Subduction zones are the primary loci for chemical exchange among the atmosphere, ocean, lithosphere, and mantle, whereby subduction of hydrated oceanic crust and its sedimentary veneer leads to flux melting of the overlying mantle wedge and generation of arc plutonism and volcanism (Richards 2011). One important consequence of this complex sequence of mass transfer events is the formation of magmatic-hydrothermal ore deposits of the porphyry-type enriched in Cu ± Au ± Mo (Sillitoe 2010;Richards 2011;Audétat and Simon 2012). Petrological and geochemical proxies suggest that subduction-like processes may have operated since 3.0 Ga or earlier (Dhuime et al. 2012), but their extent and duration were likely limited due to the low rheological strength of subducting slabs in a hotter, melt-rich Archean mantle (Sizova et al. 2010;Moyen and van Hunen 2012;Laurent et al. 2014). Oxidized S-rich magmas have recently been identified to correlate with subduction events during the later stages of Archean cratonization in the southeastern Superior Province (<2.7 Ga; Meng et al. 2022). Prior to the later stages of Archean cratonization (>2.7 Ga of the southeastern Superior Province), short-term episodic proto-subduction (∼5 Myr) is suggested to have operated based on evidence from whole-rock geochemistry and numerical modeling (Wyman and Kerrich 2009;Moyen and van Hunen 2012). However, it remains unclear as to whether oxidized S-rich sodic magmas and porphyry Cu deposits could have formed before ∼2.7 Ga in the Archean. In this study, we investigated the redox state of sodic magmas formed between ∼2750-2695 Ma in the Neoarchean Abitibi Greenstone Belt of the southeastern Superior Province (Fig. 1a) (Mole et al. 2021;Meng et al. 2021aMeng et al. , 2022)). This belt comprises arc-like, calcalkaline-dominated volcanic sequences intercalated with dominantly tholeiitic ± komatiitic assemblages that formed as a part of the assembly of volcano-plutonic terranes from ∼2750-2695 Ma prior to interpreted subduction-collision since ∼2695 Ma (Figs. 1a and 1b) (Thurston et al. 2008;Beakhouse 2011;Meng et al. 2021a;Mole et al. 2021). We measured the sulfur (S) concentration and relative abundances of S 6+ , S 4+ , and S 2-in apatite [Ca5(PO4)3(F,OH,Cl)], of which the S is incorporated during apatite crystallization from silicate melts (Kim et al. 2022) that are faithfully preserved due to being armored in zircon. Previous studies reported low-S 6+ /ΣS ratios in magmatic apatite grains from the lunar basalts and ∼2.35 Ga Na-rich tonalite-trondhjemite-granite (TTG), but the S concentrations in these apatite grains are generally low (Brounce et al. 2019;Moreira et al. 2023). In contrast, apatites with high-S concentrations but negligible S 6+ and dominant S 4+ and S 2-have only been reported where hydrothermal replacement processes are evident (Sadove et al. 2019). The results presented here reveal, instead, the rare presence of pristine S-rich apatite with relatively low S 6+ /ΣS ratios. Our results suggest, therefore, that the rarity of porphyry-type Cu deposits in well-preserved Archean volcano-plutonic terranes might be attributed to the limited extent and duration of the proto-subduction process with low productivity of oxidized S-rich arc magmas. Figure 1.
