Origin of Late Cenozoic Basaltic Magmatism in Inner Mongolia, NE China: Constraints From Sr–Nd–Hf–Pb–Mo–He Isotopes

This paper presents a study of the late Cenozoic Chifeng basalts (CBs) of NE China, including their olivine He isotopic compositions, whole‐rock major‐ and trace‐element contents, and whole‐rock Sr–Nd–Hf–Pb–Mo isotopic compositions, with the aim of constraining their mantle source. Results show that the basalts have high MgO, low CaO contents, and high FeOT/MnO values, which indicate that their mantle lithology was most likely pyroxenite. The CBs also exhibit ocean‐island‐basalt‐like trace‐element patterns (e.g., enrichment in light rare earth elements and high‐field‐strength elements) and have depleted Sr–Nd–Hf and relatively radiogenic Pb isotopic compositions, requiring both depleted and enriched components in their mantle. The low olivine He (3He/4He = 0.8–5.5 Ra) and whole‐rock Mo (δ98/95Mo = −0.71‰ to −0.18‰) isotopic values of the CBs, together with geophysical evidence, indicate that the rocks were derived from a depleted MORB mantle (DMM) enriched by recycled oceanic crust that was sourced from the mantle transition zone (MTZ). During the late Cenozoic, ascending wet mantle plumes triggered by dehydration of a stagnant Pacific oceanic slab are inferred to have transported preexisting recycled Pacific oceanic crust from the MTZ into the overlying asthenosphere mantle. The upwelling Pacific oceanic crust reacted with asthenospheric mantle peridotite (i.e., DMM) to produce mantle pyroxenite, whose partial melting at shallow depths generated the CBs. Considering the low δ98/95Mo values of both the CBs and coeval potassic basalts from NE China, we speculate that there may be a low δ98/95Mo reservoir in the MTZ beneath NE China.

During the late Cenozoic, extensive intra-plate basaltic volcanism related to continental rifting occurred in the passive continental margin of NE China.This volcanism has generated mainly potassic and sodic OIB-like basalts in NE China (e.g., J. Liu et al., 2001).Previous studies of Sr-Nd-Hf-Pb radiogenic isotopic compositions have demonstrated that the mantle source of OIB-like basalts in NE China requires both depleted and enriched components (e.g., Basu et al., 1991;Xu et al., 2018;H. Zou et al., 2000).However, the use of conventional Sr-Nd-Hf-Pb radiogenic isotopes may not effectively constrain these two end-members in the mantle source of OIB-like basalts in NE China, which has led to different hypotheses regarding their compositions.For example, it has been proposed that the depleted end-member in the mantle source of late Cenozoic OIB-like basalts in NE China is depleted MORB mantle (DMM, Kuritani et al., 2011;Xu et al., 2018;H. Zou et al., 2000), FOZO (Y.Chen et al., 2007;J.-Q. Liu et al., 2022), or ancient primordial mantle (Xue et al., 2019), whereas it has been suggested that the enriched end-member is continental lithospheric mantle veined by hornblendite (Guo et al., 2016(Guo et al., , 2020;;Yu et al., 2018) or recycled oceanic crust with or without sediments (H.Chen et al., 2017;Choi et al., 2017;Kuritani et al., 2011;Lei, Guo, Sun, et al., 2020;Lei, Guo, Zhao, et al., 2020;X.-C. Wang et al., 2015;M. Zhang & Guo, 2016).Owing to this uncertainty regarding their source end-members, the genesis of these OIB-like basalts remains poorly understood.
Helium has two stable isotopes; that is, 3 He, which is of primordial origin, and 4 He, which has both a radiogenic origin due to U and Th decay and a primordial origin.Basaltic rocks with high 3 He/ 4 He values are expected to result from relatively undegassed mantle regions with low (Th-U)/He ratios.In contrast, basaltic rocks with low 3 He/ 4 He values are expected to originate from relatively degassed mantle regions with high (Th-U)/He ratios and subsequently acquire low 3 He/ 4 He values over geologic time (e.g., Graham, 2002;Kurz et al., 1982).On the basis of this rationale, it is further argued that DMM-derived basalts (MORBs) have a nearly constant 3 He/ 4 He value (8 ± 2 Ra, where Ra is the atmospheric 3 He/ 4 He ratio; Graham, 2002).In contrast, basalts derived from FOZO or ancient primordial mantle have high 3 He/ 4 He values (>30 Ra, Farley et al., 1992;Garapić et al., 2015;Jackson et al., 2014;Stuart et al., 2003).Therefore, He isotopes provide a robust means of discriminating between DMM, FOZO, and the ancient primordial mantle.
The Chifeng basaltic lavas are one of several major late Cenozoic OIB-like basaltic units in NE China.A detailed study of the petrogenesis of the Chifeng basalts (CBs) should provide insights into the nature of mantle end-members of late Cenozoic OIB-like basalts in NE China.This study presents olivine He isotopic compositions, whole-rock major-and trace-element contents, and whole-rock Sr-Nd-Hf-Pb-Mo isotopic compositions of the Chifeng basaltic lavas to place constraints on the mantle end-members of NE China basaltic magmas and to establish a petrogenetic model for the basaltic rocks.

