Trace Elements Note magnitude of major element changes wt % Element Distribution...as a tool in the interpretation of the history of igneous rocks Different elements have diff. affinities for environments to reside: Si later melts - Mg early xls. Remember from Chapter 1: A. Goldschmidt: some elements metals "Siderophile" Fe, Pt, Mo some elements sulfides "Chalcophile" S,Cu,Zn some elements silicates "Lithophile" Si,K,Ca,REE SIMPLISTIC Xl Field Theory best results. Study of elec. envir. in lattice & melt... Trace Elements: very low conc. TE's don't govern the appearance of a phase (as K req. Ksp or Bi), but enter various phases by substitution. Compare Harkers Major E usually vary by < 101 Figure 8-2. Harker variation diagram for 310 analyzed volcanic rocks from Crater Lake (Mt. Mazama), Oregon Cascades. Data compiled by Rick Conrey (personal communication). From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Trace Elements Note magnitude of trace element changes ppm ppm TE often vary by > 103 very useful since so sensitive to distr. & fractionation ppm Figure 9-1. Harker Diagram for Crater Lake. From data compiled by Rick Conrey. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Element Distribution Goldschmidt’s rules (simplistic, but useful) 1. Two ions with the same valence and radius should exchange easily and enter a solid solution in amounts equal to their overall proportions How does Rb behave? Ni? therefore some TE's will follow similar major E Periodic Table is next slide

Rb follows K & conc. in Ksp, mica, & late melt Ni follows Mg & conc in olivine

Goldschmidt’s rules 2. If two ions have a similar radius and the same valence: the smaller ion is preferentially incorporated into the solid over the liquid smaller ion preferentially -> solid (Mg is smaller than Fe so more Mg in Ol than in melt) Fig. 6-10. Isobaric T-X phase diagram at atmospheric pressure After Bowen and Shairer (1932), Amer. J. Sci. 5th Ser., 24, 177-213. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Relative ionic radii for common valences and coordination numbers

Ionic charge vs. radius Preference for melt Preference for mineral phase Plot of ionic radius vs. ionic charge for trace elements of geological interest. Ionic radii are quoted for eight-fold coordination to allow for comparison between elements. From Rollinson (1993).

3. If two ions have a similar radius, but different valence: the ion with the higher charge is preferentially incorporated into the solid over the liquid (Cr+3 and Ti+4 are always preferred in solids/liquids)

Chemical Fractionation The uneven distribution of an ion between two competing (equilibrium) phases

K = equilibrium constant Exchange equilibrium of a component i between two phases (solid and liquid) i (liquid) = i (solid) KD = = K = equilibrium constant a solid a liquid  X solid  X liquid i i i i i i

Thus if XNi in the system doubles the XNi in all phases will double Trace element concentrations are in the Henry’s Law region of concentration, so their activity varies in direct relation to their concentration in the system, where [a] = (c) Thus if XNi in the system doubles the XNi in all phases will double This does not mean that XNi in all phases is the same, since trace elements do fractionate. Rather the XNi within each phase will vary in proportion to the system concentration For example: suppose C(Ni) = 20 ppm in a system C(Ni) in olivine may be 100 ppm C(Ni) in plagioclase may be 1 ppm C(Ni) in liquid may be 10 ppm Double C(Ni) in system to 40 ppm: Ol -> 200 ppm, Plag -> 2 ppm and liquid -> 20 ppm

incompatible elements are concentrated in the melt (KD or D) « 1 compatible elements are concentrated in the solid KD or D » 1 where D is the partition coefficient for any given trace element between phases; D is a constant for dilute concentrations of elements

For dilute solutions can substitute D for KD: D = Where CS = the concentration of some element in the solid phase CS CL Since the concentration of a trace element (unlike major elements--note Fo-Fa) in any phase is proportional to the overall concentration of that element, a more convenient constant is commonly used for them. It is commonly referred to as "D" for TE's (although some authors still use KD) and is called a partition coefficient:   D = C(S)/C(L) Where CS and CL are the concentration of a trace element in the solid and liquid phases, respectively (the wt% or ppm value directly). As long as the element occurs in very dilute concentrations, D is a constant.

