Crustal contamination inverted: some preliminary results

Wall rock assimilation and fractional crystal lization (AFC) as Bowen has pointed out in 1928, should be co nsidered toge ther because the heat req uired for assimilation can be provided by the latent heat of crystalliza tion of the magmas. Thi s process (AFC) was disc ussed by several authors, who sought to reproduce observed tren ds on isotope-i sotope an d isotope-trace element diagrams. In general , these models were loosely constrained because of the pratical difficulti es in han dlin g large num bers of elements and in exp loring the relationships between diffe rent varia bles for any data set. In this work. the approac h has been to cons ider many trace elements and isotope ratios simultan eous ly. to calculate the best fi t surface through the analy tical data for suites of related samples and to eval uate the re lationships between variables in the AFC mode l for the range of solutions which are consistent with the best-fit surface. This new analysis of the AFC model is illu stra ted with selected samp les from a detai led section through basalts and rhyoli tes of the Serra Geral Forma tion in the Parand Basin seq uence. Results appear to be inconsistent with the bulk assimi latio n of average crustal compositions, but show exce lle nt agreeme nt with the trace element pattern of typical upper crustal mel ts. RESUMO CONTAMINA<;:AOCRUSTAL INVERSA: ALGUNS RESU LTADOS PRELIMI NAR ES. o processo que co nsidera a assimilacjo da roche encaixante co nco mitanteme nte Acristal izacjio fracionada por urn determinado magma foi inicialmente enunciado por Bowen e recentemente equacionado por De Paolo, sendo conhecido como processo AFC. Os modelos AFC disponfveis na literatura lem-se baseado nas tend8ncias observadas em diagramas isolados que relacionam de vari es formas concentracfes de elementos-tracos e raz des isotdpicas, Em geraJ, esses modelo s nao sAo bern vincul ado s devido ao grande nWnero de elementos envolvidos e, conseqilente me nte, de varid veis em carla co njunto de dados, No presen te trabalho co nsideram-se simul teneemente vWios elementos bern como raroes isotdpicas e calc ula-se a super flc ie de melbor ajuste por mei o de todos os dados anaItti cos dlsponfveis de urna seqtle ncla de amostras geneticamente correlacionadas, Os parametres obtidos sao consi sten tes com urna das solucoes que salisfazem diferencas mfnimas entre os dado s analfticos e 0 mod ele calculado. 0 tratamento numerico utilizado nesta nova anali se de modele AFC ~ aqui descrito, testado com dados sinteticoe e apJicado a uma seqnencta besdlti ce da Forma. ~Ao Serra Geral, obt endo-se como resu ltado a composlcao de urn possrvet contam inante, Slio fomecidos todos os parAmetros do mod elo, que resultou compatfvel com urn processo AFC tendo co mo co ntam inante urn material proveni ente da fusao parcial da cros ta co ntinental . INTRODUCTION The extent to which continental basalts are affected by interaction with the continental crust, and in particular how such interaction should be established, remains one of the more perplexing problems in modem igneous petrology. Many continental alkali basalt suites have isotope and trace element ratios similar to their oceanic counterparts and would Iherefore appear to have suffered little contamination en route through the lithosphere. Continental flood basalt provinces. in contrast, exhibit a range of Sr-, Nd, and Ph-isotopes far greater Ihan that reported from oceanic islands. They equilibrated wiIh manlle source rocks at shallower pressures, and in many cases the range in radiogenic isotope compositions probably reflects contributions from both the mantle and crustal portions of the continental liIhosphere. Early isotope studies were set in a framework which regarded the continental crusI as Ihe sole repository of old high RblSr and low SmlNd ratios and hence as Ihe source of high 87Sr/8 6Sr and low I4 3Ndl I44Nd. However, studies of manlle xenoliIhs (Menzies & MurIhy 1980; Erlank et aI. 1987) and inclusions in diamonds (Richardson et al, 1984) have since demonstra ted that segments of the sub-continental mantle are boIh old, perhaps locally 3 Ga, and have variable RblSr and Sml Nd which with time have also generated a wide range in 87S r/86Sr and 143Ndl l44Nd. Such data highlight Ihe ambiguities of isotope ratios in magmatic rocks and the risks involved in interp reting isotope analyses in isolation from major and trace element results on the same samples. Contamination of mantle derived melts by crustal material may be considered in two ways: a) the more primitive rocks will tend to be more contaminated because they are hotter and can therefore assimilate more material, and b) the effec ts of contamination increase with differentiation and so are most marked in the more evolved rock types. The former has shown to be significant in komatiites from the Archaean (Huppert & Sparks, 1985). However, in Ihe matrity of contaminated Phanerozoic and Recent suites 87Srl 6Sr, for example, tends to increase with Si02 suggesting that contamination processes are linked to differentiation. Bowen (1928) pointed out that wall rock assimilation and * Insti tuto Astroncmicc e Geo ffsico, Departamento de Geoftsica, USP. Cilia Postal 30627 ,CEP OJ051. S30 Paulo. SP, Brasil ** Departament of Earth Sciences , Th e Open Uni versit y. Walt on Hall. Milton Keynes MK7 6AA, Unite d Kingdom 28 Revista Brosileira de Geocisncias, Volum e 18. 1988 and as F is constant for any sample, the best fit solutions of all elements in a sample will tend to a common value of F. Fur thermo re an approximate lower limit of F is available from closed system fractional crystallisation models based on major elements (e.g, the XLFRAC programme of Stormer & Nicholls 1978). Isotope ratios are incorporated from equation (l5a) of De Paolo (1981) which can be rewritten: