Silicate and iron phosphate melt immiscibility promotes REE enrichment
Affiliations | Corresponding Author | Cite as | Funding information- Share this article
-
Article views:237Cumulative count of HTML views and PDF downloads.
- Download Citation
- Rights & Permissions
top
Abstract
Figures and Tables
Figure 1 (a) Previous immiscibility experiments doped with trace elements, overlain by our new experiments (triangles, squares, and pentagons). Binodal curve and previous dry and hydrous experimental miscibility ranges are from Kamenetsky et al. (2013) and Hou et al. (2018). The width of the miscibility gap is affected by many variables (e.g., T, aH2O, and others), and this projection captures the combined effect. (b) The layered set-up of representative experiments. (c) Reflected light image of experiment 55F, with analysis spots for Si and FeP liquids marked. Abbreviations: nIM, 0.5 ≤ DLFe–LSiSiO2 ; mIM, | Figure 2 Backscattered images of mineral phases and melt textures from representative experiments. Silicate liquids contain μm-sized Fe-rich droplets at the highest temperatures and pressures (0.8 GPa, 1150 °C; i). FeP liquids contain (or mix with) minerals and silicate liquid droplets, but such contaminants decrease with IM degree, temperature, and pressure (i to c). FeP liquids crystallise substantially between 900 and 800 °C. Scale bar is 20 μm. Abbreviations: liq, liquid; Mt, magnetite; Hem, hematite; F-Ap, fluorapatite; crys, crystal. | Figure 3 (a, b) CaO–P2O5 and SiO2–FeOt plot of two liquids and FeP crystals; (c) major element partitioning of representative experiments, A = TiO2 + FeOt + MgO + CaO + P2O5 (wt. %); (d) REE partitioning between immiscible melts of 11 runs (Table 1); and (e) trace element partitioning between immiscible Si- and FeP-rich melts of representative experiments, with the inset (f) showing immiscibility widths. Starting felsic silicate and FeP mix were plotted. *Partition coefficients might be closer to 1 due to impurities in FeP liquids for nIM and mIM runs (Fig. 2), see text for details. | Table 1 Summary of experimental conditions and products. |
Figure 1 | Figure 2 | Figure 3 | Table 1 |
top
Introduction
Rare earth elements (REEs; lanthanides + Y) are required for modern society to achieve a green energy transition. Iron oxide-apatite (IOA) deposits are globally distributed and formed from the Paleoproterozoic to the Pleistocene (Reich et al., 2022
Reich, M., Simon, A.C., Barra, F., Palma, G., Hou, T., Bilenker, L.D. (2022) Formation of iron oxide–apatite deposits. Nature Reviews Earth & Environment 3, 758–775. https://doi.org/10.1038/s43017-022-00335-3
; Yan and Liu, 2022Yan, S., Liu, W. (2022) Rare earth elements in the iron-oxide apatite (IOA) deposit: insights from apatite. International Geology Review 64, 3230–3247. https://doi.org/10.1080/00206814.2022.2028198
). In IOA deposits, ubiquitous monazite was found in the ∼1.88 Ga Kiruna deposit (Pålsson et al., 2014Pålsson, B.I., Martinsson, O., Wanhainen, C., Fredriksson, A. (2014) Unlocking rare earth elements from European apatite‐iron ores. Proceedings of the 1st European Rare Earth Resources Conference, 4–7 September 2014, Milos, Greece, 211–220. https://eurare.org/docs/eres2014/fifthSession/BertilPalsson.pdf
; Yan et al., 2023bYan, S., Wan, B., Andersson, U.B. (2023b) Hydrothermal circulation at 1.8 Ga in the Kiruna area, northern Sweden, as revealed by apatite geochemical systematics. Precambrian Research 395, 107151. https://doi.org/10.1016/j.precamres.2023.107151
), and a significant REE resource (>1 Mt rare earth metals) at Kiruna was recently reported (LKAB, 2023LKAB (2023) Europe’s largest deposit of rare earth metals is located in the Kiruna area. Press release, 12 January 2023, Luossavaara Kiirunavaara AB, Luleå, Sweden. https://mb.cision.com/Main/11419/3696865/1777621.pdf
). In addition, iron oxide-phosphate (FeP) tephra containing monazite at the ∼2 Ma El Laco deposit (Mungall et al., 2018Mungall, J.E., Long, K., Brenan, J.M., Smythe, D., Naslund, H.R. (2018) Immiscible shoshonitic and Fe-P-oxide melts preserved in unconsolidated tephra at El Laco volcano, Chile. Geology 46, 255–258. https://doi.org/10.1130/g39707.1
), REE-rich tailings in ∼1.0 Ga deposits in the Adirondacks of New York (Taylor et al., 2019Taylor, R.D., Shah, A.K., Walsh, G.J., Taylor, C.D. (2019) Geochemistry and geophysics of iron oxide-apatite deposits and associated waste piles with implications for potential rare earth element resources from ore and historical mine waste in the eastern Adirondack highlands, New York, USA. Economic Geology 114, 1569–1598. https://doi.org/10.5382/econgeo.4689
), and a REE-rich breccia pipe containing ∼12 wt. % REO at the ∼1.4 Ga Pea Ridge deposit (Aleinikoff et al., 2016Aleinikoff, J.N., Selby, D., Slack, J.F., Day, W.C., Pillers, R.M., et al. (2016) U-Pb, Re-Os, and Ar/Ar geochronology of rare earth element (REE)-rich breccia pipes and associated host rocks from the Mesoproterozoic Pea Ridge Fe-REE-Au deposit, St. Francois Mountains, Missouri. Economic Geology 111, 1883–1914. https://doi.org/10.2113/econgeo.111.8.1883
) further demonstrate that IOA deposits typically contain substantial REE potential (Yan and Liu, 2022Yan, S., Liu, W. (2022) Rare earth elements in the iron-oxide apatite (IOA) deposit: insights from apatite. International Geology Review 64, 3230–3247. https://doi.org/10.1080/00206814.2022.2028198
).IOA deposits are related to magmatic rocks (e.g., Troll et al., 2019
Troll, V.R., Weis, F.A., Jonsson, E., Andersson, U.B., Majidi, S.A., et al. (2019) Global Fe-O isotope correlation reveals magmatic origin of Kiruna-type apatite-iron-oxide ores. Nature Communications 10, 1712. https://doi.org/10.1038/s41467-019-09244-4
), yet exactly how they form is unclear. Previous studies suggest the involvement of iron-rich molten salts or hydrosaline liquids (e.g., Bain et al., 2020Bain, W.M., Steele-MacInnis, M., Li, K., Li, L., Mazdab, F.K., Marsh, E.E. (2020) A fundamental role of carbonate–sulfate melts in the formation of iron oxide–apatite deposits. Nature Geoscience 13, 751–757. https://doi.org/10.1038/s41561-020-0635-9
; Zeng et al., 2024Zeng, L.-P., Zhao, X.-F., Spandler, C., Mavrogenes, J.A., Mernagh, T.P., et al. (2024) The role of iron-rich hydrosaline liquids in the formation of Kiruna-type iron oxide–apatite deposits. Science Advances 10, eadk2174. https://doi.org/10.1126/sciadv.adk2174
), magnetite-bubble flotation (e.g., Knipping et al., 2015Knipping, J.L., Bilenker, L.D., Simon, A.C., Reich, M., Barra, F., et al. (2015) Giant Kiruna-type deposits form by efficient flotation of magmatic magnetite suspensions. Geology 43, 591–594. https://doi.org/10.1130/g36650.1
), hydrothermal metasomatic replacement of lava flows (e.g., Sillitoe and Burrows, 2002Sillitoe, R.H., Burrows, D.R. (2002) New field evidence bearing on the origin of the El Laco magnetite deposit, northern Chile. Economic Geology 97, 1101–1109. https://doi.org/10.2113/gsecongeo.97.5.1101
), with or without magmatic-hydrothermal fluids (e.g., Rojas et al., 2018Rojas, P.A., Barra, F., Deditius, A., Reich, M., Simon, A., et al. (2018) New contributions to the understanding of Kiruna-type iron oxide-apatite deposits revealed by magnetite ore and gangue mineral geochemistry at the El Romeral deposit, Chile. Ore Geology Reviews 93, 413–435. https://doi.org/10.1016/j.oregeorev.2018.01.003
). In addition, the immiscibility between iron oxide phosphate liquids and silicate magma (FeP–Si) is a commonly invoked process for the formation of IOA deposits (Naslund, 1983Naslund, H.R. (1983) The effect of oxygen fugacity on liquid immiscibility in iron-bearing silicate melts. American Journal of Science 283, 1034–1059. https://doi.org/10.2475/ajs.283.10.1034
; Tornos et al., 2016Tornos, F., Velasco, F., Hanchar, J.M. (2016) Iron-rich melts, magmatic magnetite, and superheated hydrothermal systems: The El Laco deposit, Chile. Geology 44, 427–430. https://doi.org/10.1130/g37705.1
, 2024Tornos, F., Hanchar, J.M., Steele-MacInnis, M., Crespo, E., Kamenetsky, V.S., Casquet, C. (2024) Formation of magnetite-(apatite) systems by crystallizing ultrabasic iron-rich melts and slag separation. Mineralium Deposita 59, 189–225. https://doi.org/10.1007/s00126-023-01203-w
; Hou et al., 2018Hou, T., Charlier, B., Holtz, F., Veksler, I., Zhang, Z., et al. (2018) Immiscible hydrous Fe–Ca–P melt and the origin of iron oxide-apatite ore deposits. Nature Communications 9, 1415. https://doi.org/10.1038/s41467-018-03761-4
; Keller et al., 2022Keller, T., Tornos, F., Hanchar, J.M., Pietruszka, D.K., Soldati, A., et al. (2022) Genetic model of the El Laco magnetite-apatite deposits by extrusion of iron-rich melt. Nature Communications 13, 6114. https://doi.org/10.1038/s41467-022-33302-z
). However, the exact processes that control the ore-forming processes are hotly debated.For FeP–Si liquid pairs, experimental studies showed a narrow miscibility gap in dry melts (i.e. the compositions of the silicate and FeP melts are closer; Fig. 1a), which widen with lower temperature, higher oxygen fugacity (fO2), aH2O, and higher F– or P2O5 contents (e.g., Kamenetsky et al., 2013
Kamenetsky, V.S., Charlier, B., Zhitova, L., Sharygin, V., Davidson, P., Feig, S. (2013) Magma chamber–scale liquid immiscibility in the Siberian Traps represented by melt pools in native iron. Geology 41, 1091–1094. https://doi.org/10.1130/g34638.1
; Hou et al., 2018Hou, T., Charlier, B., Holtz, F., Veksler, I., Zhang, Z., et al. (2018) Immiscible hydrous Fe–Ca–P melt and the origin of iron oxide-apatite ore deposits. Nature Communications 9, 1415. https://doi.org/10.1038/s41467-018-03761-4
). For example, crustal inputs of S6+, F−, H2O, Fe and phosphate (e.g., via sulfate-evaporites or ironstones) are suggested to widen the Fe–Si immiscibility width in IOA deposits (Lledo et al., 2020Lledo, H.L., Naslund, H.R., Jenkins, D.M. (2020) Experiments on phosphate–silicate liquid immiscibility with potential links to iron oxide apatite and nelsonite deposits. Contributions to Mineralogy and Petrology 175, 111. https://doi.org/10.1007/s00410-020-01751-8
; Pietruszka et al., 2023Pietruszka, D.K., Hanchar, J.M., Tornos, F., Wirth, R., Graham, N.A., et al. (2023) Magmatic immiscibility and the origin of magnetite-(apatite) iron deposits. Nature Communications 14, 8424. https://doi.org/10.1038/s41467-023-43655-8
; Tornos et al., 2024Tornos, F., Hanchar, J.M., Steele-MacInnis, M., Crespo, E., Kamenetsky, V.S., Casquet, C. (2024) Formation of magnetite-(apatite) systems by crystallizing ultrabasic iron-rich melts and slag separation. Mineralium Deposita 59, 189–225. https://doi.org/10.1007/s00126-023-01203-w
). Preliminary work (e.g., Schmidt et al., 2006Schmidt, M.W., Connolly, J.A.D., Günther, D., Bogaerts, M. (2006) Element partitioning: the role of melt structure and composition. Science 312, 1646–1650. https://doi.org/10.1126/science.1126690
; Veksler et al., 2006Veksler, I.V., Dorfman, A.M., Danyushevsky, L.V., Jakobsen, J.K., Dingwell, D.B. (2006) Immiscible silicate liquid partition coefficients: implications for crystal-melt element partitioning and basalt petrogenesis. Contributions to Mineralogy and Petrology 152, 685–702. https://doi.org/10.1007/s00410-006-0127-y
; Lester et al., 2013Lester, G.W., Kyser, T.K., Clark, A.H., Layton-Matthews, D. (2013) Trace element partitioning between immiscible silicate melts with H2O, P, S, F, and Cl. Chemical Geology 357, 178–185. https://doi.org/10.1016/j.chemgeo.2013.08.021
) indicates that immiscible FeP melts incorporate more REEs than coexisting silicate melts (Fig. 1a). Reliable partition coefficients are only known for narrow immiscibility (nIM, Fig. 1a) conditions. Lester et al. (2013)Lester, G.W., Kyser, T.K., Clark, A.H., Layton-Matthews, D. (2013) Trace element partitioning between immiscible silicate melts with H2O, P, S, F, and Cl. Chemical Geology 357, 178–185. https://doi.org/10.1016/j.chemgeo.2013.08.021
obtained DLFe–LSiREE of ∼3–40 at DLFe–LSiSiO2 of ∼0.14–0.49 for moderate miscibility gaps (mIM). However, in the Lester et al. (2013)Lester, G.W., Kyser, T.K., Clark, A.H., Layton-Matthews, D. (2013) Trace element partitioning between immiscible silicate melts with H2O, P, S, F, and Cl. Chemical Geology 357, 178–185. https://doi.org/10.1016/j.chemgeo.2013.08.021
phosphate-bearing experiments, lower DLFe–LSiSiO2 values did not always correspond to higher DLFe–LSiREE values (Fig. 1a). For wide miscibility gaps (wIM), Lledo (2005)Lledo, H.L. (2005) Experimental studies on the origin of iron deposits; and mineralization of Sierra La Bandera, Chile. Ph.D. thesis, State University of New York at Binghamton.
derived DLFe–LSiEr of 13, but a full set of REE partition coefficients for wide miscibility gaps is still undetermined. The main reasons are (1) few Fe–Si immiscibility experiments contain REEs (Yan and Liu, 2022Yan, S., Liu, W. (2022) Rare earth elements in the iron-oxide apatite (IOA) deposit: insights from apatite. International Geology Review 64, 3230–3247. https://doi.org/10.1080/00206814.2022.2028198
), and (2) experimentally generated immiscible droplets are typically too small (unless a centrifuge setup is used), complicating trace element analysis using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) (Schmidt et al., 2006Schmidt, M.W., Connolly, J.A.D., Günther, D., Bogaerts, M. (2006) Element partitioning: the role of melt structure and composition. Science 312, 1646–1650. https://doi.org/10.1126/science.1126690
; Veksler et al., 2006Veksler, I.V., Dorfman, A.M., Danyushevsky, L.V., Jakobsen, J.K., Dingwell, D.B. (2006) Immiscible silicate liquid partition coefficients: implications for crystal-melt element partitioning and basalt petrogenesis. Contributions to Mineralogy and Petrology 152, 685–702. https://doi.org/10.1007/s00410-006-0127-y
).In this study, we address the origins and REE enrichment mechanisms of IOA deposits via a novel piston cylinder layered experimental design (Fig. 1b, c) followed by in situ analyses (Fig. 1c).
top
Methods
We conducted 17 FeP–Si immiscibility experiments over 800–1150 °C for 2–5 days (Fig. 1, Table 1), using a piston cylinder apparatus at the Research School of Earth Sciences, Australian National University (RSES, ANU). Starting materials included felsic silicate mix, FeP, and FeP–F compositions (Table S-1). Water was added in the form of solid hydroxides. These compositions were loaded into noble metal capsules (either Pt or AuPd) together with commercially available hydroxyapatite. Preliminary experiments using homogeneous starting materials resulted in tiny (mostly <5 μm) immiscible liquids. Therefore, subsequent experiments employed layered starting materials to obtain larger immiscible melts for easier geochemical analysis. Natural IOA deposits often contain coexisting hematite and magnetite (Tornos et al., 2024
Tornos, F., Hanchar, J.M., Steele-MacInnis, M., Crespo, E., Kamenetsky, V.S., Casquet, C. (2024) Formation of magnetite-(apatite) systems by crystallizing ultrabasic iron-rich melts and slag separation. Mineralium Deposita 59, 189–225. https://doi.org/10.1007/s00126-023-01203-w
; Xu et al., 2024Xu, X., Bain, W.M., Tornos, F., Hanchar, J.M., Lamadrid, H.M., et al. (2024) Magnetite-apatite ores record widespread involvement of molten salts. Geology 52, 417–422. https://doi.org/10.1130/g51887.1
). Therefore, we aimed to buffer oxygen fugacity at the magnetite–hematite (MH) oxygen buffer by adding layers of Fe3+2O3 and Fe0 powders, such that some hematite is reduced to magnetite in situ. Major and trace elements of immiscible liquids (three to eight points each) were measured using an electron probe microanalyser (EPMA) and laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) at the Institute of Geology and Geophysics, Chinese Academy of Sciences (IGGCAS). Analytical spot sizes were typically 20 μm for EPMA and 32 μm for LA-ICP-MS.Table 1 Summary of experimental conditions and products.
