AGeochemical Classification for Feldspathic Frost2 Classification for Feldspathic Igneous Rocks B. RONALD FROST* AND CAROL D. FROST DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY

  • View
    222

  • Download
    2

Embed Size (px)

Text of AGeochemical Classification for Feldspathic Frost2 Classification for Feldspathic Igneous Rocks B....

  • AGeochemical Classification forFeldspathic Igneous Rocks

    B. RONALD FROST* AND CAROL D. FROSTDEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF WYOMING, LARAMIE, WY 82072, USA

    RECEIVED MAY 13, 2008; ACCEPTED OCTOBER 3, 2008ADVANCE ACCESS PUBLICATION NOVEMBER 20, 2008

    In this paper we classify the range of feldspathic igneous rocksusing five geochemical variables: the FeO/(FeOMgO) ratio orFe-index, the modified alkali^lime index, the aluminum-saturationindex, the alkalinity index, and the feldspathoid silica-saturationindex.The Fe-index distinguishes between melts that have undergoneextensive iron enrichment during differentiation from those that havenot.The transition from tholeiite to ferrobasalt allows us to extendthis boundary to silica values as low as 48 wt %.We introduce thefeldspathoid silica-saturation index, which, coupled with the alkali-nity index, allows us to extend the geochemical classification toalkaline rocks. We show that most alkaline rocks are ferroan andthat this probably reflects extensive fractional crystallization ofolivine and pyroxene with minimal participation of Fe^Ti oxides.The expanded classification allows us to illustrate the geochemicaland petrogenetic relationship of the plutonic rocks from ferroan gran-ites to nepheline syenites that commonly occur in intracratonic envir-onments. It also allows us to distinguish four families of feldspathicrocks: (1) magnesian rocks, which are exemplified by Caledonianand Cordilleran batholiths and are characterized by differentiationunder oxidizing and relatively hydrous conditions; (2) ferroan rocks,which include fayalite granites, alkali granites, and nepheline sye-nites and are characterized by differentiation under reducing andrelatively dry conditions; (3) leucogranites, which commonly formby crustal melting; (4) potassic and ultrapotassic rocks, which origi-nate from mantle that has been enriched in K2O.

    KEY WORDS: granite; rhyolite; geochemistry; classification; nephelinesyenite; alkaline rocks; phonolite

    I NTRODUCTIONSeveral years ago we introduced a geochemical classifica-tion for granitic rocks (Frost et al., 2001). In that schemewe suggested that granitic rocks could be classified using

    three compositional variables, FeO/(FeOMgO) (orFe-index), Na2OK2O ^ CaO (or the modified alkali^lime index, MALI), and the aluminum-saturation index[ASI; molecular Al/(Ca ^ 167PNaK)]. The schemehas achieved wide use but several issues remain unad-dressed. One is whether the ferroan^magnesian boundarycan be extended to intermediate and basic rocks. Anotheris the petrological significance of the alkalic, alkali^calcic,calc-alkalic and calcic boundaries in the MALI diagrams.In addition to addressing these questions, we extendour classification scheme by introducing two additionalindices: the alkalinity index (AI) and feldspathoid silica-saturation index (FSSI). These indices allow for the discri-mination of metaluminous from peralkaline rocks andsilica-saturated from silica-undersaturated rocks, andthereby allow the geochemical classification scheme ofFrost et al. (2001) to be extended to alkaline rocks. Theenlarged classification scheme can be applied to the wholerange of feldspathic rocks; that is, rocks in which feldspars( quartz or feldspathoids) are the dominant minerals.

    REV I S IONS TO THEGEOCHEMICAL CLASS I F ICAT IONOF GRANITESFe-index: the boundary between ferroanand magnesian rocksThe FeO/(FeOMgO) ratio of rocks is an importantindication of the fractionation history of a suite of rocks.If the rocks are reduced [FMQ (fayalite^magnetite^quartz) or below, Frost & Lindsley,1992] fractional crystal-lization results in iron enrichment, whereas if the rocks arerelatively oxidized (FMQ 2 or more, Frost & Lindsley,1992) the crystallization of magnetite inhibits iron

    *Corresponding author. E-mail: rfrost@uwyo.edu

    The Author 2008. Published by Oxford University Press. Allrights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org

    JOURNALOFPETROLOGY VOLUME 49 NUMBER11 PAGES1955^1969 2008 doi:10.1093/petrology/egn054

  • enrichment (Osbourn, 1959). Miyashiro (1974) establisheda boundary between volcanic rocks that underwentan iron-enrichment trend and those that did not, whichhe identified as tholeiitic and calc-alkalic respectively.(Note: to eliminate confusion, we place the terms tho-leiitic and calc-alkalic in quotation marks when they areapplied senso lato rather than sensu stricto.) Miyashirosboundary was determined from a suite of arc-related vol-canic rocks from northeastern Japan, plotted on a diagramof FeO/MgO (where FeOFeO 09Fe2O3) againstSiO2. He showed that the calc-alkalic series could be sepa-rated from the tholeiitic series by a straight line of theform FeO/MgO 0157SiO26719. This boundary,which is linear in a plot of FeO/MgO vs SiO2, is stronglycurved in a plot of FeO/(FeOMgO) vs SiO2 (Fig. 1).Frost et al. (2001) established their boundary between

    ferroan and magnesian granites as a straight line that sepa-rated a population of A-type granites from Cordillerangranites. They recognized two boundaries: Feno, which isthe boundary determined from rocks in which both FeOand Fe2O3 have been analyzed, and Fe

    , which applies torocks in which only the total amount of FeO (or Fe2O3) hasbeen determined (Frost et al., 2001; Fig. 1). Frost et al. (2001)drew their Fe boundary so that at high silica contents itcoincided with the boundary of Miyashiro (1974). Becausethe boundary proposed by Miyashiro (1974) and that byFrost et al. (2001) diverge at SiO2 560% the questionarises which should be used for rocks with low silica.The analyses that Frost et al. (2001) used to establish their

    boundary generally had SiO2 4600%. To extend theferroan^magnesian boundary to lower silica values we

    plot ferrobasalts and basalts from the Galapagos, thetype area where ferrobasalt was defined (McBirney &Williams, 1969). The ferrobasalt^basalt boundary from theGalapagos, which occurs in rocks with 48^50% SiO2,more than 13% total iron and less than 6% MgO, agreesremarkably well with the extrapolation of the Frost et al.(2001) boundary. Our revised boundary [calculated onthe basis of total iron in the rock; FeOFeO 09Fe2O3/(FeO 09Fe2O3MgO)] has a slightly steeper slopeand fits the equation FeO 046 0005SiO2. Becauseit is defined at low silica by the ferrobasalt^basalt transi-tion, this boundary is applicable to rocks with silica aslow as 48%.

    The modified alkali^lime index (MALI)Frost et al. (2001) defined the modified alkali^lime indexfrom a plot of Na2OK2O ^ CaO vs SiO2. They plottedcompositions from the Peninsular Ranges batholith,Tuolumne intrusive suite, the Sherman batholith, andBjerkreim^Sokndal intrusion on this diagram and usedthem to draw boundaries between calcic, calc-alkalic,alkali^calcic, and alkalic series. Each boundary is con-strained to go through MALI 0 at the value definedby Peacock (1934) (namely, alkalic ^ alkali^calcic atSiO2510, alkali^calcic ^ calc-alkalic at SiO2560,and calc-alkalic ^ calcic at SiO2610). From these con-straints, the boundaries were drawn by eye to separate asmuch as possible the individual suites. Below we discusswhy the boundaries have the shape that they do and whymafic rocks commonly plot with trends that show largechanges in MALI with small changes in silica.

    MALI and igneous mineralsThe first step to understand how MALI varies in rocks isto note where common igneous minerals plot on a MALIdiagram (Fig. 2). The MALI value of plutonic rocks is thesum of the MALI values of the constituent minerals.The fractionation trend of a volcanic suite is controlled bythe MALI of the mineral assemblages that are crystallizedand extracted from the melt. As Fig. 2 shows, the mineralsthat contribute most to produce rocks with high MALIvalues are K-feldspar, albite, and nepheline (Fig. 2),whereas augite has the lowest MALI values. It is evidentfrom Fig. 2 that, for rocks with more than about 60%SiO2, MALI is controlled by the abundances and composi-tions of feldspars and quartz, whereas at lower silica theextraction of augite during fractionation of more maficrocks will have a powerful effect in increasing MALI inthe residual magma.

    Role of feldsparsTo illustrate the role of feldspars in MALI we show anumber of model rock compositions (Table 1) on a diagramof SiO2 vs MALI (Fig. 3) The suite of model granitoidsfrom diorite to trondjhemite follows a trend roughly

    1.0

    0.8

    0.6

    0.440 50 60 70 80

    SiO2

    FeO

    /(Fe

    O+

    MgO

    )

    TH-CA boundary

    Fe*

    Feno

    Galapagos ferrobasalt

    Galapagos basalt

    Fig. 1. Comparison of the ferroan^magnesian boundaries (Fe andFeno) of Frost et al. (2001; dashed lines) with the revised bound-ary proposed here (continuous line described by FeO/(FeOMgO) 046 0005SiO2) and the TH^CA boundary ofMiyashiro (1974). Ferrobasalt^basalt transition from the Galapagosis after McBirney & Williams (1969). TH, tholeiitic; CA, calc-alkalic; FeFeO 09Fe2O3/(FeO 09Fe2O3MgO); FenoFeO/(FeOMgO).

    JOURNAL OF PETROLOGY VOLUME 49 NUMBER 11 NOVEMBER 2008

    1956

  • parallel to the boundary between the calcic and calc-alka-lic fields. In contrast, those granitoids that have increasingproportions of K-feldspar to plagioclase lie at progressivelyhigher MALI values. Our simple calculations suggest thatthe shape of the boundaries in the MALI diagram reflectsthe increases in the abundance of Kspar and in the albitecomponent of plagioclase with increasing silica in plutonicrocks. For volcanic suites, the trend reflects the changes innormative abundances of these two feldspar end-members.To further emphasize the role of feldspars in the alkali^

    lime index we have plotted the modes of some of the suitesthat we used to define the MALI boundaries. Becausemodal mineralogy data are sparse fo