The tectonomagmatic setting and geological map of the Abitibi Greenstone Belt in the southeastern Superior Province, along with the spatial distribution of samples with measured zircon δ
The Neoarchean Abitibi Greenstone Belt comprises ∼2.75-2.67 Ga east-trending volcanoplutonic assemblages that preserve the following magmatic record: (1) pre-tectonic sodic volcanic rocks and the tonalite-trondhjemite-diorite thought to have formed from "plumearc" interaction or episodic short-term subduction; (2) early syn-tectonic tonalitegranodiorite and minor volcanic rocks derived from melting of subducted slabs; (3) syn-to late-tectonic sanukitoid suites (sensu lato); and (4) late-tectonic alkalic rocks (i.e., syenite suites) derived from a metasomatized mantle (Percival 2007;Thurston et al. 2008;Beakhouse 2011;Dubé and Mercier-Langevin 2020;Meng et al. 2021a;Mole et al. 2021). The syn-to late-tectonic sodic and potassic rocks that are interpreted to form in subduction-collision settings contain S-rich apatite grains with high-S 6+ /ΣS ratios, reflecting their oxidized S-rich feature of the magmas (Fig. 1a) (Meng et al. 2022). In comparison, the pre-tectonic (i.e., pre-deformation, foliated) sodic volcano-plutonic rocks are of significant interest because they are much more abundant, with some interpreted to have formed in a short-term episodic subduction setting (Wyman and Kerrich 2009; Moyen and van Hunen 2012). Twenty-eight zircon separate samples of pre-tectonic (i.e., >2695 Ma) volcanic rocks of dacitic to rhyolitic composition were previously collected from the Abitibi Greenstone Belt (Online Materials 1 Table S1). These samples yielded narrow ranges of zircon εHf(t) and δ 18 O values, based on our previous studies (i.e.,+3.1 to +4.9 and 4.2 to 6.2‰ respectively; Fig. 1c; Online Materials 1 Table S1) (Mole et al. 2021). We defined "arc-like" features as the enrichment of fluid-mobile elements (e.g., large-ion lithophile, U, and light rare earth elements [LREE]) and relative depletion of high field strength elements (e.g., Nb, Y, heavy rare earth elements [HREE]) (Richards 2011), which importantly can be recorded in zircon geochemistry (Grimes et al. 2015), as discussed below. These inferred arc-like rocks are distinguished by their high-Ui/Nb ratios of ≥40 from the mantle-derived magmas with Ui/Nb ratios of <40 using the previously proposed discrimination criteria (Grimes et al. 2015;Drabon et al. 2021), in which Ui represents the initial U concentrations in the zircon grains at the time of crystallization (Online Materials 1 Table S1). Apatite inclusions in the mounted zircon grains used in Mole et al. (2021) were identified using a scanning electron microscope equipped with an energy-dispersive spectrometer (SEM-EDS) at Laurentian University (Sudbury, Canada). The S concentration and peak integrated areas of S 6+ , S 4+ , and S 2-in primary apatite inclusions were measured using an electron microprobe analyzer (EMPA) in the GeoLabs of the Ontario Geological Survey (Sudbury, Canada) and synchrotron-based micro-X-ray absorption near-edge structure spectroscopy (μ-XANES) at S K-edge at Advanced Photon Sources (Illinois, U.S.A.). The archived samples for representative volcanic rocks were reused for lithogeochemical analyses at Australian Laboratory Services (ALS, Vancouver, Canada). Details of the methods for EMPA, μ-XANES, and whole-rock geochemistry are provided in Online Materials 1 .
