Trace Element Geochemistry Lecture 24 Geochemical Classification The Rare Earth Elements • • • • • The rare earths and Y are strongly electropositive. As a result, they form predominantly ionic bonds, and behave as hard charged spheres. The lanthanide rare earths are in the +3 valence state over a wide range of oxygen fugacities. In the transition metals, the s orbital of the outermost shell is filled before filling of lower electron shells is complete so the configuration of the valence electrons is similar in all the rare earth, hence all exhibit similar chemical behavior. Ionic radius, which decreases progressively from La3to Lu3+ (93 pm) governs their relative behavior. Because of their high charge and large radii, the rare earths are incompatible elements. o o The degree of incompatibility of the lanthanides varies with atomic number. Highly charged U and Th are highly incompatible elements, as are the lightest rare earths. However, the heavy rare earths have sufficiently small radii that they can be accommodated to some degree in many common minerals such as Lu for Al in garnet. Eu2+ can substitute for Ca in plagioclase. Rare Earth Diagrams • The systematic variation in lathanide rare earth behavior is best illustrated by plotting the log of the relative abundances as a function of atomic. Relative abundances are calculated by dividing the concentration of each rare earth by its concentration in a set of normalizing values, such as the concentrations of rare earths in chondritic meteorites. o • • Rare earths are also refractory elements, so that their relative abundances are the same in most primitive meteorites - and presumably (to a first approximation) in the Earth. Why do we use relative abundances? The abundances of even-numbered elements in the solar system are greater than those of neighboring odd-numbered elements and abundances generally decrease with increasing atomic number, leading to a saw-toothed abundance pattern. Normalizing eliminates this. Abundances in chondritic meteorites are generally used for normalization. However, other normalizations are possible: sediments (and waters) are often normalized to average shale. Partition Coefficients The Partition Coefficient • Geochemists find it convenient to define a partition or distribution coefficient of element i between phases α and β: Dia - b Cia = b Ci • Where one phase is a liquid, the convention is the solid is placed on top: s s-ℓ i D Ci = ℓ Ci • Incompatible elements are those with Ds/l ≪ 1. Compatible elements are those with Ds/l ≥ 1. These terms refer to partitioning between silicate melts and phases common to mantle rocks (peridotite). It is this phase assemblage that dictates whether lithophile trace elements are concentrated in the Earth’s crust, hence the significance of these terms. Thermodynamic Basis • The chemical potentials of element i in phases α and β are µia = µia o + RT ln lia Xia • At equilibrium: µib = µib o + RT ln lib Xib µia = µib lia Xia µi - µi = RT ln b b li X i ao o • bo Since trace elements obey Henry’s Law, we can replace the activity coefficient with h. The left hand side is ∆G˚, so that Xia hib -∆ G˚/ RT = ae b Xi hi • • • hib -∆ G˚/ RT Di = C a e hi where C is simply a constant converting concentration units (usually ppm) to mole fraction. Thus the distribution coefficient is a kind of equilibrium constant. and a -b Relationship among distribution coefficients • In a system with three phases, α, β, and γ, if α and β are in equilibrium and α and γ are in equilibrium, then β and γ must also be in equilibrium. It follows that: Dα-β=Dα-γ/Dα-β • This relationship has practical use. For example, if we can determine the partition coefficient for an element between pyroxene and melt and between garnet and pyroxene, we can then calculate the garnet–melt partition coefficient for this element. Temperature and Pressure Dependence • • • • • • In ideal solutions, the temperature dependence of the partition coefficient is the same as that of the equilibrium constant: a -b -∆ G˚/ RT i ∆G can be expanded into entropy and enthalpy terms: D =e ∆ H˚+(P - P˚)DV æ ¶ln Di ö çè ÷ø = ¶T P RT 2 In ideal solution, and assuming again that ∆V is independent of temperature and pressure, the pressure dependence is also the same as that of the equilibrium constant: -∆ V æ ¶ln Di ö çè ÷ø = ¶P T RT We would predict a strong pressure dependence when the ionic radius of an element differs greatly from that of the available crystal lattice site. Thus, for example, we would predict the partition coefficient for K between pyroxene and melt would be strongly pressure-dependent since the ionic radius of K is 150 pm and is much larger that the size of the M2 site in clinopyroxene, which is normally occupied by Ca, with a radius of about 100 