No. | Starting layers’ mass (mg) FeP, Si, Hem, Fe, OH-Ap | P GPa | T °C | Duration days | Resultant Phases | Imm. width | Ce of Si liq (max/min) | Ce of FeP liq (max/min) | Equilibrated? |
58F^ | 4.4(F^), 4.3, 2.7, 1.5, 4.1 | 0.4 | 800’ | 5 | Mt, F-Ap, whit, Si liq, FeP liq, FeP crystal | wIM | n.a. | n.a. | n.a. |
57F | 4.4(F), 4.3, 2.5, 1.7, 4.2 | 0.8 | 800’ | 5 | Mt, F-Ap, Si liq, FeP liq, FeP crystal | wIM | n.a. | n.a. | n.a. |
22 | 5.9, 6.0, 3.0, 1.0, 4.0 | 0.4 | 900 | 5 | Mt, Si liq, FeP liq | nIM | 1.1 ± 0.1# | 1.18 ± 0.04 | Yes |
53F | 4.0(F), 4.0, 2.7, 1.3, 4.0 | 0.4 | 900’ | 4 | Mt, Hem, F-Ap, q, Si liq, FeP liq, FeP crystal | wIM | 5.5 ± 0.3 | 1.01 ± 0.04 | No |
26 | 4.8. 5.1, 3.3, 1, 5.1 | 0.8 | 900’ | 4 | Hem, whit, Si liq, FeP liq | wIM | 1.1 ± 0.1 | 1.1 ± 0.1 | Yes |
52F | 4.4(F), 4.1, 3.2, 0.9, 4.1 | 0.8 | 900’ | 4 | Mt, F-Ap, Si liq, FeP liq | wIM | 1.3 ± 0.3 | 1.02 ± 0.08 | Yes |
20 | 4.9, 5.1, 3.7, 1.4, 5 | 0.4 | 1000’ | 3 | Mt, Hem, OH-Ap, q, Si liq, FeP liq | wIM | n.a. | n.a. | n.a. |
54F | 4.3(F), 4.3, 3.3, 1.0, 4.1 | 0.4 | 1000’ | 3 | Hem, F-Ap, Si liq, FeP liq | wIM | 4.1 ± 0.2 | 1.04 ± 0.02 | No |
59F | 4,2(F), 4.1, 2.6, 1.4, 4.1 | 0.4 | 1000’ | 3 | Mt, Hem, F-Ap, q, Si liq, FeP liq | wIM | 2.3 ± 0.2 | 1.03 ± 0.04 | No |
55F | 4.6(F), 4.6, 3.5, 1.0, 4.5 | 0.8 | 1000’ | 3 | Mt, Hem, F-Ap, Si liq, FeP liq | wIM | 1.1 ± 0.1 | 1.1 ± 0.1 | Yes |
56 | 4.2, 4.2, 2.9, 1.2, 4.1 | 0.8 | 1000’ | 3 | Mt, Whit, OH-Ap, Si liq, FeP liq | wIM | 1.9 ± 0.18 | 1.05 ± 0.03 | Yes |
47F | 4.2(F), 4, 3.2, 1.0, 4.0 | 0.4 | 1100’ | 2 | Mt, F-Ap, Si liq, FeP liq | mIM | 1.5 ± 0.1 | 1.08 ± 0.05 | Yes |
24 | 4.0, 4.0, 3.0, 1.1, 4.0 | 0.8 | 1100 | 4 | Mt, OH-Ap, Si liq, FeP liq | mIM | 1.03 ± 0.05 | 1.04 ± 0.08 | Yes |
50F | 3.6 (F), 3.2, 2.4, 0.7, 3.2 | 0.8 | 1100’ | 2 | Mt, F-Ap, Si liq, FeP liq | mIM | 1.1 ± 0.1 | 1.03 ± 0.03 | Yes |
32 | 4.4, 5.5, 1.2, 3.5(Mt), 4.9 | 0.4 | 1150 | 3 | Mt, Hem, whit, Si liq, FeP liq | mIM | 1.07 ± 0.08 | 1.2 ± 0.1 | Yes |
33 | 4.2, 5.3, 0.8, 4(Mt), 4.7 | 0.8 | 1150 | 3 | Hem, OH-Ap, Si liq, FeP liq | mIM | 1.2 ± 0.1 | 1.04 ± 0.03 | Yes |
39 | 3.7, 5.6, 3.1, 1.5, 4.8 | 0.8 | 1150 | 3 | Mt, whit, Si liq, FeP liq | mIM | 1.05 ± 0.05 | 1.02 ± 0.03 | Yes |
Fe-oxide phases as an indicator for fO2 are marked in bold.
^ ‘F’ suffix indicates fluoride-bearing experiments by using the FeP–F composition (e.g., 58F, 52F, 50F).
’ Two-step experiments: first heated to 1200 at 150 °C/min and held 20 min for melt homogenisation, then cooled to the target temperature at 40 °C/min (2.5–10 min) and held for 2–5 days.
# Cemax/min shown as max/min value ± 2 s.e., with 2 s.e. derived from measurement uncertainty.
Abbreviations: Mt, magnetite; OH-Ap, hydroxyapatite; F-Ap, fluorapatite; Hem, hematite; liq, liquid; whit, whitlockite; q, quartz; wIM, wide immiscibility (DLFe–LSiSiO2 ≤ 0.12); nIM, narrow immiscibility (0.5 ≤ DLFe–LSiSiO2 ); mIM, moderate immiscibility (0.12 < DLFe–LSiSiO2 < 0.5); n.a., not applicable.
Full details of the starting compositions, experimental procedures, geochemical analytical methods, resultant phases, their compositions, and calculated partition coefficients are available in the Supplementary Information.
top
Assessment of Equilibrium
FeP and silicate liquids were observed in all experiments (Fig. 2). Immiscible liquid droplets were typically well separated (Figs. 1c, 2). Solid crystals were likewise well defined and sized from several to dozens of μm (Figs. 2, S-4). Liquid compositions are shown in Figures 1 and 3 and in the Supplementary Information. Typically, equilibrium between melts is confirmed by reversals, time-series comparisons, or by spatially consistent phase compositions within a single experimental capsule. Previous REE-bearing Fe-Si melt immiscibility experiments containing homogenised starting materials reached equilibrium in less than 1 h at ∼1200 °C and 0.2 GPa (Lester et al., 2013
Lester, G.W., Kyser, T.K., Clark, A.H., Layton-Matthews, D. (2013) Trace element partitioning between immiscible silicate melts with H2O, P, S, F, and Cl. Chemical Geology 357, 178–185. https://doi.org/10.1016/j.chemgeo.2013.08.021
). Here, we use spatially consistent compositions as an equilibrium indicator, because manually loaded miniscule powder layers are nearly impossible to consistently reproduce for a constant-composition reversal or time-series experiment. The ratio between the maximum and minimum contents of elements within liquid phase droplets, measured in different places (Figs. 1c, 2b, c, S-6, S-7) within the capsule, is used to show compositional consistency. For example, Table 1 shows this ratio for Ce. The Cemax/min ratio across all FeP liquids ranges from 1.01 to 1.2 and is very close to unity (1.08 ± 0.07, 1 s.d.) considering the measurement uncertainty of 0.01–0.1 (2 s.e.), demonstrating their homogeneous composition. For silicate liquids, the Cemax/min ratios are slightly more variable, presumably because of slower diffusion in the polymerised melt. Nevertheless, excluding the three most variable experiments (53F, 54F, and 59F), the Cemax/min ratio across all silicate liquids is 1.22 ± 0.26 (1 s.d.), again showing homogeneous compositions considering the measurement uncertainty of 0.01–0.3 (2 s.e.). Examples for additional consistent major and trace elements other than Ce are available in the Supplementary Information. The overall consistency of major and trace elements (Table 1, Figs. S-6, S-7), predictable temperature dependence of DLFeP–LSi values for wIM and mIM runs (Fig. S-8), and deviation of final liquid elemental contents from starting compositions (Fig. 3) indicate adequate chemical exchange and equilibrium between the two liquids, despite the layered experimental design. According to this criterion, immiscible liquids in nIM (22), mIM (24, 33, 39, 32, 50F, 47F) and most of the wIM (52F, 26, 55F, 56) runs reached or were very close (47F, 56) to equilibrium (Table 1). The immiscible liquids of experiments 53F, 54F and 59F showed complex semi-equilibrated conditions using this criterion (Figs. S-9, S-10). Three experiments (20, 57F, 58F) did not contain liquid phases with sufficient size for multiple analyses and were excluded from D calculations. However, their textures were informative.In Fe–Si immiscibility experiments at 0.1 GPa and ∼1000 °C by Hou et al. (2018)
Hou, T., Charlier, B., Holtz, F., Veksler, I., Zhang, Z., et al. (2018) Immiscible hydrous Fe–Ca–P melt and the origin of iron oxide-apatite ore deposits. Nature Communications 9, 1415. https://doi.org/10.1038/s41467-018-03761-4
—which excluded REEs and contained homogenised starting materials—immiscible FeP liquids contained μm-scale unidentified droplets (their Figs. 1a, d). We encountered similar textures. Rarely, small inclusions (either crystals or liquid droplets) interfered with EPMA and LA-ICP-MS analyses. For nIM run 22, μm-sized silicate droplets were included within FeP liquids (Fig. 2d). For the mIM runs, FeP liquids contained μm-sized minerals, silicate droplets, and bubbles (Fig. 2g–j). Similarly, for wIM runs, FeP liquids contained μm-sized silicate droplets and bubbles (Fig. 2a–c, e, f). This cross-contamination resulted in some partition coefficients being closer to unity, making the real DLFe–LSiREE values even higher than the already high values we are reporting (Fig. 3d, e). DLFe–LSiREE values of our nIM and mIM runs (Fig. 3d, e) thereby represent lower bound values. For wIM experiments, their low-REE silicate droplets and bubbles were included in a high-REE FeP host, and there is an order of magnitude difference for most element concentrations between the two liquids. Therefore, we consider the contamination-driven effect of lowering DLFe–LSiREE to be negligible for wIM runs. In contrast to the common silicate inclusions within FeP liquids, the opposite was rare. Analysed regions of silicate liquids contained few, if any, FeP liquid droplet impurities (Fig. 2). When present, they were sufficiently large and rare (Fig. 2d, h) to facilitate straightforward analyses of silicate liquid without compromising measurements with spuriously high REE contents.Five experiments resulted in obvious coexisting magnetite and hematite, and hence their fO2 is buffered by MH, around 4.5 log units higher than the nickel–nickel oxide (NNO) buffer at our run conditions based on buffer curves determined using the online oxygen fugacity buffer calculator (https://fo2.rses.anu.edu.au/fo2app/, accessed 15 August 2024). Magnetite and hematite were distinguished based on contrasting optical properties in reflected light and Raman spectra (Figs. 1c, S-1). Other single-iron oxide experiments (Table 1) were likely to be close to the MH-buffer since they were prepared in a similar way, such that redox-sensitive elements were still expected to behave as if the runs were MH-buffered. The high oxidation state of the experiments prevented substantial Fe loss to the noble metal capsules, with typical values of ∼2 wt. % Fe, and up to a maximum of ∼8.4 wt. % Fe, alloyed with capsule materials at the capsule–experiment contact, whereas Fe was below detection limit further away (Fig. S-3).
