavodsk of this massif, dominated by the gneissic granitic rocks (Khanchuk et al. 2010;Tsutsumi et al. 2014). The late Palaeozoic-early Mesozoic can be further divided into four periods: early Permian, late Permian, late Triassic, and early Jurassic. The early Permian magmatism was represented by the Chaoxiantun pluton of Mishan region and the Qianshan pluton in the southern Hunchun area (Cao et al. 2011;Yang et al. 2015). The late Permian magmatism is mainly distributed in the western part of Khanka lake and the Vladivostok area, as represented by granitic rocks (Kovalenko 2006;Khanchuk et al. 2010).
This paper focuses on three samples from the eastern margin of the Russian part of the KM (Figure 1(c,d). Samples 19R5 and 19R6 (43°4ʹ41.09"N,133°02ʹ16.43"E) are rhyolites collected from rocks initially considered to have formed in the early Palaeogene ( Figure 3; Kovalenko 2006) and were collected ~93 km east of Vladivostok. The samples are sufficiently fresh, dark grey, and porphyritic with typical felsitic texture. They contain quartz and minor alkali feldspar (sanidine) phenocrysts within a cryptocrystalline groundmass (Figure 2(a,b).

Zircon U-Pb dating
U-Pb dating and trace elements determination of zircons were concurrently measured by LA-ICP-MS at the Experimental Test Center of Jilin University. The Agilent 7500a ICP-MS was fitted with a 193 nm laser and Helium as carrier gas. Zircon 91,500 was taken as exterior calibration for zircon dating, the NIST 610 was applied for trace element corrections. Each sample analysis consisted of a background collection of around 5 s, which was then 40 s of data capture. The data analysis and correction were performed using ICPMSDataCal and the Isoplot/Ex_ver3 (Ludwig 2003;Liu et al. 2008Liu et al. , 2010. Additionally, the experiments of the 19R6 sample were performed at the Institute of Geology and Geophysics, Chinese Academy of Sciences in Beijing, using a Cameca 1280 SIMS with similar operation and data processing procedures as Li et al. (2009).

Major and trace elements analyses
Samples used for whole-rock analysis were cleaned and altered materials were removed, then crushed to approximately 200 mesh in the agate mill. Major and trace element abundances of whole-rock were acquired at the Wuhan Sample Solutions Analytical Technology Co., Ltd. Major elements were analysed by X-ray fluorescence (XRF) and trace elements were conducted by Agilent 7700e ICP-MS in Teflon bombs after acid digestion of the samples.

Zircon situ Hf isotope analyses
By using the Neptune Plus MC-ICP-MS and Geolas HD excimer ArF laser ablation system, in situ Hf isotope ratio analyses of zircons were finished at the Wuhan Sample Solution Analytical Technology Co., Ltd. The analysis process was also equipped with a signal smoothing device to improve signal stability and isotope ratio measurement precision (Hu et al. 2015). Using helium as the carrier gas, and a small amount of nitrogen was introduced after the ablation cell to improve the sensitivities of hafnium isotopes (Hu et al. 2012). In order to ensure the reliability of the data, the 91,500 was used for external calibration to further optimize the analytical results and using the Gj-1 as the second standard sample to detect the quality of data correction. Refer to Hu et al. (2012) for specific operating requirements of the laser ablation system as well as the MC-ICP-MS apparatus.