Incompatible elements commonly  two subgroups Smaller, highly charged high field strength (HFS) elements (REE, Th, U, Ce, Pb4+, Zr, Hf, Ti, Nb, Ta) Low field strength large ion lithophile (LIL) elements (K, Rb, Cs, Ba, Pb2+, Sr, Eu2+) are more mobile, particularly if a fluid phase is involved Depends on the minerals involved! Sr -> melt as ol & px separate -> plag (Ca) & not melt if plag is phenocryst phase Commonly standardized to mantle compositions (olivine, pyroxenes, and perhaps garnet) Thus the major elements Mg and Fe would usually be referred to as compatible, while K and Na as incompatible

High field strength (HFS) elements Smaller, highly charged

Large Ion Lithophiles (LILs) Low field strength (large ions, lower charge), more mobile Rb follows K & conc. in Ksp, mica, & late melt Ni follows Mg & conc in olivine

Compatibility depends on minerals and melts involved. Which are incompatible? Why? Not exact, since D varies with the composition of mins & melt

For a rock, determine the bulk distribution coefficient D for an element by calculating the contribution for each mineral Di =  WA DiA WA = weight % of mineral A in the rock Di = partition coefficient of element i in mineral A A

Note that 85% Ol + Opx, but 5% Grt raises bulk D to 0.366 Example: hypothetical garnet lherzolite = 60% olivine, 25% orthopyroxene, 10% clinopyroxene, and 5% garnet (all by weight), using the data in Table 9-1, is: DEr = (0.6 · 0.026) + (0.25 · 0.23) + (0.10 · 0.583) + (0.05 · 4.7) = 0.366

Trace elements strongly partitioned into a single mineral Ni - olivine = 14 Figure 9-1a. Ni Harker Diagram for Crater Lake. From data compiled by Rick Conrey. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Incompatible trace elements concentrate  liquid Reflect the proportion of liquid at a given state of crystallization or melting Figure 9-1b. Zr Harker Diagram for Crater Lake. From data compiled by Rick Conrey. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Trace Element Behavior The concentration of a major element in a phase is usually buffered by the system, so that it varies little in a phase as the system composition changes At a given T we could vary Xmelt from 20  60 % Mg/Fe without changing the composition of the melt or the olivine

Trace element concentrations are in the Henry’s Law region of concentration, so their activity varies in direct relation to their concentration in the system

Trace element concentrations are in the Henry’s Law region of concentration, so their activity varies in direct relation to their concentration in the system Thus if XNi in the system doubles the XNi in all phases will double

Trace element concentrations are in the Henry’s Law region of concentration, so their activity varies in direct relation to their concentration in the system Thus if XNi in the system doubles the XNi in all phases will double Because of this, the ratios of trace elements are often superior to the concentration of a single element in identifying the role of a specific mineral

K/Rb often used  the importance of amphibole in a source rock K & Rb behave very similarly, so K/Rb should be ~ constant If amphibole, almost all K and Rb reside in it Amphibole has a D of about 1.0 for K and 0.3 for Rb K/Rb for amphibole: Usually behave similarly, so ~ constant ratio Unless amphibole which has a D of about 1.0 for K and 0.3 for Rb almost all K and Rb reside in it Melt of amphibole-bearing rock will -> decrease K/Rb in the partial melt Other factors being equal, a magma produced by partial melting of an amphibole-bearing source rock would have a lower K/Rb than one derived from amphibole-free source High absolute K or Rb could also  an amphibole-bearing source, but may result from other causes (high phlogopite, or an alkali-enriched fluid) The ratio is more indicative of amphibole due to the different D values Fractional crystallization of amphibole would also -> low K/Rb ratio in the evolved liquid

