Many of the details of the differentiation process remain uncertain, but it is clear that during core formation the silicate material that would later become the mantle and crust was in chemical equilibrium with iron metal. The evi-dence for this is found in the measured abundances of elements present in the Earth’s mantle and crust (bulk silicate Earth, or BSE), compared to chondritic meteorites, which are our best estimate for the bulk composition of the solar system.
1.1.1 Chondritic Earth
The Earth is, as a whole, believed to be chemically similar to chondritic me-teorites, which, collectively, are fragments of undifferentiated planetesimals
that appear to have remained largely unprocessed from the very beginning of the solar system (e.g. Brearley and Jones, 1998; Wood, 1988). Though the chondritic meteorite groups vary somewhat in composition, they have in common identical ratios of refractory lithophile elements, i.e., elements that would not fractionate due to their volatility or preference for a metallic or sul-fide phase (e.g., Rubin, 2011). These abundance ratios are also the same as those measured in the solar photosphere (e.g., Palme and O’Neill, 2003). It has therefore long been assumed that the suite of chondritic metorities represents the bulk composition of the solar system, and that the variations between the groups reflect different accretion histories of the parent asteroids and/or lo-cal compositional variability within the solar nebula (Allègre et al., 1995). CI carbonaceous chondrites, specifically, have elemental abundances that most closely match the solar composition, and so are commonly used as a proxy for the primitive bulk composition of the solar system (e.g., Anders and Grevesse, 1989).
Measurements of the relative abundances of elements in the bulk silicate Earth show that refractory lithophile elements are also present at the same relative abundances (figure 1.1) as chondrites. This is generally accepted as strong evidence that the bulk Earth is, at least to a first order approximation, chondritic in composition and that depleted elements reflect fractionation pro-cesses during or after accretion (e.g. Ringwood, 1979; Wänke et al., 1984;
McDonough and Sun, 1995). Indeed, most of the constraints on fractionation processes that occurred throughout Earth’s history derive from comparing the Earth’s mantle composition to that of chondrites (e.g., Palme and O’Neill, 2003).
Figure 1.1: Relative abundances of elements in the bulk silicate Earth.
Lithophile elements that would not be expected to fractionate due to volatility, i.e. those more refractory than Mg, are present in the same ratios as CI chon-drites, indicating that the bulk Earth is chondritic. Refractory siderophile elements are therefore depleated from the mantle due to core formation. The highly siderophile elements (blue symbols) have a flat (chondritic relative) abundance pattern, despite having vastly different metal/silicate partition coefficients. They were therefore probably completely stripped from the man-tle and then added back after the end of core formation. Figure is modified from Frost et al. (2008).
1.1.2 Differentiation
The most consequential post-accretion process on the Earth was the physical separation of the metallic iron-nickel material that is now the core from the silicates that comprise the mantle. There is substantial evidence from achon-dritic meteorites, which are remnants of differentiated planetesimals that were collisionally disrupted and broken apart, that this occurred very rapidly after the beginning of the solar system (e.g., Weiss and Elkins-Tanton, 2013). Iron meteorites are pieces of metallic cores, while stony achondrites represent the mantles. Isotopes of hafnium and tungsten in these planetesimal fragments provide a strong constraint on the timing of differentiation. 182W is a decay
product of 182Hf, and the two elements have different geochemical behaviour:
tungsten is siderophile (i.e. prefers a metallic phase) and hafnium is lithophile (i.e. prefers a silicate phase). If metal-silicate differentiation happens before
182Hf has fully decayed (<45 Ma after the beginning of the solar system), the Hf would remain in the silicate, and later decay to W. This tunsten would then be stranded in the silicate and visible as a positive 182W anomaly in the silicate with a corresponding negative anomaly in the metal compared to chondrite (e.g., Kleine and Walker, 2017).
Studies of achondrites using this method have determined that small plan-etesimals had grown to large enough size to retain sufficient heat (supplied primarily by gravitational release and the decay of short-lived radionuclides) to melt throughout, allowing the heavier metallic iron alloy to sink and form a core by 5 Mya after the origin of the solar system (Kleine et al., 2002). Futher, Hf-W systematics reveal that the Earth, which accreted largely from planetes-imals that had already themselves separated into a mantle and a core, had fully differentiated by 30 Mya after the beginning of the solar system (Kleine et al., 2002).
As can be seen in figure 1.1, refractory siderophile elements (i.e., elements that preferentially partition into a metallic phase but would not be expected to fractionate due to volatility) are depleted from mantle rocks. The most plausible interpretation of this is that during differentiation, the core-forming metallic alloy equilibrated with the mantle silicates as it sank. The distribution of an element M between the fractionating metal and silicate was governed by the partition coefficient
Dmet/sil(M) = XMmetal XM Osilicate
n/2
where X is the mole fraction of the element M or its oxide in the metal or silicate and n is its valence state. Siderophile elements, which are defined as having a metal/silicate partition coefficient of greater than 1, would have then partitioned into the metallic phase and thus been sequestered in the core.
While this much is relatively clear, there remains vigourous debate regarding some of the details of core separation beyond this first-order approximation.
It has, for instance, long been recognised that a single-stage low-pressure
differentiation event could not reproduce the elemental abundance pattern in the mantle today (e.g., Ringwood, 1966; Wänke, 1981). For example, mantle abundances of the slightly siderophile elements, such as V and Cr, indicate much more reducing conditions than the depletions of moderately or highly siderophile elements (e.g., Frost et al., 2008).
This led to suggestions of heterogeneous accretion models, in which the na-ture of accreting material changes with time, and the oxygen fugacity at which metal equilbration occurrs increases throughout accretion. (Wänke, 1981;
O’Neill, 1991; Wade and Wood, 2005; Wood et al., 2006). O’Neill (1991) sug-gested a secondary, more oxidised stage of core formation in which the core-forming liquid was no longer metallic iron but a FeS sulfide (the “Hadean matte”, see section 1.3.2 for a more detailed review).
Later in the 1990s, it was recognised that some elements become less siderophile with pressure, and that it was possible to reproduce the abundance pattern of many elements with a homogenous accretion model if metal-silicate parti-tioning had occurred at high pressures and temperatures (e.g., Li and Agee, 1996).
In figure 1.1, it can be seen that the highly siderophile elements (i.e., the plat-inum group elements or PGE) exist in chondritic ratios relative to each other (e.g., Day et al., 2016). Experiments to determine the metal/silicate parti-tion coefficients of the PGEs, however, have found vastly different partiparti-tioning behaviour of the individual elements (Mann et al., 2012). In other words, there is no set of metal-silicate equilibration conditions that can reproduce the observed flat abundance pattern of the highly siderophile elements. These observations led to the suggestion that the PGEs must have been completely stripped from the mantle during core formation, and then subsequently re-placed by the addition of more primitive, chondrite-like material that has been termed the “late veneer” (e.g., O’Neill, 1991).
As yet, no model can successfully reproduce every constraint and many de-tails of the differentiation process remain uncertain, however the suggestion that high-pressure metal-silicate equilibration was a feature of core fomation has amassed a fair bit of supporting evidence (Li and Agee, 1996; Rubie et
al., 2003; Chabot et al., 2005; Siebert et al., 2013; Mann et al., 2009). Re-cent experimental work on sulfide/silicate partitioning behaviour of the highly siderophile elements has also provided further support for the Hadean matte (Laurenz et al., 2016; Rubie et al., 2016), though it should be noted that these possibilities are by no means mutually exclusive.