Introduction

Rare-earth elements (REEs) are essential for high-technology industries and crucial defense systems. China hosts one-third of the known REE reserves and produces 97% of the REE + Y1 in the world. A wide variety of REE deposits are found in China2,3, among which the carbonatite-associated REE deposits (CARDs), formed by REE-rich fluids exsolved from carbonatitic melts, are the most significant, accounting for approximately 65% of China’s REE reserves. The known CARDs include Bayan Obo (the world’s largest LREE-Fe-Nb deposit4), Maoniuping (the giant LREE deposit5) and numerous medium-large LREE deposits6,7, which are all located along cratonic margins (Fig. 1a) and share salient features, listed in Supplementary Table S1.

Figure 1
figure 1

(a) Tectonic framework showing the spatial distribution of CARDs in China. (b) Sketch geological map showing the Cenozoic carbonatite–syenite complexes with zircon U-Pb ages (for syenite) in the MD REE belt, which is controlled by Cenozoic strike-slip faults. The Jinpingshan orogen was formed by subduction of the downgoing Proto-Tethyan oceanic lithosphere in the Proterozoic11,12. The Proterozoic metamorphic basement was locally covered by Paleozoic-Mesozoic sedimentary strata. NCC: North China Craton; TM: Tarim Block; YC: Yangtze Craton; CC: Cathaysia Craton; CAO: Central Asian Orogen; CCO: Central China Orogen; AHO: Alps-Himalayan Orogen; SGO: Songpan-Ganzi Orogen. The figure was generated using CorelDRAWX4 and the map will not have a copyright dispute.

Full size image

The Cenozoic Mianning-Dechang (MD) REE carbonatite belt in southwest China8 (Fig. 1b) is the best example of CARDs that developed along a cratonic edge; thus, it is an ideal location to test the hypothesis of REE-rich fertile carbonatites. The remnants of Proterozoic meta-volcanic arc rocks and arc granitoids (1000–740 Ma)9,10 exposed along the western margin of the Yangtze Craton suggest the subduction of a Proto-Tethyan oceanic lithosphere beneath the craton in the Neoproterozoic9,10,11,12. A series of Cenozoic strike-slip faults8 and extensive potassic (lamprophyre) magmatism, which peaked at ~35 Ma13 along the western margin of the craton (Fig. 1b), reflect overprinting and reworking associated with the Indo-Asian continental collision that began at 65 Ma14. The MD carbonatites occur as dykes and stocks associated with syenite intrusions that intruded Precambrian crystalline basement15. The spatial distribution of the carbonatites was controlled by Cenozoic strike-slip faults (Fig. 1b). The available age data indicate that the carbonatites and syenites were magmatically active from 40 to 12 Ma15,16, identical to the mineralization ages (34–11 Ma) of the CARDs in southwest China (Table S1)17. These deposits are genetically associated with carbonatitic dykes or stocks with fenitization alteration and have a wide variety of mineralization styles with identical assemblages of bastnaesite + calcite + barite + fluorite5,8, varying from stockwork or vein systems (e.g., Maoniuping) and breccia pipe-hosted systems (e.g., Dalucao) to carbonatite stock systems (e.g., Lizhuang)8.

A previous study found that the MD carbonatites with CARDs have high radiogenic Sr isotopic compositions with (87Sr/86Sr)i values >0.705515, which are extremely rare among the world’s carbonatites18. The sources of the fertile carbonatite have been debate18,19,20,21 and it remains unclear whether the extreme REE enrichment of carbonatites as CARDs is caused primrily by intra-crustal processes or an anomalously REE-rich metasomatic mantle source. The answer to these questions bears on the origin of CARDs. Our new evidence suggests that the MD carbonatites were formed by melting of REE-refertilized metasomatic SCLM. The recycling of marine sediments, a process that introduces abundant LREEs into the SCLM via CO2-rich fluids during mantle metasomatism, is critical to the formation of CARDs.