Sr and Ba (also incompatible elements) Sr is excluded from most common minerals except plagioclase Ba similarly excluded except in alkali feldspar Ba/Sr will help identify Kspar vs Plag Both are incompatible -> first partial melts (or residual liquids of FX) Effect depends on the mineral phases involved Sr is excluded from most common minerals except plagioclase Ba similarly excluded except in alkali feldspar Thus the ratio Ba/Sr increases with crystallization of plagioclase, but may decrease when orthoclase begins to crystallize

Compatible example: Ni strongly fractionated  olivine > pyroxene Cr and Sc  pyroxenes » olivine Ni/Cr or Ni/Sc can distinguish the effects of olivine and augite in a partial melt or a suite of rocks produced by fractional crystallization In all of the above cases using ratios, the idea is to find a mineral with a unique pair of elements for which it alone has a relatively high value of D for one element and a relatively low value of D for the other. The ratio of these elements is then sensitive only to liquid/crystal fractionation associated with that particular mineral

Models of Magma Evolution Batch Melting The melt remains resident until at some point it is released and moves upward Equilibrium melting process with variable % melting

Models of Magma Evolution Batch Melting CL = trace element concentration in the liquid CO = trace element concentration in the original rock before melting began F = wt fraction of melt produced = melt/(melt + rock) C 1 Di (1 F) F L O = - +

A plot of CL/CO vs. F for various values of Di using the Batch Melting A plot of CL/CO vs. F for various values of Di using the previous equation Di = 1.0 D = 1.0 No fractionation so CL/CO = 1 for all values of F Figure 9-2. Variation in the relative concentration of a trace element in a liquid vs. source rock as a fiunction of D and the fraction melted, using equation (9-5) for equilibrium batch melting. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Di » 1.0 (compatible element) Very low concentration in melt Especially for low % melting (low F) Values of F > 0.4 unlikely for batch melting since greater amounts should separate and rise Figure 9-2. Variation in the relative concentration of a trace element in a liquid vs. source rock as a fiunction of D and the fraction melted, using equation (9-5) for equilibrium batch melting. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Highly incompatible elements • Greatly concentrated in the initial small fraction of melt produced by partial melting • Subsequently diluted as F increases Highly incompatible elements are greatly concentrated in the initial small fraction of melt that is produced by partial melting, and subsequently get diluted as F increases Figure 9-2. Variation in the relative concentration of a trace element in a liquid vs. source rock as a fiunction of D and the fraction melted, using equation (9-5) for equilibrium batch melting. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

As F  1 the concentration of every trace element in the liquid = the source rock (CL/CO 1) Di (1 F) F L O = - + All -> 1.0 because all of the source is melted Figure 9-2. Variation in the relative concentration of a trace element in a liquid vs. source rock as a fiunction of D and the fraction melted, using equation (9-5) for equilibrium batch melting. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

C 1 Di (1 F) F L O = - + As F  0 CL/CO  1/Di If we know CL of a magma derived by a small degree of batch melting, and we know Di we can estimate the concentration of that element in the source region (CO) If know (CL) for magma derived by a small degree of batch melting, and we know D, we can estimate the concentration of that element in the source region (CO). This can provide very valuable information in constraining and characterizing the source region. Figure 9-2. Variation in the relative concentration of a trace element in a liquid vs. source rock as a fiunction of D and the fraction melted, using equation (9-5) for equilibrium batch melting. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

For very incompatible elements as Di  0 reduces to: 1 Di (1 F) F L O = - + C 1 F L O = If we know the concentration of a very incompatible element in both a magma and the source rock, we can determine the fraction of partial melt produced

Worked Example of Batch Melting: Rb and Sr Basalt with the mode: 1. Convert to weight % minerals (Wol Wcpx etc.) Table 9-2 . Conversion from mode to weight percent Mineral Mode Density Wt prop Wt% ol 15 3.6 54 0.18 cpx 33 3.4 112.2 0.37 plag 51 2.7 137.7 0.45 Sum 303.9 1.00 Suppose that a source rock with a mode of 51% plagioclase, 33% clinopyroxene, and 18% olivine undergoes batch melting