Whole-rock Sr-Nd-Pb-Hf isotopes analyses
All chemical procedures including sample dissolution and ion-exchange chromatography of Sr-Nd-Pb-Hf were performed at the Wuhan Sample Solution Analytical Technology Co., Ltd. Whole-rock powders (>200 mesh) were digested in sealed Teflon beakers for column chemistry. After centrifuging, the supernatant solution was charged into an ion-exchange column packed with AG50W resin. After complete draining of the sample solution, columns were rinsed with 2.5 N HCl to rid it of undesirable matrix elements. The Sr-Nd-Pb-Hf isotope analysis were done on a Neptune Plus MC-ICP-MS. Originally exploited for TIMS measuring which is still extensively adapted and used by MC-ICP-MS, the exponential law was used to evaluate instrumental analysis in the present study (Russell et al. 1978). The 88 Sr/ 86 Sr ratio of 8.375209 and the 146 Nd / 144 Nd ratio of 0.7219 were the results of internal normalization and thus mass discrimination correction (Lin et al. 2016). The NIST SRM 987 was evaluated once for every ten samples processed. Analysis of NIST SRM 987 resulted in the same 87 Sr/ 86 Sr ratio (0.710244 ± 22) within the error range as the announced values (0.710241 ± 12; Thirlwall 1991). The JNdi-1 was determined once for every ten samples processed, and residual interferences of 144 Sm + were corrected by Lin et al. (2016). Analysis of JNdi-1 resulted in the same 143 Nd / 144 Nd ratio (0.512118 ± 15) within the error range as the announced values (0.512115 ± 07; Tanaka et al. 2000). Mass discrimination correction for Pb isotopes analyses was obtained by normalizing to a 205 Tl/ 203 Tl ratio of 2.38714 (the certified value of NIST SRM 997). Due to the difference in mass bias behaviours of Pb and Tl, all measured 20x Pb/ 204 Pb ratios of unknown samples were normalized to the well-accepted NBS SRM 981 values of 208 Pb/ 204 Pb = 36.7262 ± 3, 207 Pb/ 204 Pb = 15.5000 ± 13, 206 Pb/ 204 Pb = 16.9416 ± 13 (Baker et al. 2004). The NBS SRM 981 was evaluated once for every ten samples processed. Analysis of NBS SRM 981 resulted in the 20x Pb/ 204 Pb ratios (0.03%, 2SD) for external accuracies. The effective sample processing in the early stage separated these interfering elements (Yb and Lu) and corrected for residual 176 Yb + and 176 Lu + interferences (Lin et al. 2016). The JMC 475 and Alfa Hf were, respectively, determined once for every seven samples processed. Analysis of JMC 475 resulted in the same 176 Hf / 177 Hf radio (0.212861 ± 13) within the error range as the announced values (0.282163 ± 21; Weis et al. 2007).

Zircon U-Pb dating
In this work, three samples were selected from different igneous rocks within the KM for zircon U-Pb dating using LA-ICP-MS and SIMS. Cathodoluminescence (CL) images of typical zircons are shown in the supplementary documents, Concordia plots are displayed in Figure 4, and the zircon U-Pb data are listed in Table S1.
The zircons from the rhyolite sample 19R5 are generally hypidiomorphic to idiomorphic and columnar, with high Th/U ratios (0.32-1.46) as well as the fine-scale oscillatory growth zoning can be seen in CL images, indicating a magmatic genesis (Corfu et al. 2003). A total of 22 analyses of yield 206 Pb/ 238 Pb ages that are split into two concordant groups with weighted mean ages of 252 ± 4 Ma (MSWD = 1.17, n = 16) and 332 ± 7 Ma (MSWD = 0.02, n = 3) and include single grain ages of ca. 347, ca. 372, and ca. 410 Ma. The age of 252 Ma is interpreted to denote the formation time of the rhyolite, with the other ages (332, 347, 372, and 410 Ma) representing the timing of crystallization of inherited or captured zircons entrained in the rhyolite prior to its eruption.
Monzogranite sample 14RF43 contains zircons that are generally euhedral-subhedral with fine-scale oscillatory growth zoning visible. This combined with their high Th/U ratios (0.41-1.45), indicates a magmatic genesis (Corfu et al. 2003). A total of 10 analyses yielded one group of concordant 206 Pb/ 238 Pb ages with a weighted mean age of 264 ± 4 Ma (MSWD = 4.5, n = 10) suggests the emplacement time of monzogranite. All of these zircon U-Pb ages indicate that the EKM underwent middle-late Permian magmatism.

Major-and trace-element geochemistry
The contents of the major and trace elements for the three samples examined in this study are shown in Table  S2  and fall in the calc-alkaline series in the K 2 O-SiO 2 diagram ( Figure 5(c,d). The rhyolites are featured by high total rare earth (REE) contents (ΣREE = 131.4-139.5 ppm) as well as enrichment in light REEs (LREEs) and a slight depletion in heavy REEs (HREEs; [La/Yb] N = 5.39-5.65). The Eu/Eu* ratios (0.55-0.65) of the samples reveal negative Eu anomalies (Figure 6(a). The N-MORB-normalized multi-element variation diagram denotes that the samples enrich in large ion lithophile elements (LILEs; e.g. Rb, Ba) and deplete in high field strength elements (HFSEs; e.g. Nb, Ta, Ti, and Sr; Figure 6( O/K 2 O. They are all classified as subalkaline series in a total alkali versus silica (TAS) diagram ( Figure 5(a) and the Zr/TiO 2 -Nb/Y diagram (Figure 5(b). The A/CNK [Al 2 O 3 / (CaO + Na 2 O + K 2 O)] ratios (1.08-1.16) of the rhyolites, which mean they are peraluminous and fall in the calc-alkaline field in the K 2 O-SiO 2 diagram ( Figure 5  (c,d). The rhyolites have high total REE contents (ΣREE = 123.03-133.14 ppm) as well as enrichment in light REEs (LREEs) and LILEs, slight loss of heavy REEs (HREEs; [La/Yb] N = 5.19-5.79) and have significant Ti depletions with minor Nb, Ta, and Sr depletions (Figure 6(b). The Eu/Eu* ratios (0.55-0.59) of the samples show negative Eu anomalies (Figure 6

Zircon Hf isotopes
The Hf isotopic content of the zircons used for U-Pb dating was determined using in situ analysis. The Hf isotopic data for samples 19R5, 19R6, and 14RF43 are given in Table S4 and are shown in Figure (9

Whole-rock Sr-Nd-Pb-Hf isotope geochemistry
The whole-rock Sr-Nd-Pb-Hf isotopic compositions of rhyolite samples 19R5 and 19R6 are provided in Table  S5. The formation ages of these rhyolites were used to calculate the initial isotopic ratios (Tables S1 and S2)

Permian magmatism within the khanka massif
The rhyolites and monzogranite sampled during this study were formerly proposed to have formed during either the early Palaeogene (for the rhyolites) or the Precambrian (for the monzogranite; Kovalenko 2006 (Cao et al. 2011;Yang et al. 2015;Wang et al. 2021). The middle Permian (ca. 268 to ca. 260 Ma) magmatism formed the granodiorites of the Majiajie pluton and the Sidaogou intrusion, the diorite in the Chunhua area, the andesites and dacites of Yanggang Formation in Jixi area, the basalts and rhyolites of Nancun Formation in southwestern Dongning city, and a series of granites of Ryazanovka complex in the southwestern Primor'e region (Khanchuk et al. 2010;Yang et al. 2015;Chen et al. 2017;Ma et al. 2019;Wang et al. 2021). The late Permian (ca. 258 to ca. 252 Ma) magmatism, as represented by a suite of granitoids to the west of Khanka Lake, south of Vladivostok (Khanchuk et al. 2010;Tsutsumi et al. 2014;Kruk et al. 2015), diorite in the Wudaogou pluton in the Hunchun area (Cao et al. 2011), granodiorite and porphyritic granite in the Yangtianzhai and Shuangyehu plutons, respectively , the granodiorite of the Yangjingou pluton (Chen et al. 2017), intermediate and felsic intrusions in the Chunhua area (Ma et al. 2019), and the andesites of Tuntianying Formation and the diorites in north Hunchun area (Wang et al. 2021). Moreover, these igneous rocks with composition varying from mafic to felsic identified along the western margin of the KM displayed a North-South trending belt (Cao et al. 2011;Chen et al. 2017). All of them respond to the active continental margin, and are considered to be related to the subduction of the Palaeo-Asian Ocean plate (Cao et al. 2011;Chen et al. 2017).
All of this suggests that the Permian magmatism in the western and eastern margins of the Khanka Massif record N-S trends in igneous activity along well-defined belts. The Permian magmatism in the western margin can be divided into early (ca. 287 to ca. 273 Ma), middle (ca. 268 to ca. 260 Ma) and late (ca. 258 to ca. 252 Ma) Permian stages, with middle (ca. 264 to ca. 263 Ma) and late (ca. 252 Ma) Permian magmatism also recorded within the eastern margin.