top
Chemical Composition of Immiscible Liquids
Elemental systematics in our experiments are consistent with FeP–Si immiscibility during the formation of IOA deposits. We find that DLFeP–LSiF , DLFeP–LSiMgO , DLFeP–LSiCaO , and DLFeP–LSiP2O5 are all above 1 (Fig. 3, Table S-2), consistent with diopside (CaMgSi2O6), actinolite [Ca2(Mg,Fe)5(Si4O11)2(OH)2], and fluorapatite [Ca5(PO4)3F] as common minerals in IOA deposits. Trace element partition coefficients (Fig. 3) likewise support Th, Co, and V (likely as V5+) enrichment in IOA deposits (Reich et al., 2022
Reich, M., Simon, A.C., Barra, F., Palma, G., Hou, T., Bilenker, L.D. (2022) Formation of iron oxide–apatite deposits. Nature Reviews Earth & Environment 3, 758–775. https://doi.org/10.1038/s43017-022-00335-3
; Tornos et al., 2024Tornos, F., Hanchar, J.M., Steele-MacInnis, M., Crespo, E., Kamenetsky, V.S., Casquet, C. (2024) Formation of magnetite-(apatite) systems by crystallizing ultrabasic iron-rich melts and slag separation. Mineralium Deposita 59, 189–225. https://doi.org/10.1007/s00126-023-01203-w
). Despite DLFeP–LSiTiO2 > 1 (Fig. 3), DLFeP–LSiTi/Fe is below 1 (Fig. 3e), causing Ti/Fe of the FeP liquid to be lower than that of the silicate liquid. Thus, FeP–Si immiscibility dilutes Ti in the FeP liquid relative to Fe, consistent with the typically low Ti contents of IOA magnetite.top
FeP–Si Immiscibility Textures
Our experiments show several textural interactions between felsic melt and Fe–Ca–P-rich phases. The experimental temperatures of 800–1150 °C are within the typical homogenisation temperature ranges for melt inclusions in IOA deposits (700–1145 °C; Xie et al., 2019
Xie, Q., Zhang, Z., Hou, T., Cheng, Z., Campos, E., et al. (2019) New insights for the formation of kiruna-type iron deposits by immiscible hydrous Fe-P melt and high-temperature hydrothermal processes: evidence from El Laco deposit. Economic Geology 114, 35–46. https://doi.org/10.5382/econgeo.2019.4618
; Bain et al., 2020Bain, W.M., Steele-MacInnis, M., Li, K., Li, L., Mazdab, F.K., Marsh, E.E. (2020) A fundamental role of carbonate–sulfate melts in the formation of iron oxide–apatite deposits. Nature Geoscience 13, 751–757. https://doi.org/10.1038/s41561-020-0635-9
; Pietruszka et al., 2023Pietruszka, D.K., Hanchar, J.M., Tornos, F., Wirth, R., Graham, N.A., et al. (2023) Magmatic immiscibility and the origin of magnetite-(apatite) iron deposits. Nature Communications 14, 8424. https://doi.org/10.1038/s41467-023-43655-8
; Xu et al., 2024Xu, X., Bain, W.M., Tornos, F., Hanchar, J.M., Lamadrid, H.M., et al. (2024) Magnetite-apatite ores record widespread involvement of molten salts. Geology 52, 417–422. https://doi.org/10.1130/g51887.1
). We observed common silicate liquid droplets in FeP liquids (Fig. 2c, e–j). If such droplets were trapped during magnetite formation (Fig. 2c), then this potentially explains the Si-rich melt inclusions observed in magnetite from the El Laco massive ores (Tornos et al., 2024Tornos, F., Hanchar, J.M., Steele-MacInnis, M., Crespo, E., Kamenetsky, V.S., Casquet, C. (2024) Formation of magnetite-(apatite) systems by crystallizing ultrabasic iron-rich melts and slag separation. Mineralium Deposita 59, 189–225. https://doi.org/10.1007/s00126-023-01203-w
) and quartz inclusions in earliest-formed magnetite in the El Romel deposit (Rojas et al., 2018Rojas, P.A., Barra, F., Deditius, A., Reich, M., Simon, A., et al. (2018) New contributions to the understanding of Kiruna-type iron oxide-apatite deposits revealed by magnetite ore and gangue mineral geochemistry at the El Romeral deposit, Chile. Ore Geology Reviews 93, 413–435. https://doi.org/10.1016/j.oregeorev.2018.01.003
). Moreover, imperfect separation of silicate liquid droplets for FeP liquids (Fig. 2) indicates that FeP melts may carry silicate even under wide immiscibility conditions where SiO2 is nearly insoluble in FeP melts. This may lead to the formation of actinolite, pyroxene, or even quartz, consistent with their occurrence in IOAs (e.g., Nyström and Henríquez, 1994Nyström, J.O., Henríquez, F. (1994) Magmatic features of iron ores of the Kiruna type in Chile and Sweden; ore textures and magnetite geochemistry. Economic Geology 89, 820–839. https://doi.org/10.2113/gsecongeo.89.4.820
; Mungall et al., 2018Mungall, J.E., Long, K., Brenan, J.M., Smythe, D., Naslund, H.R. (2018) Immiscible shoshonitic and Fe-P-oxide melts preserved in unconsolidated tephra at El Laco volcano, Chile. Geology 46, 255–258. https://doi.org/10.1130/g39707.1
; Reich et al., 2022Reich, M., Simon, A.C., Barra, F., Palma, G., Hou, T., Bilenker, L.D. (2022) Formation of iron oxide–apatite deposits. Nature Reviews Earth & Environment 3, 758–775. https://doi.org/10.1038/s43017-022-00335-3
), and particularly in their REE-rich portions (e.g., “Per Geijer” type ore bodies, e.g., Henry, Rektorn and Nukutus, in the Kiruna deposit; Nyström and Henríquez, 1994Nyström, J.O., Henríquez, F. (1994) Magmatic features of iron ores of the Kiruna type in Chile and Sweden; ore textures and magnetite geochemistry. Economic Geology 89, 820–839. https://doi.org/10.2113/gsecongeo.89.4.820
; LKAB, 2023LKAB (2023) Europe’s largest deposit of rare earth metals is located in the Kiruna area. Press release, 12 January 2023, Luossavaara Kiirunavaara AB, Luleå, Sweden. https://mb.cision.com/Main/11419/3696865/1777621.pdf
), which are otherwise expected to be silicate-poor according to wIM equilibrium compositions. FeP liquid droplets surrounded by felsic liquid (Fig. 2d) resemble melt inclusions in feldspar from El Laco (Tornos et al., 2016Tornos, F., Velasco, F., Hanchar, J.M. (2016) Iron-rich melts, magmatic magnetite, and superheated hydrothermal systems: The El Laco deposit, Chile. Geology 44, 427–430. https://doi.org/10.1130/g37705.1
; Pietruszka et al., 2023Pietruszka, D.K., Hanchar, J.M., Tornos, F., Wirth, R., Graham, N.A., et al. (2023) Magmatic immiscibility and the origin of magnetite-(apatite) iron deposits. Nature Communications 14, 8424. https://doi.org/10.1038/s41467-023-43655-8
).Skeletal or dendritic magnetite (Fig. 2a, b), generated by rapid quenching of 800 °C runs, remarkably similar to magnetite textures observed in the El Laco and Kiruna IOA deposits (Nyström and Henríquez, 1994
Nyström, J.O., Henríquez, F. (1994) Magmatic features of iron ores of the Kiruna type in Chile and Sweden; ore textures and magnetite geochemistry. Economic Geology 89, 820–839. https://doi.org/10.2113/gsecongeo.89.4.820
), indicating that quenching (and, by implication, probable eruption) of FeP-rich melts occurred during IOA formation. Experiments 53F (0.4 GPa, 900 °C), 57F (0.8 GPa, 800 °C), and 58F (0.4 GPa, 800 °C) yielded FeP liquids and FeP crystals (Figs. 2a, b, 3a, b, S-4), similar to two unidentified FeP phases of El Laco (Xie et al., 2019Xie, Q., Zhang, Z., Hou, T., Cheng, Z., Campos, E., et al. (2019) New insights for the formation of kiruna-type iron deposits by immiscible hydrous Fe-P melt and high-temperature hydrothermal processes: evidence from El Laco deposit. Economic Geology 114, 35–46. https://doi.org/10.5382/econgeo.2019.4618
).These textural similarities (Fig. S-11) and temperature overlaps suggest our experimental scenarios (narrow to wide FeP–Si immiscibility) may accurately mimic natural processes. The experiments also contain bubbles and iron oxide-bubble pairs indicative of hydrothermal fluid saturation (e.g., Fig. 2c, e, i). Sulfate-evaporite assimilation is hypothesised to be conducive to FeP–Si immiscibility (Pietruszka et al., 2023
