The dikes are dark gray in color, massive to poorly vesiculated, and show a trachytic texture with plagioclases phenocrysts representing 80–85 vol% of the total minerals (Figures 3E,F). Clinopyroxene and biotite account for 5–10 vol% of the phenocrysts. The groundmass includes fine-grained plagioclase, pyroxene, amphibole, zircon (Figures 3G,H), apatite, and magnetite microlites. No significant differences in mineral assemblage, texture, and crystal content have been recognized between the two dikes.

Zircon was mounted in epoxy discs, polished to expose the grains, cleaned ultrasonically in ultrapure water, and then cleaned again prior to the analysis using AR grade methanol. Pre-ablation was conducted for each spot analysis using 5 laser shots (∼0.3 μm in depth) to remove potential surface contamination. The analysis was performed using 30 μm diameter spot at 5 Hz and a fluence of 2 J/cm². Zircon U–Pb dating was conducted using LA-MC-ICP-MS (laser ablation multiple collector inductively coupled plasma mass spectrometry) in Nanjing Hong Chuang Geological Exploration Technology Service Company Limited (NHGETSCL). The Resolution SE model laser ablation system (Applied Spectra, United States) was equipped with ATLEX 300 excimer laser and a two volume S155 ablation cell. The laser ablation system was coupled to a Neptune Plus MC-ICP-MS (Wrexham, Wales, United Kingdom). Detailed tuning parameters can be seen in Thompson et al. (2018). Iolite software package was used for data reduction (Paton et al., 2010). Zircon 91500 and GJ-1 were used as primary and secondary reference materials, respectively. 91500 was analyzed twice and GJ-1 analyzed once every 10–12 analyses of the sample. The age of GJ-1 (604 ± 6 Ma, 1 σ) measured during the experiment was consistent with the recommended values within the uncertainty range, and the preferred value of 91500 is in Wiedenbeck et al. (1995); the ²⁰⁷Pb/²⁰⁶Pb and ²⁰⁶Pb/²³⁸U age of 91500 is 1065.4 and 1062.4 Ma. Based on the decay coefficient, Th/U ratio, and the error of the measurements in young zircons, we corrected the age loss caused by ²⁰³Th depletion (Zou et al., 2010). The LA-MC-ICP-MS U–Pb isotopic data are listed in Table 1.

Experiments of in situ Hf isotope ratio analysis were conducted using the Teledyne Cetac Technologies Analyte Excite laser ablation system (Bozeman, Montana, United States) and the Nu Instruments Nu Plasma II MC-ICP-MS (Wrexham, Wales, United Kingdom) that were hosted at the State Key Laboratory for Mineral Deposits Research, Nanjing University. The 193 nm ArF excimer laser, homogenized by a set of beam delivery systems, was focused on zircon surface with fluence of 6.0 J/cm². Ablation protocol employed a spot diameter of 50 um at 8 Hz repetition rate for 40 s. Helium was applied as carrier gas to efficiently transport aerosol to MC-ICP-MS; the results of the zircon Hf isotopic analyses are given in Table 2. The precision of sample test is 0.000005–0.000012 (0.015–0.04‰, 2 s), and the accuracy is better than 0.000 012 (∼0.04‰).

The geochemical sample analyses were performed at the State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences. The major element oxides were analyzed by a Panalytical PW4400 X-ray fluorescence spectrometer (AxiosmAX, Holland). The detailed sample-digesting procedure was as follows: 1) sample powder (200 mesh) were placed in an oven at 105°C for drying of 12 h. 2) ∼1.0g dried sample was accurately weighted and placed in the ceramic crucible and then heated in a muffle furnace at 1000°C for 2 h. After cooling to 400°C, this sample was placed in the drying vessel and weighted again in order to calculate the loss on ignition (LOI). 3) 0.6 g sample powder was mixed with 6.0 g cosolvent and 0.3 g oxidant in a Pt crucible, which was placed in the furnace at 1150°C for 14 min. Then, the sample was quenched with air for 1 min to produce flat discs on the 5fire brick for the XRF analyses. Trace elements and rare earth elements (REEs) were analyzed by Perkin-Elmer ELAN 300D ICP-MS (United States). The detailed sample-digestion procedure refers to Sun et al. (2013). The analytical precisions for major elements and trace elements were better than 3 and 5%, respectively, and the major and trace element compositions of the trachybasalt dikes (CBS-5a, CBS-5b, CBS-5c, CBS-5a-1, CBS-5b-1, CBS-6a, and CBS-6b) are given in Table 3.