Petrogenesis
The influences of alteration are required to be considered before the petrogenesis of the felsic rocks in the study area can be evaluated. The samples that were analysed in this study are fresh and have low loss on ignition (LOI) ratios (0.29-1.41 wt.%; Table S2), excluding the significant secondary alteration. The monzogranite and rhyolite samples are geochemically similar in that they contain high SiO 2 and Al 2 O 3 in content and low TiO 2 , total Fe 2 O 3 , and MgO in content. However, these rocks do not share a common petrogenetic history, as outlined below.
Middle Permian monzogranite sample 14RF43 has a high differentiation index value (DI = 94; Table S2) that is indicative of formation from a highly differentiated magma (Thornton and Tuttle 1960). This, combined with the low MgO, Fe 2 O 3 , and P 2 O 5 contents of the sample, are also indicative of formation from an evolved magma (Table S2). The monzogranite has total REE contents  ppm; Table  S2) and low LREE to HREE ratios (LREE/HREE = 3.03-3.68; (La/Yb) N = 2.49-2.94) together with significantly negative Eu anomalies (Eu/Eu* = 0.06-0.08). The monzogranite has Rb/Sr ratios of 17.7-19.4, K/Rb ratios of 125-138, Zr/Hf ratios of 12.9-17.1 (<25), and low concentrations of siderophile elements such as V and Co, all of which provide further evidence of the differentiated nature of this intrusion (Halliday et al. 1991;Bau 1996;Claiborne et al. 2006;Breiter et al. 2014;Gelman et al. 2014;Dostal et al. 2015;Ballouard et al. 2016). This is consistent with the fact that the monzogranite sample plots in the highly fractionated field of (total FeO /MgO) versus (Zr + Nb + Ce + Y) and (Na 2 O + K 2 O)/CaO versus (Zr + Nb + Ce + Y) diagrams (Figure 7(a,b). The highly fractionated nature of this intrusion is compatible with the zircon Hf isotopic compositions of the sample, which are highly variable (ε Hf (t) = +0.2 to +5.2; Figure (9). This could be interpreted as the disequilibrium partial melting in the crustal source that did not reach equilibrium during the process (Sparks and Matshall 1986;Wu et al. 2003Wu et al. , 2017Tang et al. 2014;Wang 2017). Highly differentiated granites that crystallize over a long period of time also usually assimilate material from the surrounding country rocks. The crust has an extremely heterogeneous composition and differences in the properties of the source region undergoing partial melting inevitably results in differences in the composition of the resulting magma. This means that highly fractionated granites do not reflect the properties of the source area and the physicochemical conditions of magma formation (Wu et al. 2017). However, the formation of highly fractionated granites is generally associated with hightemperature conditions and volatile-rich components, both of which are associated with extensional tectonic settings (Wu et al. 2017).
The middle-late Permian rhyolites (samples 19R5 and 19R6) have high SiO 2 (73.01-74.32 wt.%) in content with low MgO (0.48-0.72 wt.%) in content and the geochemical features are similar to those of the upper continental crust, suggesting they were derived from a crustal rather than mantle source region (Rudnick and Gao 2003). The rhyolites are enriched in LREEs and LILEs (Rb, Ba) and depleted in HFSEs, with Ti strongly depleted and Nb, Ta, and Sr only relatively depleted, and they have notable negative Eu anomalies (Figure (6a,b) that are largely inherited from their continental crustal source region. The whole-rock Pb isotopic compositions of these rhyolites plot within the mature Island arc section of the lead isotopic source diagram (Figure 10(c); after Doe and Zartman 1979;Tong et al. 2006), similar to the values expected for typical crust-derived felsic rocks (Zartman and Haines 1988). These rhyolites have positive ε Hf (t) values (+2.1 to +13.9). Combining these with their T DM2 of 823-479 and 1036-385 Ma indicates that these rhyolites formed from magmas generated as a result of the partial melting of juvenile crustal material derived from depleted mantle during the early Palaeozoic and the Neoproterozoic. Remarkably, these rhyolites also have decoupled Nd-Hf isotopic compositions (Figure 10(b). Previous researches have revealed that the anomalous behaviour of Hf isotopes and their decoupling from other isotopic systems may be the result of disequilibrium melting of metasedimentary crustal material (Yu et al. 2017;Zhang et al. 2017Zhang et al. , 2018Zhang et al. , 2019Kong et al. 2018). The rhyolites have not experienced the highly fractionated evolution, the decoupling of Nd-Hf, combined with the presence of zircon xenocrysts, jointly reveal a metasedimentary crustal source region (Zhang et al. 2020). This is further supported by their high A/CNK ratios and low ε Nd (t) values (1.08-1.26 and −0.56 to −0.16, respectively; Tables S2 and S5) that contrast with the values for other coeval felsic rocks within this region (Figure 10(a); Ma et al. 2019;Wang et al. 2021). Furthermore, these middle-late Permian rhyolites have high zircon saturation temperatures (796-817°C, with an average value of 807°C; Table S2).
These data imply that the magmas, which formed these rhyolites are the result of the fluid-absent melting of metasedimentary rocks (Brown 2013;Zhang et al. 2019Zhang et al. , 2020. In summary, we draw the conclusion that the middle Permian monzogranite in the study area is a highly fractionated granite, and the primary magmas of the middle-late rhyolites were derived from the partial melting of metasedimentary crustal material.  (Whalen et al. 1987). FG: Fractionated felsic granites; OGT: unfractionated M-, I-, and S-type granites; A: A-type granites; FeO t : as total FeO. Data sources: the samples for the southeastern part of the Jiamusi Massif (Meng et al. 2008). Data sources in the following figures are the same as those in this figure.

