Introduction

Rare earth elements (REE) are critical elements in a variety of modern high-tech applications due to their unique physical and chemical characteristics. The global demand for REE has increased significantly during recent decades (Haxel 2002; Hoshino et al. 2016). Fluorapatite is a common constituent occurring in iron-oxide copper gold (IOCG) deposits (Li et al. 2015; Krneta et al. 2017; Cherry et al. 2018), iron-oxide fluorapatite (IOA) deposits (Harlov et al. 2016; Zeng et al. 2016), and carbonatite-related REE deposits (Broom-Fendley et al. 2016; Chakhmouradian et al. 2017; Song et al. 2018; Hu et al. 2019; Ying et al. 2020). It generally accommodates substantial amounts of REE in its crystal lattice (from several thousands of ppm to several wt%; Hoshino et al. 2016), making it a potential target for economic REE mineralization.

In addition, fluorapatite is capable of incorporating a wide variety of trace elements (Pan and Fleet 2002; Hughes and Rakovan 2015; Engi 2017), making it a good indicator of fluid evolution and mineralization processes (Harlov 2015; Andersson et al. 2019). Moreover, fluorapatite often contains elevated U and Th contents, and is widely used in U–Pb thermochronometry (Chew and Spikings 2015). Previous investigations have shown that REE can be released to form REE-rich inclusions when fluorapatite interacts with fluids during late hydrothermal alteration in IOCG, IOA and carbonatite-related REE deposits, which served as one of the dominant sources of REE mineralization (Li and Zhou 2015; Harlov et al. 2016; Broom-Fendley et al. 2016; Zeng et al. 2016; Krneta et al. 2017; Hu et al. 2019; Cherry et al. 2018; Ying et al. 2020). Nevertheless, our understanding of the mechanisms behind REE transportation and enrichment during hydrothermal metasomatism is still inadequate (Harlov et al. 2005; Bonyadi et al. 2011).

The Yinachang Fe–Cu–REE deposit, one of the important Fe–Cu–REE deposits in Southwest China (Fig. 1a), contains approximately 20 Mt ore at 41.9–44.5 wt % Fe and 15 Mt ore at 0.85–0.97 wt % Cu, with abundant REE (Zhao et al. 2015). Li et al. (2015) and Zhao et al. (2015) have shown that features of the Yinachang deposit are similar to those of IOCG deposits, and its REE mineralization is related to fluid-induced metasomatic alteration of REE-rich fluorapatite by late stage hydrothermal fluids associated with Cu mineralization. Due to the lack of suitable materials for high quality dating and a limited previous study on the internal structures of fluorapatite, timing the alteration of the fluorapatite as well as REE transport and enrichment mechanism of REE during fluorapatite alteration in the Yinachang deposit remain unclear.

Fig. 1
figure 1

modified from Li et al. 2015)

a Simplified tectonic map of South China. b Geological map of the Yinachang deposit (

Full size image

In this study, we present a detailed documentation of fluorapatite textures from different stages of the Yinachang deposit using combined cathodoluminescence (CL), back-scattered electron (BSE), and TESCAN integrated mineral analyzer (TIMA) imaging. Integrating these textural relationships with electron microprobe analysis (EMPA), laser ablation inductively coupled plasma mass spectrometry (LA–ICP–MS), micro-Raman spectroscopy, and U–Pb dating, we aim to evaluate the processes behind REE transport and enrichment during fluorapatite alteration and constrain the timing of REE mineralization in the Yinachang deposit.

Deposit geology and petrography

The Yinachang deposit is situated in the central part of the Kangdian metallogenic province, southwest of the Yangtze Block (Fig. 1a; Zhao et al. 2013; Zhou et al. 2014). The main lithostratigraphic units in the deposit area are the Proterozoic Yinmin, Luoxue, and E’touchang formations, and Mesozoic strata (Fig. 1b). Ore bodies, controlled by NNE-trending faults, are hosted in slate, schist, and dolostone of the Yinmin Formation, which was intruded by Paleoproterozoic dolerite intrusions (zircon 207Pb/206Pb U–Pb age of 1690 ± 32 Ma, Zhao et al. 2010). The detailed geology and metallogeny of the Yinachang deposit have been reviewed by Zhao et al. (2013) and Li and Zhou et al. (2015).