Geological Setting and Samples
The Xing'an-Mongolia Orogenic Belt (XMOB) is the eastern segment of the Central Asian Orogenic Belt, which was amalgamated with several minor blocks (e.g., Xing'an, Songliao, Jiamusi) between the Siberia and Baltica cratons to the north and the Tarim and North China cratons to the south (e.g., Xiao et al., 2015).The NE China lies within the eastern portion of the Paleozoic Central Asian Orogenic Belt XMOB (Figure 1a).One of the widely distributed regions of late Cenozoic volcanic rocks surrounds the Songliao Basin and runs along the Yilan-Yitong and Fushun-Mishan faults in NE China, including the Jingpohu, Changbaishan, Longgang, Wudalianchi, Nuominhe, Chifeng, Chaihe-aershan, Halaha, and Abaga volcanic fields (Figure 1b, J. Liu et al., 2001).These volcanic rocks are predominantly basalts in association with minor andesites (Y.Chen et al., 2007;Ho et al., 2013;Kuritani et al., 2011;J. Liu et al., 2001).
The Chifeng volcanic field lies in the southern part of Greater Xing'An Mountains (Figure 1a).The basaltic lavas in the Chifeng area mainly consist of 100-450 m-thick flows overlain by Quaternary sediments and cover a total area of 3,000 km 2 (Han et al., 1999).The Ar-Ar dating results have shown that except for the volcanic activities in the southern part of the Chifeng volcanic field that erupted at ∼25 Ma, the dominant volcanic activities erupted during 10 ∼3 Ma (e.g., X.-C.Wang et al., 2015).Twenty-one samples of CBs in the volcanic field were sampled in this study, which show typical porphyritic texture and contain less than 10% phenocrysts comprising olivine and some clinopyroxene (Figures S1a and S1b in Supporting Information S1).The details of locations and ages for these studied samples are summarized in Table S1 in Supporting Information S1.

Helium Isotopes Analysis
He isotopes were measured on glasses separated from seven volcanic rocks at the Atmosphere and Ocean Research Institute, the University of Tokyo, Japan.For He analysis, small chips of glass from the chilled margin of lavas were selected to minimize the occurrence of phenocrysts and optimize their purity.The selected phenocryst crystals were cleaned ultrasonically with successive treatments in diluted acid (6.5% HNO 3 ), deionized water, and high-purity acetone.A sieved and weighed aliquot of crystals (1.0-2.0 g) was put in a stainless-steel bowl and placed in a crusher.He isotopic compositions in fluid inclusions that were trapped in phenocrysts of volcanic rocks were released by invacuo single-step crushing at about 200 bar.He isotopes ( 3 He and 4 He) were measured using a Noblesse mass spectrometer (Helix SFT, Thermo Scientific).The olivine samples exhibited a range of helium gas concentrations, varying from 1.2 × 10 −9 to 2.6 × 10 −8 cm 3 4 HeSTP/g.The precision of the measurements for the 3 He/ 4 He ratios on olivines spanned from ±0.03 to ±0.56 Ra.The detailed analytical procedures are provided in M. Zhang et al. (2021).

Whole-Rock Major, Trace Elements and Sr-Nd-Pb-Hf Isotopic Geochemistry Analysis
After removing the weathered or altered surfaces, the samples were broken into thin chips.Then, these chips were powdered in an agate mill to 200 mesh.Whole-rock major and trace element analyses were conducted at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (GIGCAS), Guangzhou, China and Guizhou Tongwei Analytical Technology Company Limited in Guizhou, China, respectively.Whole-rock major and trace element analyses were determined by X-ray fluorescence spectrometry and inductively coupled plasma mass spectrometry respectively.For major elements, the analytical errors are within 5%.Precision for trace elements is better than 10% based on repeated analysis of USGS standards BCR-2 and BHVO-2.The measured values for rock standards are provided in the Table S2 in Supporting Information S1.
The Micromass Isoprobe MC-ICP-MS at Guizhou Tongwei Analytical Technology Company Limited in Guizhou, China, was used to conduct whole-rock Sr-Nd-Hf-Pb isotope analyses.Approximately 50-100 mg of the sample powders were carefully weighed into a round-bottom Teflon capsule.The weighed sample powders were then digested using a concentrated mixture of HF + HNO 3 .The Teflon capsule containing the sample was opened, and the solution was evaporated on a hot plate.Once the solution was evaporated, HNO 3 was added to the sample, and the mixture was evaporated to dryness again.The resulting dry sample was mixed with a combination of HCl + HNO 3 , and the sealed capsule was left on a hot plate overnight for further processing.The detailed chemical separation for Sr-Nd-Hf-Pb isotopes can be found in Yu et al. (2022).Sr and Nd were separated using conventional ion exchange columns, and the Nd fraction was further separated using di-(2-ethylhexyl) phosphoric acid (HDEHP)-coated Kef columns.Hf separation was performed using a modified ion exchange single-column containing LN-Spec resin.Pb separation was achieved using anion resin columns in HBr media The measured Sr, Nd, and Hf isotope ratios were normalized to 86 Sr/ 88 Sr = 0.1194, 146 Nd/ 144 Nd = 0 .7219,and 179 Hf/ 177 Hf = 0.7325, respectively.Samples of Pb isotopes were doped with Tl, and the fractionation correction was performed using the 205 Tl/ 203 Tl ratio of 0.23875.The isotopic ratios of two geological reference materials (BCR-2 and BHVO-2) measured during the analytical procedure can be found in Table S3 in Supporting Information S1.

