Apatite and Monazite Geochemistry Record Magmatic and Metasomatic Processes in Rare Earth Element Mineralization at Mountain Pass, California

Carbonatites are an unusual class of igneous rocks with >30 to 50% carbonate minerals (Le Maitre et al., 2002; Mitchell, 2005). Some contain economic rare earth element (REE) mineralization, such as that observed in the Mountain Pass carbonatite orebody in California (Castor, 2008). Many carbonatites are spatially and temporally associated with alkaline silicate rocks (e.g., Woolley and Kjarsgaard, 2008), but this association may not represent a genetic link (e.g., Gittins and Harmer, 2003). The formation of carbonatites remains heavily debated. Three major formation models indicate that carbonatites may (1) originate as melts directly from a metasomatized mantle source region (e.g., Harmer and Gittins, 1998), (2) originate by liquid immiscibility in alkaline silicate melts from a metasomatized mantle source region (e.g., Le Bas, 1987; Berndt and Klemme, 2022), and (3) evolve from silica-undersaturated alkaline silicate melts by fractional crystallization (Mitchell, 2005). In reality, there may be many carbonatite types and formation mechanisms (Bell et al., 1998; Mitchell, 2005).

The past decades of carbonatite research have highlighted the important role of fluids in forming carbonatite REE deposits. Fluids may concentrate the REEs, either directly by partitioning of REEs into the fluid from the magma or indirectly by fluid leaching and remobilization of REE from primary igneous minerals. Fluids may be of magmatic origin (Costanzo et al., 2006; Ruberti et al., 2008; Nadeau et al., 2015; Xie et al., 2015; Trofanenko et al., 2016; Cheng et al., 2018; Smith et al., 2018; Beland and Williams-Jones, 2021b), from an external source such as the country rock (Cangelosi et al., 2020; Ma et al., 2021), or a combination of the two (Williams-Jones et al., 2000; Giebel et al., 2017).

Apatite [Ca₁₀(PO₄)₆(F,Cl,OH)₂] is ubiquitous accessory mineral in both silicate rocks and most carbonatites, usually occurring as fluorapatite. Because apatite is a mineral tolerant of a wide range of element substitution (Pan and Fleet, 2002; Piccoli and Candela, 2002), its trace element composition can reveal the magmatic and metasomatic history of the rock (e.g., Harlov et al., 2005; Harlov, 2015; Webster and Piccoli, 2015; Chakhmouradian et al., 2017). Apatite trace element compositions can also be used to discriminate types of ore deposits, magmatic environments, and host rocks (Belousova et al., 2002; Mao et al., 2016; O’Sullivan et al., 2020). Furthermore, experimental and natural studies have demonstrated a link between apatite and monazite (a light rare earth element [LREE] phosphate, [(LREE)PO₄]), wherein metasomatic alteration of apatite grains can result in precipitation of monazite inclusions, monazite pseudomorphs, or rims of monazite nucleating on the apatite surface (Harlov, 2002; Harlov and Förster, 2003; Harlov et al., 2005; Bonyadi et al., 2011; Beland and Williams-Jones, 2021a; Su et al., 2021; Chappell et al., 2023).

The Mountain Pass alkaline silicate suite contains apatite as an accessory mineral. Although the Mountain Pass carbonatite is unusually phosphorus-poor, monazite is the dominant REE mineral in some parts of the composite intrusion and apatite is present in trace quantities (Castor, 2008; Chakhmouradian et al., 2017). In this study, the texture and geochemistry of apatite and monazite grains in alkaline silicate and carbonatite intrusions at Mountain Pass are examined in detail to constrain the magmatic and metasomatic history of the rocks and parse genetic models of their formation.

Mountain Pass is located within the Mojave Province, which spans most of southeastern California and portions of Nevada, Arizona, and Utah (e.g., Bennett and DePaolo, 1987; Wooden and Miller, 1990; Whitmeyer and Karlstrom, 2007). Mesoproterozoic (ca. 1.45–1.37 Ga) igneous rocks of the Mountain Pass Suite crosscut Paleoproterozoic (ca. 1.8–1.6 Ga) crystalline basement rocks of the Ivanpah Mountains, which consist primarily of metasedimentary and metaigneous gneisses (Strickland et al., 2013; Watts et al., 2022). Granodioritic, tonalitic, and pelitic gneisses have foliations that trend northwest-southeast, roughly along the same strike as the Mountain Pass intrusions. The Mountain Pass intrusive suite occurs as part of a 130-km NW-SE belt of alkaline silicate intrusive rocks (Castor, 2008), several of which are geochemically and temporally similar to Mountain Pass (Gleason et al., 1994; Castor, 2008; Watts et al., 2024).

Eight map-scale stocks and hundreds of mapped dikes of carbonatite and alkaline silicate compositions compose the Mountain Pass intrusive suite (Fig. 1). The Birthday stock is the largest intrusion, dominantly composed of coarse-grained shonkinite, and crosscut by numerous fine-grained shonkinite, syenite, and alkali granite dikes, as well as carbonatite dikes hosting coarse-grained REE mineralization. Bordering the southern margin of the Birthday stock is the sole carbonatite stock in the district, the Sulphide Queen, a composite intrusion that ranges from calcite- to dolomite-dominant with abundant REE mineralization (Castor, 2008). To the south of these two largest intrusions are the Tors, Corral, Wheaton/Pops (here referenced as Wheaton), Groaner, Mexican Well, and Mineral Hill stocks. Tors is composed of melasyenite with crosscutting syenite and shonkinite dikes. It is well exposed and preserves abundant textures indicative of magma mingling (Haxel, 2005; Poletti et al., 2016). Corral and Wheaton are mixed stocks containing both syenite and shonkinite rock types, which grade into one another. Mineralogy and textures associated with fenitization are observed at Wheaton, most conspicuously as fibrous blue crocidolite (magnesioriebeckite). Groaner and Mexican Well are smaller syenite bodies and Mineral Hill is the southernmost and only alkali granite stock.

The mineralogy of the alkaline silicate rocks has been previously detailed (Olson et al., 1954; Castor, 2008) and is briefly described here. Shonkinite and melasyenite are dark gray rocks with modally abundant phlogopite, diopside, and varying amounts of blue-green sodic amphibole and K-feldspar. Apatite, monazite, zircon, titanite, allanite, barite, fluorite, pyrite, magnetite, and rutile occur as accessory and trace minerals. Potassium feldspar in these rocks is microcline and typically occurs interstitially. Syenite and leucosyenite, distinguished from the mafic members of the series by the presence of quartz, and from one another by modal abundance of quartz, contain abundant K-feldspar, variable phlogopite and sodic amphibole, and minor plagioclase and sodic clinopyroxene. Accessory and trace minerals include apatite, zircon, pyrite, galena, fluorite, thorite, monazite, and barite. Alkali granite contains quartz and K-feldspar with minor phlogopite, sodic pyroxene, and amphibole. Accessory and trace minerals include zircon, fluorite, monazite, thorite, pyrite, galena, and apatite.

Dolomite carbonatite constitutes the majority of the Sulphide Queen orebody, although calcite carbonatite is more common in the shallow portions of the intrusion (Castor, 2008). Barite is ubiquitous in Mountain Pass carbonatite, constituting roughly 20 to 30% of most ore rocks. Barite commonly gives the carbonatite a porphyritic texture, and individual phenocrysts can be centimeters in size. Bastnäsite [(LREE)CO₃F] is the dominant rare earth mineral (REM), typically as bastnäsite-(Ce), and occurs as both a primary and secondary mineral. Monazite is present as an REM in parts of the Mountain Pass carbonatite. In the dolomite carbonatite, monazite forms as much as 5 vol % of the rock and typically occurs as microcrystalline “bone” monazite intergrown with secondary bastnäsite. In the calcite carbonatite, monazite may occur as bone monazite or as primary, euhedral monazite. A small-volume monazite-bearing calcite carbonatite member contains little to no bastnäsite and is commonly associated with breccia (Castor, 2008). Other REMs include hydroxyl-bastnäsite [(LREE)CO₃OH], parisite [Ca(LREE)₂(CO₃)₃F₂], synchysite [Ca(LREE)(CO₃)F], and sahamalite [(Mg,Fe) (REE)₂(CO₃)₄], which are typically secondary in the carbonatite. Sulfides (galena, pyrite, and rare chalcopyrite), thorite, celestine, strontianite, and unidentified Fe oxides are present in accessory to trace quantities. Apatite is scarce to absent in all compositional members of the Mountain Pass carbonatite. Fenitized gneissic host rocks surrounding the Sulphide Queen stock typically contain brick-red K-feldspar, phlogopite, carbonate, magnesioriebeckite, chlorite, and veinlets of fluorite (Castor, 2008), as well as minor albite alteration of earlier feldspars (Olson et al., 1954).