Pietruszka, D.K., Hanchar, J.M., Tornos, F., Wirth, R., Graham, N.A., et al. (2023) Magmatic immiscibility and the origin of magnetite-(apatite) iron deposits. Nature Communications 14, 8424. https://doi.org/10.1038/s41467-023-43655-8
; Tornos et al., 2024Tornos, F., Hanchar, J.M., Steele-MacInnis, M., Crespo, E., Kamenetsky, V.S., Casquet, C. (2024) Formation of magnetite-(apatite) systems by crystallizing ultrabasic iron-rich melts and slag separation. Mineralium Deposita 59, 189–225. https://doi.org/10.1007/s00126-023-01203-w
). Iron-bearing hydrosaline liquids, recently proposed to be important for IOA formation (Zeng et al., 2024Zeng, L.-P., Zhao, X.-F., Spandler, C., Mavrogenes, J.A., Mernagh, T.P., et al. (2024) The role of iron-rich hydrosaline liquids in the formation of Kiruna-type iron oxide–apatite deposits. Science Advances 10, eadk2174. https://doi.org/10.1126/sciadv.adk2174
), were also observed to coexist with immiscible FeP- and Si-rich liquids at El Laco deposit (Pietruszka et al., 2023Pietruszka, D.K., Hanchar, J.M., Tornos, F., Wirth, R., Graham, N.A., et al. (2023) Magmatic immiscibility and the origin of magnetite-(apatite) iron deposits. Nature Communications 14, 8424. https://doi.org/10.1038/s41467-023-43655-8
). Our suggested processes do not preclude—and are in fact compatible with—other proposed processes, such as Fe-rich magmatic hydrothermal fluids (Rojas et al., 2018Rojas, P.A., Barra, F., Deditius, A., Reich, M., Simon, A., et al. (2018) New contributions to the understanding of Kiruna-type iron oxide-apatite deposits revealed by magnetite ore and gangue mineral geochemistry at the El Romeral deposit, Chile. Ore Geology Reviews 93, 413–435. https://doi.org/10.1016/j.oregeorev.2018.01.003
), iron oxide-bubble pairs (Knipping et al., 2015Knipping, J.L., Bilenker, L.D., Simon, A.C., Reich, M., Barra, F., et al. (2015) Giant Kiruna-type deposits form by efficient flotation of magmatic magnetite suspensions. Geology 43, 591–594. https://doi.org/10.1130/g36650.1
), Fe-bearing hydrosaline liquids (Zeng et al., 2024Zeng, L.-P., Zhao, X.-F., Spandler, C., Mavrogenes, J.A., Mernagh, T.P., et al. (2024) The role of iron-rich hydrosaline liquids in the formation of Kiruna-type iron oxide–apatite deposits. Science Advances 10, eadk2174. https://doi.org/10.1126/sciadv.adk2174
), and possibly evaporite assimilation (Pietruszka et al., 2023Pietruszka, D.K., Hanchar, J.M., Tornos, F., Wirth, R., Graham, N.A., et al. (2023) Magmatic immiscibility and the origin of magnetite-(apatite) iron deposits. Nature Communications 14, 8424. https://doi.org/10.1038/s41467-023-43655-8
).top
REE Enrichment in IOAs and Fe-REE Associations
Figure 3c–e shows DLFeP–LSi values of different immiscibility widths, where the gap widens with evolution or cooling (Fig. 1). The general trend is that, with a wider miscibility gap, REE partition more strongly into FeP relative to silicate melts. Our results show that DLFeP–LSiLREE and DLFeP–LSiHREE can reach above 100 for wide immiscibility (wIM in Fig. 3d, e), so the previously reported DLFeP–LSiEr of 13 for wide immiscibility (Lledo, 2005
Lledo, H.L. (2005) Experimental studies on the origin of iron deposits; and mineralization of Sierra La Bandera, Chile. Ph.D. thesis, State University of New York at Binghamton.
) is unrealistically low. We find that DLFeP–LSiLREE values are higher than DLFeP–LSiHREE , regardless of immiscibility width (Fig. 3d), similar to the findings of Lester et al. (2013)Lester, G.W., Kyser, T.K., Clark, A.H., Layton-Matthews, D. (2013) Trace element partitioning between immiscible silicate melts with H2O, P, S, F, and Cl. Chemical Geology 357, 178–185. https://doi.org/10.1016/j.chemgeo.2013.08.021
. The ubiquitous enrichment of LREE relative to HREE in IOA deposits is consistent with this finding and indicates FeP–Si immiscibility is likely a dominant process in IOA deposit formation. Magnetite does not host significant REE; hence, REE in IOA deposits are mainly hosted in phosphates, such as apatite, monazite, and xenotime (Yan and Liu, 2022Yan, S., Liu, W. (2022) Rare earth elements in the iron-oxide apatite (IOA) deposit: insights from apatite. International Geology Review 64, 3230–3247. https://doi.org/10.1080/00206814.2022.2028198
). Generally, IOA deposits with large iron tonnages and highly evolved portions of IOA deposits (e.g., Pea Ridge, Kiruna, Carmen, and the Fresia deposits) should have higher phosphate contents, and thereby higher REE potential (Yan et al., 2023aYan, S., Wan, B., Andersson, U.B. (2023a) Apatite age and composition: A key to the geological history of the Malmberget Iron-Oxide-Apatite (IOA) deposit and the region. Journal of Geochemical Exploration 252, 107267. https://doi.org/10.1016/j.gexplo.2023.107267
).Other FeP-rich rocks understood to be related to magmatic Fe–Si immiscibility (e.g., nelsonites, Fe-Ti-V deposit such as Panzhihua) experienced a lower immiscibility degree (Lledo et al., 2020
Lledo, H.L., Naslund, H.R., Jenkins, D.M. (2020) Experiments on phosphate–silicate liquid immiscibility with potential links to iron oxide apatite and nelsonite deposits. Contributions to Mineralogy and Petrology 175, 111. https://doi.org/10.1007/s00410-020-01751-8
) and therefore REE in their Fe-rich parts were less enriched compared to IOA deposits.top
Acknowledgements
We appreciate David Clark for aiding piston cylinder experiments, Vivian for Raman analyses, and the support of Microscopy Australia at the CAM, ANU. We also thank Lixin Gu and Dr. Lihui Jia, Jiangyan Yuan, Shitou Wu for helping with geochemical analyses. The work was supported by NSFC 42325206, the Strategy Priority Research Program (Category B) of the CAS (XDB0710000), and ARC Linkage Project LP190100635. Yan appreciates Chinese Scholarship Council (202104910432) for funding the visit to ANU.
Editor: Raul O.C. Fonseca
top
References
Aleinikoff, J.N., Selby, D., Slack, J.F., Day, W.C., Pillers, R.M., et al. (2016) U-Pb, Re-Os, and Ar/Ar geochronology of rare earth element (REE)-rich breccia pipes and associated host rocks from the Mesoproterozoic Pea Ridge Fe-REE-Au deposit, St. Francois Mountains, Missouri. Economic Geology 111, 1883–1914. https://doi.org/10.2113/econgeo.111.8.1883
Show in context
In addition, iron oxide-phosphate (FeP) tephra containing monazite at the ∼2 Ma El Laco deposit (Mungall et al., 2018), REE-rich tailings in ∼1.0 Ga deposits in the Adirondacks of New York (Taylor et al., 2019), and a REE-rich breccia pipe containing ∼12 wt. % REO at the ∼1.4 Ga Pea Ridge deposit (Aleinikoff et al., 2016) further demonstrate that IOA deposits typically contain substantial REE potential (Yan and Liu, 2022).
View in article
Bain, W.M., Steele-MacInnis, M., Li, K., Li, L., Mazdab, F.K., Marsh, E.E. (2020) A fundamental role of carbonate–sulfate melts in the formation of iron oxide–apatite deposits. Nature Geoscience 13, 751–757. https://doi.org/10.1038/s41561-020-0635-9
Show in context
Previous studies suggest the involvement of iron-rich molten salts or hydrosaline liquids (e.g., Bain et al., 2020; Zeng et al., 2024), magnetite-bubble flotation (e.g., Knipping et al., 2015), hydrothermal metasomatic replacement of lava flows (e.g., Sillitoe and Burrows, 2002), with or without magmatic-hydrothermal fluids (e.g., Rojas et al., 2018).
View in article
The experimental temperatures of 800–1150 °C are within the typical homogenisation temperature ranges for melt inclusions in IOA deposits (700–1145 °C; Xie et al., 2019; Bain et al., 2020; Pietruszka et al., 2023; Xu et al., 2024).