Whole-Rock Mo Isotopes Analysis
The chemical separation of Mo was conducted at the Guizhou Tongwei Analytical Technology Company Limited in Guizhou, China, following the protocols outlined in J. Li et al. (2014).To begin, 0.1-0.2g of sample powder was spiked with a double isotope tracer ( 97 Mo-100 Mo) and digested in HF + HNO 3 in beakers at 150°C for 3 days.N-benzoyl-N-phenylhydroxylamine (BPHA) resin was utilized for Mo separation.Molybdenum isotope data were acquired with a Neptune Plus MC-ICP-MS at GIGCAS and the double-spike method was used to correct instrumental mass bias.Results were expressed as δ 98/95 Mo relative to the NIST SRM 3134 standard with external reproducibility of ±0.06‰ (2SD; n = 34).Mo contents were determined using the isotope dilution method.The Mo contents and Mo isotopes of IAPSO seawater and geological reference materials GSR-3 and AGV-2 are given in the Table S4 in Supporting Information S1.

Results
The results of olivine He isotopes (Table S5 in Supporting Information S1), whole rock major-trace element, and whole rock Sr-Nd-Pb-Hf-Mo isotopic compositions (Table S6 in Supporting Information S1) are given in the Supporting Information.

Whole-Rock Major, Trace Elements and Sr-Nd-Hf-Pb Isotopic Geochemistry
The CBs have high MgO (7.0-9.9 wt.%), low SiO 2 (45.7-51.0wt.%), and high Na 2 O + K 2 O (3.5-5.4 wt.%) contents, which are predominantly concentrated in the fields of trachybasalt and basalt (Figure 3).In the chondrite-normalized REE diagram (Figure 4a), the CBs exhibit the enrichment of LREEs relative to heavy rare earth elements with (La/Yb) N ratios of 5.7-16.9 and have no obvious Eu anomalies (Eu* = 0.96-1.05).In the primitive mantle-normalized trace element pattern (Figure 4b), the CBs are characterized by the enrichment of the large ion lithophile elements (LILEs; e.g., Ba, Sr) and have the enrichment of the high field-strength elements (HFSEs; e.g., Nb, Ta), resembling the geochemical characteristics of OIBs.

Role of Crustal Contamination and Fractional Crystallization in the Formation of the CBs
Unlike the parental magmas of oceanic basalts, which do not typically undergo crustal contamination during ascent, those of continental basalts may undergo some degree of crustal contamination as a result of passing through the thick continental lithosphere.However, the following lines of evidence suggest that crustal contamination did not play a significant role    in the generation of the CBs.The Nb/U ratios of continental basalts are similar to those of MORB and OIB (47 ± 10, Hofmann et al., 1986) and much higher than those of the continental crust (3.9, Rudnick & Gao, 2014), precluding the substantial crustal contamination in the formation of the CBs.
In addition, considering that an ancient crustal basement is found in the XMOB (e.g., Pang et al., 2019) and assuming that the parental magmas of the CBs were affected by crustal contamination, the MgO content of the CBs should display a negative correlation with 87 Sr/ 86 Sr ratios and positive correlations with εNd and εHf values, which is not observed (Figures S2a-S2c in Supporting Information S1).Furthermore, given that continental crust has high La/Nb values (Rudnick & Gao, 2014), a negative correlation between La/Nb and 206 Pb/ 204 Pb ratios would be expected if the crustal contamination had played an important role in the petrogenesis of the CBs; however, this is not observed (Figure S2d in Supporting Information S1).
Moderate variations in the MgO contents of the CBs suggest that their parental magmas may have undergone fractionation of mafic minerals (e.g., olivine and clinopyroxene).In addition, the positive correlation between Ni and MgO (Figure S3a in Supporting Information S1) implies that the CBs underwent fractional crystallization of olivine.Furthermore, the positive correlation between CaO/Al 2 O 3 and MgO (Figure S3b in Supporting Information S1) suggests that clinopyroxene fractionation may have occurred during the formation of the CBs.Fractionation of plagioclase was negligible, as demonstrated by the lack of a positive correlation between Al 2 O 3 and MgO (Figure S3c in Supporting Information S1) and the absence of negative Eu and Sr anomalies in REE patterns (Figures 4a  and 4b).
In summary, geochemical evidence indicates that parental magmas of the CBs underwent fractional crystallization of olivine and clinopyroxene, with negligible fractionation of plagioclase.These results are consistent with the petrographic observations presented above (Figures S1a and S1b in Supporting Information S1).

