( "heroical Geology, 92 ( 199 [ ) 107-114 107 Elsevier Science Publishers B.V., Amsterdam Modelling the petrogenesis of high Rb/Sr silicic magmas A.N. Halliday a, j.p. Davidson ~"", W. Hildreth b and P. Holden ~"~ ~' D~7~arll~tepll qt Geo/o~ica/ Sciences, The University q/,llichi~an, Inn. lrhor, UI 4S I ()~- I 0~ 3, (.S'I " ('S. Geological SI,'rVE')', MSglO, 345 )diddlefiehl Road, Menlo t'ark. (;,1 94025. ( ,S\-I " D~7~artment o/Earlh and Space Sciences, ~ )zivetwitr q/('alih;rnta. Los ..t n£,eh'~. (' I 90024, ~ ',S'. I (Received February 22, 1990: revised and accepted November 7, 1991)) ABSTRACT Halliday, A.N., Davidson, J.P., Hildreth, W. and Holden, P., 1991. Modelling the petrogenesis o1 high Rh/Sr silicic mag- mas. In: A. Peccerillo (Guest-Editor), Geochemistry of Granitoid Rocks. Chem. Geol., 92: 107-114. Rhyolites can be highly evolved with Sr contents as low as 0.1 ppm and Rb/Sr > 2,0(t0. In contrast, granite batholiths are commonly comprised of rocks with Rb/Sr < l 0 and only rarely > 100. Mass-balance modelling of source compositions, differentiation and contamination using the trace-element geochemistry of granites are therefi)re commonly in error be- cause of the failure to account for evolved differentiates that may have been erupted from the system. Rhyolitic magmas with very low Sr concentrations ( ~< 1 ppm ) cannot be explained by any partial melting models involving typical crustal source compositions. The only plausible mechanism for the production of such rhyolites is Rayleigh fiactional crystalli- zalion involving substantial volumes of cumulates. A variety of methods ['or modelling the differentiation of magmas with extremely high Rb/Sr is discussed. In each case it is concluded that the bulk partition coefficients for Sr have to be largc. In the simplest models, the bulk D sr of the most evolved types is modelled as > 50. Evidence from phenocryst/glass/ whole-rock concentrations supports high Sr partition coefficients in feldspars from high silica rhyolites. However, the low modal abundance of plagioclase commonly observed in such rocks is difficult to reconcile with such simple fractionation models of the observed trace-element trends. In certain cases, this may be because the apparent trace-element trend defined by the suite of cogenelic rhyolites is the product of different batches of magma with separate differentiation histories accumulating in the magma chamber roof zone. 1. Introduction fractionated magmas with less than 1 ppm Sr (0.5 ppm at Yellowstone, 0.3 ppm at It has been known for a long time (e.g. La Primavera and 0.1 ppm at Glass Mountain, Nockolds and Mitchell, 1946) that most SiO2- Long Valley). In this paper we explain these rich igneous rocks have much higher Rb/Sr differentiated magmas in terms of extreme than SiO2-poor igneous rocks and that this ra- Rayleigh fractionation and discuss the con- tio can therefore be used as an index of differ- straints placed thereby on models of rhyolite entiation. This is obviously because Rb is more genesis. incompatible than Sr in nearly all magmatic systems. Accurate isotope dilution measure- 2. Rb and Sr concentration trends ments are now illustrating the remarkable ex- tent of this fractionation. Studies of rhyolitic Rb and Sr concentration data for some highly suites from Yellowstone (Leeman and Phelps, fractionated silicic magmas are shown in Fig. 1981, Hildreth et al., 1991 ), La Primavera 1 together with data for major granitic suites (Mahood and Halliday, 1988 ) and Long Val- from around the world. The individual Cale- ley (Halliday et al., 1989) have revealed highly donian granitic suites shown (from Donegal 0009-2541/91/$03.50 ¢~ 1991 Elsevier Science Publishers B.V. All rights reserved. 108 +~.N. [IH LII)AY ET-\L. .... Yellowstone ,Trawenagh Bay Rosses Complex Main Glass Mountain ..... Rhyolites / Pluton Donegal Rhyolites ~ ~ I ('~ ~ "G4 ~ P9 PI" ~ " °n 2,.e, i I O0 - ,,'< / ' Bishop , '\ "~-...:~.."X ~New ~,~ La Primavere Tuff Peru - ~ , ~.~Brunswick E e-"v Rhyolites Coastal ~ 'k , i Mmnettes Batholith , ;,/[~ ch / A::) Kosciousko ',, ~ /~fi El2 BatholHh ~ln 5 ly~'-^e ' "~ I0 -~' Yei owstone ~,'' ' Ranges I Basalt '/ ,,. a Primaver~j © I B,)S<] It i i ~-'~ :* ,H , ' -t , 1.0 C~l l O 16- ido " I000 Sr p pm Fig. 1. Rb vs, Sr for rhyolites and associated basalts from western North America ( Halliday et al., 1984, 1989: Mahood and Halliday, 1988; Hildreth et el., 1991 ), and granitic suites from around the world (Halliday, unpublished data on Donegal; Hine et el., 1978: Bachinski and Scott, 1979; Atherton et al., 1979: Bateman and Chappell, 1979; Halliday el al., 1980: DePaolo, 1981 b ). The data for Rosses (Donegal) are divided into separate phases ES (early sheets ) and G l to ~;4. HI Sr concentrations < 50 ppm determined by isotope dilution. and S. Scotland) were selected