Results

We performed an integrated geochemical study of key samples from the MD carbonatites and CARDs in southwest China, including analyses of the chemical composition and Sr-Nd-Pb and C-O isotopic composition of the calcite separates and their host rocks. The major and trace element compositions and isotopic compositions are listed in Supplementary Tables S2–5 and presented in Figs. S1–6.

Based on the extensive petrographic study, the ratios of trace elements with similar geochemical behaviors (e.g., Sr/Ba, Rb/Sr, Nb/Y) and isotopic composition of Sr-Nd-Pb and C-O in the fresh bulk-rock samples and calcite separates (Tables S2–4) were used to characterize the MD carbonatites. The analyses show that the MD carbonatites have low SiO2 (<20.92%), FeO (<0.93%) and MgO (<1.50 %) concentrations and a wide range of CaO concentrations (21.41–55.4%) similar to that of calciocarbonatite22 (Fig. S1) but distinct from primary calcite carbonatites derived from direct melting of carbonated mantle sources23. These carbonatites have relatively low but variable concentrations of high field-strength elements (HFSEs; Fig. S2) with invariant Nb/Y ratios (1.0 ± 0.5; Fig. 2a). The carbonatites are extremely enriched in LREEs, with highly variable ratios of (La/Yb)N (29–2889; Fig. S3), high concentrations of Sr (5100–56,300 ppm) and Ba (1070–60,800 ppm), relatively low ratios of Sr/Ba (Fig. 2b) and high ratios of Ba/Th (Fig. 2c). These features are consistent with other CARD-hosted carbonatites (e.g., Bayan Obo24,25, Mountain Pass26 and Laiwu6) but differ from most barren carbonatites worldwide (Figs. 2 and S4).

Figure 2
figure 2

Plots of (a) Ba vs. Nb/Y, (b) Sr/Ba vs. REEs and (c) Ba/Th vs. 87Sr/86Sr for the MD carbonatites and most global carbonatites. The fresh and least-altered carbonatite samples were plotted after eliminating altered and mineralized samples from the dataset (Supplementary Tables S2–5). (a) The majority of the CARD-hosted carbonatites have much lower Nb/Y ratios (1.0 ± 0.5) than most of the barren carbonatites, suggesting that their source enrichments are related to fluid metasomatism because fluid metasomatism does not strongly fractionate chemically similar elements, such as Nb and Y. Certain barren carbonatites have low Nb/Y ratios (≤1.0) and variable Ba concentrations (10–1000 ppm) close to that of ocean island basalts (OIBs) or mid-ocean ridge basalts (MORBs), implying derivation from a mantle plume source and/or interaction with lithospheric material during the generation of these magmas18. (b) The fresh samples show a weak negative correlation between Sr/Ba ratios and REE concentrations. (c) The much higher Ba/Th ratios of the MD carbonatites require the involvement of CO2-rich fluids derived from subducted sediments and oceanic crust. Certain carbonatites (e.g., Laiwu) have 87Sr/86Sr and Ba/Th ratios that overlap with those of marine sediments, suggesting the involvement of subducted sediments during carbonatite formation6. All data from this study (Tables S2–3) and Ying et al. (2004) and Yang et al. (2011).

Full size image

A remarkable feature of the MD carbonatites is that their Sr-Nd isotope compositions are more radiogenic than the majority of global carbonatites with Sr-Nd isotopic ratios similar to that of ocean-island basalts (OIBs) involving high-μ U/Pb mantle (HIMU), enriched mantle I (EMI) and focal zone (FOZO) mantle components27,28,29,30,31. Both the bulk-rock samples and calcite separates yield Sr-Nd isotopic arrays shifted toward marine sediments (Fig. 3a) and away from East African carbonatite HIMU-EMI arrays31 and isotopically overlap with the Cenozoic REE-rich potassic rocks in southwest China13 (Fig. 3a). Such high radiogenic carbonatites are also found in other ore districts, e.g., the Bayan Obo24 and Laiwu6 (Figs. 1a and 3a).

