Figure 10: Primitive-mantle normalized REE (McDonough and Sun, 1995) and HSE (Becker et al., 2006) patterns of Isua (panel a) and Pilbara (panel b) ultramafic rocks (this study; Szilas et al., 2015; Waterton et al., 2022), compared to modelled and compiled mineral and/or whole-rock compositions that may explain the REE–HSE systematics of Isua and Pilbara ultramafic rocks. We use binary mixing models between a relatively REE, Pt and Pd depleted component and a relatively REE, Pt and Pd enriched component trying to reproduce the trace element characteristics of Isua and Pilbara ultramafic rocks. For REEs, we mix modelled cumulate olivines (Waterton et al., 2022) with potentially co-genetic melts, as olivines in Isua dunites, which show relatively enriched LREEs and Eu-anomalies (Friend and Nutman, 2011), may not retain their igneous trace element abundances (Waterton et al., 2022). For HSEs, due to the lack of HSE data for modelled cumulate olivines (Waterton et al., 2022), we mix the most HSE depleted ultramafic sample in Isua or Pilbara with potentially co-genetic melts. Because the HSE data for Pilbara andesites are not available, we use the HSE composition of andesite studied by Day et al. (2013). Mixing models show that Isua and Pilbara ultramafic rocks can be best explained by mixing between cumulates and co-genetic HSE depleted melts. Such HSE depleted melts can be formed by sulphide removal or fractional crystallization in magma chambers, or deep mantle melting. Additional data sources: average REE values of Isua tholeiitic basalts are taken from the least altered samples measured by Polat and Hofmann (2003); HSE compositions of Isua tholeiitic basalts are taken from a single sample measured by Szilas et al. (2015); average Pilbara tholeiitic basalts and andesites are from the ~3.35 Ga Euro Basalt Formation and/or the ~3.43 Ga Panorama Formation of the East Pilbara Terrane (Smithies et al. 2007); continental flood basalts after sulphide removal are from Lightfoot et al. (2005).
Isua ultramafic rocks may have mixed with coeval tholeiitic basalts with generally unfractionated REE patterns (e.g., Van de Löcht et al. 2020; Waterton et al. 2022) but also low Pt and Pd contents (potentially ≤1 ppb). Although basaltic melts typically have primitive mantle-like or higher Pt and Pd due to their preferential partitioning to melts during mantle melting (e.g., Bockrath et al. 2004; Fiorentini et al. 2010; Brenan et al., 2016), because HSEs are extremely siderophile (Lesher et al. 2001; Fiorentini et al. 2010), sulphide removal during crystallization (e.g., Lightfoot et al. 2005) and retaining sulphide as residues in the mantle sources (e.g., Fiorentini et al. 2011) are two mechanisms that can lower the Pt and Pd abundances of the co-genetic melts. In Isua, some meta-basalt layers were found to contain sulphide minerals which crystallized coevally with the basalts (Appel, 1979). Therefore, at least some Isua basalts may have undergone sulphide exhaustion and consequently have sufficiently low HSE concentrations (see Fig. 10a for low HSE contents of basalts after sulphide removal; Lightfoot et al. 2005). Furthermore, because S is less soluble in melts under higher pressures (O’Neill and Mavrogenes, 2002), mantle partial melts with low HSE abundances indicate deep mantle sources (e.g., low HSE komattites studied by Fiorentini et al., 2011). Therefore, the potential presence of low Pt and Pd Isua tholeiitic basalts may be explained by deep mantle melting, which is not compatible with a plate tectonic mantle wedge origin (e.g., Figure 8 of Nutman et al., 2013a), but would be consistent with a recent geochemical model for Isua basalts that imply a heat-pipe origin (Rollinson, 2021).
Alternatively, Isua ultramafic rocks may have interacted with other co-exisitng melt / rock types instead of the tholeiitic basalts. TTGs can be excluded because these crust partial melting products (e.g., Nagel et al. 2012) would have radiogenic 187Os/188Os signatures (e.g., Gannoun et al. 2016), which is inconsistent with the overall unradiogenic 187Os/188Os values of Isua meta-peridotite lens samples (Waterton et al. 2022). Instead, mixing with evolved melts formed by fractional crystallization of basaltic magmas can explain the geochemistry of Isua and Pilbara ultramafic rocks (Fig. 10b ). For example, in the East Pilbara Terrane, minor andesites can be found in thick (ultra)mafic to felsic volcanic successions (Smithies et al. 2007), which may be co-genetic with lava fractional crystallization. These andesites (SiO2 ≈ 54 to 61 wt.%) have flat to mildly fractionated REE patterns (Fig. 10b ). Simple binary mixing models show that 10% to 15% mixing between cumulate olivine (Waterton et al. 2022) and a component with Pilbara andesite REE compositions can successfully reproduce REE systematics of Pilbara ultramafic rocks, especially their mildly fractionated LREE and negative Eu anomalies (Fig. 10b ) which cannot be accounted only by mixing with potentially coeval Pilbara tholeiitic basalts (Fig. 10b ; cf. Smithies et al. 2009 showing that some ~3.5 Ga Pilbara basalts yield enriched and fractionated LREE patterns). Although no HSE data are available for these andesites, limited data obtained so far for melts with similar andesitic composition (Day et al. 2013) indicate generally low HSE contents in such melt type (e.g., Pt and Pd <3.1 ppb). Thus, we envision that REE/HSE abundances of Pilbara ultramafic rocks can be best explained by mixing with Pilbara andesites (Fig. 10b ). Although such evolved components are not currently observed in the Isua supracrustal belt, they are speculated to exist (Szilas et al. 2015) and together with the observation that they may have been volumetrically minor (as in Pilbara), they might be unexposed or remain undetected due to ~90% thinning by regional deformation (e.g., Webb et al., 2020). Indeed, several Isua ultramafic samples have enriched, fractionated REEs and negative Eu anomalies (Figs. 7, 10a ), indicating the presence of evolved components that may have compositions similar to Pilbara andesites.
To conclude, in contrast to Waterton et al. (2022) who only interpret that Isua ultramafic rocks formed by cumulates mixing with unspecified co-genetic tholeiitic basaltic magmas, we specify that such melts must be enriched in REEs but depleted in HSEs (especially Pt and Pd). Possible candidates are (1) tholeiitic basalts that derived from deep mantle and/or experienced sulphide removal in magma chambers; and (2) evolved melts derived from fractional crystallization of basaltic magmas. Because Isua and Pilbara ultramafic rocks show similar REE and HSE systematics (Figs. 7, 8 ), mixing REE-enriched but HSE-depleted melts with cumulus olivine ± chromite ± orthopyroxene can explain the formation of both Isua and Pilbara ultramafic rocks, along with potentially other Eoarchean ultramafic cumulates which show similar REE and HSE geochemistry (Coggon et al. 2015; McIntyre et al. 2019).