Implications for the tectonic nature of the eastern margin of the khanka massif in the permian
The KM is the easternmost massif of the eastern CAOB and represents an ideal region for research into the timing and processes involved in the transition between the tectonic systems of the Palaeo-Asian Ocean and Palaeo-Pacific Ocean tectonic systems. Nevertheless, the Permian tectonic nature of the massif is still controversial (Cao et al. 2011;Yang et al. 2015;Chen et al. 2017;Xing et al. 2019;Wang et al. 2021). In the present study, the Permian felsic rocks were identified and provided the powerful evidence in this regard.
Firstly, the middle Permian monzogranite with highly fractionated features reflected an extensional tectonic setting. However, it is unclear whether this extensional setting was a tectonic response to the subduction of the Palaeo-Asian oceanic plate beneath the WKM or the subduction of the Palaeo-Pacific Plate beneath the EKM Wang et al. 2021). Spatially, this middle Permian monzogranite is located in the northeastern part of the massif and was originally adjacent to the southeastern Jiamusi Massif to the north (Figure 1(a,b). Coeval felsic rocks have also been recognized within the southeastern part of the Jiamusi Massif, representing a magmatic response to the amalgamation of the Jiamusi and Khanka massifs at this time (Figure 8(a,b); Meng et al. 2008). Thus, we consider that the middle Permian monzogranite formed in an N-S extensional setting associated with the collision of the Jiamusi-Khanka massifs rather than in response to subduction along either the eastern or western margins of the Khanka Massif ( Figure 12).
Secondly, the middle-late Permian rhyolites response to the moderate-high temperature and fluid-absent characteristics of melting condition. As indicated by the saturation temperature of zircons, the formation of these rhyolites theoretically occurred at 796-817°C (Table S2; Figure 11), similar to the 'hot' granite that the extra heat source was required (Miller et al. 2003). However, both post-collisional and ridge-subduction settings provide extra heat source for melting of shallow crust because of the upwelling of asthenospheric mantle material Wang et al. 2021). The Permian eastward-directed ridge subduction of the Palaeo-Asian Ocean beneath the WKM (Wang et al. 2021) caused normal arc magmatism to migrate away from the subduction zone, and the spatial scale of the inward migrated arc igneous rocks associated with ridge subduction is extensive accompanying the heat source ( Figure 12; Thorkelson and Breitsprecher 2005;Huangfu et al. 2016). Previous studies have revealed that the felsic magmatism related to the post-collision tectonic settings generally holds the higher zircon saturation temperatures than those in ridge-subduction settings ( Figure 11). In addition, no contemporaneous subduction-related accretionary complexes or arc-type mafic rocks have been reported along the EKM. On the regional scale, previous studies have also identified arcfeatured granites and the contemporaneous mafic rocks at the western margin of the Jiamusi Massif in the Permian, implying that possibly the branch of the Palaeo-Asian Ocean was subducted beneath the Jiamusi-Khanka terrane (Figure 12; Dong et al. 2017aDong et al. , 2017b. Some previous studies have identified the active continental margin settings at the eastern margin of Jiamusi massif in Permian (Meng et al. 2008;Sun et al. 2015). But combined with the absence of the subduction-related response from the eastern margin of KM, it is unconvincing to suggest that this is the westward subduction of the Palaeo-Pacific Ocean beneath the Eurasian plate. Consequently, we conclude that the ridge subduction in west is most likely responsible for the formation of the middle-late Permian rhyolites within the EKM. In summary, the middle Permian granitic rocks identified within the EKM were considered as the magmatic response to the amalgamation of the Khanka and Jiamusi massifs, whereas the formation of the middle-late Permian rhyolites was reasonably related to the ridge subduction of the Palaeo-Asian oceanic plate beneath the WKM. Our findings further reinforce the existence of the ridge subduction of the Palaeo-Asian Figure 10. (a) ε Nd (t) versus Initial 87 Sr/ 86 Sr, (b) whole-rock ε Hf (t) versus ε Nd (t) Terrestrial array equations are from (Vervoort et al. 2011) and (c) 208 Pb/ 204 Pb(t) versus 206 Pb/ 204 Pb(t) diagrams for the middle-late Permian rhyolites (after Doe and Zartman 1979). Data Sources: the samples plot in the MORB field (Hofmann 1997), the samples plot in the N-MORB of west CAOB field  and references therein), the samples plot in the Permian arc-type rocks from the WKM field (Ma et al. 2019). The fields of Chaihe pluton and Chushan pluton are from Wu et al. (2000) .   Cao et al. 2011;Yang et al. 2015;Chen et al. 2017;Dong et al. 2017aDong et al. , 2017bWang et al. 2021) .
Ocean at the western margin of Khanka Massif in the Permian, while excluding the possibility of Permian Palaeo-Pacific oceanic plate subduction occurred beneath the eastern margin of the Khanka Massif in the Permian.