Mantle Lithology of the CBs
It has been shown that the whole-rock major-element compositions of primitive basalts can be used to infer their mantle lithology (e.g., Lee et al., 2009).The least evolved basaltic samples of the CBs (i.e., those with MgO contents of >8.5 wt.%) were selected to infer mantle lithology.Previous studies have demonstrated that basaltic melts derived from peridotite should have higher CaO contents than those of pyroxenite-derived basaltic melts at given MgO contents, reflecting the higher compatibility of CaO within pyroxenite relative to peridotite (Herzberg, 2011).The CBs have CaO contents similar to those of melts derived from mantle pyroxenite at given MgO contents and lower than those expected for melts derived from mantle peridotite (Figure 7a).These characteristics suggest that the CBs might have been derived from pyroxenite.It has been argued that the FeO T / MnO ratios of basalts can be used to constrain their mantle lithology, as Fe/Mn ratios show fractionation between garnet (or clinopyroxene) and melt but not between olivine (or orthopyroxene) and melt when these minerals are involved in melting or crystallization (Herzberg, 2011).On this basis, it is inferred that peridotite-derived melts have lower FeO T /MnO ratios (50-60) than those of melts derived from pyroxenite (>60, Herzberg, 2011).The CBs have high FeO T /MnO ratios of 65-83 (Figure 7b), which also suggests that the study rocks were derived from pyroxenite.The FC3MS parameter (defined as FeO T /CaO − 3MgO/SiO 2 ) of basalts can be used to infer their mantle lithology (pyroxenite or peridotite) and is not affected by temperature or pressure (Z.-F.Yang & Zhou, 2013).Experimental melts derived from pyroxenite and peridotite have average FC3MS values of 0.46 and −0.07, respectively (Z.-F.Yang & Zhou, 2013).In general, the FC3MS values of melts originating from peridotite do not exceed 0.65 (Z.-F.Yang & Zhou, 2013).The CBs have FC3MS values of 0.69-0.93,which are within the range of pyroxenite-derived melts (Figure 7c), further implying that their mantle source was pyroxenite.As Fe and Zn exhibit markedly different partition behaviors between pyroxene and olivine during partial melting (Le Roux et al., 2010), it has been proposed that basaltic melts derived from pyroxenite have higher 10,000 × Zn/ Fe ratios (>12.5)than those derived from peridotite (8.5-12.5, Le Roux et al., 2010).The CBs have Zn/Fe ratios of 13.1-14.4,which are highly similar to those of melts derived from a pyroxenite source and higher than melts derived from a peridotite source (Figure 7d), also indicating that the mantle lithology of the CBs was mantle pyroxenite.
The 3 He/ 4 He values of the CBs are slightly lower than those of MORBs but much lower than those of basalts derived from FOZO or ancient primordial mantle (Figure 2), indicating that the depleted mantle in the mantle source of the CBs is most likely DMM rather than FOZO or ancient primordial mantle.Both FOZO and ancient primordial mantle are sourced from the lower mantle, possibly the core-mantle boundary, through mantle plumes (Farley et al., 1992;Garapić et al., 2015;Jackson et al., 2014).However, to date, no geophysical evidence has supported the presence of a mantle plume rooted in the lower mantle beneath NE China during the late Cenozoic (e.g., Zhao et al., 2009), suggesting that the depleted component in the mantle source of the CBs is more likely to be DMM rather than FOZO or ancient primordial mantle.
In comparison with DMM-derived MORBs, the CBs have enriched incompatible elements (LILEs and LREEs) and Sr-Nd-Hf isotopic compositions (Figures 5a-5d), the formation of which requires an enriched component in their source.In the following section, we focus on the enriched end-member in the mantle source of the CBs.