Samples were collected from each of the alkaline silicate stocks of the Mountain Pass intrusive suite (Table 1). Apatite was separated from 16 of the 17 samples using conventional techniques; no apatite grains were recovered from the alkali granite stock. Samples were crushed and sieved to <425 μm, then passed over a water density table. This was followed by magnetic and heavy liquid separations. All samples were separated using bromoform, and additional methylene iodide separations were conducted for samples with substantial zircon fractions as identified using a binocular microscope. Apatite grains were handpicked under a binocular microscope and mounted in epoxy. To ensure that picked apatite grains were not biased toward a single visually distinct population, a bulk epoxy mount was also made for each sample. In addition to samples of the alkaline silicate rocks, carbonatite was sampled from an archival drill core collection at Mountain Pass. Two core samples contained apatite in thin section, and six contained monazite grains of sufficient size for analytical purposes. Apatite and monazite from these samples, and monazite in sample MIN-09, were analyzed in situ in polished thin sections.

Whole-rock geochemistry: Clean rock chips (>35 g), free from weathering rinds or alteration, were collected in the field for geochemical analysis. Whole-rock geochemical analyses were performed under U.S. Geological Survey (USGS) contract by SGS Canada Inc. or by AGAT Laboratories (as indicated in Table 3). Rock chips were prepared for analysis by grinding to <200 mesh using an agate pulverizer. Major elements were analyzed by X-ray fluorescence (XRF), and trace elements were analyzed by inductively coupled plasma optical emission (ICP-OES) and ICP-mass spectrometry (ICP-MS). For carbonatite samples, light REE (LREE) concentrations were also determined by XRF. Concentrations of Cl and F in the whole rock were determined by ion-selective electrode (ISE). To analyze S and CO₂, samples were combusted, and concentrations were measured by an infrared detector.

Electron microprobe analysis (EMPA): Minerals in grain mounts and thin sections were imaged by backscattered electrons (BSE) and major and minor element data were collected at the USGS Denver Microbeam Laboratory using a JEOL JXA-8530FPlus electron microprobe. For apatite analyses, the instrument was operated with an accelerating voltage of 15 kV and a beam current of 10 nA with a spot diameter of 10 μm. Although the beam current was higher than that suggested by Goldoff et al. (2012) to prevent halogen migration (Stormer et al., 1993; Goldoff et al., 2012; Stock et al., 2015), our time-dependent intensity curves indicated minimal migration occurred. Replicate analyses of the fluorine-rich Wilberforce apatite reference material (n = 85) were used throughout the analytical session to monitor accuracy. Manganese, Ba, and Cl are typically below the instrument detection limits (averaging respectively 0.03, 0.32, and 0.01 wt %) in this reference material (App. Table A1). Analytical uncertainty is higher for analyses approaching the lower limit of detection for an element. Based on previous monazite analyses in the Denver Microbeam Laboratory, the instrument was operated with an accelerating voltage of 20 kV and a beam current of 50 nA with a 10-µm spot diameter for monazite analysis. Replicate analyses of the NAM monazite reference material (n = 25) were used to assess analytical accuracy (Aleinikoff et al., 2012). This reference material is less homogeneous than the Wilberforce apatite. Detection limits and replicate analyses of the Wilberforce apatite and NAM monazite are provided in Appendix Tables A1 and A2. Inclusions in apatite and monazite were later identified qualitatively by energy dispersive spectroscopy (EDS) on a JEOL JXA-8500F hyperprobe field emission electron microprobe at the Washington State University GeoAnalytical Laboratory.

Hyperspectral cathodoluminescence (CL) data for apatite and monazite in grain mounts and thin sections were collected and analyzed using the xCLent Analysis software package (v. 5.0.39.0) and a CL detector attached to the USGS JEOL JXA-8530 FPlus electron microprobe. Acquisition parameters for apatite CL data were 15 kV accelerating voltage, 30 nA cup current, 30 ms dwell, and 1 µm step size. Acquisition parameters for monazite CL data were 20 kV accelerating voltage, 50 nA cup current, 30 ms dwell, and 1 µm step size. Element maps for major and minor elements in each mineral phase were collected simultaneously with CL data. For assessment of zoning within apatite, false-color CL images were produced from the raw hyperspectral data by defining three regions of interest, setting luminescence range 400 to 495 nm as blue, 495 to 590 nm as green, and 590 to 750 nm as red to produce a mixed RGB image (Steadman et al., 2022). The CL response of monazite is most intense outside the visible spectrum. False-color CL images for monazite were produced by setting luminescence range 290 to 400 nm (ultraviolet) as blue, 400 to 750 nm (the visible spectrum) as green, and 750 to 1,000 nm (infrared) as red.

Laser ablation (LA)-ICP-MS analysis: Trace element concentrations were measured at the USGS Denver USGS-LTRACE laboratory using an Applied Spectra Resolution S155 laser ablation system with an ATL excimer ArF laser source operating at 193 nm wavelength and ~5 ns pulse width coupled to an Agilent 8900 ICP-MS instrument. Ablation was performed in a He atmosphere at a flow rate of 400 mL/min. Ablation products were immediately mixed with Ar carrier gas at a flow rate of 1,150 mL/min after the ablation. A squid signal smoother device was used to remove laser shot noise (Müller et al., 2009). A suite of 38 elements was analyzed. The lists of isotopes and dwell times for apatite and monazite are presented in Appendix Tables A3 and A4. Spot ablation was carried out using a 24 µm spot size in the epoxy-grain-mount apatite samples and a 10 µm spot size in the thin-section apatite samples to avoid abundant inclusions. A 14 µm spot size was used for monazite analyses. Spot ablation was carried out at 3.5 J/cm² laser fluence, firing the laser at 5 Hz for 30 s. Primary calibration used large (50-µm diameter) spot analyses of NIST612 reference glass for all elements except P, which used the Durango apatite. Calibration reference materials were run in duplicate at the beginning, end, and every ~60 min to correct for instrument drift. Reference values for the NIST612 glass were from GeoReM preferred (Jochum et al., 2005). Reference values for the Durango apatite were from Jarosewich et al. (1980). USGS basaltic glass reference materials BCR-2g and GSD-1g were used for a secondary correction in all runs and matched at the spot size of the unknowns to correct for differences in spot size between primary calibration and unknowns (Thompson et al., 2022). Arsenic, Ta, and W were corrected for REE interferences (doubly-charged and oxide interferences, respectively) using a monazite reference material assumed to contain no As, Ta, or W. Data were processed using the LADR software, v. 1.1.07 (Norris and Danyushevsky, 2018). Small inclusion-induced signals were removed from the analytical signal, and analyses were discarded if fluctuations caused by inclusions were prevalent throughout the ablation signal. The ⁴³Ca isotope was used as an internal standard for apatite and all cations were stoichiometrically combined with oxygen in typical oxidation states (e.g., Ca²⁺, Fe²⁺). Then, all cations + oxygen were normalized to 96% total for each analysis, assuming 4% F + Cl + OH in apatite. Normalization to 96% total, rather than use of the EMPA data as an internal standard, avoids errors due to compositional zoning, because LA-ICP-MS samples up to 10 times the volume as that sampled by EMPA. The 96% normalization used in this study is an approximation that imparts a maximum bias of 1.6% (at endmember fluorapatite composition) in all data between LA-ICP-MS and EMPA analyses. The presence of Cl in apatite lowers this bias. This is a minor systematic offset given methodological uncertainties. Analytical uncertainties were exported at the “full analytical uncertainty” level, and replicate analyses for LA-ICP-MS reference materials (BHVO-2g and GSE-1g), which were analyzed as unknowns for quality control, have been provided in Appendix Tables A5 and A6. Replicate analyses of the Durango apatite, analyzed for quality control, are also presented. The Durango apatite is not homogeneous. For these analyses, a fixed Ca concentration was used instead of normalizing. Phosphorus was calibrated using the Durango apatite, thus P values for the replicate analyses are not provided in Appendix Table A5.

Mineral Stoichiometry: Using geochemical data collected by EPMA and LA-ICP-MS, apatite stoichiometry was calculated on the basis of cations in the Ca and P sites (Chakhmouradian et al., 2017; Walters, 2022) to account for nonstoichiometry of the X site. Silicon concentrations were determined by LA-ICP-MS and added to the stoichiometric calculation. Of the REE, only Ce was analyzed by EMPA; La, Pr, Nd, Sm, and Y were analyzed by LA-ICP-MS and added to the stoichiometric calculation. Monazite stoichiometry was calculated on an anhydrous four-oxygen basis. Strontium was added to the stoichiometric calculation of monazite from the LA-ICP-MS data.