Combined with the study of Li and Zhou (2015) and Li et al. (2015), we divided hydrothermal alteration and mineralization at the Yinachang deposit into three stages, including pre-ore Na–Fe alteration, Fe–Cu (–REE) mineralization and REE (–Cu) remobilization (Fig. 2). The Fe–Cu (–REE) mineralization stage contains early Fe (–REE) and late Cu (–REE) mineralization sub-stages (Li et al. 2015). The earliest Na–Fe alteration minerals in ore-hosting rocks are albite, quartz, magnetite, and hematite. The Fe (–REE) mineralization is characterized by the formation of magnetite, siderite, quartz, REE-rich fluorapatite, and less amounts of fluorite and ankerite (Figs. 3a and 4a). The subsequent Cu (–REE) mineralization is featured by abundant sulfide minerals (mainly chalcopyrite) and quartz, accompanied with subordinate biotite, calcite, fluorapatite (Figs. 3b, c and 4b), and REE-rich minerals [e.g., allanite, monazite–(Ce), bastnaesite-(Ce) and synchysite–(Ce); Li et al. 2015]. The REE (–Cu) remobilization stage, which is characterized by the formation of REE-rich minerals [e.g., monazite–(Ce) and synchysite–(Ce)], can be subdivided into two sub-stages (Fig. 2). The early REE (–Cu) remobilization is recorded by metasomatic alteration of REE-rich fluorapatite with abundant mineral inclusions [monazite–(Ce), quartz, hematite, bastnaesite–(Ce)] occurring in the alteration domains (Figs. 3d, e and 4c). The late REE (–Cu) remobilization is characterized by hydrothermal veins containing chalcopyrite, molybdenite, arsenopyrite, fluorapatite, monazite–(Ce), synchysite–(Ce) and magnetite (Figs. 3f and 4d). In this remobilization stage, the late chalcopyrite clearly replaced and cut the early altered REE-rich fluorapatite (Fig. 3e).

Fig. 2
figure 2

modified from Li et al. (2015) and Li and Zhou (2015)

Alteration and mineralization paragenesis of the Yinachang deposit,

Full size image
Fig. 3
figure 3

Representative hand specimen photographs and microphotographs showing representative mineral assemblages, textural features of the Yinachang deposit. a Banded Fe ore, which contains sub-parallel, chalcopyrite-rich bands. b Quartz + chalcopyrite + biotite + fluorapatite veins intersect dolostone. c Fluorapatite grains (Ap2) in Cu ore, coeval with quartz, chalcopyrite, and biotite. d A wide quartz + chalcopyrite + ankerite + fluorapatite + fluorite + molybdenite + chlorite + pyrite vein intersecting dolostone. e TIMA image showing the mineral inclusions and micro-pores in altered domains of the original Ap1. f TIMA image showing a chalcopyrite + magnetite + fluorapatite (Ap4) + synchysite-(Ce) + monazite-(Ce) + molybdenite vein intersecting a banded Fe ore. Mineral abbreviations: Ank ankerite, Ap fluorapatite, Apy arsenopyrite, Bast bastnaesite–(Ce), Bt biotite, Ccp chalcopyrite, Cal calcite, Chl chlorite, Col columbite, Fl fluorite, Hol hole, Hem hematite, Mag magnetite, Mnz monazite–(Ce), Mod modderite, Mol molybdenite, Py pyrite, Qz quartz, Sd siderite, Syn synchysite–(Ce), Xen xenotime

Full size image
Fig. 4
figure 4

Representative BSE and CL images showing textural features of fluorapatite in the Yinachang deposit. a BSE image of unaltered REE-rich fluorapatite grains (Ap1) in banded Fe ore, intimately associated with coeval siderite and magnetite. b CL image of fluorapatite in Cu ore showing a bright and irregular core (Ap2-1) and a dark mantle (Ap2-2). c CL image showing Ap3-1 formed during fluid-induced metasomatic alteration of Ap1. Ap3-2 occurs as an overgrowth margin around Ap3-1. Note that pervasive REE-rich mineral inclusions or micro-pores are present in altered domains from the original Ap1. d BSE image of a chalcopyrite + magnetite + fluorapatite (Ap4) + synchysite-(Ce) + monazite-(Ce) + molybdenite vein intersecting a banded Fe ore. Mineral abbreviations: Ap fluorapatite, Ccp chalcopyrite, Mag magnetite, Mnz monazite–(Ce), Mol molybdenite, Sd siderite, Syn synchysite–(Ce)

Full size image

Sampling and analytical procedures

In this study, a total of 20 samples, 10 from the ore bodies and 10 from the country rocks (mainly schist and dolostone) were collected. Polished thin sections of these samples were first examined using transmitted and reflected light microscopy, and then BSE and CL imaging to characterize textural relationships and establish different stages of metasomatic alteration, i.e., Fe-REE mineralization stage (Ap1), Cu mineralization stage (Ap2), early REE (–Cu) remobilization stage (Ap3), and late REE (–Cu) remobilization stage (Ap4). Representative samples were then analyzed by in-situ Raman, TIMA, EMPA, and LA–ICP–MS to determine compositional and structural characteristics. EPMA, LA–ICP–MS and U–Pb dating results are listed in Electronic Supplementary Material 1.

BSE and CL imaging was carried out on polished thin sections using a TESCAN MIRA3 field-emission scanning electron microprobe (FE-SEM) at the Testing Center, Tuoyan Analytical Technology Co. Ltd., Guangzhou, China. After the samples were carbon-coated, SEM-CL images were acquired under an acceleration voltage of 10 kV and a beam current of 15 nA. Mineral compositional mapping was obtained on carbon-coated thin sections using a MIRA3 scanning electron microscope equipped with two energy dispersive X-ray spectrometers (EDS, EDAX Element) (TIMA) at the State Key Laboratory of Continental Dynamics, Department of Geology, Northwest University, Xian, China. An acceleration voltage of 25 kV and a beam current of 9 nA were used. The current and BSE signal intensity were calibrated on a platinum Faraday cup using the automated procedure. The EDS intensities were then normalized using a Mn standard. The samples were also scanned using a TIMA liberation analysis module.

Major and minor element analyses of selected fluorapatite grains were conducted using a Cameca SX100 electron microprobe at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences (GIGCAS), Guangzhou, China. The following conditions were used for measurements of all standards and samples: 15 kV accelerating voltage, 20 nA beam current, 1 µm beam diameter, and 10 s counting time. To correct for the matrix effects during EPMA, the ZAF correction was applied to transform the relative peak intensities into the elemental weight composition (Bernard et al. 1986). The relative precisions of EPMA are < 2% for major elements and < 5% for minor elements.

In situ U–Pb and trace element analyses of fluorapatite were made using a NWR laser ablation system coupled with an inductively coupled plasma mass spectroscopy (ICP-MS, model: iCAP RQ) at the Tuoyan Analytical Technology Co. Ltd., Guangzhou, China. The details of LA-ICP-MS analyses have been given as tables in Electronic Supplementary Material 2. Analyses were calibrated using two external standards (SRM-610 and SRM-612) for trace elements analysis, and Madagascar (MAD) and Durango (DUR) apatite as internal and external standards for U–Pb dating, respectively. Repeated analyses yielded 206Pb/238U ages of 474.5 ± 4.5 Ma for MAD and 30.9 ± 0.6 Ma for DUR, which were in good agreement with the recommended MAD 206Pb/238U age of 474.3 ± 0.4 Ma (Thomson et al. 2012) and DUR 206Pb/238U age of 31.13 ± 1.01 Ma (McDowell et al. 2005), respectively.

Raman spectra from fluorapatite were collected using a RM2000 laser micro-Raman spectrometer at GIGCAS. A 514.5 nm Ar laser was used, with a 2 μm diameter laser spot is 2 μm in diameter. The scanning range was between 100 and 1500 cm–1 with a spectral resolution of 1 cm−1. The laser power reaching the sample surface was 10 mW and the typical acquisition time was 60 s to avoid laser-induced thermal effects and oxidation.

