Figure 10 : Schreinemarkers diagrams showing the relevant talc + magnesite forming reaction and the stability field of these phases in temperature (T) versus X(CO2) space at two different pressures. Calculations were performed with Perple_X (Connolly, 2005) (version 6.9.0) and an updated version of the internally consistent Holland and Powell (2011) dataset (ds63 update). The stability fields of Ti-humite phases cannot be calculated due to a lack of relevant thermodynamic data. Nonetheless, co-existing phases of Ti-humite (i.e., magnesite, forsteritic olivine and talc) in Isua sample AW17724-2C are consistent with metamorphism under crustal conditions and 500–650 °C in the presence of CO2. This is potentially consistent with the regional amphibolite metamorphism affecting the Isua supracrustal belt (Ramírez-Salazar et al., 2021). Abbreviations: en: enstatite; fo: forsterite; mag: magnesite; qz: quartz; per: periclase; tlc: talc.
Whole-rock and trace element characteristics of ultramafic rocks may help to discern crustal igneous rocks from tectonic mantle slices. This is because crustal igneous rocks should have geochemical signatures corresponding to mantle melts, their crystal precipitates assimilation effects and/or post-cumulus modifications, whereas mantle slices should show geochemical evidence reflecting melt-depletion and re-enrichment under mantle conditions. To facilitate petrogenetic interpretation for rocks with potentially complex alteration histories (see section 5.1), we compare Isua and Pilbara ultramafic rocks with (1) modelled melt-refertilized abyssal peridotites generated by combining batch melting models (calculated using pMELTS; Ghiorso et al., 2002) and mixing models (Chin et al., 2014); (2) variably altered (including melt infiltrated and fluid infiltrated) arc peridotites; (3) variably altered abyssal peridotites; (4) cumulates (which largely experienced serpentinization) from Archean terranes and Phanerzoic settings; (5) modelled cumulates of hydrous and anhydrous Phanerzoic settings (calculated using alphaMELTS; Ghiorso and Sack, 1995; Smith and Asimow, 2005) as presented in Chin et al. (2018); and (6) komatiites (seeFigs 5–9 captions for references from which the literature data were compiled ).
We find that both Isua and Pilbara ultramafic rocks show similar whole-rock major element geochemistry to the compiled mantle peridotites (e.g., melt-refertilized forearc peridotites from mantle wedges and abyssal peridotites from mid-ocean ridges; Fig. 5 ) and the most Mg-rich cumulates that represent <10% fractional crystallization products of basaltic melts (Fig. 6 ). Although the Isua and Pilbara ultramafic rocks generally have systematically less CaO compared with mantle peridotites, possibly due to low to zero abundances of clinopyroxene (Fig. 5 ), this can also be explained by alteration effects (see section 5.1). In fact, small clinopyroxene inclusions occur in olivine grains of some Isua ultramafic rocks from lens A, which have been explained as indicative of olivine participation coupled with clinopyroxene dissolution during reactions between mantle peridotites and ascending melts (Nutman et al., 2021). However, we note that clinopyroxene undersaturation and olivine saturation is possible across a range of pressure-temperature-composition combinations (Chen and Zhang, 2009 and references therein) and could happen under crustal conditions during magma crystallization in the presence of water, crustal assimilation and/or magma recharge (e.g., Kelemen, 1990; Gordeychik et al., 2018). Because Al2O3 is unlikely to be significantly mobilized during fluid-assisted alterations (see section 5.1), the negatively correlated MgO and Al2O3 found in these Isua and Pilbara ultramafic samples should reflect primary igneous features or melt-assisted interactions (note that MgO concentrations for most samples may not be significantly altered, Fig. 4b ). We therefore interpret that the observed major element geochemical systematics reflect either (1) depleted mantle peridotites that were variably altered by percolating melts (e.g., Friend and Nutman, 2011; Nutman et al., 2021; Van de Löcht et al., 2020) or (2) Mg-rich cumulates that were variably contaminated by co-existing, more evolved melts (e.g., Szilas et al., 2015).
Compared to Isua and Pilbara ultramafic samples, most depleted mantle rocks from plate tectonic settings generally show much stronger depletions in many trace elements versus primitive mantle values (Fig. 7 ), as highlighted by their (La/Sm)PM, (Gd/Yb)PM, and Th values, possibly because of strong melt depletion as expected for tectonically-emplaced mantle residues. Elevation of trace element concentrations is expected during some alterations (see section 5.1), but fluid-assisted alterations (including serpentinization and talc carbonate alteration) alone cannot produce the observed trace element geochemistry of Isua and Pilbara ultramafic samples (Fig. 8a–b ). Instead, some mantle rocks (modified by melt-rock interactions), and also cumulate rocks, have comparable trace element geochemistry (Fig. 7 ). Indeed, similarities between trace element patterns of Isua ultramafic rocks and those of nearby basalts (Friend and Nutman, 2011; Szilas et al., 2015; Van de Löcht et al., 2020) were explained by (1) reactions between mantle resitites and melts (Friend and Nutman, 2011; Nutman et al., 2021; Van de Löcht et al., 2020) or (2) reactions between melt components (e.g., evolving basaltic melts) and cumulus minerals (Szilas et al., 2015). Therefore, the observed generally flat, primitive-mantle-like trace element characteristics found in new and compiled Isua and Pilbara samples are consistent with both depleted mantle residue or cumulate origins.
Relatively high primitive mantle-normalized Os, Ir, and Ru versus Pt, Pd and Re [e.g., (Pt/Ir)PM < 1] is often used to discriminate depleted mantle rocks because Pt, Pd and Re behave as moderately incompatible elements during mantle melting (e.g., Bockrath et al., 2004; Wang et al., 2013). However, we note that during mantle melting, I-PGEs typically show similar compatibility [such that (Ru/Ir)PM ≈ 1], whereas in the studied Pilbara samples, significant fractionation among I-PGEs [i.e., >2 (Ru/Ir)PM values] are observed (Fig. 8 ). Although mantle rocks which suffered extensive melt-rock interactions could also have strongly fractionated HSEs (e.g., Büchl et al., 2002; Ackerman et al., 2009), their overall HSE trends [including (Pt/Ir)PM and (Ru/Ir)PM values] are dissimilar to those of Pilbara samples (Fig. 8b ). In contrast, some chromite-bearing peridotite cumulates from layered intrusions show similar HSE fractionation trends versus Pilbara samples (Fig. 8c ). Therefore, HSE patterns of the Pilbara samples are consistent with the proposed cumulate origin of these rocks depicted from their rock textures (see above, Fig. 3 ) and spinel geochemistry (see below, Fig. 9 ). The elevated Pd and/or Re as well as unrealistic Re-Os systematics in some Pilbara samples may result from an addition of Fe-sulphides during later alteration events (see section 5; Lorand and Luguet, 2016). These Fe-sulphides then would have been altered to magnetite (which commonly occurs in alteration veins or triple junction points of serpentine clusters of Pilbara samples,Fig. 3b ) over billions of years. In comparison, ultramafic rocks from the Isua supracrustal belt (including dunites from the lenses; Fig. 8a ; Szilas et al., 2015) and meta-tonalite enclaves south of the Isua supracrustal belt (Fig. 8a , Van de Löcht et al., 2018) show a range of HSE patterns, highlighted by their <0.5 to >10 (Ru/Ir)PM and (Pt/Ir)PM values. Significantly elevated (Ru/Ir)PM and (Pt/Ir)PM values may be explained by interactions with melts; such interactions could happen to both cumulates and mantle peridotites (e.g., Ackerman et al., 2009; Gannoun et al., 2016; Szilas et al., 2015). Although some Isua ultramafic rocks have (Ru/Ir)PM close to 1 and (Pt/Ir)PM less than 1 [e.g., “group 1” peridotites sampled from meta-tonalite enclaves (Van de Löcht et al., 2018) and a portion of ultramafic rocks studied by Szilas et al. (2015)], which are similar to depleted abyssal mantle rocks (Wang et al., 2013), such patterns also occur in cumulates (Fig. 8c ) (McIntyre et al., 2019; Szilas et al., 2014; 2018). McIntyre et al. (2019) argue that these specific HSE patterns may be alternatively explained by preferential partitioning of I-PGEs into cumulus phases such as olivine and chromite. Therefore, HSEs may not be as discriminative as previously thought in terms of recognizing depleted mantle rocks. Accordingly, HSE patterns of Pilbara and Isua ultramafic rocks, including those from meta-tonalite enclaves south of the Isua supracrustal belt, can be explained by cumulate origins (± modifications potentially by melts).