View in article
Hou, T., Charlier, B., Holtz, F., Veksler, I., Zhang, Z., et al. (2018) Immiscible hydrous Fe–Ca–P melt and the origin of iron oxide-apatite ore deposits. Nature Communications 9, 1415. https://doi.org/10.1038/s41467-018-03761-4
Show in context
In addition, the immiscibility between iron oxide phosphate liquids and silicate magma (FeP–Si) is a commonly invoked process for the formation of IOA deposits (Naslund, 1983; Tornos et al., 2016, 2024; Hou et al., 2018; Keller et al., 2022).
View in article
For FeP–Si liquid pairs, experimental studies showed a narrow miscibility gap in dry melts (i.e. the compositions of the silicate and FeP melts are closer; Fig. 1a), which widen with lower temperature, higher oxygen fugacity (fO2), aH2O, and higher F– or P2O5 contents (e.g., Kamenetsky et al., 2013; Hou et al., 2018).
View in article
Binodal curve and previous dry and hydrous experimental miscibility ranges are from Kamenetsky et al. (2013) and Hou et al. (2018).
View in article
In Fe–Si immiscibility experiments at 0.1 GPa and ∼1000 °C by Hou et al. (2018)—which excluded REEs and contained homogenised starting materials—immiscible FeP liquids contained μm-scale unidentified droplets (their Figs. 1a, d).
View in article
Kamenetsky, V.S., Charlier, B., Zhitova, L., Sharygin, V., Davidson, P., Feig, S. (2013) Magma chamber–scale liquid immiscibility in the Siberian Traps represented by melt pools in native iron. Geology 41, 1091–1094. https://doi.org/10.1130/g34638.1
Show in context
For FeP–Si liquid pairs, experimental studies showed a narrow miscibility gap in dry melts (i.e. the compositions of the silicate and FeP melts are closer; Fig. 1a), which widen with lower temperature, higher oxygen fugacity (fO2), aH2O, and higher F– or P2O5 contents (e.g., Kamenetsky et al., 2013; Hou et al., 2018).
View in article
Binodal curve and previous dry and hydrous experimental miscibility ranges are from Kamenetsky et al. (2013) and Hou et al. (2018).
View in article
Keller, T., Tornos, F., Hanchar, J.M., Pietruszka, D.K., Soldati, A., et al. (2022) Genetic model of the El Laco magnetite-apatite deposits by extrusion of iron-rich melt. Nature Communications 13, 6114. https://doi.org/10.1038/s41467-022-33302-z
Show in context
In addition, the immiscibility between iron oxide phosphate liquids and silicate magma (FeP–Si) is a commonly invoked process for the formation of IOA deposits (Naslund, 1983; Tornos et al., 2016, 2024; Hou et al., 2018; Keller et al., 2022).
View in article
Knipping, J.L., Bilenker, L.D., Simon, A.C., Reich, M., Barra, F., et al. (2015) Giant Kiruna-type deposits form by efficient flotation of magmatic magnetite suspensions. Geology 43, 591–594. https://doi.org/10.1130/g36650.1
Show in context
Previous studies suggest the involvement of iron-rich molten salts or hydrosaline liquids (e.g., Bain et al., 2020; Zeng et al., 2024), magnetite-bubble flotation (e.g., Knipping et al., 2015), hydrothermal metasomatic replacement of lava flows (e.g., Sillitoe and Burrows, 2002), with or without magmatic-hydrothermal fluids (e.g., Rojas et al., 2018).
View in article
Our suggested processes do not preclude—and are in fact compatible with—other proposed processes, such as Fe-rich magmatic hydrothermal fluids (Rojas et al., 2018), iron oxide-bubble pairs (Knipping et al., 2015), Fe-bearing hydrosaline liquids (Zeng et al., 2024), and possibly evaporite assimilation (Pietruszka et al., 2023).
View in article
Lester, G.W., Kyser, T.K., Clark, A.H., Layton-Matthews, D. (2013) Trace element partitioning between immiscible silicate melts with H2O, P, S, F, and Cl. Chemical Geology 357, 178–185. https://doi.org/10.1016/j.chemgeo.2013.08.021
Show in context
Preliminary work (e.g., Schmidt et al., 2006; Veksler et al., 2006; Lester et al., 2013) indicates that immiscible FeP melts incorporate more REEs than coexisting silicate melts (Fig. 1a).
View in article
Lester et al. (2013) obtained DLFe–LSiREE of ∼3–40 at DLFe–LSiSiO2 of ∼0.14–0.49 for moderate miscibility gaps (mIM).
View in article
However, in the Lester et al. (2013) phosphate-bearing experiments, lower DLFe–LSiSiO2 values did not always correspond to higher DLFe–LSiREE values (Fig. 1a).
View in article
Previous REE-bearing Fe-Si melt immiscibility experiments containing homogenised starting materials reached equilibrium in less than 1 h at ∼1200 °C and 0.2 GPa (Lester et al., 2013).
View in article
We find that DLFeP–LSiLREE values are higher than DLFeP–LSiHREE , regardless of immiscibility width (Fig. 3d), similar to the findings of Lester et al. (2013).
View in article
LKAB (2023) Europe’s largest deposit of rare earth metals is located in the Kiruna area. Press release, 12 January 2023, Luossavaara Kiirunavaara AB, Luleå, Sweden. https://mb.cision.com/Main/11419/3696865/1777621.pdf
Show in context
In IOA deposits, ubiquitous monazite was found in the ∼1.88 Ga Kiruna deposit (Pålsson et al., 2014; Yan et al., 2023b), and a significant REE resource (>1 Mt rare earth metals) at Kiruna was recently reported (LKAB, 2023).
View in article
This may lead to the formation of actinolite, pyroxene, or even quartz, consistent with their occurrence in IOAs (e.g., Nyström and Henríquez, 1994; Mungall et al., 2018; Reich et al., 2022), and particularly in their REE-rich portions (e.g., “Per Geijer” type ore bodies, e.g., Henry, Rektorn and Nukutus, in the Kiruna deposit; Nyström and Henríquez, 1994; LKAB, 2023), which are otherwise expected to be silicate-poor according to wIM equilibrium compositions.
View in article
Lledo, H.L. (2005) Experimental studies on the origin of iron deposits; and mineralization of Sierra La Bandera, Chile. Ph.D. thesis, State University of New York at Binghamton.
Show in context
For wide miscibility gaps (wIM), Lledo (2005) derived DLFe–LSiEr of 13, but a full set of REE partition coefficients for wide miscibility gaps is still undetermined.
View in article
Our results show that DLFeP–LSiLREE and DLFeP–LSiHREE can reach above 100 for wide immiscibility (wIM in Fig. 3d, e), so the previously reported DLFeP–LSiEr of 13 for wide immiscibility (Lledo, 2005) is unrealistically low.
View in article
Lledo, H.L., Naslund, H.R., Jenkins, D.M. (2020) Experiments on phosphate–silicate liquid immiscibility with potential links to iron oxide apatite and nelsonite deposits. Contributions to Mineralogy and Petrology 175, 111. https://doi.org/10.1007/s00410-020-01751-8
Show in context
For example, crustal inputs of S6+, F−, H2O, Fe and phosphate (e.g., via sulfate-evaporites or ironstones) are suggested to widen the Fe–Si immiscibility width in IOA deposits (Lledo et al., 2020; Pietruszka et al., 2023; Tornos et al., 2024).
View in article
Other FeP-rich rocks understood to be related to magmatic Fe–Si immiscibility (e.g., nelsonites, Fe-Ti-V deposit such as Panzhihua) experienced a lower immiscibility degree (Lledo et al., 2020) and therefore REE in their Fe-rich parts were less enriched compared to IOA deposits.
View in article
Mungall, J.E., Long, K., Brenan, J.M., Smythe, D., Naslund, H.R. (2018) Immiscible shoshonitic and Fe-P-oxide melts preserved in unconsolidated tephra at El Laco volcano, Chile. Geology 46, 255–258. https://doi.org/10.1130/g39707.1
Show in context
In addition, iron oxide-phosphate (FeP) tephra containing monazite at the ∼2 Ma El Laco deposit (Mungall et al., 2018), REE-rich tailings in ∼1.0 Ga deposits in the Adirondacks of New York (Taylor et al., 2019), and a REE-rich breccia pipe containing ∼12 wt. % REO at the ∼1.4 Ga Pea Ridge deposit (Aleinikoff et al., 2016) further demonstrate that IOA deposits typically contain substantial REE potential (Yan and Liu, 2022).
View in article
This may lead to the formation of actinolite, pyroxene, or even quartz, consistent with their occurrence in IOAs (e.g., Nyström and Henríquez, 1994; Mungall et al., 2018; Reich et al., 2022), and particularly in their REE-rich portions (e.g., “Per Geijer” type ore bodies, e.g., Henry, Rektorn and Nukutus, in the Kiruna deposit; Nyström and Henríquez, 1994; LKAB, 2023), which are otherwise expected to be silicate-poor according to wIM equilibrium compositions.
View in article
Naslund, H.R. (1983) The effect of oxygen fugacity on liquid immiscibility in iron-bearing silicate melts. American Journal of Science 283, 1034–1059. https://doi.org/10.2475/ajs.283.10.1034
Show in context
In addition, the immiscibility between iron oxide phosphate liquids and silicate magma (FeP–Si) is a commonly invoked process for the formation of IOA deposits (Naslund, 1983; Tornos et al., 2016, 2024; Hou et al., 2018; Keller et al., 2022).