The monophase, equant to sub-equant apatite inclusions hosted in zircon grains with oscillatory or sector zoning were divided into two groups based on their S concentration (Table 1; Fig. 2). Group-I apatite grains yielded S concentrations of 0.14 ± 0.03 wt% and S 6+ /ΣS ratios of 0.56 ± 0.17 (1σ; Tables 1 and 2; Fig. 2), while Group-I whole rock samples have La/Yb ratios of 49 ± 15 (1σ), zircon 1000•(Eu/Eu*)/Yb ratios of 26 ± 7, and mantle-like zircon δ 18 O isotopic values of 5.8 ± 0.1‰ (1σ; Table 1; Fig. 2). In contrast, Group-II apatite grains yielded S contents of <0.01 wt%, which incidentally, were too low for XANES to measure the abundances of S 6+ , S 4+ , and S 2-. For these Group-II samples, the whole-rock samples hosting Group-II apatite yielded variable zircon δ 18 O values of 4.0-6.5 ‰, as well as La/Yb ratios and zircon 1000•(Eu/Eu*)/Yb ratios that skew to low values of <30 (except for one sample with a ratio of 49) and 11 ± 8, respectively (1σ; Table 1; Online Materials 1 Tables S1 and S2; Fig. 2) 1 σ La/ Yb rat io Samples with high-S apatite and arc-like zircon Ui/Nb ratios of ≥40 93H NB-208 Volca nic tuff 273 0 ± 1 .0 0. 1 7 1 0 . 0 5. 9 0 . 3 27. 9 9 . 4 0 . 8 3 0 . 6 0 N/ A Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io 6 8 96JA A-041 Felsic volca nic 271 8 ± 2 .3 0. 1 2 3 0 . 0 5 3 5. 8 0 . 1 30. 6 6 . 1 0 . 8 4 0 . 3 8 96JA A-011 Felsic volca nic 269 4 ± 4 .5 0. 1 4 5 0 . 0 8 4 5. 9 0 . 1 29. 9 9 . 5 2 . 2 8 0 . 2 1 06-BRB-246 Quart zphyric volca nic 272 2 ± 1 .7 0. 1 0 9 0 . 0 0 0 5. 7 0 . 0 16. 3 4 . 4 2 . 2 2 0 . 5 3 N/ A Aver age 0. 1 4 5. 8 26. 2 1 . 5 Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io 1σ 0. 0 3 0. 1 6.7 0 . 8 15 Samples with low-S apatite and arc-like zircon Ui/Nb ratios of ≥40 92H NB-0083 Felsic Lapilli tuff 270 1 ± 2 .3 0. 0 0 4 B DL 0 . 0 0 2 5. 7 0 . 2 12. 9 4 . 9 0 . 1 9 0 . 5 0 N/ A 94H NB0 059 Rhyoli te lapilli tuff 273 5 ± 6 .0 0. 0 0 8 BD L 0 . 0 0 2 4. 9 0 . 2 16. 0 3 . 8 0 . 0 4 0 . 5 4 N/ A 96JA A-086 Felsic Lapilli tuff 271 0 ± 1 .9 0. 0 1 1 BD L 0 . 0 0 1 5. 5 0 . 2 4.7 3 . 0 -0 . 6 5 0 . 5 6 10 Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io 98JA A-0019 Felsic tuff brecc ia 271 7 ± 1 .7 0. 0 0 0 BD L N / A 5. 5 0 . 1 14. 0 4 . 3 0 . 0 8 0 . 4 3 C-82-8 Dacit e tuff 270 1 ± 2 .0 0. 0 1 4 0 . 0 0 2 5. 2 0 . 1 23. 5 4 . 6 0 . 8 7 0 . 3 7 93H NB-087A Felsic lapilli tuff 273 1 ± 2 .0 0. 0 3 9 0 . 0 3 3 5. 6 0 . 6 17. 6 5 . 7 0 . 0 9 0 . 3 4 N/ A C88-17 Rhyoli te 272 5 ± 1 .0 0. 0 0 4 BD L 0 . 0 0 2 5. 5 0 . 2 5.3 1 . 1 -0 . 3 7 0 . 1 8 Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io 93H NB-B Felsic tuff 272 4 ± 1 .0 0. 0 1 2 0 . 0 0 9 5. 7 0 . 2 23. 4 8 . 7 -0 . 1 8 0 . 3 0 N/ A SGN O -07 Felsic tuff 272 1 ± 0 .8 0. 0 0 0 BD L N / A 5. 7 0 . 1 16. 6 6 . 0 -0 . 2 2 0 . 1 5 18 94H NB-Rhyoli te 271 8 ± 2 .0 0. 0 0 0 BD L N / A 5. 1 0 . 3 6.2 7 . 2 -0 . 3 8 0 . 5 7 N/ A C-82-7 Daciti c flow 271 1 ± 3 .0 0. 0 0 8 BD L 0 . 0 0 0 5. 8 0 . 1 12. 9 0 . 8 0 . 3 4 1 . 1 2 N/ A Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io 94H NB-0115 Felds parquart z tuff 270 7 ± 1 .0 0. 0 0 6 BD L 0 . 0 0 6 5. 5 0 . 3 6.1 1 . 7 -0 . 0 3 0 . 3 3 N/ A 96TB 082 Dacit e brecc ia 270 5 ± 1 .5 0. 0 1 2 0 . 0 0 2 5. 7 0 . 1 9.8 2 . 9 -0 . 0 7 0 . 6 3 N/ A LAPL -146-2000 Rhyo dacit e 269 8 ± 0 .8 0. 0 0 4 BD L 0 . 0 0 1 4. 8 0 . 2 3.2 0 . 5 0 . 3 3 0 . 4 0 17 03A SP0 179. 1.1 Rhyoli te 269 7 ± 1 .3 0. 