Worked Example of Batch Melting: Rb and Sr Basalt with the mode: 1. Convert to weight % minerals (Wol Wcpx etc.) 2. Use: Di =  WA Di and the table of D values for Rb and Sr in each mineral to calculate the bulk distribution coefficients: DRb = 0.045 and DSr = 0.848 Table 9-2 . Conversion from mode to weight percent Mineral Mode Density Wt prop Wt% ol 15 3.6 54 0.18 cpx 33 3.4 112.2 0.37 plag 51 2.7 137.7 0.45 Sum 303.9 1.00 Rb is incompatible and Sr only slightly so, but near unity

3. Use the batch melting equation to calculate CL/CO 3. Use the batch melting equation to calculate CL/CO for various values of F Table 9-3 . Batch Fractionation Model for Rb and Sr C L /C O = 1/(D(1-F)+F) D Rb Sr F 0.045 0.848 Rb/Sr 0.05 9.35 1.14 8.19 0.1 6.49 1.13 5.73 0.15 4.98 1.12 4.43 0.2 4.03 3.61 0.3 2.92 1.10 2.66 0.4 2.29 1.08 2.11 0.5 1.89 1.07 1.76 0.6 1.60 1.05 1.52 0.7 1.39 1.04 1.34 0.8 1.23 1.03 1.20 0.9 1.01 1.09 From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

4. Plot CL/CO vs. F for each element Figure 9-3. Change in the concentration of Rb and Sr in the melt derived by progressive batch melting of a basaltic rock consisting of plagioclase, augite, and olivine. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall. Results: Incompatible element Rb (no K minerals) strongly concentrated in the early small melt proportions (low F) Thus  a sensitive measure of the progress of fractional crystallization (at least until rock half melted) As melting proceeds, the incompatible element is gradually diluted by more compatible ones Since D(Sr) is close to 1.0, the ratio Rb/Sr vs. F is nearly the same as Rb alone Any ratio of incompatible to compatible element should then be sensitive to the degree of partial melting (at least in the initial stages). Important for Rb/Sr isotopic systems Note that can create a series of melts from a single source each with diff Rb/Sr

Incremental Batch Melting Calculate batch melting for successive batches (same equation) Must recalculate Di as solids change as minerals are selectively melted (computer)

Fractional Crystallization 1. Crystals remain in equilibrium with each melt increment Incremental batch melting in reverse Equation 9-5 would still apply Batch FX: F = proportion of liquid remaining

Rayleigh fractionation The other extreme: separation of each crystal as it formed = perfectly continuous fractional crystallization in a magma chamber Rayleigh: Crystals form and accumulate -> isolated from reaction with the remaining liquid

Rayleigh fractionation The other extreme: separation of each crystal as it formed = perfectly continuous fractional crystallization in a magma chamber Concentration of some element in the residual liquid, CL is modeled by the Rayleigh equation: CL/CO = F (D -1) Rayleigh Fractionation Can also apply the Rayleigh equation to Rayleigh fractional melting

Other models are used to analyze Mixing of magmas Wall-rock assimilation Zone refining Combinations of processes

The Rare Earth Elements (REE) Members of Group IIIA of the Periodic Table La -> to Lutetium (Z = 57  71) All have very similar chemical and physical properties -> behave as a coherent series All have a 3+ oxidation state Ionic radius decreases steadily with increasing atomic number (lanthanide contraction)

Contrasts and similarities in the D values: All are incompatible Also Note: HREE are less incompatible Especially in garnet Eu can  2+ which conc. in plagioclase

La Ce Nd Sm Eu Tb Er Dy Yb Lu REE Diagrams Plots of concentration as the ordinate (y-axis) against increasing atomic number Degree of compatibility increases from left to right across the diagram Concentration La Ce Nd Sm Eu Tb Er Dy Yb Lu

estimates of primordial mantle REE chondrite meteorite concentrations Chondrite normalization: Normalize to chondrite meteorites to: 1) avoid Oddo-Harkins (even Z more common) 2) think chondrite = primitive earth, so can compare to initial distribution Eliminate Oddo-Harkins effect and make y-scale more functional by normalizing to a standard estimates of primordial mantle REE chondrite meteorite concentrations

What would an REE diagram look like for an analysis of a chondrite meteorite? 0.00 2.00 4.00 6.00 8.00 10.00 56 58 60 62 64 66 68 70 72 sample/chondrite L La Ce Nd Sm Eu Tb Er Yb Lu ?