Figure 3
figure 3

(a) Sr–Nd isotopic compositions of the fresh MD carbonatites and their calcite separates. Other CARD-hosted carbonatites (e.g., Bayan Obo24 and Laiwu6) and associated Cenozoic-Mesozoic igneous rocks in China13,38 are shown for comparison with carbonatites from southwest China, East Africa (EACL)27, North America26, Brazil22 and India29. The dotted line outlines the global marine sediment field32. The MD samples yield an Sr-Nd isotopic array shifted toward marine sediment. The DM (depleted mantle), HIMU (high-μ mantle), EMI (enriched mantle I) and EMII (enriched mantle II) fields are distinct mantle end-members39. (b) 207Pb/204Pb vs. 206Pb/204Pb correlation diagram for the fresh MD carbonatites and calcite separates. Carbonatites from East Africa31 and Cenozoic potassic rocks from southwest China13 are shown for comparison. (c) δ13C vs. δ18O correlation diagram for the calcite separates and fresh carbonatites. Previously collected calcite data were re-examined by extensive microscope observations and plotted for comparison with the fresh carbonatites. All the plotted data fall into the fields of mantle-derived primary carbonatites35. Fresh samples from the Dalucao and Lizhang districts feature two distinct coherent trends with weak positive correlations between δ18OVSMOW and δ13CVPDB values. All data from Tables S2–4.

Full size image

This radiogenic signature is also demonstrated by the Pb isotopic compositions of the MD calcite separates and host carbonatites. Their compositions vary between EMI and EMII components but are shifted toward marine sediments32 (Fig. 3b; Fig. S5). The calcite separates have a wide range of 207Pb/204Pb (15.53–15.71) and 208Pb/204Pb (38.32–38.92) ratios15,33,34 that exhibit positive correlations with the 87Sr/86Sr values of the calcite separates (Table S3). However, the bulk-rock samples yield relatively uniform Pb isotopic compositions (Fig. 3b), which differ from most barren carbonatites because of relatively low 206Pb/204Pb ratios27,28 (Fig. S5). Twenty-three calcite separates and 22 fresh carbonatites yielded δ18OV-SMOW values from 6.4% to 9.8% and δ13CV-PDB values between −4.4% and −8.8%, which plot in the field of mantle-derived primary carbonatites35 (Fig. 3c). The samples from individual districts (e.g., Lizhuang and Dalucao) show distinct coherent trends with a weak positive correlation between the δ18O and δ13C values (Fig. 3c).

Discussion

Carbonatites can become enriched in Sr, Ba and REEs through intra-crustal processes, e.g., subsolidus remobilization4 and liquid immiscibility25,26. However, these processes do not explain the REE-Sr-Ba enrichment observed in the MD carbonatites. Our analyses indicate that the least-altered and freshest samples have extremely high concentrations of REEs (>2000 ppm), Sr (>5,000 ppm) and Ba (>1,000 ppm), although minor leaching of Sr and Ba from the MD carbonatites occurs by fenitization alteration. Experiments confirmed that liquid immiscibility can lead to strong partitioning of Sr, Ba and REE into an immiscible carbonatite melts36,37; however, the MD syenites formed by immiscibility15 still yield high concentrations of Sr (up to 4,920 ppm), Ba (up to 7,663 ppm) and REEs (Fig. S3; Table S2). Using the partitioning coefficients of these elements in immiscible carbonatitic-silicate melts36,37, we estimated that their parent magmas have been extremely rich in Sr (>2000 ppm), Ba (>2000 ppm) and REEs (>5000 ppm), suggesting that such high concentrations may be a prerequisite for forming giant CARDs.

The high Sr and low Nd isotopic ratios in CARD-hosted carbonatites have been attributed to three processes: (1) crustal assimilation15, (2) sedimentary carbonate contamination40 and (3) heterogeneous mantle sources6,15,24. Crustal assimilation could elevate 87Sr/86Sr values and Rb/Sr ratios in carbonatites because of high Rb abundance41 and high 87Sr/86Sr ratios in the crust42. However, a positive correlation between 87Sr/86Sr and Rb/Sr ratios is not observed in the MD carbonatites (Fig. S6). The extremely high Sr concentrations (>5000 ppm) in the MD carbonatites may effectively buffer the effect of crustal assimilation on 87Sr/86Sr values18. Pb isotope is the most sensitive indictor for crustal assimilation33, however, the MD carbonatites yield relatively uniform Pb isotopic compositions (Fig. 3b), thus refuting the crustal assimilation hypothesis. Contamination by marine sediments with high 87Sr/86Sr ratios (>0.712)42 and high δ13C values (>−2%)6 could lead to a synchronous increase in 87Sr/86Sr, δ18O and δ13C values in carbonatites (ref. 40). However, such contamination cannot explain why the Dalucao samples have the highest (87Sr/86Sr)i values (Fig. 3a) and lowest δ13C values among the MD carbonatites (Fig. 3c). Although δ13C values are strongly temperature dependent (with heavier values found with decreasing temperatures of carbonatite formation40), the carbonatites from the three ore districts had similar formation temperatures15. Therefore, we contend that these highly radiogenic signatures reflect the heterogeneity of the mantle source. This argument is supported by two facts observed in the MD carbonatites: (1) a wide range of Li isotopic values (δ7Li: −4.5 to +10.8%) for carbonatite and calcite, which were probably produced by the mixture of a mantle component with subducted oceanic crust and marine sediments43; and (2) Sr-Nd-Pb isotopic compositions that overlap with those of associated Cenozoic potassic rocks, which were derived from a heterogeneous enriched mantle13 (Fig. 3a,b).

Previous studies indicate that most of the world’s carbonatites usually have Sr-Nd-Pb isotope ratios similar to that of OIBs. Thus, these carbonatites are widely regarded to be derived from the melting of sub-lithospheric mantle triggered by either asthenospheric upwelling or mantle plume activity18. However, this model cannot explain the Sr-Nd-Pb isotopic arrays for the MD carbonatites (Fig. 3) or extreme enrichment of Sr, Ba and REEs (Fig. 2). Our data suggest that the MD carbonatites were derived from a metasomatic-enriched SCLM with involvement of subducted marine sediments, which is consistent with modeling results for the Sr-Nd isotopic data15 and Li isotopic data43.

The metasomatism of fluids derived from the subducted oceanic slab (i.e., MORB) is capable of hydrating, oxidizing and refertilizing the SCLM with ore-forming elements44,45,46, but it would be difficult for this process to sufficiently carbonate the mantle to generate carbonatite melts18, thus requiring the recycling of marine sediments into the mantle source47,48. Direct melting of the recycled carbonate sediments would produce primary carbonatitic melts with more radiogenic Sr-Nd isotopic compositions (87Sr/86Sr >0.706)48 and relatively low Ba/Th ratios (<200)49. Compared with such carbonatitic melts, the MD carbonatites have similar 87Sr/86Sr values (0.705–0.708) but unusually high Ba/Th ratios (>200), suggesting the involvement of CO2-rich fluids from the recycled marine sediments (Fig. 2c).

CO2-rich fluids generally have high concentrations of large-ion-lithophile elements (LILE) and high LILE/HFSE ratios, which is primarily a result of the high mobility of LILEs and relative insolubility of HFSEs and Y in aqueous fluids50,51. The SCLM metasomatism by such CO2-rich fluids would produce a carbonated peridotite refertilized with respect to the LILEs (e.g., Ba and Sr). Subsequent melting of this carbonated peridotite would concentrate the alkaline-earth elements (e.g., Ca, Sr and Ba) into the carbonated melt52,53,54; the HFSEs (Nb, Ta, Zr and Hf), heavy REEs and Y would be retained in Ti-oxides because of their high Dmin/melt values55. This scenario is consistent with certain geochemical features of the MD carbonatites, such as their extreme enrichment in Sr and Ba, relative depletion in HFSEs and low ratios of Nb/Y (Fig. 2c).

