Figure 7. Trace element characteristics for Isua and Pilbara ultramafic samples in comparison with compiled cumulates and variably altered mantle peridotites. a, Primitive mantle (PM) normalized Gd/Yb and La/Sm ratios [i.e., (Gd/Yb)PM and (La/Sm)PM] of investigated samples and compiled rocks.b, Th and Gd/Yb ratios of investigated samples and compiled rocks. c, PM-normalized spider diagrams showing trace element patterns of investigated samples and compiled rocks (see Figure S4 for spider diagrams with sample locations). These diagrams show that new and compiled data for ultramafic rocks from the Isua supracrustal belt have similar trace element characteristics to dunite enclaves from the south Isua meta-tonalites, Pilbara ultramafic samples and ultramafic cumulates. Only some abyssal peridotites which experienced serpentinization and melt-rock interactions have comparable trace element patterns. Other mantle peridotites have lower Th, Gd/Yb, (Gd/Yb)PM, and/or (La/Sm)PM values. Data sources: compiled cumulates involve samples from the Permian Lubei intrusion of NW China (Chen et a., 2018), the late Proterozoic Ntaka Ultramafic Complex of Tanzania (Barnes et al., 2016), the Mesoarchean Nuasahi Massif of India (Khatun et al., 2014), the Mesoarchean Tartoq Group of SW Greenland (Szilas et al., 2014), the Mesoarchean Seqi Ultramafic Complex of SW Greenland (Szilas et al., 2018), and the Eoarchean Tussapp Ultramafic Complex of SW Greenland (McIntyre et al., 2019); compiled Eoarchean ultramafic samples are rocks from the Isua supracrustal belt (Szilas et al., 2015) and the enclaves in meta-tonalite south of the Isua supracrustal belt (Van de Löcht et al., 2020); fresh arc peridotites are from the Kamchatka arc (Ionov, 2010); arc peridotites that experienced serpentnization, talc/tremolite alteration, and/or melt-rock interactions are from the Loma Caribe peridotite body of Dominican Republic (Marchesi et al., 2016) and the Izu-Bonin-Mariana forearc (Parkinson and Pearce, 1998); abyssal peridotites that experienced serpentinization are from the Oman ophiolite (Hanghøj et al., 2010); variably altered abyssal peridotites from the Mid-Atlantic Ridge are summarized by Paulick et al. (2006). Primitive mantle values are from McDonough and Sun (1995).
Assessment of alteration impactsPetrological and geochemical information obtained from Isua and Pilbara ultramafic rocks represents the combined effects of petrogenetic processes and alterations. Below we discuss potential types and impacts of alteration on the petrology and geochemistry of these rocks. High-grade (e.g., granulite facies) metamorphism can lead to partial melting. The partial melting process and subsequent melt-rock interactions could strongly disturb the geochemistry and mineral assemblages of affected rocks. However, the Isua supracrustal belt and the area from which Pilbara ultramafic samples were collected (Fig. 1 ) have only experienced amphibolite facies metamorphism (or lower conditions) (e.g., Hickman, 2021; Ramírez-Salazar et al., 2021). In general, for an anhydrous system larger than a hand specimen, amphibolite facies or lower metamorphism is usually considered isochemical. However, both Isua and Pilbara samples show evidence of hydrothermal alterations, as indicated by the dominance of serpentine minerals (Figs. 2–3 ). Therefore, some whole-rock geochemical changes are possible (see below). In addition, at mineral scales, some chemical changes during metamorphism are possible. For example, Cr-spinel could be altered to magnetite during metamorphism (e.g., Barnes and Roeder, 2001). Therefore, care must be taken when interpreting petrogenesis using spinel data. Fluid assisted alterations could result in changes in mineral assemblages and element concentrations, especially for fluid-mobile elements (e.g., K, Ca, Si, Rb, Ba and Sr, etc.). LOI contents (Fig. 4a ), and the presence of serpentine, talc, and/or magnesite (Figs. 2–3 ) in Isua and Pilbara ultramafic samples show that these rocks have experienced variable degrees of serpentinization and carbonitization (including talc -alteration; although some magnesite crystals could be primary minerals crystallized from fluid-rich magmas, see Fig. 2a for magnesite crystals as olivine inclusions; also Smithies et al., 2021). A ternary plot of anhydrous SiO2, LOI, and other oxides (e.g., MgO, TiO2, see Fig. 4 caption) shows that serpentinization is the dominant controlling factor for their geochemistry as these samples plot near the serpentine mineral composition. Effects of other alterations on major element concentrations and LOI (e.g., Deschamps et al., 2013; Paulick et al., 2006) in most samples seem to be secondary with the exception of sample AW17724-1 which has a high anhydrous CaO concentration (10.4 wt.%). Hydrothermal fluids may mobilize some elements such as Mg, Si, and trace elements including REEs (e.g., Deschamps et al., 2013; Malvoisin et al., 2015; Paulick et al., 2006). Nonetheless, the potential MgO and SiO2 loss/gain could be insignificant (i.e., <10%) compared to other factors (e.g., melt depletion, melt-rock interaction, or talc-alteration, seeFig. 4b ; Snow and Dick, 1995). Some HSEs like Os, Ir, Ru and Pt are largely immobile during fluid assisted alterations, but Pd and Re could be mobile (e.g., Büchl et al., 2002; Deschamps et al., 2013). Spinel Al and Cr concentrations can be increased or reduced during fluid-rock interaction, respectively (e.g., El Dien et al., 2019). Melt-rock interaction is commonly observed in mantle rocks (e.g., Ackerman et al., 2009; Büchl et al., 2002; Deschamps et al., 2013; Niu, 2004; Paulick et al., 2006) where ascending melts react with wall rocks. This process is similar to reactions between cumulate phases and trapped/evolving melts during crystallization or post-cumulus processes (e.g., Borghini and Rampone, 2007; Goodrich et al., 2001; Wager and Brown, 1967). In general, melt-rock interaction can alter the geochemistry of affected rocks towards those of melts at increasing rock/melt ratios (e.g., Kelemen et al., 1992; Paulick et al., 2006). For peridotites interacting with basalts or more evolved melts, the elevation of elements that are relatively enriched in melts (e.g., Si, Ca, Th, Al, Fe, Ti, LREE, Pt, Pd, and Re) is significant (Figs. 4–7 ; e.g., Deschamps et al., 2013; Hanghøj et al., 2010). Other effects include changes in mineral modes and/or mineral geochemistries (e.g., olivine Mg# reduction; spinel Cr-loss and Al-gain) (e.g., El Dien et al., 2019; Niu and Hekinian, 1997). Two Isua supracrustal belt ultramafic samples (AW17725-2B, AW17806-1) which were sampled near the meta-tonalite bodies have the highest Al2O3 and lowest MgO concentrations among all samples (Figs. 4b, 5 ), which may be explained by reactions with relatively Al2O3 rich components (fluids and/or melts). In summary, fluid/melt rock interaction might partly control the observed geochemistry and petrology of studied Isua and Pilbara samples. Therefore, for petrogenetic interpretation, we compare the observed geochemistry and petrology of Isua and Pilbara ultramafic samples with those of cumulates and mantle peridotites that potentially experienced similar alterations (including serpentinization, talc/tremolite alteration, and melt-rock interaction).