Whole-rock Pb isotopic measurements were conducted by a MAT-261 thermal ionization mass spectrometer. All chemical preparations were performed on class 100 work benches within a class 1000 over-pressured clean laboratory. Sample digestion is as follows: 1) sample powder (200 mesh) was placed in an oven at 105°C for drying of 12 h; 2) 50–200 mg sample powder was accurately weighed and placed in a Teflon bomb; 3) 1–3 ml HNO₃ and 1–3 ml HF were added into the Teflon bomb; 4) Teflon bomb was put in a stainless steel pressure jacket and heated to 190 °C in an oven for >24 h; 5) after cooling, the Teflon bomb was opened and placed on a hotplate at 140°C and evaporated to incipient dryness, and then 1 ml HNO3 was added and evaporated to dryness again; and 6) the sample was dissolved in 1.0 ml of 1.0 M HBr. After centrifugation, the supernatant solution was loaded into an ion-exchange column packed with AG resin. After complete draining of the sample solution, columns were rinsed with 1.0 M HBr to remove undesirable matrix elements. Finally, the Pb fraction was eluted using 6.0 M HCl and gently evaporated to dryness prior to mass-spectrometric measurement. The sample analyses were performed at the Analytical Laboratory of the Beijing Research Institute of Uranium Geology (BRIUG), the precision of ^20xPb/²⁰⁴Pb is (0.002–0.025%, 1σ), and the accuracy is better than 0.03%. The results of the whole-rock Pb isotopic analyses are given in Table 4.

The results of Sr–Nd isotope analyses are reported in Table 5. All chemical preparations were performed on class 100 work benches within a class 1000 over-pressured clean laboratory; the experimental process refers to Pan et al. (2020). The REE solution was used to separate the Nd fraction by the Nd-column method. Sr and Nd isotopes analyses were performed on a Nu Plasma II MC-ICP-MS (Wrexham, Wales, UK) at the Analytical Laboratory of the BRIUG, Beijing, China. The precision of the test sample = 0.000005–0.00002 (0.01–0.05‰, 1σ), better than 0.000025 (∼0.05‰). Raw data of isotopic ratios were corrected for mass fractionation by normalizing to ⁸⁶Sr/⁸⁸Sr = 0.1194 for Sr, ¹⁴⁶Nd/¹⁴⁴Nd = 0.7219 for Nd, and ²⁰⁵Tl/²⁰³Tl = 2.3885 for Pb with exponential law. International isotopic standards (NIST SRM 987 for Sr, JNdi-1 for Nd, and NIST SRM 981 for Pb) were periodically analyzed to correct instrumental drift. Geochemical reference materials of USGS BCR-2, BHVO-2, AVG-2, and RGM-2 were treated as quality control.

Zircons from the trachybasalt dikes (sample CBS-5a and CBS-6a) are light gray with no inclusions. They are euhedral–subhedral columnar crystals with size of 50–200 μm (aspect ratio of 1:1 to 3:1). In cathodoluminescence (CL) images, the zircons exhibit low, homogeneous luminescence and occasionally display oscillatory zoning (Figure 4). Th/U is 0.8–2.5, a range of values consistent with a magmatic origin (Hoskin and Schaltegger, 2003). The 25 analyses of zircon grains from CBS-5a and CBS-6a give weighted mean ²⁰⁶Pb/²³⁸U ages of 1.19 ± 0.0057 Ma (MSWD = 2.00, n = 015) and 1.20 ± 0.0048 Ma (MSWD = 0.89, n = 10), respectively (Figures 5B,D). Because the magmatism in the Tianchi area occurred in the late Cenozoic, the content of common Th, U, and Pb in the Th-U-Pb system of zircons may be considered high, and the content of ²³⁵U decays to ²⁰⁷Pb is low. Therefore, ²⁰⁶Pb/²⁰⁸U is selected as the weighted average age. Tera-Wasserburg concordia diagrams were used to calculate the concordant age (1.19 ± 0.0030 Ma and 1.20 ± 0.0038, respectively, Figures 5A,C). This age is taken to represent the emplacement age of the studied dikes.

The spots used for LA-MC-ICP-MS U–Pb dating are the same as those for in situ zircon Lu–Hf isotopic analysis. Twenty-five analyses of the trachybasalt dikes (sample CBS-5a and CBS-6a) yielded ¹⁷⁶Hf/¹⁷⁷Hf of 0.282700–0.282769, εHf(t) = −2.5 to −0.1, and T_DM2 = 1548–1768 Ma (Figure 8A).