Experiments indicate that CO2-rich fluids could transport and add LREEs into metasomatized SCLM at pressures of >25 kbar56 because of the high partitioning coefficients of LREEs in CO2-rich fluids/melts54, thus explaining the common REE enrichments in most of the world’s carbonatites18. However, the extremely high REE concentrations in the MD carbonatites indicate that LREE enrichments can be controlled by additional factors, including variations in the amount of metasomatically introduced components, phases of subducted sediments and total volatile concentrations within the magmatic source37.

Research on subducting sediments indicates that REE and Ba abundances are closely linked to seafloor hydrothermal sediments1 and that Sr contents are controlled by carbonate phase abundances32. Recent observations also demonstrate that REEs are strongly concentrated in metalliferous, Fe-oxyhydroxide-bearing sedimentary muds on the seafloor, such as in the eastern part of the southern Pacific (>1000 ppm total REE + Y) and northern Pacific (>400 ppm total REE + Y)1. These concentrations are in turn diluted by increases in the proportion of carbonate phases within the marine sediments1, which indicates that REE concentrations in subducted sediments are controlled by the ratio of hydrothermal mud to carbonate phases. As a first-order approximation, the Sr/Ba ratio of marine sediments may reflect the quantitative proportions of carbonate and hydrothermal phases and amount of REEs within the sediments. This relationship predicts that carbonatitic melts derived from carbonated mantle metasomatized by CO2-rich fluids from recycled marine sediments should have Sr/Ba ratios similar to those of the subducted sediments because of the similar geochemical behaviors of Sr and Ba. The negative correlation found between Sr/Ba ratios and REE concentrations in global carbonatites (Fig. 2a) also suggests that the carbonate/hydrothermal phase proportion in the subducted sediments exerts important control on the REE concentration of carbonatite melts. The proportion of these two phases probably controls the quantity of REEs released into the CO2-rich fluids, thereby controlling the quantity of REEs that is metasomatically introduced into the SCLM. Higher proportions of hydrothermal phases in the subducted sediments increases the amount of LREEs that are introduced into the SCLM; and the higher fluxes of LREE- and CO2-rich fluids during metasomatism of the SCLM, the higher concentration of REEs in the carbonatite melts. We therefore conclude that giant CARDs form from LREE-rich carbonatitic melts derived from REE-refertilized mantle sources related to the recycling of REE-rich marine sediments.

This recycling of subducted marine sediments for CARD formation is consistent with the continuous subduction of a Proto-Tethyan oceanic lithosphere beneath the Yangtze Craton in the Neoproterozoic9,10,11,12 and development of a 1000-km long, Neoproterozoic magmatic arc along the western margin of the Yangtze Craton10. The Nd model ages for the MD carbonatites (0.69–1.63 Ga; Table S3) and associated Cenozoic potassic rocks (0.54–1.60 Ga)13 suggest that the enrichment of SCLM metasomatized by subducted slab fluids was likely multi-phase, with most of the enrichment clustering in the Neoproterozoic. Similar metasomatic enrichment has also been recognized in the Bayan Obo and Laiwu districts, where an Archean-depleted SCLM was metasomatized by subducted sediments at 1.9–2.0 Ga57 and a Paleozoic-enriched SCLM was replaced by EMII-type mantle because of subduction of the paleo-Pacific plate since the Late Jurassic38, respectively.

