By Rick

In the widely used pyrolite model for mantle composition developed by Ted Ringwood (a former Professor at RSES), the chemistry of the Earth’s mantle is dominated by a small handful of oxides: SiO2, MgO, FeO, CaO, and Al2O3, with smaller amounts of Na2O, TiO2, and Cr2O3. The study of the physical and chemical properties of the Earth’s mantle is predominantly the study of the mineral phases formed by these oxides. However, chemical elements occurring at minor and even trace abundance can be almost as important for the properties of the mantle.

One way in which low-abundance elements can influence the properties of the mantle is through the formation of separate phases, which may provide a chemical environment quite different than those available in the major mantle silicates and oxides. This potentially allows them to incorporate trace elements at far higher concentrations than the pyrolitic mantle, so that they can dominate the chemical budget for certain elements. For example, calcium phosphate minerals, with their large cation sites and “floppy” Ca-O bonds, make excellent hosts for all manner of chemical impurities, including the rare earth elements (REEs), strontium, and the actinides. Indeed, the solubility of REEs in phosphate minerals is so high that the presence of apatite in igneous rocks can significantly perturb observed REE trends (see eg. Watson and Capobianco, 1981). Another example is to be found in the siderophilic (iron-loving) elements, whose abundance in peridotite (an ultramafic rock with a vibrant green colour) is almost entirely controlled by volumetrically insignificant sulfides and metal alloys (eg. König et al. 2015).

Trace elements can also change the way that mantle minerals flow under applied stress. Although many mantle silicates, like olivine and ringwoodite are nominally anhydrous (ie. dry) they can host water (actually H, in the form of OH- groups) at low to moderate concentrations under mantle conditions. However, even at these low concentrations, water has been shown to enhance diffusion, change deformation fabrics, decrease the shear modulus at low frequency (Fig. 1.), and increase strain rates. This last effect, commonly referred to as ‘water weakening,’ increases mantle flow rates in some parts of the Earth’s interior, such as the mantle wedges above subduction zones.

Fig. 1. The shear modulus, a measured of the resistance of a material to applied shear stress, for hydrous (filled) and dry (empty) dunite (a type of ultramafic rock). Reproduced from Aizawa et al. (2008)

How does this happen? There are two primary mechanisms by which water may increase slip rates. Hydrogen ions are small, and so the diffusivity of hydrous defects is comparable to that of vacancies. As such, the presence of hydrous defects increases the effective vacancy concentration, and hence the rate at which slow moving cations (like Si) can diffuse through the lattice. In the dislocation creep regime, strain rates are controlled by dislocation climb, except at high stress (higher than usually found in the mantle), where dislocation glide predominates. This process occurs through the emission/adsorption of defects to dislocation line, and so it is rate-limited by the diffusivity of the slowest species, usually Si. By increasing the diffusivity of these species, water can thus increase the rate at which these linear defects move and multiply.

Fig. 2. The (poorly named) titano-clinohumite defect in olivine.

This interpretation has recently been called into question by Fei et al. (2013), who found little effect of water on Si diffusivity in olivine, which suggests that water should not appreciably increase strain rates, even in the dislocation creep regime. However, the water weakening effect is well attested in natural samples. This apparent contradiction may be resolved by appealing to extrinsic defects (ie. chemical impurities). One example is the so-called ‘titano-clinohumite’ defect in olivine (Fig. 2), which consists of Ti4+ ion substituting for an Mg2+ or Fe2+ ion, charge balanced by two H+ ions in a nearby Si vacancy (the name is misleading; this defect bears little resemblance to clinohumite). This is perhaps the most extreme illustration of the central thesis of this post: one trace element  (Ti) can determine the substitution mechanism of another (H), which in turn affects the dynamics of the Earth’s upper mantle!


[1] Watson EB, Capobianco CJ (1981) Phosphorus and the rare earth elements in felsic magmas: an assessment of the role of apatite. Geochimica et Cosmochimica Acta 45:2349–2358. doi: 10.1016/0016-7037(81)90088-0
[2] König S, Lissner M, Lorand J-P, et al (2015) Mineralogical control of selenium, tellurium and highly siderophile elements in the Earth’s mantle: Evidence from mineral separates of ultra-depleted mantle residues. Chemical Geology 396:16–24. doi: 10.1016/j.chemgeo.2014.12.015
[3] Aizawa Y, Barnhoorn A, Faul UH, et al (2008) Seismic Properties of Anita Bay Dunite: an Exploratory Study of the Influence of Water. J Petrology 49:841–855. doi: 10.1093/petrology/egn007
[4] Fei H, Wiedenbeck M, Yamazaki D, Katsura T (2013) Small effect of water on upper-mantle rheology based on silicon self-diffusion coefficients. Nature 498:213–215. doi: 10.1038/nature12193