View in article
Nyström, J.O., Henríquez, F. (1994) Magmatic features of iron ores of the Kiruna type in Chile and Sweden; ore textures and magnetite geochemistry. Economic Geology 89, 820–839. https://doi.org/10.2113/gsecongeo.89.4.820
Show in context
This may lead to the formation of actinolite, pyroxene, or even quartz, consistent with their occurrence in IOAs (e.g., Nyström and Henríquez, 1994; Mungall et al., 2018; Reich et al., 2022), and particularly in their REE-rich portions (e.g., “Per Geijer” type ore bodies, e.g., Henry, Rektorn and Nukutus, in the Kiruna deposit; Nyström and Henríquez, 1994; LKAB, 2023), which are otherwise expected to be silicate-poor according to wIM equilibrium compositions.
View in article
Skeletal or dendritic magnetite (Fig. 2a, b), generated by rapid quenching of 800 °C runs, remarkably similar to magnetite textures observed in the El Laco and Kiruna IOA deposits (Nyström and Henríquez, 1994), indicating that quenching (and, by implication, probable eruption) of FeP-rich melts occurred during IOA formation.
View in article
Pålsson, B.I., Martinsson, O., Wanhainen, C., Fredriksson, A. (2014) Unlocking rare earth elements from European apatite‐iron ores. Proceedings of the 1st European Rare Earth Resources Conference, 4–7 September 2014, Milos, Greece, 211–220. https://eurare.org/docs/eres2014/fifthSession/BertilPalsson.pdf
Show in context
In IOA deposits, ubiquitous monazite was found in the ∼1.88 Ga Kiruna deposit (Pålsson et al., 2014; Yan et al., 2023b), and a significant REE resource (>1 Mt rare earth metals) at Kiruna was recently reported (LKAB, 2023).
View in article
Pietruszka, D.K., Hanchar, J.M., Tornos, F., Wirth, R., Graham, N.A., et al. (2023) Magmatic immiscibility and the origin of magnetite-(apatite) iron deposits. Nature Communications 14, 8424. https://doi.org/10.1038/s41467-023-43655-8
Show in context
For example, crustal inputs of S6+, F−, H2O, Fe and phosphate (e.g., via sulfate-evaporites or ironstones) are suggested to widen the Fe–Si immiscibility width in IOA deposits (Lledo et al., 2020; Pietruszka et al., 2023; Tornos et al., 2024).
View in article
The experimental temperatures of 800–1150 °C are within the typical homogenisation temperature ranges for melt inclusions in IOA deposits (700–1145 °C; Xie et al., 2019; Bain et al., 2020; Pietruszka et al., 2023; Xu et al., 2024).
View in article
FeP liquid droplets surrounded by felsic liquid (Fig. 2d) resemble melt inclusions in feldspar from El Laco (Tornos et al., 2016; Pietruszka et al., 2023).
View in article
Sulfate-evaporite assimilation is hypothesised to be conducive to FeP–Si immiscibility (Pietruszka et al., 2023; Tornos et al., 2024).
View in article
Iron-bearing hydrosaline liquids, recently proposed to be important for IOA formation (Zeng et al., 2024), were also observed to coexist with immiscible FeP- and Si-rich liquids at El Laco deposit (Pietruszka et al., 2023).
View in article
Our suggested processes do not preclude—and are in fact compatible with—other proposed processes, such as Fe-rich magmatic hydrothermal fluids (Rojas et al., 2018), iron oxide-bubble pairs (Knipping et al., 2015), Fe-bearing hydrosaline liquids (Zeng et al., 2024), and possibly evaporite assimilation (Pietruszka et al., 2023).
View in article
Reich, M., Simon, A.C., Barra, F., Palma, G., Hou, T., Bilenker, L.D. (2022) Formation of iron oxide–apatite deposits. Nature Reviews Earth & Environment 3, 758–775. https://doi.org/10.1038/s43017-022-00335-3
Show in context
Rare earth elements (REEs; lanthanides + Y) are required for modern society to achieve a green energy transition. Iron oxide-apatite (IOA) deposits are globally distributed and formed from the Paleoproterozoic to the Pleistocene (Reich et al., 2022; Yan and Liu, 2022).
View in article
Trace element partition coefficients (Fig. 3) likewise support Th, Co, and V (likely as V5+) enrichment in IOA deposits (Reich et al., 2022; Tornos et al., 2024).
View in article
This may lead to the formation of actinolite, pyroxene, or even quartz, consistent with their occurrence in IOAs (e.g., Nyström and Henríquez, 1994; Mungall et al., 2018; Reich et al., 2022), and particularly in their REE-rich portions (e.g., “Per Geijer” type ore bodies, e.g., Henry, Rektorn and Nukutus, in the Kiruna deposit; Nyström and Henríquez, 1994; LKAB, 2023), which are otherwise expected to be silicate-poor according to wIM equilibrium compositions.
View in article
Rojas, P.A., Barra, F., Deditius, A., Reich, M., Simon, A., et al. (2018) New contributions to the understanding of Kiruna-type iron oxide-apatite deposits revealed by magnetite ore and gangue mineral geochemistry at the El Romeral deposit, Chile. Ore Geology Reviews 93, 413–435. https://doi.org/10.1016/j.oregeorev.2018.01.003
Show in context
Previous studies suggest the involvement of iron-rich molten salts or hydrosaline liquids (e.g., Bain et al., 2020; Zeng et al., 2024), magnetite-bubble flotation (e.g., Knipping et al., 2015), hydrothermal metasomatic replacement of lava flows (e.g., Sillitoe and Burrows, 2002), with or without magmatic-hydrothermal fluids (e.g., Rojas et al., 2018).
View in article
If such droplets were trapped during magnetite formation (Fig. 2c), then this potentially explains the Si-rich melt inclusions observed in magnetite from the El Laco massive ores (Tornos et al., 2024) and quartz inclusions in earliest-formed magnetite in the El Romel deposit (Rojas et al., 2018).
View in article
Our suggested processes do not preclude—and are in fact compatible with—other proposed processes, such as Fe-rich magmatic hydrothermal fluids (Rojas et al., 2018), iron oxide-bubble pairs (Knipping et al., 2015), Fe-bearing hydrosaline liquids (Zeng et al., 2024), and possibly evaporite assimilation (Pietruszka et al., 2023).
View in article
Schmidt, M.W., Connolly, J.A.D., Günther, D., Bogaerts, M. (2006) Element partitioning: the role of melt structure and composition. Science 312, 1646–1650. https://doi.org/10.1126/science.1126690
Show in context
Preliminary work (e.g., Schmidt et al., 2006; Veksler et al., 2006; Lester et al., 2013) indicates that immiscible FeP melts incorporate more REEs than coexisting silicate melts (Fig. 1a).
View in article
The main reasons are (1) few Fe–Si immiscibility experiments contain REEs (Yan and Liu, 2022), and (2) experimentally generated immiscible droplets are typically too small (unless a centrifuge setup is used), complicating trace element analysis using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) (Schmidt et al., 2006; Veksler et al., 2006).
View in article
Sillitoe, R.H., Burrows, D.R. (2002) New field evidence bearing on the origin of the El Laco magnetite deposit, northern Chile. Economic Geology 97, 1101–1109. https://doi.org/10.2113/gsecongeo.97.5.1101
Show in context
Previous studies suggest the involvement of iron-rich molten salts or hydrosaline liquids (e.g., Bain et al., 2020; Zeng et al., 2024), magnetite-bubble flotation (e.g., Knipping et al., 2015), hydrothermal metasomatic replacement of lava flows (e.g., Sillitoe and Burrows, 2002), with or without magmatic-hydrothermal fluids (e.g., Rojas et al., 2018).
View in article
Taylor, R.D., Shah, A.K., Walsh, G.J., Taylor, C.D. (2019) Geochemistry and geophysics of iron oxide-apatite deposits and associated waste piles with implications for potential rare earth element resources from ore and historical mine waste in the eastern Adirondack highlands, New York, USA. Economic Geology 114, 1569–1598. https://doi.org/10.5382/econgeo.4689
Show in context
In addition, iron oxide-phosphate (FeP) tephra containing monazite at the ∼2 Ma El Laco deposit (Mungall et al., 2018), REE-rich tailings in ∼1.0 Ga deposits in the Adirondacks of New York (Taylor et al., 2019), and a REE-rich breccia pipe containing ∼12 wt. % REO at the ∼1.4 Ga Pea Ridge deposit (Aleinikoff et al., 2016) further demonstrate that IOA deposits typically contain substantial REE potential (Yan and Liu, 2022).
View in article
Tornos, F., Velasco, F., Hanchar, J.M. (2016) Iron-rich melts, magmatic magnetite, and superheated hydrothermal systems: The El Laco deposit, Chile. Geology 44, 427–430. https://doi.org/10.1130/g37705.1
Show in context
In addition, the immiscibility between iron oxide phosphate liquids and silicate magma (FeP–Si) is a commonly invoked process for the formation of IOA deposits (Naslund, 1983; Tornos et al., 2016, 2024; Hou et al., 2018; Keller et al., 2022).
View in article
Preliminary work (e.g., Schmidt et al., 2006; Veksler et al., 2006; Lester et al., 2013) indicates that immiscible FeP melts incorporate more REEs than coexisting silicate melts (Fig. 1a).