Conclusions
(1) Permian magmatic events in the EKM can be subdivided into middle (ca. 264 to ca. 263 Ma) and late (ca. 252 Ma) Permian stages.
(2) The middle Permian monzogranite is formed as a result of the cessation of the amalgamation of the Khanka and Jiamusi massifs. In comparison, the middle-late Permian rhyolites within the EKM are formed from magmas generated by the partial melting of metasedimentary crustal rocks as a result of the subduction of a ridge within the Palaeo-Asian oceanic plate beneath the WKM. Our findings further reinforce the existence of the ridge subduction of the Palaeo-Asian Ocean at the western margin of Khanka Massif in the Permian, while excluding the possibility of Permian Palaeo-Pacific oceanic plate subduction occurred beneath the eastern margin of the Khanka Massif in the Permian.

Research Highlight
• Permian felsic rocks were identified in the eastern margin of Khanka Massif. • P 2 monzogranite was related to the collision of Khanka-Jiamusi massifs in the north. • P 2-3 rhyolites were related to the ridge-subduction in the west. • The subduction of Palaeo-Pacific plate was excluded in the Permian.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This work was financially supported by the National Natural Science Foundation of China (Grants 42022013) and the Key Special Project for Introduced Talents Team of Southern Marine Science and Engineering Guangdong Laboratory (Guangzhou) (Grant GML2019ZD0202).