New data acquired during this study were integrated with available published whole-rock data (Castor, 2008; Poletti et al., 2016; Watts et al., 2022) for Mountain Pass alkaline silicate intrusions and metamorphic basement rocks of the Ivanpah Mountains (Fig. 2; Table 2). All alkaline silicate samples are ultrapotassic, with K₂O contents from ~6 to 12 wt % and K₂O/Na₂O ratios of ~4 to 7 (Fig. 2; Table 2). There are no discernable compositional gaps for the whole-rock array, which varies in SiO₂ content (~45–75 wt %) and MgO (~0.1–13 wt %) (Fig. 2). Gradation from mafic to silicic samples progresses through a sequence of shonkinite, melasyenite, syenite, leucosyenite, and alkali granite (Fig. 2). Whole-rock P₂O₅ content is a good proxy for apatite abundance in the samples, with the most mafic samples having the greatest abundance of apatite and the highest P₂O₅ contents (Table 2). The low P₂O₅ concentration in the Mineral Hill sample, which lacks apatite, is attributed to trace amounts of monazite.

One of the most distinctive characteristics of the ultrapotassic rock suite is the uniformly high LREE concentrations, with La up to ~8,000 times the chondrite value, greatly exceeding the LREE concentrations of typical crustal rock such as the metamorphic basement (Fig. 3). The REE concentrations in the most REE-enriched alkaline silicate sample (BIR-16) are comparable to those in the least REE-enriched carbonatite sample. In the carbonatite, La concentrations range from ~10,000 to over 100,000 times greater than the chondrite concentration (Fig. 3, Table 3). Mountain Pass alkaline silicate and carbonatite rocks do not exhibit Eu anomalies, in contrast to the positive and negative Eu anomalies present in the surrounding crustal basement rocks (Fig. 3). Carbonatite and ultrapotassic rocks are both enriched in LREE compared to heavy REE (HREE), with average chondrite-normalized La/Lu = 61 for the ultrapotassic rocks and La/Lu = 3,500 for the carbonatites.

Inclusions are relatively common in Mountain Pass apatite but vary between the apatite groups and between intrusions. In all samples, Ap1 and Ap2 host inclusions of the major rock-forming minerals, including phlogopite, richterite, sodic pyroxene, and K-feldspar. Notably, one Ap2 grain from Corral contains an inclusion of bastnäsite, which is the dominant ore mineral in the carbonatite REE body. Inclusions of monazite, barite, and calcite are variably identified at each ultrapotassic silicate stock (Table 5). Zircon inclusions are exclusively hosted by Ap3 at Birthday and are not identified elsewhere. Ap3 cores do not contain inclusions of minerals aside from zircon, but other inclusions are sometimes observed in the Ap1 rim.

Forty to 50 apatite grains from each sample were analyzed for a total of 919 analytical points. One to three points were analyzed per grain, depending on the complexity of zoning observed in BSE images. After excluding points affected by mineral inclusions or epoxy signals, a total of 874 apatite analyses from the alkaline silicate rocks were evaluated for this study (complete dataset in App. Table A7 and Benson and Watts, 2024; Table 4 shows representative analyses). All grains from this study are fluorapatite (App. Fig. A1). Fluorine contents range from 3.1 to 5.3%, which corresponds to 1.6 to 2.8 atoms per formula unit (apfu) (App. Table A7). Apatite with F exceeding the stoichiometric value of 2 apfu may be the result of halogen migration (Stormer et al., 1993; Goldoff et al., 2012; Stock et al., 2015) or of overfilled F sites (Binder and Troll, 1989; Fleet and Liu, 2008; Yi et al., 2013), although possibly only as much as 0.2 apfu excess F can occur in natural apatite (Mason et al., 2009). Because OH was calculated stoichiometrically, only samples with F below ~3.8% are assumed to contain OH groups. There are 151 OH-bearing analytical points with an average OH content of 0.1 apfu. These apatites are approximately evenly distributed among the studied samples. Apatite with a small amount of Cl is relatively common in the analyzed apatite. Of the analyzed apatite grains, 731 (82%) points contain Cl above the detection limit, although 109 of these analyses contain low Cl (<150 ppm) with correspondingly higher analytical errors. The average Cl content is 0.02 apfu. Unlike F and OH, Cl is not evenly distributed across stocks. The concentration is notably higher in apatite from Birthday, up to 0.13 apfu (average = 0.07 apfu). Only 29 of 66 total carbonatite apatite analyses contain measurable Cl; the average is 0.006 apfu.

Zoning patterns in BSE and CL images reveal three apatite populations (Fig. 4; Benson and Watts, 2024). The Ap1 group includes grains that are homogeneous or concentrically zoned in BSE. Apatite grains that appear homogeneous in BSE may exhibit minor zoning in CL images due to variations in REE content. In false-colored RGB (see Methods), Ap1 grains are generally pink (Fig. 4A, C–E). Ap1 grains occur across all alkaline silicate stocks. Ap2 includes points on Ap1 grains with overgrowth rims or conversion textures as well as grains with patchy zoning (Fig. 4B, C). Ap2 grains in CL images exhibit similar, but more complex, patchy zoning compared to what is observed in BSE. The patchy zoning is generally blue-green or orange in false-color CL images. Ap2 grains occur in all alkaline silicate stocks except Birthday. The final group, Ap3, is the least abundant group. It occurs as partially resorbed cores (Fig. 4D), observed in about 8% of apatite grains from Wheaton and Corral, and as rounded cores (Fig. 4E), observed in about 47% of apatite grains in dikes hosted by Birthday (samples BIR-11, BIR-14, and BIR-16). Ap3 cores have an Ap1 rim. In BSE, Ap3 cores can mimic patchy Ap2 textures, and Ap1 rims around Ap3 cores can appear similar to Ap2 overgrowth textures. Ap3 cores must be distinguished compositionally rather than optically. The three apatite groups are variably distributed among intrusions (Fig. 5).

Apatite groups classified on the basis of BSE and CL images also exhibit differences in composition (Fig. 6). The Ap1 group has higher concentrations of Si, Mg, Sr, Ba, LREE, Th, and U than the Ap2 group, which conversely contains higher levels of Ca and P (Fig. 6A, B; Table 4). The Ap3 compositional group contains lower Sr concentrations (~400–8,000 ppm; Fig. 6D). Arsenic (~20–150 ppm) is observed commonly in the Ap2 group, although a handful of Ap1 and Ap3 analyses also contain high As (Fig. 6C). Sulfur contents are similar across all apatite compositional groups (Fig. 6C).

One of the most diagnostic chemical aspects that separates the three apatite populations are the relative REE concentrations (Fig. 7) and chondrite-normalized REE profiles (Fig. 8) as well as Eu anomalies [Eu/Eu* = Eu_CN/(Sm_CN × Gd_CN)^0.5, in which CN = chondrite normalized]. A bivariate plot of summed middle REE (MREE = Sm + Eu + Gd) vs. summed LREE (La + Ce + Pr + Nd) demonstrates different arrays for the three apatite populations (Fig. 7). Ap1 grains have distinctly higher abundances of LREE, but comparable abundances of MREE to Ap2 grains. The Ap3 population is distinguished by a strong negative Eu anomaly (Eu/Eu* = 0.16–0.69), which is less pronounced in Ap1 and Ap2 (Eu/Eu* = 0.62–0.98) (Fig. 8).

Apatite was identified in thin section in dolomite carbonatite sample 83-4-630 and dolomite-calcite carbonatite sample 85-12-304. In both samples, grains occur in granular aggregates (Fig. 9). Individual grains consist of generally homogenous unzoned cores with rims that are slightly darker in BSE. Some grains contain abundant barite and galena inclusions, which also occur along grain boundaries. Inclusions of monazite are also observed. Grain cores are generally pink in false-color CL images, and some grains appear blue near the edges of the apatite aggregates (Fig 9A, B). Apatite in sample 85-12-304 accounts for much of the mesostasis between calcite and barite grains. Apatite grains are patchy in BSE and host monazite and barite inclusions. In false-color CL images, grains are generally yellow to blue (Fig. 9C, D). Element maps highlight the lack of Sr and Na in most grains from sample 85-12-304, although some patches contain more abundant Sr and Na compared to the bulk of the grain (Fig. 9C, D).