The dikes show a trachybasaltic composition (SiO₂ = 49.08–49.35 wt%; Na₂O = 3.54–3.83 wt%, and K₂O = 2.34–2.55 wt%) with high Al₂O₃ (15.67–16.05 wt%) and P₂O₅ (0.70–0.74 wt%) contents and have a potassic affinity (Figures 6A,B). Mantle-normalized trace element (REE) and chondrite-normalized rare earth element patterns are reported in Figures 7A,B. The samples display parallel trace element patterns. They show relative enrichments in LILE and HSFE. The Zr/Nb (6.31–6.35), La/Nb (0.91–0.95), Ba/Nb (15.65–16.21), Ba/Th (149.22–157.57), Rb/Nb (0.94–1.14), K/Nb (421.50–431.55), Th/Nb (0.10–0.11), and Ba/La (16.92–17.33) values are higher than those of MORB, whereas the chondrite-normalized REEs show enrichments in LREE and depletion in HREE; LREE/HREE ranges from 8.34 to 8.48. The dikes are characterized by a low differentiated REE pattern (La_N/Yb_N = 11.49–11.75) and weak Eu negative anomalies (Eu/Eu* = δEu = 0.96–0.97).

The dikes have moderately high ²⁰⁶Pb/²⁰⁴Pb (17.7047–17.7241) values with ²⁰⁷Pb/²⁰⁴Pb = 15.5525–15.5580 and ²⁰⁸Pb^/204Pb = 38.0870–38.0993. In the ²⁰⁷Pb/²⁰⁴Pb vs. ²⁰⁶Pb/²⁰⁴Pb plots (Figure 8B), they lie within or close to the values representative of the lower crust.

Five samples of the trachybasalt dikes were analyzed for their whole-rock Sr–Nd isotopic composition. They have evolved radiogenic isotope compositions with (⁸⁷Sr/⁸⁶Sr)_i = 0.705165–0.705324 and (¹⁴³Nd/¹⁴⁴Nd)_i = 0.512552–0.512607 corresponding to εNd (t) values of −0.58 to −1.65. According to the data reported in Figure 8C, the Sr–Nd isotope composition of the trachybasaltic dikes approaches that of the coeval basaltic rocks in the CMVF region.

Application of U–Pb disequilibrium dating of zircons with contrasting stabilities and compositions can provide a unique perspective to study the evolution of magma within the framework of absolute time (Wu et al., 2004). All dike outcrops in the field consist of hypabyssal basalts (Figures 3A,B). All dated zircons from the dikes are euhedral–subhedral columnar crystals and display oscillatory zoning; their Th/U is 0.8–2.5, a value consistent with a magmatic origin. Although the zircon crystals of the two dikes could be interpreted as antecrysts, however, their petrographic features do not support such hypothesis. The growth of zircons in poorly evolved magmas may be due to different processes including (Borisova et al., 2020) the partial melting of hydrated peridotite, the felsic crustal involvement/recycling or zircon sealing, total zircon dissolution, and reprecipitation and/or a contribution of metasomatic Si-F-rich fluid. Our data do not allow us to better constrain the origin of the CMVF zircons studied here. As preliminary hypothesis, crustal recycling (see Discussion) or metasomatic processes in a hypabyssal environment could be responsible for the formation of the zircon crystals of CMVF dikes. Field observations suggest that dikes were formed by basaltic magma hypabyssal process (Figures 3A,B), consistent with the euhedral–subhedral columnar morphology and oscillatory zoning of all the dated zircon crystals. Single grain LA-MC-ICP-MS zircon U–Pb concordance ages of 1.19–1.20 Ma (n = 25) and ²⁰⁶Pb/²³⁸U ages (1.18–1.20 Ma) are here obtained from the trachybasaltic dikes and confirm that the age is 1.18–1.20 Ma. We conclude that studied dikes emplaced in the Calabrian stage of Pleistocene. As previously reported, the last three main stages of the Tianchi volcanism are 2.8–0.3 Ma (shield stage), 1.5–0.01 Ma (cone construction stage), and 0.2 Ma-present (caldera-forming stage) (Cheong et al., 2019; Zou et al., 2021). Therefore, our new geochronological data show that the trachybasalt dikes emplaced between the shield and cone forming stages (Figures 9A-C). The trachybasalt dikes, which are coeval with the less evolved basaltic magmatism, evidence an early phase of evolution processes within the CMVF magmatic system, at least in the Tianchi norther sector. Our field data better constrain the role of regional tectonics in controlling the CMVF volcanism during Calabrian. In fact, the NW-SE strike of dikes mirrors that of the major structural discontinuity affecting the CMVF area, i.e., the Baishan-Jince fault, and the preferred alignment of the monogenetic volcanoes (Figure 1). As a result, the CVMF Calabrian magmatism developed along a deep discontinuity probably affecting the whole crust along which basaltic magma upwelled and partly differentiated. The SiO₂ content and the values of A/CNK, MgO, Fe₂O₃, TiO₂, P₂O₅, and V of the dikes confirm that they represent poorly evolved products of mantle melts.

As known, intraplate magmatism can be due to different processes including 1) asthenosphere upwelling carried by mantle plume associated or not to a hot spot, 2) lithosphere thinning, continental stretching, and consequent upwelling of the lithospheric mantle, 3) fluid release from a subducting plate and mantle metasomatism, and 4) crust-mantle interaction (Defant and Drummond., 1990; Puffer, 2001; Sisson et al., 2005). The Sr, Nd, Pb, and Hf isotopic rations (Figure 7C; Figures 8B,D), Zr/Hf (41.35–42.69), Nb/Ta (16.41–16.98), Th/U (4.53–4.56), Th/La (0.11), Nb/U (43.49–44.58), and Lu/Hf (0.05), and the ratio of other trace elements (Figure 7A,B; Figure 8; Zhao, 2016) of the Calabrian CVMF dikes (Figure 8B) are consistent with an OIB-type mantle source (Figure 10B). In particular, the values of the Sr and Nd isotopic ratios (Figure 7C) overlap those of the CVMF basaltic rocks, for which an OIB source has been recognized (Zhang et al., 2018; Choi et al., 2020). However, the trace element pattern indicates that the source is more complex. The (Na₂O + K₂O)/δEu (6.52–6.63), Nb/La (1.05–1.09), La/Yb (10.36–17.61), Th/Nd (0.11–0.15), and δCe (0.96–0.97) values and incompatible trace element abundances such as Th, U, Nb, and Ta suggest the involvement of an enriched mantle type I (EMI) (Figures 10C,D) and some interactions with crustal material (Figure 10A). The CVMF Pleistocene basaltic rocks consist of early sodium-rich rocks to later potassic rocks, and the data obtained in this study indeed indicate an alkaline potassic affinity of the rocks emplaced during Calabrian. According to Wang et al. (2017) and Lei et al. (2020), the EMI signature of the NE China potassic volcanism could be due to fluids released by the Pacific sediments of the subducted Pacific slab. More in general, potassic rocks as those studied here are generally considered to be derived from the partial melting of a mantle modified by slab-derived fluids or melts from the subcontinental lithospheric mantle (SCLM) (Foley and Peccerillo, 1992). The last stage of the Pacific subduction plate between NE China and Japan may favor the dehydration of the oceanic crust allowing the partial melting of the mantle. In this framework, the interaction between hydrous fluid and the mantle can produce a phlogopite-bearing source whose partial melting may produce potassic magmas (Wyllie and Sekine, 1982; Condamine and Médard, 2014). Generally, melts in equilibrium with phlogopite are expected to have higher Rb/Sr (>0.1) and lower Ba/Rb (<20) values than those formed from amphibole-bearing mantle sources (Furman and Graham, 1999), and the early Pleistocene basaltic rocks of the study area have low Rb/Sr (0.07–0.09) and Ba/Rb (14.24–16.59) values. This suggests that potassic amphibole is the main K-rich phase. Because the mantle material is depleted of Zr and Hf, it is impossible for the mantle derived magma to crystallize zircon without the addition of crustal material. As a result, the occurrence of zircon in the mantle derived magma indicates that some crust-mantle interactions occurred during the magma emplacement (Zheng et al., 2006). The studied trachybasaltic dikes are characterized by negative εHf(t) values (−0.1 to −2.5), ¹⁷⁶Hf/¹⁷⁷Hf = 0.000588–0.003769, and the Mesoproterozoic two-stage Hf model ages are from 1548 Ma to 1768 Ma (Figure 6A). This indicates that there is some crustal material involved in the magma generation processes, i.e., crustal contamination. On the other hand, the consistent negative Nd isotopic compositions (εNd(t) = −0.58 to −1.67) of the trachybasalt dikes preclude their derivation from the melting of an old lower continental crust (T_DM1 = 1021–1139 Ma). The variation ranges of the (⁸⁷Sr/⁸⁷Sr)_I and εNd(t) values in the mafic rocks in the region are 0.70485–0.70619 and +0.02 to −1.65, respectively. These values indicate an initial magmatic source consisting of OIB-EMI type melts (Figure 8C). Furthermore, the Pb-Pb isotopic data (Figure 8B) of the trachybasalt dikes and CVMF basaltic rocks support a mixed source consisting of mantle and ancient lower crust components. In the εHf(t)-εNd(t) diagram (Figure 8D), all the samples fall into the overlapping area of the global lower crust and OIB. The results of the modeling of assimilation of lower crust material in Figure 8C suggest that the studied trachybasalts assimilated about 10–20% of crustal material.

The late Cenozoic basaltic rocks in the NE China continental margin have geochemical compositions consistent with OIB-type and EMI sources (Zhou and Armstrong, 1982; Xu et al., 2005; Zhi et al., 1990; Liu et al., 2015; Wang et al., 2015; Mao and Wang, 2000; Liu et al., 2008; Chen et al., 2009; Andreeva et al., 2014; Zhang et al., 2009; Ge et al., 2001; Wang et al., 2011). However, the CMVF Calabrian dikes studied here have moderate Mg^# (27.85–28.09) and higher Na₂O and K₂O contents with respect to the less evolved CMVF basaltic rocks (Figure 6). Therefore, the dikes do not represent primary magmas and experienced some crystal fractionation. Despite these features, they show geochemical features, e.g., values of Na₂O/K₂O and isotopic ratios, and some trace element contents, comparable with those of the CMVF less evolved rocks (Figure 6, Figure 7B, and Figures 8B–D). This allows us to discuss the significance of the OIB and EMI signatures of the Pleistocene studied dike in the intraplate geodynamic context of NE China. Seismic tomography studies show that there is a high velocity anomaly in the mantle transition zone (MTZ) below NE China at 410–660 km depth, which is interpreted as the slab of the Pacific plate in the MTZ, and a low velocity body above the flat portion of the slab interpreted as a big mantle wedge (BMW) (Li and Hilst, 2010; Tian et al., 2016; Wang et al., 2020). From the middle Cenozoic, the subducted plate gradually subsided into the MTZ and partly melts. The resulting water-rich melt and the slab-derived fluids may metasomatize the overlying mantle. However, the EMI signature of the studied CMVF dikes cannot be explained by slab-derived sediments and associated metasomatizing fluids (Xu et al., 2012) because of their moderate Sr and low Pb isotopic ratios. In accordance with the geochemical signature of the gas emitted in NE China (Xu S. et al., 2013b) and of the Cenozoic basalts (Zhang et al., 2009), the EM1 source could be the subcontinental lithospheric mantle (SCLM) (Figure 10D; Zhu et al., 2011). Regarding the source of the late Cenozoic initial OIB magma in the continental margin of NE China, this source could be a response to the back-arc extension and rifting caused by the roll-back of the Pacific plate (Li and Hilst, 2010; Zheng et al., 2018). Some geophysical studies suggest, however, that the Pacific slab is broken and an OIB-type mantle plume upwells from the lower mantle crossing the slab and entering MTZ (Tang et al., 2014). This model is not supported by shear-wave splitting studies (Lynner and Long, 2014) which show trench-parallel sub-slab fast splitting directions and not, as expected for a mantle plume from the lower mantle, a vertical direction. Based on geochemical data, Kuritani et al. (2009) propose that a back-arc type rifting associated to the slab retreat and progressive flattening may favor the decompression melting of the asthenosphere. This model could explain the OIB signature of the CMVF basalts including the trachybasaltic dikes studied here. In this scenario, the CMVF OIB melts upwell from the asthenosphere to the lithosphere, where they mix to SCLM melts. These melts accumulate at the base of the lower crust in the early Pleistocene, where they may assimilate lower crust material of the ancient Asia continent (Figure 8A and Figure 10A). However, assimilation processes could affect the mantle melts during their upwelling and storage in the crust. Tomographic data in NE China (Zhu et al., 2019) clearly show a thickening of the Moho due to the accumulation of mantle melts at the base of the crust below CMVF. These data suggest active underplating process where these mantle melts may promote the melting of the lower crust. Therefore, we propose that most of the assimilation of crustal material underwent by the CMVF trachybasalts occurs at the Moho depth and not during their ascent and storage in the crust.