0 0 4 BD L 0 . 0 0 3 4. 7 0 . 2 4.1 2 . 8 -0 . 8 7 0 . 5 4 5 Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io 95H NB-0273 Felsic volca nic 269 5 ± 2 .0 0. 0 1 4 0 . 0 0 0 6. 2 0 . 2 24. 7 6 . 3 0 . 8 9 0 . 3 5 N/ A SGN O-Bou sque t 2 Rhyoli te 269 8 ± 1 .0 0. 0 0 0 BD L 0 . 0 0 0 4. 6 0 . 2 3.7 2 . 0 N / A N / A 26 SGN O99-10 Rhyoli te 270 3 ± 0 .9 0. 0 0 6 BD L 0 . 0 0 0 5. 2 0 . 2 1.3 0 . 4 N / A N / A 3 94H NB-267 Quart zphyric volca nic 271 5 ± 2 .0 0. 0 0 7 BD L 0 . 0 1 0 5. 7 0 . 2 N/ A N / A N / A N / A N/ A Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io C83-16 Daciti c tuff 272 9 ± 3 .0 0. 0 0 7 BD L 0 . 0 0 0 6. 1 0 . 2 N/ A N / A N / A N / A N/ A Aver age 0. 0 0 8 5. 4 11. 4 0 . 0 18 1σ 0. 0 0 9 0. 4 7.6 0 . 5 14 Samples with low-S apatite and low-zircon Ui/Nb ratios of <40 R-16 Rhyoli te tuff 270 6 ± 2 .0 0. 0 1 0 BD L 0 . 0 1 2 5. 7 0 . 2 2.4 70 0 . 7 5 0 1 . 1 9 0 . 5 6 Sam ple no. Rock type Zirc on U-Pb age a Apatite Zircon b W hol ero ck S ( w t % ) 1 σ δ 1 8 O ( ‰ ) 1 σ 10 00 0• (Eu /E u*) /Y b 1 σ Δ F M Q 1 σ La/ Yb rat io SGN O--02 Dacit e tuff 270 6 ± 3 .3 0. 0 0 3 BD L 0 . 0 0 3 5. 3 0 . 3 6.4 64 5 . 1 3 8 -0 . 4 4 0 . 4 1 SGN O -07 Rhyoli te 269 7 ± 0 .8 0. 0 0 0 BD L N / A 4. 2 0 . 1 2.1 68 0 . 5 4 5 0 . 9 1 0 . 4 2 SGN O -08 Rhyoli te View largeDownload slide The normalized μ-XANES spectra at S K-edge for the S-rich apatite grains as well as the analyzed apatite S concentration vs. (b) whole-rock La/Yb ratios, (c) zircon 10 000 • (Eu/Eu*)/Yb ratios (a proxy of magma hydration state, see Loucks and Fiorentini 2023), and (d) zircon δ 18 O values from the >2695 Ma volcanic rocks in the Abitibi Greenstone Belt (from Mole et al. 2021). Group-I and Group-II apatite grains and the host rocks are shown in red and blue, respectively. The mantle value for zircon O isotopes is 5.3 ± 0.3 wt% (1σ; Valley et al. 1998). The apatite data for intrusive rocks were reported by Meng et al. (2022). The S concentrations in primary apatite grains from syn-mineralization intrusive rocks for porphyry Cu deposits are compiled in Online Materials 1
Table S5. Note that the zircon δ 18 O values and 10 000•(Eu/Eu*)/Yb ratios are recalculated based on data sets reported by Mole et al. (2021). Error bars represent standard deviation. (Color online.) 2019), whereas the melt S concentrations are estimated using the model for the partitioning of S between apatite and melt using existing apatite/melt partition coefficient values as function of magmatic fO2 and temperature (Parat and Holtz 2004;Konecke et al. 2019;Meng et al. 2021b). The zircon-hosted apatite crystals from the volcanic rocks are assumed to crystallize at pressure over a range from that for the pre-tectonic TTG batholith (300-700 MPa and locally 100 MPa; Beakhouse et al. 2005Beakhouse et al. , 2011) ) to near-atmospheric pressure. The modeled range of temperature (892 ± 49 °C; 1σ, n = 1372) is estimated using the apatite saturation thermometer (Piccoli and Candela 2002). The standard deviations for the corrected magmatic fO2 in ΔFMQ involve the intra-sample standard deviation and those derived for the pressure and temperature estimation. For simplicity, the standard deviation of intra-sample apatite S 6+ /ΣS ratios is not considered in estimating the melt S concentration. See details of the relevant methods in Online Materials 1 .