Apatite grains from the two carbonatite samples are geochemically distinct from one another (Fig. 6; Table 4). Compared to sample 83-4-630, sample 85-12-304 has higher Ca and P concentrations (Fig. 6A), and lower abundances of Na, Sr, and Y (Fig. 6B, D). Essentially no As is present in apatite from sample 83-4-630, whereas small amounts are observed in apatite from sample 85-12-304 (Fig. 6C). In contrast, S was observed in most analyses from sample 83-4-630 but below detection limit in almost all analyses from sample 85-12-304 (Fig. 6C). Apatite REE concentrations also differ among the samples (Fig. 7), despite their similar whole-rock REE contents (Fig. 3). Sample 83-4-630 apatite is enriched in REE, typically containing 3 to 5 wt % LREE and 0.2 to 0.3 wt % MREE, whereas sample 85-12-304 apatite is typically REE-depleted, averaging 0.15 wt % LREE and 0.04 wt % MREE. The REE profiles underline the differences in REE composition for apatite from the two carbonatites (Fig. 10). Apatite grains from both carbonatite samples lack Eu anomalies (Eu/Eu* = 0.90 ± 0.05).

Apatite-group minerals are defined by the general formula M₁₀(ZO₄)₆X₂. In fluorapatite, the M-site is generally occupied by Ca, the Z site by P, and the X site by F. However, apatite-group minerals are tolerant of structural distortion and chemical substitution, resulting in compositional variation (Pan and Fleet, 2002). In addition to direct substitution for Ca²⁺ by other divalent cations (e.g., Sr²⁺), other relevant coupled substitutions include the following:

Where the above coupled substitutions occur in apatite, the apfu of the elements involved in the coupled substitution should yield a positive correlation. In the Mountain Pass apatite, none of these substitutions individually are well correlated, but when combined (i.e., Na + Si and S + REE) they demonstrate good correlation, indicating all four coupled substitutions have occurred (Fig. 11A–E).

In the apatite found in the alkaline silicate rocks, Sr is the most common substituent for Ca, although substitutions detailed in equations (1) and (2) also occur. Substitutions in Ap2 are similar to Ap1, but Ap2 contains lower concentrations of Si, REE, and Sr when compared with Ap1 (Figs. 6B, D, 7). The REE substitution in Ap2 is primarily balanced by the mechanisms detailed in equation (1) rather than equation (2) (Fig. 11B, C). Ap3 contains similar Si, S, and Na concentrations when compared with Ap1, but the former group often hosts lower REE and Sr contents.

High-REE apatite from carbonatite sample 83-4-630 dominantly exhibits coupled substitution detailed in equation (1) as well as direct substitution of Sr²⁺ for Ca²⁺ (Fig. 11B, F). Conversely, low-REE apatite from carbonatite sample 85-12-304 exhibits little substitution and contains some Si that does not appear to be paired with an appropriately charge-balanced companion (Fig. 11C, E). One possibility is that Si in apatite from this sample is charge-balanced by substitution of $CO32−$⁠; however, carbon contents of apatite were not analyzed in this study. Apatite from both carbonatite samples contains low levels of S (Fig. 6C).

Monazite was analyzed in six carbonatite thin sections and one alkali granite sample (MIN-09), which contained no apatite. A total of 54 monazite analyses were acquired. After excluding two points affected by the ablation of mineral inclusions or epoxy, 52 analyses are presented in this study (Table 6, App. Table A8; Benson and Watts, 2024). In three samples, monazite grains were sparse, and only two to three points were analyzed. In the other four samples, monazite grains were relatively abundant, and 10 to 15 grains were analyzed in each. Most monazite grains exhibit zoning in BSE and CL images (Fig. 12; Benson and Watts, 2024). Grains may exhibit oscillatory (Fig. 12B) or patchy (Fig. 12A, C, D) zoning, and some grains are porous (Fig. 12E). In monazite with oscillatory zoning, composition varies between relatively Th- and Ca-enriched bands. Broadly, Mountain Pass monazite contains few inclusions. Mineral associations vary by sample. The Mineral Hill monazite grains are substantially smaller than monazite grains from the carbonatite. Texturally associated minerals include hematite, fluorite, quartz, zircon, phlogopite, and K-feldspar. The two monazite analyses from Mineral Hill are compositionally distinct from those in the carbonatite (Fig. 13), with much higher Th and U contents (Table 6).

Monazite is defined by the general formula MZO₄. The M site hosts REE, predominantly LREE, in contrast to HREE-dominant xenotime (Boatner, 2002; Clavier et al., 2011). The two dominant substitutions are the huttonite-type substitution (eq. 5) and the cheralite-type substitution (eq. 6), so named for the endmembers isostructural with monazite (Kucha, 1980):

In the examined monazite, huttonite-type substitution is dominant, but minor cheralite-type substitution is required to balance the observed Th contents in the monazite from both the Mineral Hill alkali granite and the Sulphide Queen carbonatite (App. Fig. A2). Thorium contents in the Mineral Hill monazite grains are roughly 0.09 apfu (or 9% of the M site), but average only 0.016 apfu in the carbonatite samples. The substitution shown in equation (5) is dominant over the substitution in equation (6) in monazite from the carbonatite body. Aside from Si, substituents for (PO₄)^3– in monazite were not analyzed. All monazite samples exhibit similar Ce:La:Nd ratios, with Ce making up ~50% of the total LREE contents, La representing ~35%, and Nd representing ~15%.

Apatite: Compositional and textural variability in apatite is the result of varying magmatic and fluid-driven processes (Rae et al., 1996; Shore and Fowler, 1996; Tepper and Kuehner, 1999; Dempster et al., 2003; Harlov and Förster, 2003; Streck, 2008; Zirner et al., 2015; Bouzari et al., 2016; Broom-Fendley et al., 2016; Ladenburger et al., 2016; Cao et al., 2021; Lu et al., 2021). Magmatic processes such as fractional crystallization and magma mixing, and post-magmatic processes such as dissolution/reprecipitation and diffusive re-equilibration during fluid metasomatism are recorded texturally and compositionally.

Homogeneous, concentric, and oscillatory zoning are typically interpreted as primary magmatic textures (Shore and Fowler, 1996; Dempster et al., 2003; Streck, 2008; Bouzari et al., 2016; Ladenburger et al., 2016). Ap1 grains exhibit these zoning patterns (e.g., Fig. 4A), consistent with crystallization from an igneous melt. These apatite grains are found as inclusions in all major phases of the shonkinite through leucosyenite samples. The relatively high Sr, Ba, LREE, and Th content of the Ap1 grains mirror the high content of these elements in corresponding whole-rock analyses (Fig. 2; Table 2), which is further evidence that these grains crystallized directly from the magma represented by the whole-rock. High chondrite-normalized (CN) REE ratios, (La/Yb)_CN and (La/Sm)_CN, are consistent with crystallization from a melt enriched in REE (Fig. 14A) (Lu et al., 2021). Ap1 analyses also contain the highest total REE and are especially LREE-enriched, with high (La/Lu)_CN (Fig. 14B).

Ap1 compositions (primary apatite) analyzed from different stocks and rock types mostly overlap within broad compositional fields (Fig. 15). Evolutionary trends are apparent for primary apatite from a single intrusion but are not defined based on rock type or between intrusions. If the Mountain Pass suite was generated from a single parental magma, apatite would exhibit continuous compositional evolution from the mafic to felsic rock types. Instead, apatite compositions between rock types and stocks broadly overlap in composition. This suggests individual magma batches were derived from a similar source region, and perhaps utilized the same conduit system, but did not share a single lineage (e.g., Chakhmouradian et al., 2017). The stocks and dikes crystallized from separate melts, resulting in compositional variation of apatite within single units but no consistent, regional-scale trends. Zircon geochronology has similarly indicated long-lived magmatism at Mountain Pass, repeatedly tapping the same mantle reservoir with discrete melts that can be differentiated at the crystal scale (Watts et al., 2022).

Patchy and overgrowth textures in apatite have been attributed to metasomatic overprinting by evolved fluids or melts (Zirner et al., 2015; Bouzari et al., 2016; Ladenburger et al., 2016; Lu et al., 2021). Fluid-induced reactions are often linked to precipitation of monazite during remobilization of REE from apatite (Harlov et al., 2005; Bonyadi et al., 2011; Lu et al., 2021; Su et al., 2021). Fluid alteration of apatite has also been associated with the formation of porous, ‘turbid’ apatite replacing primary magmatic apatite (Broom-Fendley et al., 2016). Dissolution-reprecipitation of apatite during reaction with a fluid is enhanced by porosity of the mineral, which is necessary for fluid propagation (Putnis, 2009; Putnis and Ruiz-Agudo, 2013). Ap2 from the Mountain Pass alkaline silicate suite is texturally similar to metasomatized apatite from other carbonatitic and alkaline silicate complexes. The observed patchy textures, zones of alteration, and compositionally distinct overgrowths are consistent with metasomatism by a fluid agent (Fig. 4B, C). The patchy-zoned apatite is porous and cracked compared to primary magmatic Ap1. Further, monazite inclusions are prevalent in Ap2 grains. Fluid-assisted removal of REE from apatite and precipitation of monazite have been demonstrated experimentally at high temperatures (e.g., Harlov and Förster, 2003; Harlov et al., 2005). Because inclusions were identified qualitatively and not by Raman spectroscopy (Zhukova et al., 2022) or quantitative EMPA (Pan et al., 1993), it is possible that some inclusions identified as monazite are in fact rhabdophane [(LREE)(PO₄)·H₂O]. Experimental results have indicated that monazite nucleates in apatite at high temperatures, and no REE-phosphate nucleates within the apatite grain at lower temperatures (Harlov and Förster, 2003). The possibility of low-temperature nucleation forming rhabdophane is beyond the scope of this investigation.