Results

Textural and structural characteristics of fluorapatite

Ap1, commonly developed in Fe ores (Fig. 3a), is typically euhedral to subhedral in shape with variable sizes of 20–200 μm in the maximum dimension. The unaltered Ap1 grains are relatively homogeneous under CL imaging (Fig. 4a). Ap2 with few REE mineral inclusions, coexisting with chalcopyrite, biotite, calcite, and quartz (Fig. 3c), exhibits elongated morphology with variable lengths of 100–1500 μm, and contains a bright and irregular core (Ap2-1) with a dark mantle (Ap2-2) in the CL image (Fig. 4b). Ap3 occurs in close association with chalcopyrite and often as a replacement after Ap1 (Fig. 4c). It can be subdivided into Ap3-1 and Ap3-2 on the basis of CL imaging. Abundant mineral inclusions [hematite, pyrite, modderite ((Co, Fe)As), monazite-(Ce), quartz, synchysite-(Ce) (CaCe(CO3)2F), bastnaesite-(Ce) (CeCO3F)] and micro-pores are developed in the reaction front (i.e., dark domains in Ap1, Fig. 4c) between Ap1 and Ap3-1 (Figs. 3e and 4c), while lesser but larger inclusions and micro-pores are developed when Ap1 was completely replaced by Ap3-1. Compared with Ap1, Ap3-1 is darker under BSE and CL imaging. Ap3-2 with brighter CL occurs as an overgrowth around Ap3-1. Ap4 is coeval with chalcopyrite, molybdenite, synchysite-(Ce), monazite-(Ce), and magnetite in the hydrothermal veins (Fig. 3f), and has homogeneous CL (Fig. 4d).

All fluorapatite grains at Yinachang exhibit Raman bands characteristic of fluorapatite, including a strong peak at ~ 965 cm–1 and a weak peak at 582–616 cm–1 (Fig. 5; Chakhmouradian et al. 2017). In addition, Ap1 and Ap3-2 have the more obvious Si–O vibration modes at 395 and 811–838 cm–1, and the S–O vibration modes at 640 and 1114–1186 cm−1 than other fluorapatite at Yinachang. The carbonate ν4 vibration mode at ~ 700 cm−1 (Awonusi et al. 2007) is also visible in Ap1 and Ap3-2 (Fig. 5).

Fig. 5
figure 5

Raman spectra measured in situ on the four kinds of fluorapatite from the Yinachang deposit. Characteristic vibration modes of carbonate, phosphate, silicate and sulfate groups are marked. Note that Ap1 and Ap3-2 have the more obvious Si–O, S–O and C–O vibration modes than Ap2-1, Ap2-2, Ap3-1 and Ap4

Full size image

Chemical composition of fluorapatite

Of the four fluorapatite types, Ap1 contains the highest SiO2 content, Ap2 has the highest Cl content, Ap4 contains the highest FeO content, and Ap2-1 has the lowest MnO content (Fig. 6). Compared with Ap2, Ap3, and Ap4, Ap1 contain the highest REE + Y content (> 1 wt%; Electronic Supplementary Material 1). Ap4 is characterized by LREE-enriched, chondrite-normalized REE patterns with slightly negative to moderately positive Eu anomalies (Fig. 7a; 0.78–4.03; Electronic Supplementary Material 1). The degree of LREE enrichment relative to HREE for Ap1, represented by the (La/Yb)N ratio, varies between 6.20 and 67 (Electronic Supplementary Material 1). Ap2-1, Ap2-2, Ap3-1, and Ap3-2 all display MREE-enriched chondrite-normalized REE patterns (Figs. 7b and c) with similar positive Eu anomalies (1.66–3.32; Electronic Supplementary Material 1). Ap4, similar to Ap1, shows LREE-enriched chondrite-normalized REE patterns with the (La/Yb)N ratio ranging from 24 to 52 (Electronic Supplementary Material 1), the most pronounced positive Eu anomalies (24–32; Electronic Supplementary Material 1), and the highest Eu contents (Fig. 7d). These positive Eu anomalies in Ap4 represent some of the highest values known for apatite-group minerals (Fleet and Pan 1995; Krneta et al. 2017; Algabri et al. 2020; Bromiley 2021).