Metasomatic refertilization of metals (Cu, Au and REEs) occurred in the SCLM with the intensity increasing toward cratonic margins above ancient subduction zones, which created lateral compositional heterogeneity46 (Fig. 4a). The formation of fertile carbonatites requires a thick lithosphere and/or high pressure (>25 kbar)56, metasomatic enriched mantle sources15 and favorable pathways for magma ascending into the overlying crust46 where REE-rich fluids exsolve from the cooling magma26,58. The optimal configuration of all three factors only occurs along the margins of a craton with a continental root, rather than in a modern subduction zone with relatively thin lithosphere. The melting of REE-refertilized mantle was triggered by upwelling of asthenosphere59 and the resulting magmas ascended along translithospheric faults along cratonic edges and emplaced at shallow crust (Fig. 4a). These magmas rapidly exsolved highly oxidized, REE-rich fluids during stress relaxation58, which led to fenitization around the carbonatite stocks60 and precipitation of bastnaesite when the temperatures decreased, thus forming CARDs with distinct styles of mineralization8 (Fig. 4b). These patterns suggest that areas of metamorphic basement bounded or cut by translithospheric faults along cratonic margins and other lithospheric boundaries have significant potential for such REE deposits.

Figure 4
figure 4

Metasomatic refertilization of the SCLM and formation of fertile carbonatites and CARDs.

(a) Upwelling of asthenosphere triggers the melting of refertilized SCLM46 that was previously metasomatized by CO2-rich fluids derived from marine sediments associated with “fossil” subduction zones. The subducted sediments released their REEs into CO2-rich fluids that metasomatized old depleted or enriched SCLM to form an unusually REE-rich, carbonated mantle source, which then produced carbonatite melts or CO2-rich silicate melts. The margins of the craton experience low degrees of partial melting and the melts ascend through fracture zones into the overriding crust. (b) Schematic illustration of models of CARD formation, including a variety of orebodies formed by fluids exsolved from REE-rich carbonatitic magmas emplaced at shallow crustal levels. Lateral migration, replacement, open-space filling and focused discharges of ore-forming fluids produced semi-stratabound (Bayan Obo-style)24,25, disseminated (Lizhuang or Mountain Pass-style)8,26,58, stringer-stockwork (Maoniuping-style)5 and breccia pipe (Dalucao-style)8 orebodies with associated fenitization and K-silicate alterations, respectively.

Full size image

Methods

Whole-rock chemical analysis

All the samples were partially analyzed at the National Research Center of Geoanalysis, CAGS, Beijing. Whole-rock powders were weighed (0.7 g) and then mixed with 5.3 Li2B4O7, 0.4 g LiF and 0.3 g NH4NO3 in a 25 ml porcelain crucible. The mixture powder was then transferred to a platinum alloy crucible and 1 mL LiBr solution was added to the crucible and then dried. The sample was then melted in an automatic flame fusion machine and the resulting cooled glass was used for the X-ray fluorescence (XRF) major element analyses. Analytical errors were <2 relative %. The analytical errors were less than 2%. For the trace element (and REE) analyses, whole-rock powders were weighed (50 mg) and then dissolved in 1 mL distilled HF and 0.5 mL HNO3 in 15 ml Savillex Teflon screw-cap capsules at 190 °C for 1 day, dried and then digested with 0.5 mL HNO3 and dried again. The capsules were digested with 0.5 mL HNO3 and dried again. The samples were then digested with 5 mL HNO3 and sealed at 130 °C for 3 hours. After cooling, the solutions were transferred to plastic bottles and diluted to 50 ml before analysis. The sample solutions were analyzed for trace elements via inductively coupled plasma mass spectrometry (ICP-MS). The analytical precision for most elements was greater than 5%.

C and O isotopic analysis

The C and O isotopic compositions of carbonates were partially analyzed via the GasBench II method on a MAT253 mass spectrometer in the Institute of Mineral Resources, CAGS (Beijing). Approximately 0.1 mg sample powder was added to a reaction bottle with a total volume of 12 ml. Less than eighty-eight samples, including eighteen Chinese isotopic standards (GBW04416, GBW04417, GBW04405 and GBW04416), were prepared for every measurement. The sample powders were flushed for 600 s with pure He (99.999%) at a flow rate of 100 ml/min. Each sample was then dissolved with 5 drops of 100% anhydrous H3PO4 at 72 °C for more than four hours to liberate CO2 from the calcite. The CO2 was injected into a loop, separated via a chromatographic column (Poraplot Q, 25 m * 0.32 mm, Varian Ltd) at a constant 70 °C, dried using a Nafion membrane and analyzed by an MAT 253 mass spectrometer. The analytical results of eighteen isotopic standards indicate that the external precision was better than 0.10% for the δ13C and δ18O values. The δ13C and δ18O results were reported in standard delta notation (% units) relative to the Vienna Pee Dee Belemnite (VPDB).

Sr, Nd and Pb isotopic analysis

The Sr-Nd-Pb isotope compositions were partially analyzed at the Laboratory of Isotope Geology, CAGS (Beijing). For the Rb-Sr and Sm-Nd isotope analyses, rock chips of less than 20 mesh size were used. The sample powders were spiked with mixed isotope tracers and then dissolved in Teflon capsules with HF + HNO3. After the samples were fully dissolved and dried to remove HF, the samples were dissolved with HCl. These samples were then dissolved with HClO4 and transferred to an HCl medium. The Sr and REE fractions were separated in solution using cationic ion-exchangers (AG 50 W(H+), 200 to 400 mesh) in columns. Then, Sr and Nd were collected in the proper sequence via leaching with 4 mol ml−1 HCl. The Sr fraction was purified again according to the above procedure for complete separation. Then, the solution was dried for isotopic measurement. The collected Nd fraction was evaporated and dissolved in 0.2 mol ml−1 HCl and the Nd was separated from the REE fractions using cationic ion-exchange columns and extraction and eluviation resin (HDEHP). The solution was then rinsed with 0.2 mol ml−1 HCl to remove Sm and Sr isotopic measurements were performed on an MAT 262 mass spectrometer. The mass fractionation corrections for the Sr isotopic ratios were based on 88Sr/86Sr = 8.37521. The international standard NBS987 yielded 87Sr/86Sr = 0.710250 ± 10 (2σ). The Nd isotopic measurement was performed on a Nu Plasma HRMC-ICP-MS (Nu Instruments) 262 mass spectrometer. The mass fractionation corrections were based on 146Nd/144Nd = 0.7219. The international standard is JMC Nd2O3, in which 143Nd/144Nd = 0.511125 ± 10 (2σ). The entire procedure blank was less than 10−9–10−9 g for Sr and 5 × 10−11 g for Nd. The analytical errors are given as 2σ. The 87Rb/86Sr and 147Sm/144Nd ratios were calculated using the Rb, Sr, Sm and Nd concentrations and the initial 87Rb/86Sr and 143Sm/144Nd ratios were calculated using SHRIMP (sensitive high-resolution microprobe) concordant ages. Initial εNd values and 87Sr/86Sr ratios were calculated according to the ages of the corresponding carbonatites or syenites.

The measurements of the Pb isotopes were performed with a Nu Instrument multi-collector ICP-MS using the method described by Belshawa et al. (1998)61. The samples were digested using a mixture of ultrapure HF and HNO3 at 800 °C for 72 h, followed by purification using conventional ion-exchange chromatography (AG1X8, 200–400 resin). A portion of the sample (0.2 ml) was added and drip washed with 1 ml of 1 mol/L HBr as eluant five times and 0.5 ml of 2 mol/L HBr as eluant once. The Pb was collected by drip washing with 1 ml of 6 mol/L HCl and 0.5 ml of 6 mol/LHCl . The entire procedure blank is less than 0.1 ng. During the period of analysis, repeat analyses of the international standard NBS981 yielded the following values: 208Pb/206Pb = 2.1674 ± 0.0004; 207Pb/206Pb = 0.91478 ± 0.00018; 206Pb/204Pb = 16.9402 ± 0.0070; 207Pb/204Pb = 15.4966 ± 0.0030; and 208Pb/204Pb = 36.7155 ± 0.0120 (2σ).

Additional Information

How to cite this article: Hou, Z. et al. Formation of carbonatite-related giant rare earth element deposits by the recycling of marine sediments. Sci. Rep. 5, 10231; doi: 10.1038/srep10231 (2015).