The typical composition of Ap2 indicates that the metasomatizing fluid removed substituent cations and anions from the apatite structure. When compared with primary magmatic apatite, Ap2 is depleted in Si and LREE, but not Na or S (Figs. 6B, 7), indicating that the fluid dominantly mobilized Si and LREE. Depletion of Ap2 in Sr, Ba, Th, and U similarly points to the preferential removal of these cations. Ap2 is enriched in As relative to Ap1, which is consistent with hydrothermal alteration (Liu et al., 2017). Lu et al. (2021) found that low (La/Yb)_CN, (La/Sm)_CN, and Sr/Y were indicative of fluid alteration of apatite. These trends are also observed in Ap2 from the Mountain Pass suite (Fig. 14). The Ap2 compositional group is hereafter referred to as metasomatized apatite. Some apatite grains with metasomatic textures in BSE images align compositionally with Ap1 or Ap3 populations. These apatite grains are grouped separately in relevant figures. The presence of such textural and geochemical outliers in the apatite population demonstrates that some areas of metasomatic alteration partially preserve the precursor apatite composition.

The third group, Ap3, is distinguished by composition, because its textures are similar to both Ap1 and Ap2. Ap3 occurs only in dikes in the Birthday stocks and in a small number of grains from the Wheaton and Corral stocks. At Wheaton and Corral, Ap3 occurs as partially resorbed cores surrounded by primary magmatic (Ap1) apatite (Fig. 4D), which appears in BSE images to be similar to patchy Ap2 textures. In samples from dikes crosscutting Birthday, Ap3 occurs as rounded cores surrounded by thin rims of primary magmatic apatite (Fig. 4E) and appears similar to magmatic Ap1 zoning in BSE images. Ap3 is compositionally distinguished from both primary magmatic and metasomatized varieties by negative Eu anomalies (Fig. 8) and low Sr contents (Figs. 4D, E, 6D), both of which indicate plagioclase crystallization. Mountain Pass alkaline silicate and carbonatite rocks contain little to no plagioclase (Olson et al., 1954; Castor, 2008; Poletti et al., 2016). Inherited (ca. 1800–1600 Ma) zircons in Mountain Pass alkaline silicate rocks with distinctly low Eu/Eu* (0.1–0.4) support plagioclase crystallization in assimilated gneissic rocks (Watts et al., 2022). Ap3, with similarly low Eu/Eu* in addition to low Sr contents, is here interpreted as inherited apatite. The apatite grains are compositionally variable, likely due to variability in the basement rocks from which they originated (Fig. 1).

Inherited apatite is rare, at both Mountain Pass and in other igneous bodies, because of the relatively rapid dissolution under igneous and metamorphic conditions, particularly in hydrous melts (Harrison and Watson, 1984). Nonetheless, inherited apatite grains have been documented in granite (Dempster et al., 2003), low-grade metamorphic rocks (Henrichs et al., 2018), and fenitized carbonatite (Slezak et al., 2018). At Mountain Pass, inherited apatite cores are most common in dike samples from Birthday. The fine-grained texture of the dikes indicates fast cooling that may have prevented dissolution of inherited apatite grains. At Corral and Wheaton, inherited grains are partially resorbed, suggesting slower cooling and possibly higher fluid content compared with the Birthday dikes. The absence of inherited apatite grains in other stocks (Groaner, Mexican Well, and Tors) may be related to more hydrous parental magmas. Using experimental observations of apatite dissolution in hydrous and dry melts, Harrison and Watson (1984) calculated geologically instantaneous dissolution of inherited apatite grains in hydrous melts, whereas apatite in dry melts required 2 Ma to dissolve. The Corral, Birthday, and Wheaton stocks contain substantially lower proportions of metasomatized apatite (Fig. 5), suggesting less hydrous melts, which preserved the inherited apatite.

Apatite populations in the two carbonatite samples are compositionally and texturally distinct from one another. Apatite from sample 83-4-630 is similar to primary apatite from the alkaline silicate suite, although apatite from sample 83-4-630 contains higher total REE and is more LREE enriched. Most apatite grains in sample 83-4-630 are interpreted to have crystallized from the carbonatite melt. The pill-like morphology of apatite grains from this sample (Fig. 9A, B) may be a morphology specific to primary apatite in plutonic carbonatites (Chakhmouradian et al., 2017). Apatite of similar morphology has been synthesized experimentally in P₂O₅-saturated, silica-undersaturated melts in the MgO-CaO-SiO₂-P₂O₅ system (Shyu and Wu, 1991). The low abundance of Si in apatite from this sample reflects the silica-poor carbonatite melt. Alkalis present in the carbonatite melt could explain the increased Na content. A small number of grains from this sample indicate metasomatism. These grains exhibit zoning in CL images and (La/Sm)_CN and (La/Yb)_CN trends similar to the metasomatized apatite (Fig. 14A). In contrast, apatite from sample 85-12-304 contains very low total REE and Na concentrations when compared with metasomatized apatite from the alkaline silicate rocks, which is the result of extensive metasomatic alteration. The low S contents of apatite from both samples are likely due to incorporation of sulfur into abundant barite rather than sparse apatite in the carbonatite.

O’Sullivan et al. (2020) presented a classification scheme for apatite on the basis of LREE and Sr/Y (Fig. 16). Primary magmatic apatite from this study falls into the ultramafic (UM, including carbonatite) and alkali-rich igneous rock (ALK) fields. Primary apatite of the Sulphide Queen carbonatite (sample 83-4-630) also falls within the ultramafic field, along the alkali-rich igneous rock border, providing support for a genetic link between the alkaline silicate rocks and the carbonatite. Metasomatized apatite of the alkaline silicate suite plots mostly within the alkali-rich igneous rock field, but in comparison to the primary apatite group, the metasomatized group is depleted in Sr and LREE, and trends toward the low-grade metamorphic/metasomatic (LM) field. The metasomatized apatite does not plot within the LM rock field, suggesting it is only partially metasomatized. Apatite from the metasomatized carbonatite sample (85-12-304) exhibits greater LREE loss. Apatite analyses from samples 83-4-630 and 85-12-304 fall along a trend from primary compositions in the ultramafic region to metasomatized compositions in the LM rock region.

Monazite: Few systematic studies of compositional variability in monazite have been conducted. Carbonatite monazite has generally been noted to be ThO₂-poor relative to other igneous monazite, possibly due to lower carbonatite melt temperatures (Overstreet, 1967; Wall and Zaitsev, 2004). Hydrothermal monazites in carbonatite may exhibit (La/Ce)_CN <1 and reduced ThO₂ content relative to primary carbonatite monazite (Wall and Zaitsev, 2004). Hydrothermal monazite in noncarbonatite epigenetic orebodies generally contains <1 wt % ThO₂ (Schandl and Gorton, 2004).

A recent synthesis of well-constrained monazite from nine localities, including monazite from the Gifford Creek carbonatite in Australia (Zi et al., 2024) also found that hydrothermal monazite generally contains <1 wt % ThO₂. Zi et al. (2024) proposed using Th and U content and the Th-to-U ratio to distinguish monazite origins, and they developed separate fields for carbonatite and other magmatic and hydrothermal monazite but did not distinguish between hydrothermal and primary monazites in carbonatite. In contrast, hydrothermal monazite from Lesnaya Varaka, Kola, Russia, which formed by replacement of primary niobates, contained ~35 wt % ThO₂, the highest ever recorded in monazite (Chakhmouradian and Mitchell, 1998). Formation of monazite after Th-rich apatite can similarly result in ThO₂-rich hydrothermal monazite (Harlov and Förster, 2003; Harlov et al., 2005). The U and Th concentrations of the monazite are compared to fields established by Zi et al. (2024) (Fig. 17B, C). Both the hydrothermal and igneous monazite grains overlap the carbonatite field (Fig. 17B), but the hydrothermal monazite analyses trend toward the hydrothermal field whereas the igneous monazite analyses trend toward the igneous field. More work is needed to synthesize existing monazite data to define the compositional changes that occur with hydrothermal alteration of monazite in carbonatites. The Th and U concentrations combined with the textural information of this study suggest that, as in more typical igneous systems, metasomatic alteration of monazite typically results in reduced Th concentrations, although instances of high-Th metasomatic monazite have been documented (e.g., Chakhmouradian and Mitchell, 1998).