pm. Conversely, where the size difference is small (e.g., Mn (83 pm) substituting for Fe (78 pm)), we would expect the pressure dependence to be smaller. In non-ideal solutions, the T and P dependencies will be more complex because the activity coefficients (the Henry’s Law coefficients) will also depend on T and P. Bottom line: partition coefficient is temperature and pressure dependent. Importance of Ionic Size and Charge • • • Ionic radius (picometers) vs. ionic charge contoured for clinopyroxene/liquid partition coefficients. Cations normally present in clinopyroxene M1 and M2 sites are Ca2+, Mg2+, and Fe2+, shown by ✱ symbols. Elements whose charge and ionic radius most closely match that of the major elements have the highest partition coefficients Much of the interest in trace elements in igneous processes centers on the elements located in the lower left portion of the periodic table (K, Rb, Cs, Sr and Ba; the rare earths, Y, Zr, Nb, Hf and Ta). One reason for this focus of attention is that these elements are all lithophile and therefore present at relatively high abundance in the Earth’s crust and mantle. There is another reason, however: their chemical behavior is comparatively simple: their behavior in igneous systems is mainly (not entirely) a function of ionic size and charge. The other trace elements that receive the most attention from igneous geochemists are the first transition series elements. Though their electronic structures and bonding behavior are considerably more complex, charge and size are also important. Many of these elements, particularly Ni, Co, and Cr, have partition coefficients greater than 1 in many Mg–Fe silicate minerals. Hence the term “compatible elements” often refers to these elements. Quantitative Treatment • Consider the substitution reaction: Mℓ + CaMgSi2O6 ⇄ Caℓ + MMgSi2O6 • The Gibbs free energy change of this reaction can be expressed as: M -Ca Di ∆ Gr = ∆ Gexchange - ∆ Gmelting • The first term is ∆G for transferring an M2+ ion from the melt to the crystal lattice and simultaneously transferring a Ca2+ ion from the lattice site to the liquid. • The second term is the ∆G associated with the melting of diopside, and governs the distribution of Ca between diopside and the liquid. The distribution coefficient for element M then depends on these two components of free energy: Di/ M D Di æ ∆ Gmelting - ∆ GexM -Ca ö = exp ç ÷ RT è ø Substitution & Lattice Strain • If we substitute an atom that larger than one normally occupying a site in a crystal lattice, for example a Ba2+ for a Mg2+ in olivine, the lattice will strain to accommodate that atom. • Strain is a form of stored energy. • We can express this as? Olivine Lattice Strain Theory • According to the lattice strain energy theory, ∆Gexchange is dominated by the energy associated with the lattice strain resulting from M2+ being a different size than Ca2+. Because the melt (at least at low pressure) has a far less rigid structure and is more compressible than the solid, any strain in the melt is essentially negligible compared with the strain in the solid. In other words: ∆ GexM -Ca @ ∆ Gstrain • The strain energy, ∆Gstrain, may be calculated as: ∆ Gstrain • 1 é r0 ù 2 = 4p EN A ê (rm - r0 ) + (rm - r0 )3 ú 3 ë2 û where r0 is the optimal radius of the lattice site, rM is the ionic radius of M, NA is Avogadro’s Number, and E is Young’s modulus, which is the ratio of stress applied to the resulting strain (change in dimension) and has units of pressure. It is a property of the material and is related to compressibility. Blundy & Wood Model • We can now substitute this expression into Di/ M D • and obtain Di/ DMDi/ = DCa Di æ ∆ Gmelting - ∆ GexM -Ca ö = exp ç ÷ RT è ø 1 ì é r0 2 3ùü -4 p EN (r r ) + (r r ) A m 0 m 0 ïï êë 2 úû ïï æ -∆ Gstrain ö 3 exp ç = exp í ý è RT ÷ø RT ï ï ïî ïþ • where • Where charge is different from the ion normally occupying the site, we much consider the coupled substitution. For example, we can balance charge by substituting both La3+ and Na+ into the Ca2+ site in diopside. In this case, our relationship is: Di/ Ca D =e Di -∆ Gmelt /RT Di La-Ca Na-Ca ∆ Gmelt -∆ Gstrain -∆ Gex RT • Di/ DLa =e The point is we can build a theoretical framework to predict how partitioning will depend on ionic size and charge (and the T and P dependency). Blundy & Wood Model We still depend heavily on experimental determination of partition coefficients. But by experimentally determining a few, we can use the Blundy & Wood model to predict others (as well as pressure dependence). Dependence on Composition Ol-liquid partition coefficient for Zn parameterized as a function of non-bridging oxygens in the melt and T. pyx-liquid partition coefficient for Sm as a function of liquid and solid composition Rare Earth Partition Coefficients