The S 6+ /ΣS ratio in apatite varies systematically with the oxygen fugacity ( fO2 ) of the magmatic system as demonstrated experimentally by Konecke et al. (2019), whereas the S concentration in apatite is a function of magmatic fO2 , crystallization temperature, and S concentration of the melt (Parat and Holtz 2004;Konecke et al. 2019;Meng et al. 2021b). Magmatic fO2 , expressed here as ΔFMQ values, were estimated using the calculated S 6+ /ΣS ratios and the P-T-corrected S-in-apatite oxybarometer (Online Materials 1 ). The results yielded a relatively low average ΔFMQ value of 0.80 ± 0.16 to 1.26 ± 0.16 (1σ) for Group-I apatite (Table 2). A model for the partitioning of S between apatite and melt ( DSap/m ) using existing apatite/melt partition coefficient values as a function of magmatic fO2 and temperature (Meng et al. 2021b) was used to calculate a model S concentration of 0.11-0.34 wt% in the silicate melt at the time of crystallization of Group-I apatite (Table 2). Discussion
Historic studies used a mass-balance approach with EPMA-determined concentrations of S and other cations in apatite, as required for coupled substitutions, to hypothesize that S 6+ is the dominant S species in the apatite structure (Konecke et al. 2019 and references therein). However, recent studies using S-μ-XANES demonstrate that the multiple oxidation states of S, including S 6+ , S 4+ , S 0 , S 1-, and S 2-, are instead incorporated into apatite crystallized from hydrothermal fluids (Sadove et al. 2019), hosted in lunar basalts (Brounce et al. 2019), as well as those formed in both experimental and terrestrial silicate melts (Konecke et al. 2019;Kim et al. 2022;Meng et al. 2021aMeng et al. , 2021bMeng et al. , 2022)). Previous experimental and empirical studies for magmatic systems generally reported high-S 6+ /ΣS ratios in apatite with a strong positive correlation between their S 6+ /ΣS ratio and S contents, in agreement with experimental data for silicate melts that document higher concentrations of S 6+ than S 2-coupled with high fO2 (Konecke et al. 2019;Tassara et al. 2020;Meng et al. 2021aMeng et al. , 2021bMeng et al. , 2022;;Moreira et al. 2023). In contrast to the previous studies suggesting covariance of S 6+ /ΣS ratio and S concentrations in magmatic apatite, the S-μ-XANES data reported here are the first to our knowledge to reveal that S-rich igneous apatite with low-S 6+ /ΣS ratios of ∼0.5 crystallized under fO2 conditions of ∼ΔFMQ +1 where the S 6+ /ΣS ratio of the silicate melt is ∼0.1 (Kleinsasser et al. 2022(Kleinsasser et al. , 2024)). Compared to silicate melt, apatite preferentially incorporates sulfate (S 6+ ) rather than the reduced-intermediate redox state of S (e.g., S 2-, S 4+ , S 1+ ) (Konecke et al. 2019). The range of the model S concentration in the melts overlaps the maximum values of S content at sulfide saturation (SCSS) for basaltic to dacitic silicate melts (0.1-0.2 wt%) (Kleinsasser et al. 2022). Hence, the S 6+ /ΣS ratio of ∼0.5 and high-S concentration of apatite in Group-I samples can only be explained by the crystallization of apatite from sulfide-saturated silicate melts where the SCSS is at its maximum value. This inferred sulfide saturation at the time of apatite crystallization is preserved in the form of pyrrhotite (Fe0.93S; based on EDS data) inclusion exposed in zircon-hosted apatite grains in the Group-I sample (sample 93HNB-208; Online Materials 1 Fig. S1g).