Fig. 6
figure 6

The ranges in the SiO2 a, Cl b, FeO c, and MnO d contents in fluorapatite fromthe Yinachang deposit. Note that Ap1 contains the highest SiO2 content, Ap2 has the highest Cl content, Ap4 contains the highest FeO content, and Ap2-1 has the lowest MnO content. The data of Ap1 are obtained from Li and Zhou (2015)

Full size image
Fig. 7
figure 7

Chondrite-normalized REE patterns for the four stages of fluorapatite from theYinachang deposit. Note that Ap1 contain the highest REE + Y content than Ap2, Ap3, and Ap4. The data of Ap1 are obtained from Li and Zhou (2015)

Full size image

U–Pb geochronology of fluorapatite

Due to the low U contents of Ap2 and the high common Pb contents of Ap1 and Ap4, only Ap3 was suitable for U–Pb dating. A total of 29 laser ablation spots from 10 Ap3-1 grains were obtained for U–Pb geochronology, which defined a regression line on the Tera-Wasserburg plot, with a lower intercept age of 939.6 ± 17.6 Ma (MSWD = 0.89, n = 27; Fig. 8). We consider the lower intercept age to indicate the timing of Ap3-1 formation.

Fig. 8
figure 8

LA–ICP–MS U–Pb Tera–Wasserburg concordia diagram for Ap3-1

Full size image

Discussion

Metasomatic alteration of fluorapatite and substitution mechanisms

In this contribution, combined BSE-CL images and TIMA data as well as Raman spectra highlight the complicated textures for the different stages of fluorapatite in the Yinachang deposit, which were not fully revealed by previous SEM imaging (Li and Zhou 2015; Zhao et al. 2015). TIMA studies show that pervasive REE-rich mineral inclusions and micro-pores are present in altered domains in the original Ap1 (Fig. 3e), which have long been documented by both natural and experimental studies (Pan et al. 1993; Harlov and Förster 2003; Li and Zhou 2015; Harlov et al. 2016; Krneta et al. 2017). In addition, a close spatial relationship and obvious reaction fronts were observed between Ap1 and Ap3-1 (Fig. 4c), indicating that extensive metasomatic alteration occurred by means of a coupled dissolution (Ap1)-reprecipitation (Ap3-1) process (Pan et al. 1993; Harlov et al. 2005; Putnis 2009). Different Raman signatures of Ap1 and Ap3-1 imply that structural and chemical changes occurred along with the leaching of REE from the fluorapatite (Fig. 5). Compared with Ap1, Ap3-1 has a lower Si and REE + Y content, indicating that Si in Ap1 was removed by a hydrothermal fluid resulting in charge imbalance, which promoted the dissolution-reprecipitation process and freed the (REE + Y) to react with PO43− and CO32− groups to form monazite–(Ce) and bastnaesite–(Ce) micro-inclusions. In addition, CL images show that Ap3-2 and Ap2-2 occur as an overgrowth around Ap3-1 and Ap2-1, respectively, without mineral inclusions or micro-pores between them.

The REE + Y contents of Ap1 have a positive correlation with Si but invariably lie slightly above the 1:1 molar ratio line (Fig. 9a). This result suggests that the coupled substitution, Si4+  + (REE + Y)3+  = P5+  + Ca2+ (Fleet and Pan 1995; Pan and Fleet 2002), is dominant. However, other substitution mechanisms such as C4+  + (REE + Y)3+  = P5+  + Ca2+ may have contributed as well, which is supported by the elevated carbonate revealed by the Raman spectrum of Ap1 (Fig. 5). However, the REE + Y values in Ap2-1, Ap2-2, Ap3-1, Ap3-2, and Ap4 are strongly positively correlated with Na (Fig. 9b), while only show a weakly positive correlation with Si, suggesting that the coupled substitution, Na+  + (REE + Y)3+  = 2Ca2+ (Fleet and Pan 1995; Pan and Fleet 2002) predominates. These results suggest that the REE substitution mechanisms in fluorapatite changed with time in the Yinachang deposit.

Fig. 9
figure 9

Binary diagrams showing the substitution mechanisms of REE in Yinachang fluorapatite a REE + Y vs. Si; b REE + Y vs. Na. Note that the 1:1 solid lines in a, which indicate the coupled substitution Si4+  + (REE + Y)3+  = P5+  + Ca2+ and the solid lines in b, which indicate the substitution Na+  + (REE + Y)3+  = 2Ca2+. The data of Ap1 are obtained from Li and Zhou (2015)

Full size image

Timing of REE remobilization and fluorapatite alteration

Li and Zhou (2015) suggested that the Sm–Nd isochron age of 1700 Ma represents the timing of Fe (–REE) mineralization (hosting Ap1), consistent with the formation age of the dolerite intrusions (zircon U–Pb age of 1690 ± 32 Ma; Zhao et al. 2010), indicating a direct genetic relationship between Fe (-REE) mineralization and the dolerite intrusions. Re-Os dating of molybdenite and chalcopyrite yields ages of 1660–1690 Ma for the main Cu (–REE) mineralization (hosting Ap2) at Yinachang (Ye et al. 2013; Zhao et al. 2013), which resulted in alteration of REE-rich fluorapatite (Ap1) and REE mineralization (Li et al. 2015; Li and Zhou 2015). In this study, in situ U–Pb analysis of Ap3 demonstrates that the REE (–Cu) remobilization stage of the Yinachang deposit occurred at ~ 940 Ma, which is obviously younger than the age of the major Fe-Cu (-REE) mineralization event. In addition, Li and Zhou (2015) also reported that monazite-(Ce) grains associated with Ap4 have a U–Pb age of 842 ± 9 Ma. Aside from the REE-rich Ap1 alteration event at the Fe–Cu (–REE) mineralization stage (1700 Ma, Li and Zhou 2015), these absolute ages indicate that intensive alteration of REE-rich apatite and remobilization of REE and Cu in the Yinachang deposit also occurred at 940–840 Ma, which is consistent with recently identified Cu mineralization events at 1000–800 Ma in this region (Zhao et al. 2017). In fact, magmatic events of 1000–800 Ma in the Kangdian metallogenic province have been widely documented over the last two decades (Zhou et al. 2002, 2014; Li et al. 2006; Zhao et al. 2017; Su et al. 2021).

Physical–chemical controls on REE mobility during fluorapatite alteration

The previously reported Sm–Nd isochron age (1700 Ma) from the Yinachang Fe ore samples (Li and Zhou 2015) is consistent with the Paleoproterozoic age of the mantle-derived dolerite intrusions in the ore district. Interestingly, Ap1 and the dolerite intrusions have similar initial 87Sr/86Sr ratios (0.704–0.710 and 0.706–0.707; Zhao et al. 2015). Fluid temperatures for Fe (–REE) mineralization estimated from oxygen isotope thermometry on the magnetite-siderite pair range from 395 to 503 °C (Li et al. 2015). Hence, Ap1 coeval with magnetite was most likely related to a high temperature magmatic-hydrothermal fluid, which was rich in REE + Y and CO2, as seen by the presence of co-existing REE-rich Ap1 and siderite (FeCO3) (Fig. 4a) as well as the presence of carbonate in the Raman spectrum of Ap1 (Fig. 5). δ34S values of the ore-forming fluids in the Cu (–REE) mineralization sub-stage and the REE (–Cu) remobilization stage have a concentrated range near 0‰ (− 2.8 to + 2.7‰; Li et al. 2015), just overlapping the compositional fields of magmatic derived S (~ 0‰; Ohmoto and Rye 1979), suggesting that Ap2, Ap3, and Ap4 are also related to a magmatic-hydrothermal fluid.