The two monazite grains from the Mineral Hill alkali granite contain substantially higher ThO₂ concentrations (~9.5 wt %) than primary monazite from the carbonatite. This may be an igneous feature, as these analyses overlap the monazite magmatic field of Zi et al. (2024) (Fig. 17B). In this case, the higher ThO₂ could be due to higher temperatures in the alkali granite melt compared to the carbonatite (Wall and Zaitsev, 2004). However, the complex zoning of these monazite grains (Fig. 12A) is probably due to metasomatism. As noted above, high ThO₂ in monazite can be associated with secondary processes (Chakhmouradian and Mitchell, 1998). One monazite grain in this sample appears to be a pseudomorphic replacement of apatite, and some very minor, porous, relict apatite occurs around that grain. Apatite in the alkaline silicate rock has generally higher Th concentrations (Table A7), and sample MIN-09 contains thorite, suggesting relatively high ThO₂ in the melt.

Birthday stock: Despite some small-scale textural features that indicate minor fluid alteration of apatite in the Birthday stock, none of the analyzed regions have compositional evidence of metasomatism (Fig. 5). This suggests that the widespread fluid circulation observed in the other intrusive bodies of the Mountain Pass suite was not present at Birthday. Of the five samples collected in the Birthday stock, two were from the coarse-grained stock itself. The REE profiles of these apatite grains are generally tightly clustered. Apatite from the shonkinite sample is slightly REE-enriched compared with that from the melasyenite. Apatite from the unaltered Birthday stock samples (BIR-12 and BIR-15) hosts 70 to 90% of the whole-rock REE (Table 7), calculated based on the average REE contents of the primary apatite and an estimated normative apatite content determined for each sample based on the whole-rock P₂O₅ content. In comparison, apatite grains from the three dikes (BIR-11, BIR-14, BIR-16) sampled at Birthday host less of the total LREE content of the rock. LREE in sample BIR-16, which contains high whole rock LREE (Fig. 3) and lower apatite Ce concentrations (Table 7), are instead hosted by allanite. Fluorapatite in allanite-bearing rocks is generally LREE-depleted (Krenn et al., 2012).

Wheaton stock: Apatite in the Wheaton stock is almost entirely primary, with a few grains containing inherited cores. The Wheaton whole-rock sample contains high REE (750 ppm), and apatite hosts roughly 50% of the whole-rock REE budget (Table 7). Minor inherited apatite is present at Wheaton, exclusively as partially resorbed cores (Fig. 4D). Sample WHE-10 is a coarse-grained rock and contains abundant phlogopite. Slow cooling of a hydrous magma should be at odds with preservation of inherited apatite (Harrison and Watson, 1984). Apatite occurs poikilitically within euhedral phlogopite and pyroxene, and does not occur within late, interstitial K-feldspar, suggesting apatite was an early liquidus mineral at Wheaton. Primary apatite (Ap1) crystallizing early around the inherited cores (Ap3) may have protected the cores from complete dissolution.

Mineral Hill stock: The Mineral Hill alkali granite stock was not found to contain any apatite. Exceedingly little phosphorus is present in the rock sample (0.05 wt % P₂O₅). Previous geochronology at Mineral Hill (Poletti et al., 2016) found all zircon to be discordant, and thus the emplacement timing of this intrusion relative to the other intrusive bodies is poorly constrained. The low P contents at Mineral Hill are similar to the low P observed in the carbonatite orebody, which was emplaced at the end of the Mountain Pass intrusive sequence. The monazite identified in the sample is a trace constituent. Mass balance calculations indicate that monazite hosts 50% of the whole-rock REE budget (Table 7). The patchy texture in BSE and CL images, combined with the possible pseudomorphic replacement of primary apatite, suggests a metasomatic origin.

Tors stock: Magma mingling textures between shonkinite/melasyenite and syenite at Tors support emplacement of compositionally diverse magmas over a relatively short period of time. Field evidence indicates that exsolution of magmatic fluids produced pegmatitic amphibole (arfvedsonite) in areas of magma mingling, such as along dike margins and in shonkinite inclusions (c.f., fig. 24 of Haxel, 2005). The outcrop-scale indications of fluid exsolution during crystallization of Tors magmas are also apparent at the crystal-scale in apatite. Complex, patchy zoning in many Tors apatite grains indicates pervasive metasomatism. Minor, fine (1–5 µm) monazite grains are observed within some metasomatized apatite grains at Tors but do not occur in primary apatite. Fluid-induced nucleation of monazite after apatite has been observed experimentally (e.g., Harlov et al., 2002a, 2005; Harlov and Förster, 2003) and in nature (e.g., Harlov et al., 2002b; Bonyadi et al., 2011; Beland and Williams-Jones, 2021a; Su et al., 2021). It is schematically described as follows (Pan et al., 1993):

The formation of only minor monazite in most Tors grains indicates metasomatism by a low-temperature fluid. Harlov and Förster (2003) found that at T ≥ 900°C, monazite will nucleate within the apatite grain following REE removal, but below this temperature, monazite will only crystallize externally on the surface, suggesting transport of LREE out of the grain. Apatite in sample TOR-19, a pegmatite at Tors, contain larger and more abundant monazite inclusions when compared to apatite from other Tors samples. Fluid interacting with this sample was a higher-temperature fluid responsible for crystallization of patches of pegmatite, which did not transport the REE out of the apatite but instead nucleated larger monazite grains. The average LREE content of primary apatite (Ap1) at Tors is 24,000 ppm, and the average LREE content of the metasomatized apatite (Ap2) is 9,600 ppm. Using the average LREE contents of the primary apatite at Tors as representative of the total LREE availability prior to metasomatism, metasomatized apatite retains on average only 40% of initial LREE contents and has lost ~60%, or 14,400 ppm. Some of the lost LREE nucleated within the apatite as monazite. However, many metasomatized apatite grains do not contain visible monazite inclusions on the polished analytical surfaces, suggesting at least some of the LREE was mobilized away from the apatite grain entirely.

Groaner stock: Apatite from the Groaner stock is almost entirely metasomatized. Although some grains preserve primary compositions, all grains appear patchy and porous. As at Tors, monazite inclusions are fine or not present in metasomatized apatite grains, suggesting REE has been lost from the apatite. All grains at Groaner have been at least partially metasomatized. A comparison of the most primary apatite composition to the average metasomatized apatite compositions suggests that, as at Tors, apatite at Groaner has lost much of its original LREE. Although it is a small, coarse-grained syenite body, it contains evidence of extensive metasomatism outside of the apatite, including the presence of riebeckite, generally associated with fenitization at Mountain Pass.

Corral stock: Corral is the only stock to contain a mixture of all three apatite compositional groups. Two samples were analyzed from Corral. Apatite from the shonkinite member lacked metasomatized apatite (Ap2) and contained distinctly inherited apatite cores (Ap3). In contrast, the syenite member of Corral contained metasomatized apatite (Ap2). Inherited apatite cores in this sample were often geochemically intermediate between primary and inherited compositions, indicating resorption and alteration, which affected the geochemical signature of the inherited apatite during its reaction with a hydrous alkali-rich silicate melt. Primary apatite in the syenite sample contains lower average LREE than the shonkinite sample that lacks metasomatized apatite, which may be the result of either fractional crystallization or partial metasomatism of primary apatite (Ap1). Roughly 35% of the LREE budget of Ap1 has been leached away by metasomatism of these grains.

Mexican Well stock: Metasomatized apatite at Mexican Well contains substantially more monazite than all the other samples, often as abundant inclusions along the edges of apatite grains (Fig. 4C). Most of the LREE removed from the apatite in this sample was reprecipitated in situ. Some metasomatized Mexican Well apatite is texturally more similar to apatite observed at Tors, Groaner, and Corral, with complete metasomatism and few to no monazite inclusions. This suggests multiple fluid generations at Mexican Well, of differing compositions and temperatures, that resulted in unique episodes of monazite precipitation and variable LREE mobilization.