View in article
FeP liquid droplets surrounded by felsic liquid (Fig. 2d) resemble melt inclusions in feldspar from El Laco (Tornos et al., 2016; Pietruszka et al., 2023).
View in article
Tornos, F., Hanchar, J.M., Steele-MacInnis, M., Crespo, E., Kamenetsky, V.S., Casquet, C. (2024) Formation of magnetite-(apatite) systems by crystallizing ultrabasic iron-rich melts and slag separation. Mineralium Deposita 59, 189–225. https://doi.org/10.1007/s00126-023-01203-w
Show in context
In addition, the immiscibility between iron oxide phosphate liquids and silicate magma (FeP–Si) is a commonly invoked process for the formation of IOA deposits (Naslund, 1983; Tornos et al., 2016, 2024; Hou et al., 2018; Keller et al., 2022).
View in article
For example, crustal inputs of S6+, F−, H2O, Fe and phosphate (e.g., via sulfate-evaporites or ironstones) are suggested to widen the Fe–Si immiscibility width in IOA deposits (Lledo et al., 2020; Pietruszka et al., 2023; Tornos et al., 2024).
View in article
Natural IOA deposits often contain coexisting hematite and magnetite (Tornos et al., 2024; Xu et al., 2024).
View in article
Trace element partition coefficients (Fig. 3) likewise support Th, Co, and V (likely as V5+) enrichment in IOA deposits (Reich et al., 2022; Tornos et al., 2024).
View in article
If such droplets were trapped during magnetite formation (Fig. 2c), then this potentially explains the Si-rich melt inclusions observed in magnetite from the El Laco massive ores (Tornos et al., 2024) and quartz inclusions in earliest-formed magnetite in the El Romel deposit (Rojas et al., 2018).
View in article
Sulfate-evaporite assimilation is hypothesised to be conducive to FeP–Si immiscibility (Pietruszka et al., 2023; Tornos et al., 2024).
View in article
Troll, V.R., Weis, F.A., Jonsson, E., Andersson, U.B., Majidi, S.A., et al. (2019) Global Fe-O isotope correlation reveals magmatic origin of Kiruna-type apatite-iron-oxide ores. Nature Communications 10, 1712. https://doi.org/10.1038/s41467-019-09244-4
Show in context
IOA deposits are related to magmatic rocks (e.g., Troll et al., 2019), yet exactly how they form is unclear.
View in article
Veksler, I.V., Dorfman, A.M., Danyushevsky, L.V., Jakobsen, J.K., Dingwell, D.B. (2006) Immiscible silicate liquid partition coefficients: implications for crystal-melt element partitioning and basalt petrogenesis. Contributions to Mineralogy and Petrology 152, 685–702. https://doi.org/10.1007/s00410-006-0127-y
Show in context
The main reasons are (1) few Fe–Si immiscibility experiments contain REEs (Yan and Liu, 2022), and (2) experimentally generated immiscible droplets are typically too small (unless a centrifuge setup is used), complicating trace element analysis using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) (Schmidt et al., 2006; Veksler et al., 2006).
View in article
Xie, Q., Zhang, Z., Hou, T., Cheng, Z., Campos, E., et al. (2019) New insights for the formation of kiruna-type iron deposits by immiscible hydrous Fe-P melt and high-temperature hydrothermal processes: evidence from El Laco deposit. Economic Geology 114, 35–46. https://doi.org/10.5382/econgeo.2019.4618
Show in context
The experimental temperatures of 800–1150 °C are within the typical homogenisation temperature ranges for melt inclusions in IOA deposits (700–1145 °C; Xie et al., 2019; Bain et al., 2020; Pietruszka et al., 2023; Xu et al., 2024).
View in article
Experiments 53F (0.4 GPa, 900 °C), 57F (0.8 GPa, 800 °C), and 58F (0.4 GPa, 800 °C) yielded FeP liquids and FeP crystals (Figs. 2a, b, 3a, b, S-4), similar to two unidentified FeP phases of El Laco (Xie et al., 2019).
View in article
Xu, X., Bain, W.M., Tornos, F., Hanchar, J.M., Lamadrid, H.M., et al. (2024) Magnetite-apatite ores record widespread involvement of molten salts. Geology 52, 417–422. https://doi.org/10.1130/g51887.1
Show in context
Natural IOA deposits often contain coexisting hematite and magnetite (Tornos et al., 2024; Xu et al., 2024).
View in article
The experimental temperatures of 800–1150 °C are within the typical homogenisation temperature ranges for melt inclusions in IOA deposits (700–1145 °C; Xie et al., 2019; Bain et al., 2020; Pietruszka et al., 2023; Xu et al., 2024).
View in article
Yan, S., Liu, W. (2022) Rare earth elements in the iron-oxide apatite (IOA) deposit: insights from apatite. International Geology Review 64, 3230–3247. https://doi.org/10.1080/00206814.2022.2028198
Show in context
Rare earth elements (REEs; lanthanides + Y) are required for modern society to achieve a green energy transition. Iron oxide-apatite (IOA) deposits are globally distributed and formed from the Paleoproterozoic to the Pleistocene (Reich et al., 2022; Yan and Liu, 2022).
View in article
In addition, iron oxide-phosphate (FeP) tephra containing monazite at the ∼2 Ma El Laco deposit (Mungall et al., 2018), REE-rich tailings in ∼1.0 Ga deposits in the Adirondacks of New York (Taylor et al., 2019), and a REE-rich breccia pipe containing ∼12 wt. % REO at the ∼1.4 Ga Pea Ridge deposit (Aleinikoff et al., 2016) further demonstrate that IOA deposits typically contain substantial REE potential (Yan and Liu, 2022).
View in article
The main reasons are (1) few Fe–Si immiscibility experiments contain REEs (Yan and Liu, 2022), and (2) experimentally generated immiscible droplets are typically too small (unless a centrifuge setup is used), complicating trace element analysis using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) (Schmidt et al., 2006; Veksler et al., 2006).
View in article
Magnetite does not host significant REE; hence, REE in IOA deposits are mainly hosted in phosphates, such as apatite, monazite, and xenotime (Yan and Liu, 2022).
View in article
Yan, S., Wan, B., Andersson, U.B. (2023a) Apatite age and composition: A key to the geological history of the Malmberget Iron-Oxide-Apatite (IOA) deposit and the region. Journal of Geochemical Exploration 252, 107267. https://doi.org/10.1016/j.gexplo.2023.107267
Show in context
Generally, IOA deposits with large iron tonnages and highly evolved portions of IOA deposits (e.g., Pea Ridge, Kiruna, Carmen, and the Fresia deposits) should have higher phosphate contents, and thereby higher REE potential (Yan et al., 2023a).
View in article
Yan, S., Wan, B., Andersson, U.B. (2023b) Hydrothermal circulation at 1.8 Ga in the Kiruna area, northern Sweden, as revealed by apatite geochemical systematics. Precambrian Research 395, 107151. https://doi.org/10.1016/j.precamres.2023.107151
Show in context
In IOA deposits, ubiquitous monazite was found in the ∼1.88 Ga Kiruna deposit (Pålsson et al., 2014; Yan et al., 2023b), and a significant REE resource (>1 Mt rare earth metals) at Kiruna was recently reported (LKAB, 2023).
View in article
Zeng, L.-P., Zhao, X.-F., Spandler, C., Mavrogenes, J.A., Mernagh, T.P., et al. (2024) The role of iron-rich hydrosaline liquids in the formation of Kiruna-type iron oxide–apatite deposits. Science Advances 10, eadk2174. https://doi.org/10.1126/sciadv.adk2174
Show in context
Previous studies suggest the involvement of iron-rich molten salts or hydrosaline liquids (e.g., Bain et al., 2020; Zeng et al., 2024), magnetite-bubble flotation (e.g., Knipping et al., 2015), hydrothermal metasomatic replacement of lava flows (e.g., Sillitoe and Burrows, 2002), with or without magmatic-hydrothermal fluids (e.g., Rojas et al., 2018).
View in article
Iron-bearing hydrosaline liquids, recently proposed to be important for IOA formation (Zeng et al., 2024), were also observed to coexist with immiscible FeP- and Si-rich liquids at El Laco deposit (Pietruszka et al., 2023).
View in article
Our suggested processes do not preclude—and are in fact compatible with—other proposed processes, such as Fe-rich magmatic hydrothermal fluids (Rojas et al., 2018), iron oxide-bubble pairs (Knipping et al., 2015), Fe-bearing hydrosaline liquids (Zeng et al., 2024), and possibly evaporite assimilation (Pietruszka et al., 2023).
View in article
top
Supplementary Information
The Supplementary Information includes:
- 1. Rationale for Starting Compositions
- 2. Piston Cylinder Experiments
• 2.1 Preparation of Starting Materials
• 2.2 High Temperature and Pressure Experiments - 3. Microscope, EDS and Raman Analyses
- 4. Major Element Analyses
- 5. Trace Element Analyses
- 6. Geochemical Characteristics
- 7. Equilibrium of Experiments
- Tables S-1 and S-2
- Figures S-1 to S-10
- Datasets S-1 to S-12
- Supplementary Information References
Download the Supplementary Information (PDF)
Download Dataset S-1 (.xlsx)
Download Dataset S-2 (.xlsx)
Download Datasets S-3 to S-12 (.xlsx files in.zip directory)