Enriched Component in the Mantle Source of the CBs
Recycled oceanic crust and metasomatized lithospheric mantle containing hornblendite veins are common enriched components in the mantle sources of OIBs (e.g., Halliday et al., 1995;Hofmann, 2014;Niu & O'Hara, 2003;Pilet et al., 2008;White & Hofmann, 1982).It has been argued that both of these components constitute the enriched component in the mantle source of the CBs (Guo et al., 2016(Guo et al., , 2020;;Pang et al., 2019;X.-C. Wang et al., 2015).On the basis of the following considerations, we propose that the enriched end-member in the mantle source of the CBs was most likely recycled oceanic crust with some sediments.
1.The model of recycled metasomatized lithospheric mantle proposes that hornblendite veins that formed during the percolation and differentiation of low-degree asthenospheric melts within the lithospheric mantle could be a candidate for the source of OIBs (Halliday et al., 1995;Niu & O'Hara, 2003;Pilet et al., 2008).However, metasomatized minerals (e.g., amphibole) are not observed in mantle peridotite xenoliths from the XMOB (e.g., D. Zou et al., 2014).In addition, because amphibole is a Ti-rich mineral in the mantle, the melting of hornblendite veins would generally result in OIBs with relatively high TiO 2 values (>3%, Pilet et al., 2011), which are inconsistent with the relatively low TiO 2 values (<3%) of the CBs.In addition, amphibole-bearing lithospheric-mantle-derived basaltic rocks normally have high Ba/Rb values (>20, Furman & Graham, 1999;Pilet et al., 2011), which are also inconsistent with the relatively low Ba/Rb values (14-20, except for one sample with a value of 22) of the CBs. 2. OIBs derived from metasomatized lithospheric mantle with hornblendite veins typically have mantle-like O isotopic compositions (δ 18 O = ∼5.3‰;Eiler, 2001), as hornblendite veins that formed by the percolation and differentiation of low-degree asthenospheric melts are generally characterized by mantle-like O isotopic compositions (e.g., Zeng et al., 2019).In contrast, relative to normal mantle, OIBs derived from a mantle source with recycled oceanic crust would display either heavier or lighter O compositions because the upper part of the recycled oceanic crust could have high δ 18 O values (>7‰) as a result of the influence of low-temperature water-rock interaction, whereas the lower part of the recycled oceanic crust could have low δ 18 O values (3‰-5‰) on account of the influence of high-temperature water-rock interaction (e.g., Eiler, 2001).A previous study has shown that the CBs have predominantly low δ 18 O values (3.0‰-5.2‰,X.-C.Wang et al., 2015), implying that the mantle source of the CBs may have included a contribution from recycled oceanic crust, more specifically, lower oceanic crust.3.During slab subduction, the slab-dehydration process occurs at sub-arc depths and supplies fluid-mobile elements (e.g., Th and U) to the overlying mantle wedge, whereas fluid-immobile elements (e.g., Nb and Ta) are retained in the residual oceanic crust (Rudnick et al., 2000).This process might be the only known process that can effectively fractionate Nb from Th and Ta from U. Therefore, basalts with values of (Ta/U) N > 1 and (Nb/Th) N > 1 might be related to the contribution of dehydrated ocean crust in their mantle source (Niu & Batiza, 1997).The high values of (Ta/U) N > 1 and (Nb/Th) N > 1 of the CBs are consistent with their mantle source containing recycled dehydrated oceanic crust (Figure 8).Fractionation of hydrous minerals (e.g., amphibole) has commonly been invoked to explain Mo isotopic variations of mafic rocks (Liang et al., 2017;Voegelin et al., 2014).The absence of an amphibole phase from the CBs may indicate that fractionation of amphibole did not occur in these basaltic magmas (Figure S1 in Supporting Information S1).In addition, amphibole has lower δ 98/95 Mo values than the melts (Wille et al., 2018), and Dy is more compatible than Yb in amphibole (e.g., Davidson et al., 2007).This means that the fractionation of amphibole would cause a decrease in the Dy/Yb values and an increase in the δ 98/95 Mo values of basaltic melts, resulting in basaltic magmas displaying a negative correlation between δ 98/95 Mo and Dy/Yb.However, a negative correlation between the δ 98/95 Mo and Dy/ Yb values of the CBs is not observed (Figure 9a), which is inconsistent with possible modification of their Mo isotopic compositions by fractional crystallization of amphibole.
Magmatic sulfides have either higher (Voegelin et al., 2012) or lower (Hin et al., 2022) δ 98/95 Mo values than those of their equilibrated melts, suggesting that sulfide fractionation would cause the melts to have lower or higher δ 98/95 Mo values, respectively.Given that Cu partitions more strongly than Mo into sulfides (Y.Li & Audétat, 2012), the fractionation of sulfides would decrease Cu/Mo ratios while decreasing or increasing δ 98/95 Mo values in melts.However, such a positive or negative correlation between Cu/Mo and δ 98/95 Mo is not observed in the CBs (Figure 9b), signifying an insignificant influence of sulfide fractionation on their Mo isotopic variations.Furthermore, it has been inferred that the mantle source beneath the XMOB is relatively oxidized (ΔFMQ up to 2.9, J.-Q. Liu et al., 2022), which is not an ideal condition for sulfide stability; therefore, the presence of residual sulfides in the mantle source of the CBs is unlikely.It is inferred that the effect of sulfides on the Mo composition of the CBs during magmatic fractionation and partial melting of the mantle was negligible.
Given the above considerations, it is concluded that the low and variable δ 98/95 Mo values of the CBs were not induced by alteration and weathering, fractional crystallization, or partial melting, but were more likely inherited from their mantle source.As demonstrated above, the mantle source of the CBs has both DMM and enriched  components, and the low and variable δ 98/95 Mo values of the CBs probably resulted from the enriched component.Given that the enriched component in the mantle source of the CBs may be either continental lithospheric mantle containing amphibole-bearing metasomes or recycled ancient oceanic crust (e.g., Guo et al., 2016Guo et al., , 2020;;Pang et al., 2019;X.-C. Wang et al., 2015), here we explore which one of these two candidates more likely accounts for the low δ 98/95 Mo component of the CBs.Chen et al., 2022;Hin et al., 2022) and could have served as the source of the CBs.However, Ce and Mo are not strongly fractionated during partial melting (especially low-degree partial melting) of the mantle, which would result in near-constant Ce/Mo ratios in mantle-derived melts (e.g., Willbold & Elliott, 2023).Therefore, such residual mantle after low-degree melt extraction would yield normal-mantle-like Mo/Ce ratios, which is also not a suitable component for the CBs displaying low Mo/Ce ratios.3. It has been demonstrated that slab residues should have a low δ 98/95 Mo signature (S.Chen et al., 2019;Freymuth et al., 2015).This is because Mo is a fluid-mobile element that is inclined to be partitioned into aqueous fluids during slab dehydration (Freymuth et al., 2015).In addition, heavy 98 Mo is preferentially transported into hydrous fluids, leading to Mo isotope fractionation between slab-derived fluids and oceanic slab residues, which would result in low δ 98/95 Mo values of oceanic slab residues (e.g., Willbold & Elliott, 2017).Indeed, it has been argued that natural eclogites (e.g., eclogites from the Raspas Complex in Ecuador), which are dehydrated and partially melted residues of subducted oceanic crust, typically have very low δ 98/95 Mo values (down to −1.01‰, Ahmad et al., 2021;S. Chen et al., 2019).Therefore, the partial melting of a mantle source with recycled oceanic crust, as is the case for oceanic eclogites, could also have the potential to generate basaltic melts with low δ 98/95 Mo values (S.Chen et al., 2019).Thus, the low δ 98/95 Mo values of the CBs may indicate that their mantle source contained recycled oceanic crust.In addition, the CBs have low Mo/Ce ratios (Figure 6), which are consistent with the contribution of eclogites to their source, as eclogites generally have extremely low Mo/ Ce ratios (0.011, S. Chen et al., 2019).This result is consistent with a recent Mo isotope study of OIBs with low δ 98/95 Mo and Mo/Ce values from La Palma and Hawaii, which have been ascribed to a recycled oceanic crust component in the mantle source of these rocks (Willbold & Elliott, 2023).Furthermore, two samples of the CBs with high δ 98/95 Mo values suggest that their source could also have contained recycled sediments, in particular black shales, which have high δ 98/95 Mo values (δ 98/95 Mo = 0-1.7‰,Freymuth et al., 2016;Xu et al., 2012) and would generally be recycled into the mantle together with the oceanic crust.This interpretation is consistent with the fact that the CBs have high contents of Ba and Sr (Figure 4b).To estimate the proportion of subducted oceanic components in the mantle source of the CBs, we conducted geochemical modeling of their Mo-Nd-Hf isotopic compositions using an approach similar to that of Yu et al. (2022).The calculations show that the mantle source of the CBs contains 5%-25% recycled oceanic material comprising 85%-98% recycled oceanic crust and 2%-15% oceanic sediments (Figures 10a and 10b).The variable Mo isotopic compositions of the CBs may indicate differing proportions of recycled oceanic crust in their source.
On the basis of the above discussion, we propose that a predominant amount of recycled oceanic crust and a subordinate amount of pelagic sediments constitute the most likely candidate for the enriched end-member with low δ 98/95 Mo in the mantle source of the CBs.