Europium L3-edge X-ray absorption near edge structure (XANES) spectroscopic analyses of natural and synthetic fluorapatite have shown that Eu exists in both the divalent and trivalent states (Rakovan et al. 2001). Previous studies indicate that fluorapatite usually displays negative or no Eu anomalies (Fleet and Pan 1995; Mao et al. 2016; Bromiley 2021), although positive Eu anomalies in fluorapatite have been reported in the literature, such as the Olympic Dam IOCG deposit (Krneta et al. 2017). In this study, the Yinachang Ap1 displays slightly negative to moderate positive Eu anomalies, whereas, Ap2, Ap3, and Ap4 all show obviously positive Eu anomalies (Fig. 7). Hence, the positive Eu anomalies of Ap2, Ap3, and Ap4 were not inherited from Ap1, but record the effects of the hydrothermal fluids during the Cu–Au–REE mineralization stage. Hydrothermal fluids with positive Eu anomalies have been reported to occur in the following two scenarios. First, magmatic feldspars, especially plagioclase, usually show strongly positive Eu anomalies due to the preferential incorporation of Eu2+, which could be transferred into the hydrothermal fluid through their dissolution. This process has been proposed to be responsible for seafloor hydrothermal fluids with remarkably uniform chondrite-normalized REE patterns characterized by LREE enrichment and large positive Eu anomalies (Klinkhammer et al. 1994; Hoshino et al. 2016). Second, previous studies also have shown that if the REE pattern of hydrothermal fluids is controlled by sorption processes, the decreasing ionic radius of the trivalent REE leads to (La/Lu)cn > 1, and the larger ionic radius of the Eu2+ ion compared to that of neighboring Sm3+ and Gd3+ causes positive Eu anomalies. But if the pattern is controlled by complex mechanisms, this decrease in radius will result in (La/Lu)cn < 1. Note that sericite or other K-rich minerals are conspicuously absent during the Cu (–REE) mineralization and REE (–Cu) remobilization stage and the (La/Lu)cn values of Ap2, Ap3, and Ap4 are all above 1. Therefore, the positive Eu anomalies in Ap2, Ap3, and Ap4 could not be related to feldspar dissolution, but may be accounted for by the sorption processes and the reduction of Eu3+ to Eu2+ (Bau 1991).

The study of Li and Zhou (2015) identified two types of fluorapatite formed through dissolution-reprecipitation processes during REE-rich fluorapatite (Ap1) alteration: Type I fluorapatite with few REE mineral inclusions formed at the Cu (–REE) mineralization stage, and Type II fluorapatite hosting abundant REE mineral inclusions formed at the REE (–Cu) remobilization stage corresponding to Ap3 in this study. Hand specimen photographs and microphotographs show that Ap2 intergrows with chalcopyrite in the hydrothermal veins (Fig. 3b and c), indicating that Ap2 was directly deposited from the aforementioned magmatic-hydrothermal fluid related to Cu mineralization. Compared with Type I fluorapatite (480–6300 ppm) in the study of Li and Zhou (2015), the total REE contents of Ap2 (~ 200 ppm; Electronic Supplementary Material 1) are obviously lower, indicating that Ap2 formed from the hydrothermal fluid before leaching REE from REE-rich Ap1. The total REE contents of Ap2 are the lowest among all kinds of fluorapatite in the Yinachang deposit (Fig. 7), indicating that the original Cu (–REE) mineralization stage fluids was REE-poor. Ap3-1 formed through the reprecipitation with dissolution of REE-rich Ap1 during metasomatic replacement. Previous studies have shown that hydrothermal fluids leaching REE from fluorapatite typically have low Na and/or Si, because these elements in the fluids can enter the fluorapatite lattices to maintain an electrovalent balance, thereby inhibiting the removal of REE from fluorapatite (Harlov et al. 2003). In addition, the coeval relationships of hematite, ankerite, and fluorite with Ap3-1 indicate a relatively high oxidation state for the magmatic-hydrothermal fluids, which are rich in sulfate, F, Ca, and CO2, and depleted in REE + Y. The relatively high oxidation state of the magmatic-hydrothermal fluids during the formation Ap3-1 is also supported by the detection of the sulfate group in their Raman spectra (Fig. 5). Compared with Ap3-1, Ap3-2 with clean homogenous CL image has higher total REE content (Fig. 7c), indicating Ap3-2 may have formed through regrowth of Ap3-1, after the end of the dissolution-reprecipitation process, which resulted in increasing REE content of hydrothermal fluid through leaching REE from REE-rich Ap1. The occurring of F-rich REE minerals [synchysite-(Ce), bastnaesite-(Ce) and fluorite] in the REE (–Cu) remobilization stage provides support for the F-rich composition of the hydrothermal fluid, which might explain the common occurrence of micro-pores in Ap1 due to F-induced dissolution (Zhou et al. 2009).

Compared with Ap2 and Ap3, Ap4 has higher positive Eu anomalies (Fig. 7d), which can be formed from a change in the oxidation state of the hydrothermal fluid and the resulting reduction of Eu3+ to Eu2+. The presence of magnetite and arsenopyrite in the hydrothermal veins containing Ap4 suggests that the oxidation state of the hydrothermal fluids at the late REE (–Cu) remobilization stage was more reduced than that of the early REE (–Cu) remobilization stage. The reduced oxygen fugacity for the hydrothermal fluid responsible for the formation of Ap4 is also consistent with the lowest sulfate intensities in its Raman spectrum among those from all the fluorapatite types in the Yinachang deposit (Fig. 5). These results show that the geochemical and textural features of the altered fluorapatite in the Yinachang deposit are controlled by both the sorption processes and the oxidation state of the hydrothermal fluids. In particular, oxidizing fluids appear to be more conducive for the alteration of fluorapatite allowing for the associated REE to be released from this mineral.

Genesis of Yinachang fluorapatite and implications for REE mineralization

Formation of Ap1 was most likely related to a high temperature magmatic-hydrothermal fluid from Paleoproterozoic dolerite intrusions during the Fe (–REE) mineralization stage. This magmatic–hydrothermal fluid was rich in (REE + Y) and had high carbonate and sulfate contents (Fig. 10a). Ap2 formed from a magmatic-hydrothermal fluid related to Cu mineralization, rich in Cl and S, and depleted in REE + Y (Fig. 10b). Ap3 with high REE contents, and containing REE-rich mineral inclusions, formed after the alteration of REE-rich Ap1 with a high oxidation state, which led to subsequent enrichment in REE in the Neoproterozoic magmatic-hydrothermal fluid (Fig. 10c). Ap4, with the most positive Eu anomalies and the highest Eu contents, was related to a late reduced hydrothermal fluid (Fig. 10d). In conclusion, this study represents a good example of how to use fluorapatite textures and geochemistry to reveal the nature and sources of ore-forming fluids.

Fig. 10
figure 10

Sketches illustrating the various stages in the formation of Ap1, Ap2, Ap3, and Ap4 in the Yinachang deposit

Full size image

Most previous studies of apatite alteration, including Li and Zhou (2015) on fluorapatite from the Yinachang deposit, have emphasized the localized nature of REE remobilization, as evidenced by the formation of REE-rich mineral inclusions within or beside the apatite grains (Pan et al. 1993; Harlov and Förster, 2003; Bonyadi et al. 2011; Broom-Fendley et al. 2016; Harlov et al. 2016; Hu et al. 2019; Ying et al. 2020). In this study, mass balance calculations for the formation of Ap3, as well as its associated REE-rich inclusions from Ap1, assuming no external contributions of Ca and P, suggests significant leaching of REE during hydrothermal alteration at the Yinachang deposit, which is further supported by the common occurrences of Ap4 and associated REE-rich minerals [monazite–(Ce), bastnaesite–(Ce), and synchysite–(Ce)] in late hydrothermal veins. These results demonstrate that REE remobilization associated with fluorapatite alteration in the Yinachang deposit was not strictly localized in extent, and REE in fluorapatite can also be transported in distance and precipitate as REE-rich minerals associated with other minerals in mineralized veins, which has important implications for upgrading of sub-economic REE mineralization in apatite-rich rocks. Specifically, REE remobilization associated with the hydrothermal alteration of apatite-rich rocks such as those in marine phosphorites, IOCG deposits, and carbonatite deposits (Li et al. 2015; Chakhmouradian et al. 2017; Krneta et al. 2017; Cherry et al. 2018; Algabri et al. 2020), are potentially promoting the mineralization of REE in fluorapatite.