specifically for which Rb and Sr are both incompatible, a pos- their relatively simple quasi-linear trends when itive slope should result in Fig. 1 (e.g., the New plotted in this manner. Certain key points can Brunswick minettes). These conditions would be made from examination of this diagram. be met in the lower crust or upper mantle at (1) In general, the rhyolite suites contain pressure-temperature conditions exceeding the highly fractionated magmas not normally in- stability of plagioclase. In basaltic systems Sr cluded in our granite sampling. There is there- is depleted (in response to plagioclase frac- fore a need for detailed studies of the relation- tionation and/or crustal assimilation) as Rb ship between what are commonly seen as late increases, corresponding to a more or less ver- stage differentiates (aplites and leucogranite tical trend of data on Fig. I (e.g., the Yellow- minor intrusives) on the one hand and granite stone basalts). In systems in which Sr deple- pluton geochemistry on the other. These mi- tion is more marked (many granites and nor intrusives are commonly regarded as eso- rhyolites, for example), a negative trend is an- teric and a minor aspect of the overall story. ticipated in Fig. 1. A cogenetic suite of com- ]'hey are, in fact, possibly of fundamental im- positions evolving from marie to felsic would portance to our understanding of the genesis of therefore be expected to display some kind of granite batholiths. Trying to mass balance an inverted "L" or recumbent "V" shape on crustal contamination, melt compositions and this diagram, which is seldom (if ever) ob- differentiation in granites using bulk geochem- served in perfection, but is possibly the reason istry is of limited value without this informa- for the less perfect inverted "L" shapes of the tion. Most large-volume high-level plutons may Peru Coastal Batholith and the Kosciusko be accumulative relative to the once associ- Batholith I-types (Lachlan foldbelt ). The trend ated, now lost, cogenetic volcanics. in Fig. 1 will be horizontal if bulk D R[' ~- 1 or if (2) In plagioclase-free differentiation in D sr 2>> D Rb. The former is only likely in assem- M()DELLING THE PETROGENESISOF HIGH Rb/Sr SILICIC MAGMAS 109 blages fractionating large amounts of biotite rocks with which this paper is mainly (Hanson, 1978 ). Many evolved granites such concerned. as the Trawenagh Bay pluton, Donegal (Fig. 1) have clearly undergone extensive K-feld- 3. Partial melting vs. fractional crystallization spar fractionation, as judged by their very high Rb/Sr, Rb/Ba, and Rb concentrations indica- Trace-element behavior as a result of sili- tive of bulk DRb< 1, bulk DS~>l and bulk cate-melt equilibria is described in a number DU~'> 1. of familiar equations (e.g., Shaw, 1970; Arth, ( 3 ) The fact that quasi-linear negative trends 1976 ). Partial melting is generally modelled by are observed in Fig. 1 for "evolved" granites an equilibrium batch melting equation such as: and rhyolites, suggests that there is some sim- ple relationship between the change in Rb and C'/C~,=I/(F +D'-FD') (1) Sr in the magma. It will be shown below that whereas fractional crystallization is described this is indeed explicable with a manipulation by the Rayleigh equation: of the Rayleigh fractionation equation. We can take several approaches to the ques- (C'/C[,)=F ')' ' (2) tion of how evolved silicic magmas such as (in which C'= concentration of element i in the those of Long Valley (with more than three or- liquid, C~,=concentration of element i in the ders of magnitude variation in Rb/Sr ratio) are starting mineral-melt assemblage, F= fraction produced. In so doing we are also attempting of liquid). If the bulk crystal/liquid distribu- to place limited constraints on what kind of tion coefficient (D) is greater than unity, the parent magmas were involved in the genesis of trace element (i) is depleted much more rap- these volcanic rocks, although from the outset idly during crystal fractionation than in melt- we would admit our prejudice that many of the ing (Fig. 2). Even assuming a high D s', it is parent magmas were complex mixtures of almost impossible to achieve low Sr concentra- mantle- and crust-derived components with an tions (and correspondingly high Rb/Sr) overall intermediate to silicic composition through equilibrium partial melting. This is (A.N. Halliday in Leake et al., 1980; Halliday true whether modal or non-modal melting et al., 1980; DePaolo, 1981a, b; Hildreth, 1981; equations are considered (as F approaches Huppert and Sparks, 1988). Isotopic studies zero, approaches 1/D'). Large deple- prove that some of the arrays shown in Fig. 1 C'/C~, are not simply the result of Rayleigh fraction- 500 as ~ o2 ~r } s2 l ation. Processes such as crustal contamination I i C~R~ : ,b ppm ], can also play a role.