Divide each element in analysis by the concentration in a chondrite standard 0.00 2.00 4.00 6.00 8.00 10.00 56 58 60 62 64 66 68 70 72 sample/chondrite L La Ce Nd Sm Eu Tb Er Yb Lu

REE diagrams using batch melting model of a garnet lherzolite for various values of F: Diagram created in similar way to Rb-Sr diagram in Example Problem Calculate D(La) for lherzolite and model conc. in melt at F = 0.1 -> point on this diagram for La … LREE are less compatible than HREE, so melts enriched in LREE -> (-) slopes Slope is steeper for low values of F (low % partial melting) As F  1 slope  0 at S/C = 1.0 since all of mantle sample is melted Note again the use of RATIOS to -> slope on REE La/Lu ratio -> REE slope Tb/Lu for HREE only (garnet) La/Sm -> LREE only Figure 9-4. Rare Earth concentrations (normalized to chondrite) for melts produced at various values of F via melting of a hypothetical garnet lherzolite using the batch melting model (equation 9-5). From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Europium anomaly when plagioclase is a fractionating phenocryst or a residual solid in source Eu* is the value Eu “should” have if Eu+2 did not -> plagioclase Another example of how RATIOS can help Eu alone is inconclusive (low REE of low Eu) Sm/Eu is slope or Eu anomaly trough (Use Eu*/Eu anyway) Figure 9-5. REE diagram for 10% batch melting of a hypothetical lherzolite with 20% plagioclase, resulting in a pronounced negative Europium anomaly. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Spider Diagrams An extension of the normalized REE technique to a broader spectrum of elements Chondrite-normalized spider diagrams are commonly organized by (the author’s estimate) of increasing incompatibility L  R Different estimates  different ordering (poor standardization) Order of elements based on estimates of increasing incompatibility from right to left in a "typical" mantle undergoing partial melting Elements are all incompatible (D<1) during most partial melting and fractional crystallization processes. The main exceptions are Sr, which may be compatible if plagioclase is involved, Y and Yb with garnet Ti with magnetite Troughs at these elements would indicate respective mineral involvement Oceanic basalts = large degrees of PM, their spider diagrams should reflect the trace element patterns of their source Less incompatible elements on the right-hand side should be less enriched during PM (particularly for small degrees of it), tilting the curve up on the left -> (-) slope Additionally, FX subsequent to magma segregation from the source should tip the pattern even further Fig. 9-6. Spider diagram for an alkaline basalt from Gough Island, southern Atlantic. After Sun and MacDonough (1989). In A. D. Saunders and M. J. Norry (eds.), Magmatism in the Ocean Basins. Geol. Soc. London Spec. Publ., 42. pp. 313-345.

MORB-normalized Spider Separates LIL and HFS MORB-normalization LIL on left and HFS on right. Incr. incompatibility toward center. Coherent magma with enrichment has hump shape Figure 9-7. Ocean island basalt plotted on a mid-ocean ridge basalt (MORB) normalized spider diagram of the type used by Pearce (1983). Data from Sun and McDonough (1989). From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Application of Trace Elements to Igneous Systems 1. Use like major elements on variation diagrams to document FX, assimilation, etc. in a suite of rocks More sensitive  larger variations as process continues Figure 9-1a. Ni Harker Diagram for Crater Lake. From data compiled by Rick Conrey. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

2. Identification of the source rock or a particular mineral involved in either partial melting or fractional crystallization processes Example: can use REE to distinguish between high pressure and low pressure sources of a mantle-derived magma In the deep continental crust, and at depths over about 100 km in the mantle, garnet and clinopyroxene are important phases, which remain as residual solids during the generation of up to 15-20% partial melting

Garnet concentrates the HREE and fractionates among them Thus if garnet is in equilibrium with the partial melt (a residual phase in the source left behind) expect a steep (-) slope in REE and HREE Shallow (< 40 km) partial melting of the mantle will have plagioclase in the resuduum and a Eu anomaly will result

Garnet and Plagioclase effect on HREE 0.00 2.00 4.00 6.00 8.00 10.00 56 58 60 62 64 66 68 70 72 sample/chondrite La Ce Nd Sm Eu Tb Er Yb Lu 67% Ol 17% Opx 17% Cpx Garnet and Plagioclase effect on HREE 0.00 2.00 4.00 6.00 8.00 10.00 sample/chondrite 60% Ol 15% Opx 15% Cpx 10%Plag La Ce Nd Sm Eu Tb Er Yb Lu 0.00 2.00 4.00 6.00 8.00 10.00 56 58 60 62 64 66 68 70 72 sample/chondrite La Ce Nd Sm Eu Tb Er Yb Lu 57% Ol 14% Opx 14% Cpx 14% Grt

Trace elements have much broader general application in petrology than indicating individual mineral involvement They have been used to indicate mama mixing, assimilation, the degree of partial melting of fractional crystallization, etc. We will see numerous examples in forthcoming chapters Figure 9-3. Change in the concentration of Rb and Sr in the melt derived by progressive batch melting of a basaltic rock consisting of plagioclase, augite, and olivine. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Use as a petrogenetic indicator Table 9-6 A brief summary of some particularly useful trace elements in igneous petrology Element Use as a petrogenetic indicator Ni, Co, Cr Highly compatible elements. Ni (and Co) are concentrated in olivine, and Cr in spinel and clinopyroxene. High concentrations indicate a mantle source. V, Ti Both show strong fractionation into Fe-Ti oxides (ilmenite or titanomagnetite). If they behave differently, Ti probably fractionates into an accessory phase, such as sphene or rutile. Zr, Hf Very incompatible elements that do not substitute into major silicate phases (although they may replace Ti in sphene or rutile). Ba, Rb Incompatible element that substitutes for K in K-feldspar, micas, or hornblende. Rb substitutes less readily in hornblende than K-spar and micas, such that the K/Ba ratio may distinguish these phases. Sr Substitutes for Ca in plagioclase (but not in pyroxene), and, to a lesser extent, for K in K- feldspar. Behaves as a compatible element at low pressure where plagioclase forms early, but as an incompatible at higher pressure where plagioclase is no longer stable. REE Garnet accommodates the HREE more than the LREE, and orthopyroxene and hornblende do so to a lesser degree. Sphene and plagioclase accommodates more LREE. Eu 2+ is strongly partitioned into plagioclase. Y Commonly incompatible (like HREE). Strongly partitioned into garnet and amphibole. Sphene and apatite also concentrate Y, so the presence of these as accessories could have a significant effect. Table 9-6. After Green (1980). Tectonophys., 63, 367-385. From Winter (2001) An Introduction to Igneous and Metamorphic Petrology. Prentice Hall.

Trace elements as a tool to determine paleotectonic environment Useful for rocks in mobile belts that are no longer recognizably in their original setting Can trace elements be discriminators of igneous environment? Approach is empirical on modern occurrences Concentrate on elements that are immobile during low/medium grade metamorphism A bit ambiguous, since many variables (country rx, %PM, %FX)   Need immobile TE's, so met has no effect. Ti, Zr, Y, Hf & apply to basic volc's to eliminate high %FX

Figure 9-8. (a) after Pearce and Cann (1973), Earth Planet, Sci. Lett Figure 9-8. (a) after Pearce and Cann (1973), Earth Planet, Sci. Lett., 19, 290-300. (b) after Pearce (1982) in Thorpe (ed.), Andesites: Orogenic andesites and related rocks. Wiley. Chichester. pp. 525-548, Coish et al. (1986), Amer. J. Sci., 286, 1-28. (c) after Mullen (1983), Earth Planet. Sci. Lett., 62, 53-62.