Sulphide Queen stock: The carbonatite orebody is notably depleted in phosphorus, particularly compared with other carbonatites, where apatite is a major modal constituent or monazite the dominant LREE host (Chakhmouradian et al., 2017). In carbonatites with high P contents, apatite is an early liquidus mineral and can incorporate substantial REE, which may prevent the carbonatite from becoming highly REE-enriched and precipitating ore-grade concentrations of REM (e.g., Huang et al., 2024). In this sense, the lack of P may have contributed to the high REE grade of the Mountain Pass orebody (Castor, 2008). Monazite-dominant carbonatite occurs only locally near breccia zones, suggesting the P required to produce phosphate minerals as the dominant REM partially originated from the brecciated country rocks or alkaline silicate rocks. The apatite-bearing carbonatite samples contain no REM beyond monazite and have the lowest whole-rock REE contents of any carbonatite samples analyzed in this study. Because the apatite-bearing carbonatite samples are not Si-enriched and are not spatially associated with brecciated country rocks, it is reasonable to assume that apatite crystallized directly from the carbonatite. The pill-like morphology of these grains further supports this assumption. Chakhmouradian et al. (2017) suggested apatite in the Sulphide Queen was commonly xenocrystic, indicating apatite was not stable in its magmatic environment. The results presented here provide evidence for small pulses of P-enriched carbonatite melt where apatite was stable. Aside from the lack of bastnäsite, these samples are mineralogically identical to more typical Mountain Pass carbonatite rocks.

Metasomatic processes of REE mobility in alkaline and carbonatite intrusions: Internal REE mass balance for apatite indicates loss of LREEs from the apatite (Fig. A3), but the distance LREEs were mobilized is unclear. Apatite in heavily metasomatized samples (TOR-01, COR-04, MEX-06, GRO-07, TOR-17, TOR-18, and TOR-19) hosts less than 30% of the LREEs in the sample. In MEX-06 and TOR-19, this imbalance is primarily due to in situ nucleation of monazite as inclusions in apatite grains. In the remaining samples, alternative REE hosts within apatite are not evident in thin section, but high LREE contents in the whole-rock data indicate that these elements may not be remobilized to greater distances than the outcrop scale. If LREEs were removed from the stock by fluids causing the alteration of apatite, the LREE concentration in the whole rock should have been reduced, which is not observed. For example, Groaner has a whole-rock analysis with ~600 ppm Ce, whereas whole-rock samples from stocks lacking metasomatized apatite contain ~500 to 700 ppm Ce. A sample of the Groaner stock analyzed by Poletti et al. (2016) contains substantially less Ce (~250 ppm), providing evidence of heterogeneity in REE contents within even a relatively small stock.

The distances over which REEs can be remobilized by fluid before being deposited remain poorly constrained by the available studies of experimental and natural systems, but likely are influenced by temperature and fluid composition. REEs are transported in fluids by forming complexes with ligands. The dominant ligands for REE transport in fluids include F, Cl, SO₄, and CO₃ (Migdisov and Williams-Jones, 2014; Migdisov et al., 2016; Louvel et al., 2022). Experimental results have found that, although REEs form the strongest complexes with fluoride ligands, carbonate and fluoride function as depositional ligands, whereas sulfate and chloride function as transport ligands (Migdisov et al., 2016). Notably, REE transport in alkaline fluids may differ from acidic fluids, and high alkalinity may promote transport of REE with fluoride and carbonate in carbonatitic systems (Louvel et al., 2022). The composition of the fluid(s) present in the Mountain Pass stocks can be estimated based on apatite geochemistry, using these experimental studies as a guide.

Metasomatized apatite (Ap2) is Si-depleted but contains similar Na compared to primary apatite (Fig. 6B). This suggests that the fluid contained Na and was Si-depleted, such that Na in the fluid was in equilibrium with the apatite but Si was not. An Na-rich fluid would be alkaline, rather than acidic. The presence of riebeckite (sodic amphibole) as an alteration product throughout the ultrapotassic stocks similarly supports the presence of a Na-rich fluid. Apatite SO₃ contents overlap between intrusions and apatite compositional groups (Fig. 6C), indicating no major change in sulfur contents during metasomatism. This suggests a sulfate-bearing fluid was in equilibrium with the apatite grains. Carbonate was not analyzed in apatite in this study, and thus cannot be compared among groups. The presence of calcite inclusions in metasomatized regions of apatite, as well as minor calcite in the groundmass of several samples at Tors, indicates the fluid may have been CO₃ bearing.

The role of chlorine (Cl) in the fluid is unclear. Chlorine is globally common in crustal fluids at depth (Bucher and Stober, 2010), and has been identified in carbonatite-related fluid inclusions in other localities (e.g., Walter et al., 2021). The fenitization assemblage in the host gneiss (Castor, 2008) suggests a Na- and K-rich fluid, as magnesioriebeckite, K-feldspar, and albite are the main components of the fenite mineral assemblage. Albitization has been associated with NaCl-rich fluids, but can also be associated with F-rich fluids (e.g., Duan et al., 2022). Metasomatized apatite at Mountain Pass is depleted in Cl relative to primary apatite, and apatite at Birthday, which lacked metasomatized apatite, contains the highest Cl content. Experimental metasomatism of apatite indicates both NaCl-rich fluids (Harlov and Förster, 2003) and F-bearing fluids (Harlov et al., 2002a) can replace Cl with F in the apatite structure. However, Cl-bearing phases are absent in the alkaline silicate rocks, the carbonatite, and the fenitized gneiss. Abundant F-bearing minerals are present, including fluorite, amphibole, and phlogopite in the alkaline silicate rocks, bastnäsite in the carbonatite orebody, and fluorite veinlets in the fenitized gneiss. Rare bastnäsite inclusions in the metasomatized apatite are also observed. There is clear evidence for abundant F in the Mountain Pass melts and fluids, and minimal evidence for the presence of Cl. However, fluorine metasomatism of chlorapatite has not been found experimentally to mobilize LREE (Harlov et al., 2002a), in contrast to NaCl metasomatism (Harlov and Förster, 2003). At T ≥ 400°C, an alkaline, carbonate-rich fluid can keep both LREEs and F mobile (Louvel et al., 2022). The role of K and Na in the fluid for enhanced LREE mobility has been emphasized in experimental work by Anenburg et al. (2020). The presence of sulfate in the fluid can enhance LREE mobility as well (Migdisov et al., 2016; Cui et al., 2020; Wan et al., 2021; Wan et al., 2023). Evidence from the apatite grains in this study supports a carbonate-, sulfate-, and fluorine-bearing alkaline fluid capable of mobilizing LREEs at Mountain Pass. The prominence of Cl in this fluid remains unclear and cannot be determined with the available apatite compositional data, although the depletion in LREEs observed in metasomatic apatite suggests at least some Cl was present. Detailed studies of the metasomatic alteration of phlogopite and amphibole, as well as investigation of the metasomatic alteration of the host gneiss, may help to quantify Cl in the fluid.

Monazite is present in the Sulphide Queen carbonatite as a volumetrically minor part of the groundmass, typically where bastnäsite is the dominant ore mineral. This bone monazite (Castor, 2008) is generally too fine to analyze. Where monazite occurs in grains large enough for accurate analysis, it is commonly the dominant, or only, REM observed in the sample. In samples 85-9-157 and 85-1-162, for instance, bastnäsite is not observed in thin section, and monazite accounts for 60 to 90% of the LREE contents of the sample (Table 7). In contrast, in samples 84-6-125 and 85-12-317, where monazite co-occurs alongside relatively abundant bastnäsite, monazite accounts for only 5 to 10% of the total LREE contents. The two samples for which monazite is the dominant REE host mineral are breccia samples, suggesting assimilation of P from the wall rock initiated monazite crystallization (e.g., Chakhmouradian et al., 2017).

Apatite is not observed in most Mountain Pass carbonatites, but where present it has an unexpectedly high modal abundance. Apatite in sample 85-12-304 accounts for roughly 10 vol % of the rock. Despite this, apatite in this sample is REE-depleted and hosts only ~5% of the whole-rock REE budget. No other modally abundant REE phases are found in the sample. Modally abundant carbonates, combined with monazite inclusions within the apatite, may dominate the REE budget of this sample. Primary calcite and dolomite minerals in the Sulphide Queen can host LREEs at 1,000 to 10,000 times chondrite values (J. Thompson, pers. commun., 2024). LREE-depleted and monazite inclusion-rich apatite in this sample indicates extensive metasomatism. Intrusion-scale circulation of LREE-enriched fluid may have contributed to higher ore grades elsewhere in the Sulphide Queen stock. Apatite hosts roughly 60% of the total REE budget of sample 83-4-630 (Table 7), indicating that it has not been heavily reworked by a fluid. These apatite-bearing samples have the two lowest whole-rock REE contents of any carbonatites analyzed in this study. They may represent unusual pulses of phosphorus-enriched carbonatite magma, containing sufficient P for crystallization of apatite which prevented the melt from becoming REE-enriched.

Rare earth mineralization in the Sulphide Queen has long been considered dominantly primary, or magmatic (e.g., Mariano, 1989). Both Mariano (1989) and Castor (2008) describe minor fluid-related mineralization, although it is unclear on what basis this distinction is made. Some modern work on carbonatites has classified Mountain Pass as a carbothermalite or carbothermal residue (formed by precipitation of solids from fluid) (Mitchell and Gittins, 2022). The presence of both igneous and metasomatic phosphates in various samples of carbonatite indicates that some phosphate minerals crystallized from a melt, whereas others were formed by fluid interaction. The same may be true for bastnäsite (Andersen and Watts, 2024); further investigation of the rare earth mineralization in the Sulphide Queen could constrain the effect fluids had in producing the dominant mineralization at Mountain Pass.

Origin of fluid in the alkaline silicate rock: One possible source of fluid is from the country rock. A crustal fluid released by contact metamorphism when the stocks were emplaced could explain the spatial distribution of the metasomatized apatite. The Mountain Pass intrusive suite is hosted by two different gneiss units (Fig. 1). The leucocratic gneiss hosts Groaner, Corral, and Tors, the three intrusions that contain substantial metasomatized apatite. Wheaton and Birthday, which contain only minor metasomatized apatite, are hosted by the granodiorite gneiss. Mexican Well, where apatite metasomatism is variable, is located at the border of these two gneiss units. Differences in fluids released by contact metamorphism in the two gneiss units could present a control on fluid availability for apatite metasomatism in the stocks. In general, gneisses have low fluid content available for release during dehydration, and the small size of most alkaline silicate stocks may not have provided sufficient heat for substantial dehydration.

A more feasible source is an alkaline magmatic fluid exsolved from the stocks. Metasomatized apatite is most prevalent at Tors, where clear magma mingling and fluid exsolution textures are observed in outcrop. Furthermore, autometasomatic fenitization by alkaline fluids is common in carbonatite and alkaline silicate intrusive bodies (e.g., Elliott et al., 2018). Secondary calcite at Tors supports interaction of the syenite with a $CO32−$-enriched fluid in the stock. High F contents would also be expected in a magmatic fluid exsolving from the alkaline silicate melt. Differences in the spatial distribution of metasomatized apatite may relate to intrusion timing. At Birthday, crosscutting dikes were emplaced after the stock had cooled; the Birthday dikes are very fine-grained and preserved abundant inherited apatite. In contrast, dikes that intruded the Tors intrusion have a coarse grain size, indicative of slower cooling, and show evidence of magma mingling, which may have caused fluid exsolution and thus apatite metasomatism.

Genetic links between the carbonatite and alkaline silicate rocks: The Sulphide Queen carbonatite contains a high barite component (typically 20–30 vol %). The presence of barite inclusions in apatite is uncommon in apatite from other localities, but barite inclusions are frequently identified in both primary and metasomatized apatite from the alkaline silicate stocks at Mountain Pass (Table 5). Furthermore, some metasomatized apatite grains contain calcite and bastnäsite inclusions, which indicate high CO₃ and F concentrations in the fluid. High concentrations of LREE, Ba, Sr, and Th in primary magmatic apatite (Figs. 6, 7) and corresponding whole-rock data for the alkaline silicate intrusions (Fig. 2) support a geochemical link to the Sulphide Queen stock, most likely related to a similar mantle source region.

The lack of apatite compositional evolution among stocks supports a model in which the carbonatite and alkaline intrusions were derived as separate melts, repeatedly tapping a similar source region over an extended (10s of millions of years) period of time (Poletti et al., 2016; Watts et al., 2022). The longevity of the Mountain Pass intrusive system (ca. 1.45–1.37 Ga) and lack of systematic evolution among intrusions are evidence against carbonatite formation by immiscibility from a hybrid carbonate-silicate parental melt. Abundant phlogopite in the alkaline silicate rocks supports the presence of hydrous, fluorine-rich melts. An altered shonkinite dike (glimmerite) at Mountain Pass was documented with matrix calcite, zircon grains with calcite inclusions, and low (≤500°C) Ti-in-zircon temperatures (Watts et al., 2022). This evidence supports carbonate-rich fluids and hydrothermal modes of mineral precipitation. Calcite has been identified in some samples from Tors in this study, and Castor (2008) identified melasyenite dikes containing carbonate minerals at the New Trail Canyon locality in the Mojave ultrapotassic belt, which he considered to be hybrid alkaline-carbonatite intrusions. One possibility for carbonate enrichment in silicate melts could be underplating of alkaline silicate intrusions by carbonatite sills or carbonatite degassing, which would result in unusually high (CO₃)^2– fluxes, the precipitation of carbonate minerals, and exsolution of carbonate-rich fluids.

The Mountain Pass carbonatite orebody is distinctly poor in phosphorus, but the associated alkaline silicate rocks contain abundant apatite grains with a textural and chemical archive of magmatic and metasomatic processes. Inherited apatite grains from the alkaline silicate rocks support previous interpretations of crustal assimilation in the intrusive suite. The compositions of primary magmatic apatite support a model in which magmatism repeatedly tapped an extensive mantle reservoir, resulting in within-stock, but not between-stock, evolution of apatite compositions. The presence of carbonatite-related inclusions, such as barite and calcite in primary apatite from the alkaline silicate rocks, and the compositional similarity of primary apatite in the silicate rocks to sparse apatite in the carbonatite, indicates the alkaline silicate rocks and the carbonatite are genetically related. A population of metasomatized apatite records the effects of an alkaline F- and CO₃-bearing fluid that mobilized LREEs from primary apatite. The presence of apatite in the carbonatite is generally correlated with REE-depleted samples. The absence of apatite in REE-rich carbonatite samples suggests low phosphorus contents may have enabled development of high-grade fluorocarbonate ore, as apatite was not an early liquidus mineral that removed REEs from the carbonatite melt. Phosphate chemistry supports a mixed igneous-hydrothermal origin for the Mountain Pass carbonatite, challenging longstanding assumptions of a primary magmatic origin. Fluid remobilization of LREEs may have contributed to the development of high-grade mineralization in the carbonatite. Additional high spatial resolution studies of its minerals could enhance understanding of the complex igneous and metasomatic processes responsible for REE mineralization in this world-class deposit.

Geochemical data, including whole-rock geochemistry and major and trace element data for apatite and monazite analyzed in this study, are available in the supplementary material for this publication. The geochemical data, back-scattered electron images, element maps, and composite cathodoluminescence maps for apatite and monazite that support this study are also available as a USGS data release from Benson and Watts (2024).

Lin Xia and Robby Ruesch of MP Materials are thanked for providing access to sites within the Mountain Pass mine property and to their archival drill core collection. Steve Castor provided additional Molycorp drill core samples and maps. Discussions with Steve Castor, Tony Mariano, Sr. and Tony Mariano, Jr. about Mountain Pass phosphate occurrences were informative and guided sample selection. We thank Heather Parks of the USGS for her assistance in generating GIS maps to guide our field work. Sharing of ideas with USGS colleagues Allen Andersen and Jake Poletti improved this work. USGS undergraduate interns Erin Toulou, Marley Ollero, and Natalie Potter are thanked for their assistance with field work and mineral separations. Heather Lowers and Jay Thompson at the USGS Denver Microbeam Laboratory and LTRACE Laboratory provided guidance during analytical sessions. Scott Boroughs at the Washington State University GeoAnalytical Lab assisted in inclusion identification in apatite and monazite. We thank Anton Chakhmouradian for his exceptionally detailed editorial handling of this manuscript. Reviews by Allen Andersen, Anton Chakhmouradian, Daniel Harlov, and an anonymous reviewer greatly improved the manuscript. This research was funded and supported by the USGS Mineral Resources Program and the USGS Mendenhall Research Fellowship Program. Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.

Erin Benson is a U.S. Geological Survey Mendenhall Postdoctoral Research Fellow in Spokane, Washington. She is part of the Geology, Minerals, Energy, and Geophysics Science Center. She received her B.S. in geology from Western Washington University. During her M.S. at Indiana University Bloomington, she investigated conduit-style Ni-Cu deposits. She received a Ph.D. in earth and climate sciences from Duke University, studying the genesis of layered mafic intrusions. Erin’s current research into carbonatite formation utilizes high spatial resolution analytical techniques to unravel the magmatic and hydrothermal processes that produce rare earth element deposits.