Some spinel crystals with ~100 Cr# and ~0 Mg# values in the new and compiled Isua ultramafic rocks reflect metamorphic modifications of primary chromite into magnetite (Fig. 3b ; Barnes and Roeder, 2001). However, igneous petrogenesis can be interpreted from primary chromite grains of both Isua and Pilbara ultramafic samples. New and compiled spinel data [from Szilas et al. (2015) which were obtained from rocks of lenses A and B] of these rocks match the Fe Ti trend in the Mg# Cr# space (Fig. 9b ). Such a trend can be produced by equilibration of spinel phases during fractional crystallization (Barnes and Roeder, 2001), and thus can be found in cumulates (Fig. 9b ). In contrast, due to equilibration with olivine, mantle spinel typically has high Mg# and varied Cr# (i.e., the Cr Al trend in Fig. 9b , Barnes and Roeder, 2001) as well as low TiO2 (Fig. 9a ) (e.g., Tamura and Arai, 2006). Although fluid/melt assisted alterations could impact spinel geochemistry in mantle rocks, expected changes include Cr# reduction and Mg# increase along the Cr Al trend (El Dien et al., 2019), which are not consistent with the observed spinel geochemistry. Therefore, we conclude that some chromite spinel crystals from Isua (Szilas et al., 2015) and Pilbara ultramafic rocks are not similar to spinel hosted in mantle rocks, but rather indicate cumulate origins (cf. Nutman et al., 2021).
Olivine oxygen isotopes of some dunites from the Isua supracrustal rocks are interpreted to be mantle-like and indicative of fluid metasomatism (likely from recycling hydrated crust) in the mantle wedge (Nutman et al., 2021). This material exchange between surface and mantle is thought to be exclusive to plate tectonic subduction settings (Nutman et al., 2020; Nutman et al., 2021). However, such material exchange is also possible for hot stagnant-lid settings, accomplished by recycling of the buried or dripping hydrated crustal materials (see tectonic models introduced in section 2; also, Moore and Webb, 2013; Smithies et al., 2007). Indeed, mantle-like oxygen isotopes are observed in zircons from some tonalites (originally lower crust partial melts) of the East Pilbara Terrane (where its geometry and structures largely suggest non-plate tectonic origins, see section 2.2; Smithies et al., 2021). This finding implies a fluid-rich early mantle, buffered by fluxing from the recycled crust, that was capable of introducing mantle-like oxygen isotope signatures to early crust and magmas (Smithies et al., 2021). Therefore, mantle-like oxygen isotopic signatures found in some dunites from the Isua supracrustal belt need not be explained by plate tectonic subduction (cf. Nutman et al., 2021).
In summary, although several features of Isua or Pilbara ultramafic samples are commonly associated with depleted mantle rocks (e.g., the B-type olivine fabrics and Ti-humite preserved in Isua ultramafic samples), these features are not inconsistent with cumulate origins. In addition, the cumulate textures of Pilbara ultramafic samples and the spinel geochemical characteristics of both Isua and Pilbara ultramafic samples are inconsistent with tectonically-emplaced depleted mantle, but instead are compatible with cumulate origins (Figs. 4–9 ). As such, both Isua and Pilbara ultramafic samples can be interpreted as crustal cumulates. Because crustal cumulates are produced by fractional crystallization of melts, these rocks are consistent with both plate tectonics and hot stagnant-lid tectonics. Thus, plate tectonics is not required to explain the petrogenesis of Isua ultramafic rocks (cf. Nutman et al., 2020; Nutman et al., 2021).