Integrated Petrogenetic Model and Implications
The mantle beneath NE China has undergone multiple subduction events during the Archean (∼2 Ga) and Phanerozoic (Kuritani et al., 2011;X.-J. Wang et al., 2017;Xiao et al., 2003).The mantle source of HIMU basalts requires recycled oceanic crust that is related to an ancient subduction event, as HIMU basalts typically have highly radiogenic Pb isotopic compositions ( 206 Pb /204 Pb > 20.0, Zindler & Hart, 1986).However, compared with HIMU basalts, the CBs have much lower 206 Pb/ 204 Pb values (Figures 5c and 5d), suggesting that the recycled oceanic crust in their mantle source is probably related to a recent rather than ancient subduction event.It has also been demonstrated that recycled oceanic crust (e.g., MORBs and gabbros) has lower Lu/Hf and Sm/Nd ratios than those of the MORB mantle source and that Lu/Hf fractionation exerts a greater control on Hf isotopic evolution than does Sm/Nd fractionation on Nd isotopic compositions (Hanyu et al., 2011;Salters & White, 1998).Consequently, if the mantle source contains ancient (>1 Ga) recycled oceanic crust, its partial melting would result in basalts with negative ΔεHf anomalies, which would plot below the mantle array in the Nd-Hf isotope diagram (Hanyu et al., 2011;Salters & White, 1998).The fact that the CBs do not show pronounced negative ΔεHf anomalies and plot approximately along the mantle array in an Nd-Hf isotope diagram (Figure 5b) also indicates that recycled oceanic crust in the mantle source of the CBs was introduced by a recent (Phanerozoic) subduction event.
It is generally accepted that NE China has undergone southward subduction of the Paleo-Asian Plate and westward subduction of the Pacific Plate during the Phanerozoic (e.g., S. Li & Wang, 2018;Q. Ma & Xu, 2021;Xiao et al., 2003).Both of these geodynamic processes could have provided the inferred recycled oceanic crust in the mantle source of the CBs.The following considerations suggest that the recycled oceanic crust in the mantle source of the CBs is composed of Pacific oceanic crust.Several tomographic studies have interpreted that the subducted Pacific Plate may have sunk and then stagnated in the mantle transition zone (MTZ, 410-660 km) during at least the late Cenozoic, as judged by the presence of a broad and large-scale high-velocity zone in the MTZ beneath eastern China (Figure 1c, Fukao et al., 1992;J. Li et al., 2013;X. Liu et al., 2017;Zhao, 2004;Zhao et al., 2009).According to geophysical data, the western edge of the stagnant Pacific slab has reached the MTZ beneath the Daxing'anling-Taihangshan gravity lineament (DTGL) in eastern China (Figure 1c).The Chifeng volcanic field is situated next to the DTGL and is closely aligned with the surface projection of the western boundary of the stagnant Pacific slab (Figure 1c, J. Huang & Zhao, 2006;Zhao et al., 2009).The ages and some geochemical features (SiO 2 contents and Sm/Nd ratios) of the CBs are correlated with distance from the western edge of the stagnant Pacific slab (Pang et al., 2019;X.-C. Wang et al., 2015), whereby the CBs display westward younging, depletion in silica contents, and lower Sm/Nd ratios (X.-C.Wang et al., 2015).These patterns show that the formation of the CBs was spatially related to the stagnant Pacific slab.Thus, the recycled oceanic crust in the mantle source of the CBs was most likely part of the Pacific Plate rather than the Paleo-Asian Plate.(Frey et al., 1991); and for sediment, δ 98/95 Mo = 0.3‰ (Freymuth et al., 2015), εNd = −10, and εHf = −12.3(Plank, 2014).
Accordingly, we propose that the enriched component in the mantle source of the CBs is recycled Pacific oceanic crust and some sediments resident in the MTZ.
Seismic tomography studies and geodynamic simulations have revealed a prominent low-velocity anomaly extending from the upper mantle down to ∼410 km beneath NE China, which may represent ascending wet mantle plumes originating from the MTZ (J.Huang & Zhao, 2006;J. Yang & Faccenda, 2020;Zhao et al., 2009).
During the late Cenozoic, ascending wet mantle plumes that were probably triggered by the continuous dehydration of the stagnant Pacific slab are inferred to have transported recycled oceanic materials from the MTZ into the upper asthenosphere (Figure 11a).Owing to the lower solidus temperatures of recycled oceanic crust compared with mantle peridotite (Hirschmann, 2000;Kogiso et al., 2004), such crust would have melted preferentially relative to ambient peridotite mantle to produce silicic melts.These silicic melts would have reacted with asthenospheric mantle peridotite (i.e., DMM) to produce pyroxenite by forming orthopyroxene at the expense of olivine, which is referred to as stage-2 pyroxenite (Herzberg, 2011).The newly produced stage-2 pyroxenite ascended to shallow depths and underwent partial melting to produce the CBs (Figure 11b), consistent with pyroxenite as their mantle lithology, as inferred above.
A previous study of late Cenozoic potassic rocks in NE China has suggested that their mantle source was also characterized by a low δ 98 / 95 Mo component; that is, recycled oceanic crust with sediment in the MTZ (L.Ma et al., 2022).Therefore, our study provides complementary evidence for the possible existence of a low δ 98/95 Mo reservoir (recycled oceanic crust with some sediments) in the MTZ beneath NE China.Further studies of Mo isotopes of other late Cenozoic continental OIB-like basalts (e.g., the Abaga and Halaha basalts) in NE China will help test this hypothesis.

Conclusion
We conducted a study of the late Cenozoic CBs of NE China, including analyses of olivine He isotopic compositions, whole-rock major-and trace-element contents, and whole-rock Sr-Nd-Pb-Hf-Mo isotopic compositions, with the aim of constraining the mantle source of the rocks.The results show that the CBs have high MgO and low CaO contents and high Fe/Mn values, indicating that their mantle lithology was pyroxenite.The CBs are also characterized by OIB-like geochemical characteristics, low 3 He/ 4 He and δ 98/95 Mo values, and depleted Sr-Nd-Hf and relatively radiogenic Pb isotopic compositions.These geochemical features require both depleted and enriched components in their mantle sources.Integrating our new geochemical results and available geophysical evidence, we argue that the depleted component in the mantle source of the CBs is DMM and that the enriched component is recycled oceanic crust that originated from the MTZ.During the late Cenozoic, upwelling wet mantle plumes triggered by dehydration of the stagnant Pacific slab ascended from the MTZ to the upper mantle.These plumes are hypothesized to have transported preexisting recycled Pacific oceanic crust from the MTZ into the overlying asthenospheric mantle, which would have reacted with asthenospheric mantle peridotite (i.e., DMM) to produce mantle pyroxenite.Subsequent partial melting of this

Figure 1 .
Figure 1.(a) Simplified geological map of eastern China, showing the spatial distribution of tectonic plates and late Cenozoic basalts; (b) simplified map of sample locations in NE China (modified from J. Liu et al. (2001)); and (c) high-resolution P-wave model showing the inferred stagnant Pacific Plate in the mantle transition zone beneath Asia along latitude (modified from J. Huang and Zhao (2006)).XMOB: Xing'an-Mongolia Orogenic Belt and DTGL: Daxing'anling-Taihangshan gravity lineament.

Figure 2 .
Figure 2. Diagram of 3 He/ 4 He ratio versus He content for olivines of the Chifeng basalts.

Figure 4 .
Figure 4. Chondrite-normalized REE and primitive-mantle-normalized multi-element variation diagrams for the Chifeng basalts.Normalizing values and data for ocean-island basalt are fromSun and McDonough (1989).Sources of previously published data are the same as in Figure3.

5. 5 .
Nature of the Low δ 98/95 Mo Component in the Source of the CBs The CBs have low and variable δ 98/95 Mo values, and several mechanisms may generate basaltic rocks with such Mo isotopic compositions, including (a) alteration and weathering, (b) fractional crystallization of hydrous minerals, (c) the influence of sulfides during magmatic fractionation and partial melting, and (d) source heterogeneity caused by recycled materials.The CBs have low loss-on-ignition (LOI) values (<3 wt.%), indicating a minimal alteration.In addition, there was no correlation between the δ 98/95 Mo values or Mo/Ce ratios of the CBs and their chemical weathering index (CIA) or LOI values (Figures S4a-S4d in Supporting Information S1), also implying that the alteration and weathering processes had only a limited impact on the Mo isotopic composition of the CBs.

Figure 9 .
Figure 9. Diagrams of (a) δ 98/95 Mo versus Dy/Yb and (b) δ 98/95 Mo versus Cu/Mo for the Chifeng basalts.Sources of previously published data are the same as in Figure 3.

Figure 8 .
Figure 8. Diagram of (Ta/U) N versus (Nb/Th) N for the Chifeng basalts.Sources of previously published data are the same as in Figure 3.
1.A recent study of K-rich mafic rocks derived from lithospheric mantle enriched by Ca-poor pelagic sediments during oceanic subduction has revealed that these mafic rocks have low δ98/95 Mo values (down to −0.45‰, F.Huang et al., 2023).This finding implies that recycled lithospheric mantle enriched by pelagic sediments could be a potential mantle source of the CBs with low δ 98/95 Mo values.However, K-rich mafic rocks derived from such lithospheric mantle have extremely enriched Sr-Nd-Hf isotopic compositions and high K 2 O/Na 2 O ratios (>1), as recycled sediments are generally characterized by highly enriched Sr-Nd-Hf isotopic compositions and high K 2 O/Na 2 O ratios (>1; e.g., F. Huang et al., 2023), which contrast with the depleted Sr-Nd-Hf isotopic compositions and low K 2 O/Na 2 O ratios (<1) of the CBs. 2. It has been argued that the continental lithospheric mantle beneath NE China is a predominantly juvenile mantle source, similar to oceanic lithospheric mantle, rather than ancient enriched lithospheric mantle (Park et al., 2017; Y.-L.Zhang et al., 2011).A study of E-MORBs from the East Pacific Rise has shown that these basalts have high δ 98/95 Mo values (δ 98/95 Mo = −0.23‰ to −0.06‰), which require two end-members in their mantle source, that is, DMM with normal-mantle-like δ 98/95 Mo values and an enriched end-member with high δ 98/95 Mo values (S.Chen et al., 2022).It has been further argued that the high δ 98/95 Mo values in the mantle source of these enriched MORBs represent recycled mantle lithosphere veined by amphibole-bearing metasomes, which are generated by the percolation and differentiation of volatile-bearing melts within the lithosphere mantle (S.Chen et al., 2022).Thus, mantle lithosphere veined by metasomes is characterized by high δ 98/95 Mo values and could not have constituted the enriched end-member in the mantle source of CBs with low δ 98/95 Mo values.It could be argued that residual mantle resulting from the extraction of low-degree melts of such mantle lithosphere veined by metasomes should have correspondingly low δ 98/95 Mo values (S.

Figure 11 .
Figure 11.Schematic diagram illustrating the proposed model accounting for the petrogenesis of the Chifeng basalts (CBs).(a) Pacific oceanic crust and sediments were transported into the mantle transition zone (MTZ) during the subduction of the Pacific Plate.(b) Upwelling wet mantle plumes transported Pacific oceanic material from the MTZ into the overlying upper asthenosphere mantle, which reacted with asthenospheric mantle peridotite (depleted MORB mantle) to form pyroxenite. Partial melting of pyroxenite at shallow depths produced the CBs during the late Cenozoic.