The high-zircon Ui/Nb ratio of ≥40 is interpreted to reflect fluid-mobile element enrichment in the source region and/or relative depletion of immobile elements during the crystal fractionation (e.g., garnet, amphibole) (Grimes et al. 2015;Drabon et al. 2021). These samples can be grouped with respect to the apatite S concentrations.
The high whole-rock La/Yb and zircon (Eu/Eu*)/Yb ratios for Group-I samples reflect a deep source and hydrous state for the sodic magmas (Moyen 2009; Loucks and Fiorentini 2023), which may be formed from either partial melting of subducting slab or metasomatized mantle (Moyen and van Hunen 2012), remelting of thickened hydrated mafic crust (Bédard 2018; Mole et al. 2021), or plagioclase ± amphibole fractionation of parental tonalitic liquids from basalt melting (Laurent et al. 2020). The slightly elevated ΔFMQ value of +0.7-1.4 estimated using the S-in-apatite oxybarometer for Group-I samples is more consistent with the former two models because: (1) slab-derived fluids are capable of oxidizing the mantle in the Archean (Meng et al. 2022); (2) magmas can be further oxidized during fluidundersaturated magma differentiation at Moho-vicinity depths (Loucks and Fiorentini 2023); or (3) crystallization of significant amounts of garnet slightly increases reduced magmas in the fO2 value by ≤0.8 log unit (Tang et al. 2018, Holycross and Cottrell 2023). The high 1000 • (Eu/Eu*)/Yb ratios of the zircon grains indicate suppression of Eu depletion in plagioclase-undersaturated silicate melt or Yb depletion due to hornblende crystallization during zircon precipitation (Loucks and Fiorentini 2023). The crystallization of hydrous amphibole, as well as suppression of anhydrous plagioclase crystallization, indicate elevated H2O concentrations in the magmas. The high 1000•(Eu/Eu*)/Yb ratios with mantle-like O isotopic values, therefore, indicate relatively hydrous magmas formed by hydration of the mantle-like source regions in a subduction-like setting (Fig. 3a), consistent with the previously proposed episodic subduction-like process (Moyen and van Hunen 2012). Fluids released from the subducted slab, without interaction with sediments, plausibly oxidize the mantle source by hydrogen incorporation in the surrounding mantle (e.g., orthopyroxene) coupled with H2O dissociation (Tollan and Hermann 2019). The estimated S concentration of 0.11-0.34 wt% in the evolved silicate melt represents the lower limit of the S concentration in the parental magmas due to sulfide saturation during magma differentiation. The high-S concentration in the parental magmas with limited oxidation state could be attributed in part to the active upwelling of sulfide-saturated Archean mantle during the frequent slab breakoff or contamination by a lower crustal reservoir enriched in reduced S (e.g., mafic cumulate). et al. 2009), so we suggest that the increase in magmatic fO2 with cooling is more consistent with the fractionation of ferromagnesian silicates.
The exceptionally rare occurrence of >2.7 Ga arc-type sodic magmas with high-apatite S concentrations and low-S 6+ /ΣS ratios (Fig. 2) may be interpreted to suggest the aforementioned oxidation process was merely localized and insufficient. In contrast, most of the other arc-like samples (Group-II) yielded relatively low apatite S concentrations, below the detection limit (0.012 wt%), as well as zircon (Eu/Eu*)/Yb ratios and whole-rock La/Yb ratios skewing to lower values. The geochemical features for Group-II samples were more consistent with their derivation from shallow intra-crustal melting of a mafic source or plagioclase ± amphibole differentiation of basalt-derived parental liquids (Fig. 3b), which could explain the relatively dry (as manifested by the generally lower zircon [Eu/Eu*]/Yb ratios) and reduced redox state of the sodic magmas (ΔFMQ value of 0.0 ± 0.5 on average estimated using zircon geochemistry; Table 1).
Plots of the whole-rock data for the sodic igneous rocks from the Abitibi subprovince (Fig. 3c; Online Materials 1 Table S4) reveal that: (1) less evolved sodic magmas (MgO content >4 wt%; a proxy for magma differentiation) yield Cu concentration of ∼50 ppm, which is comparable to Phanerozoic arc magmas (Richards 2015); and (2) with decreasing whole-rock MgO content, the whole-rock La/Yb ratio increases and the concentration of Cu decreases, consistent with loss of Cu to magmatic sulfides during amphibole±garnetdominated fractionation or by early volatile exsolution (Fig. 3c). Magmatic sulfides in hydrous upper crustal magma reservoirs are thought to be capable of temporarily retaining S and Cu that can subsequently be destabilized by the exsolved magmatic-hydrothermal fluids (Audétat and Simon 2012; Chelle-Michou and Rottier 2021) in tandem with a concomitant increase in magmatic fO2 (Loucks and Fiorentini 2023). We suggest this scenario might have operated locally by considering the rare occurrence of slightly oxidized and S-rich magmas with sulfide-rich apatite grains. In comparison, for most of the relatively dry, reduced pre-tectonic sodic magmas, the S-poor apatite grains may reflect sulfide saturation at a greater depth (e.g., lower crust) and/or the ineffective capacity of many relatively dry sodic magmas to exsolve oxidized S-rich hydrothermal fluids. Because syn-mineralization magmas associated with medium-to large-size porphyry-type Cu deposits typically yielded ΔFMQ values of +1 to +2 and S-bearing apatite grains (S content >0.02 wt%; Fig. 2; Online Materials 1 Table S5), we suggest that the pre-tectonic sodic magmas, including those with arc affinity may have limited, although not totally excluded (e.g., Coté gold; Katz et al. 2021), the formation of porphyry-type Cu deposits in the Archean Greenstone Belt.
The results presented here demonstrate the presence of S-rich apatite grains with a low-S 6+ /ΣS ratio crystallized from >2.7 Ga slightly oxidized and S-rich terrestrial silicate melts, which may have formed in association with the short-lived proto-subduction in the Neoarchean Abitibi Greenstone Belt. In contrast, most of the >2.7 Ga pre-tectonic arc-like igneous rocks, as classified using zircon Ui/Nb ratios of ≥40, yielded S-poor apatite grains that may reflect relatively reduced or S-poor features of the magmas. The results indicate either local or weak emission of oxidized S from sodic magma reservoirs to the nearsurface environment during early Earth's history. These conclusions can explain the rarity of porphyry-type Cu deposits associated with syn-volcanic TTG rocks that were prevalent in >2.7 Ga in the Abitibi Greenstone Belt and possibly in other cratons.
The research was funded by the National Natural Science Foundation of China (Grant # 41820104010, J.M.), Canada First Research Excellence Fund via Metal Earth (CFREF-2015-00005), U.S. National Science Foundation EAR (Grant # 2233425 and 2214119, A.C.S.), and a start-up research grant from China University of Geosciences (Grant # 3-8-2023-008, X.M.).
Accepted manuscript online July 16, 2024 Manuscript handled by Claire E. Bucholz Endnote: