The chemical compositions of igneous racks that form by these processes are generally wot explainable as simple mixtures of two components because of the re ‘moval of cumulate minerals that form during fractional rystallization of hybrid magmas. The isotope compo sitions of Sr, Nd, Pb, et, of hybrid magmas are not ap- preclably changed by crystallization of minerals because the differences in their isotopic masses are small, be- «ause isotope fractionation effects decrease with increas ing temperatures,and becanse-all measured isotope ra tios are routinely corrected for isotope fractionation tegardless ofits origin (Sect, 1.2.3). Even the isotope cam. Position of © of hybrid magmas is primarily determined by the differences in the 60 values of the components and by the proportions of mixing rather than by subse quent fractional crystallization (Taylor 1980). The iso- tope composition of O in thehighly differentiated gabbra intrusion at Skaergaard in East Greenland (Sect. 5.2.6) twas investigated and interpreted by Taylor and Epstein (0963), Taylor and Forester (1979),and Norton and Taylor (1979), ee 1.10 Fractional Crystallization of Magma When minerals crystallize from a cooling magma, the concentrations of Rb and Sr of the remaining residual liquid change depending on whether the minerals ac cept or exclude these elements, The transfer of Rb-and St from sificate melt into the minerals that are crys tallizing from it can be described by means of distribu tion costiicients combined with the Rayleigh ecuation. a 1.10.1 Distribution Coefficients ‘Trace elements may enter a mineral crystallizing from a silicate met by substituting for the ion of a major ele ment. The extent of substitution depends on many fac tors pertaining 10 the ions, theenvironmental conditions, and the properties of the site in the crystal lattice. In ‘general, these tactors combine to give rise to three pos- sible scenarios: L. Theions ofa trace element may be captured by grove ing crystals in preference 0 the ions of the major el ements Crystal may camoufage itto enter without discrimination; } Atraceclement may be excluded andis adiitted only toa limited exten trace clement by allowing Goldschmidt (1937) originally attempted to predict the replacement of ions in crystals solely on the basts of their radii and chatyes. However, the resulting "rules" ere only qualitative at best and were violated in many 1.10 = Fractional Crystallization of Magma cases (Shaw 195% Ringwood 1955; Burns and Fyfe 1967; Philpotts 1978). Consequently, the partitioning of trace elements between crystals and silicate liquids is ex- pressed more effectively by means of the distribution coefficient (D) defined by che relation: (139) where sand I stun for solid and liquid, respectively, Cis the concentration of any trace element and the replace- ‘ment reaction is assumed to be at equilibrium. Distr. bution coefficients are determined either by analyzing phenocrysts and matrix in porphyritic volcanic rocks or by analyzing ceystals grown from melts in the labo ratory Numerical values of distribution coettcients of Rb* and Sr** and many other trace elements in the common rock-forming minerals hate been compiled by Irving (0978), allegre and Hart (2978), Cox etal. (1979),Hender son (1982), Wilson (1989), Foley andl Van der Laan (1 and other recently published textbooks in igneous pe trology Trace elements having D> 1.0 ate sid to be compar ible whereas thosehawing D elements which are compatible in the crystals forming from a magma are enriched in the solids and depleted in the residual liq ments tend to be concentrated into the temaining liq- uid Rubidium is an incompatible trace element in all ma- Jor rock-forming minerals. Consequently, itis progres sively concentrated into the residual Liquid during frac tional erystallization of magma. Late forming crystals of certain minerals (e.g, mica and feldspar) therefore have higher Rb concentrations than carly-formed crys tals of the same minerals Strontium is compatible in K-Fekispan, plagioclase, apatite; and sphene but incompatible in micas, most amphiboles, pyroxenes, and olivine, Therefore, the Sr concentration of residual magmas increases whoh lv ineor pyroxene are the only crystallizing minerals. When plagioclase begins to crystallize, it takes up Sr from the ‘magma and causesits concentration inthe residual lig uid to decrease The measured values ofthe distribution coefficients of RbandSr in some minerals listed in Table 1.10 vary by a factor of ten or more because their magnitudes depend not only’on the properties of the ons and on the charac- ter of the lattice sitesi the erystals,but aso on thechemi- sal composition of the melt, s well as on the tempe \ureand pressure. The chemical composition ofthe melt becomes the dominant factor in silica-rch magmas b sause of the increasing polymerization of S-AL-O units Consequently, such liquids reject trace elements, caus ing their crystalliquid distribution coefficients to in- hore iacompatible. Trace whereas incompatible tzace eleGuaeienT + Chemical Properties and Isotope Systematics Table 1.10, Average distribution cvelficietts for Rb and Sin minerals crying from magmas of diferen! composition ‘Mineral A Basat and andesite Olivine Orthopytexenie Diopside Clinopyrovene Augite Amiphiboles Homblende Mica Phiogeite Garnet Apatite Whitlokte Peroussite 8. Dache ond rhyolite Onhopyroxene Hypersthene Hornblende Plagioclase alka fs Guan Gamet Biotite Muscovite Chat rocks OWine Onhopyoxene Glinopyroxene Magnette Amphibole Hotnbende Koersutte Plagiocsse Alka fl Sanigine Anorthoctase ‘Metite(Ak 121090) Bitte Apatite D.Unaiferentianed Pagiociase efelsspar popiroxene Onthoproxene Micas Hombiende DiRb) ono79 06: 01s 031 ons 031 32 308 00 009 ‘gae7 p14 970 oot 0087 32) 154 ots 0.922 ons 3s 19 a6 021 015: fens: 1 Art and Hanae 6 LeMachand e: 0111987), 7 1983), 18 Dupuy (1972), 19.5 1 1980 26, Fae Range Bish Range (00179-0011 amr?) oacorat ~ one 0022-0061 apiz p06 a0 oat 0078 - a2 oo — 03 a3 99s, — 060 0,12 0027-91 ops ats = 100 029-033 oa7 ode 04g ‘90202 - 050 199 Me — 306 02) 098 012 505 37 = 68 100()) 0734 0005-029, 20%0 009 — 005 oes 932-000 D516 oa 24-046, 8. Is -40 a1) = 104 545 27-6 oo -o016 = 008s -0909 omy 1s ~ oe 24-53 pat 098 = 053, 0308 02 - 008 012 gs” = ope 0017 oot - vos oad 08 a ous - 47 059 oss a0 009-014 os - 038 5 16-028 is ag 28 on - 018 49 28-30 oss a2 12 19 12 096 03 0. poze - 0188 2 0129 0.166 00187 ~ 0516 oor 017? 90108. — 028 0936 0231 opi? - 0872 os 045s Ole — sat ‘ons 900839 90113 90138000037 ~ onie! 1962), teu et aL (V9. vg and Pe vote, (1989, Shimizu (1978), 10. Sunt vid We (1925), 15 vi (D878, 16, Sta and Ceecratt 1990), 20 Long (1998) 21 Mahone: andl Hic 13), sist Neue (1984) 25 Betin and. Henderson 178), 29 Phdgowsand Schroer (970 24 Manend and sti. Reference” 1,2.3:28 123, 78 2.3,5,9,10,11,12,28 1 ae un 12,10,41,13 14.26 5 12 1 8 2 1 i 1 1.2.16 17.18.19 132.17,20, 16 12 1.2:16.17,21 : 46, 6 4.6.23,28 623,28 2 62428 6.24.25 24,25,26 ar 423 aa 1 2 > 6 of) 11 Dostal et 1985). 17 Mitlefeat anc Mi Leeman and Phelps