American Mineralogist, Volume 71, pages 6047, 1986

Mg-rich hollow sanidine in partially melted granite xenoliths in amica peridotite at Rose domeo Woodson County, Kansas

Mrcn,lBr- P. SrrlrrHt eNo Peur C. FneNxs

Department of G eosciencesUniversity of Tulsa, Tulsa, Oklahoma 74104

Abstract

Hollow sanidine as much as 100 pm long enclosed in micropoikilitic quartz in Precam-brian Rose dome granite, woodson county, Kansas, probably grew from a supersaturatedmelt. The melt formed by the addition of heat and water to granite xenoliths from a LateCretaceous mica-peridotite host. The granitic melt probably became supersaturated insanidine due to a substantial loss of water from both the peridotite and granite melts uponemplacement. Universal-stage 2V, measurements in the range 37-44 are consistent witha high-sanidine structure. Microprobe analyses show that the sanidine contains relativelyhigh concentrations of Fe, Mg, and Si, and low Al. In contrast, relict K-feldspar has a morenormal composition: MgO below detection limits and 0.04 wt.o/o FeO. The hollow sanidines,which contain more than three silica cations per eight oxygens, are, to the best of ourknowledge, the most siliceous alkali feldspars ever analyzed.

Hollow calcic plagioclase has been reported from lunar maria and glassy mid-ocean ridgebasalts, but, as far as we know, naturally occurring hollow alkali feldspar has not beenreported previously. The morphology of both the hollow sanidine and hollow plagioclaseis indicative of rapid $owth from supersaturated melts. Unlike the lunar basalts andmid-ocean ridge basalts, loss ofvolatiles rather than heat probably induced supersaturationof the granitic melt. In contrast to the high Fe and Mg in hollow plagioclase, Fe and Mgin the hollow sanidine cannot be attributed to a Ca(Fe,Mg)SirO, component. To rationalizethe sanidine analyses a K, (Fe,Mg)SirO,u feldspar component is proposed. If the proposedcomponent is written in a more usual formulation based on 8 oxygens, it must be expressedas the sum of [K(Mg,Fer+)Si3O8]- and [KSioOr]*. Charge balance of the component requires(K,Na)* occupancy of an octahedral site within a [SioOr]o structural unit to compensate forthe excess negative charge left unbalanced by (K,Na)* occupancy of an octahedral sitewithin a [(Fe,Mg)SirOr]r- unit.

The sanidine/granitic melt partition coefrcient for Mg is estimated to be locally greaterthan five. The ratio of Mg in the sanidine to Mg in the mica peridotite is about 0.022.This value is close to previously reported feldspar/liquid Mg partition coefficients andimplies that mica peridotite and granite melts achieved at least local metastable equilib-rium.

Introduction

Mg-rich hollow sanidine, as much as I 00 pm long, occurin blocks of Precambrian granite that are xenoliths in aCretaceous mica peridotite exposed at Rose dome, Wood-son County, Kansas (SEYr, sec. 13, T.26S., R.l5E.) (Bick-ford et al.,l97l1' Franks et al., l97l). Insofar as we know,this is the first reported natural occurrence ofhollow san-idine. Petrographic textures provide evidence that thegranite was partially melted by the peridotite (Franks etal., l97l). The skeletal morphology of the sanidine mi-crolites suggests growth from a melt that was supersatu-

-;Er."J

"ddress: Amoco Production Company, Research

Center, P.O. Box 3385, Tulsa, Oklahoma74102.

0003-004x/86/0102{060$02.00 60

rated with sanidine. Geological evidence indicates thatsupersaturation of the granitic partial melt primarily wasinduced by loss of volatiles from the peridotite-granitesystem, rather than by loss of heat. The hollow sanidinecontains about 0.5 wt.o/o MgO. The content of MgO ismuch higher than that of other alkali feldspars, and sug-gests ion-exchange between the granitic melt and the pe-ridotite. Moreover, the hollow sanidines not only containabout 1.5 wt.o/o FeO, but they also are unusually rich inSi and poor in Al, containing more than three Si cationsper eight oxygens. The unique chemical composition im-plies local charge imbalance in the feldspar structure. Thecomposition of the sanidine offers insight into the char-acter ofthe interaction between the granitic partial rneltand the host mica peridotite. The Mg-rich nature of the

SMITH AND FRANKS: Mg-RICH HOLLOW SANIDINE 6r

Fig. l. Photomicrograph of Rose dome granite showingrounded relict grains ofquartz (Q) and alkali feldspar (F) enclosedin microcrystalline matrix (M) of quartz and alkali feldspar crys-tallized from partial melt. Crossed polars.

sanidine suggests that Mg-rich metasomatic fluids derivedfrom the mica peridotite played an important role in theirformation.

Iextures

Melt textures

In parts ofthe granite, large subrounded to subangulargrains ofquartz and feldspar are separated by a fine-grainedmatrix (Fig. l). The fine-grained matrix forms channelsbetween the relict feldspar and quartz grains (Fig. l) thatappear similar to melt channels produced by experimentalpartial fusion of gneisses and granulites (Mehnert et al.,1973) and albite-quartz mixtures (Jurewicz and Watson,

Fig.2. Photomicrograph of Rose dome granite showing fret-ted, oriented overgrowths of sanidine (dark) near center of field.Fretted overgrowths are nucleated on and extended parallel totwin lamellae in relict plagioclase (P) at left of center. H, hollowsanidine; K relict K-feldspar; R, overgrowth rim of sanidine onrelict plagioclase (P) in lower left; Q, quartz; S, equant and lathlikesanidine. Crossed polars.

Fig. 3. Photomicrograph of Rose dome granite showing hol-Iow, equant, and lathlike sanidine (S) in felted matrix (M) com-posed chiefly of acicular feldspar but containing blebs of "li-monite" (L) and scarce needlelike apatite (A). Plane-polarizedhCht.

1984). The observed distribution of relict grains and ma-trix is also similar to partially melted granulite xenolithsin the Lashaine kimberlite (Jones et al., 1983), and sup-ports the conclusions ofFranks et al. (1971) that the Rosedome granite was partially melted by the mica peridotite.

Crystallizatio n textures

Much of the fine-grained matrix consists of micropoi-kilitic quanz that encloses tabular, lathlike, and skeletalsanidine microlites. Fretted sanidine laths occur on themargins of some remnant plagioclase (Fig. 2). Albite twin-ning is observed in some of the remnant grains, and thesanidine laths have nucleated and grown on alternate twins'Sanidine also forms rims on remnant K-feldspar grains,

some of which show relict microcline twinning. Thosesanidine crystals that did not nucleate on the margins ofremnant feldspars form lathlike to equant and hollow

Fig. 4. Photomicrograph of Rose dome granite showing elon-gate, hollow sanidine (arrow) and equant to lathlike sanidine (S)poikilitically enclosed in matrix quartz (M) that is bordered byrelict K-feldspar (K) and quartz (Q). Note sanidine rim (R) onrelict K-feldspar. Crossed polars.

62 SMITH AND FRANKS: Mg-RICH HOLLOW SANIDINE

Table l. K-feldspar compositions and optical properties. Points analyzed: A, B, hollow sanidine; C, D, E, F, hollow sanidine shownin Figures 4 and 5; G, hollow sanidine; H, relict K-feldspar

f A B

2v 39+30X -

G H

42+60 <6604o+20

Mic rop robe ana l yses i n we igh t pe r cen ! ox i des

s i o 2 6 6 . 4 6 7 . 5 6 6 . 1 6 7 . 5 6 6 . 2 6 6 . s 6 6 . 5

o t , 0 ,

I 5 . 2 I 5 . 2 r 4 . 7 1 5 . 4 I 5 . 2 I 4 . 8 1 5 . 7

c a o o . g 7 0 . o 3 6 . 6 7 o . o g o . 1 8 0 . q 4 0 - o I

F e O ' 1 . 5 3 1 . 3 6 1 . 3 5 I . 5 2 1 . 3 3 1 . s 5 t . 0 1

M g o 4 . 4 9 0 . 4 6 o . 4 4 0 . 5 2 O . 4 5 6 . 5 3 0 . 5 2

^ r o

1 6 . r I 5 . 2 1 s . 4 L 5 . 2 1 s . 5 1 5 . 7 1 5 . 4

N a 2 O 0 . 2 7 0 . 9 7 r . t 5 t . o 3 o . g 2 l . o o 0 . 6 5

T o r a l 1 0 0 . 1 I o o . 7 9 9 . 2 r g r . 2 g g . 7 g g . 6 g g . g

* a1 l i r on repo r t ed as Feo

S t ruc tu ra l f o rnu lae i n ca t i ons pe r g oxygens

s i 4 + 3 . 0 8 3 3 . o g j 3 . o g 4 3 . o a 6 3 . o 7 8 3 . 0 8 3 3 . g 7 g

A l 3 + 6 . 8 3 2 0 . 8 2 2 o . 8 r r 6 . 8 3 0 0 . 8 3 3 g . B r 5 o . a s : .

F e 2 + o . o 5 g o - o 5 2 o . o s 3 o . q 5 g o . o 5 2 o . q 6 r o . o 3 g

r 4 g 2 + 9 . 6 3 4 o . o 3 r o . o 3 r g . 0 3 5 6 . o 3 r s . o 3 7 g . 0 3 6

c a 2 + 6 . 0 b 3 o . g o L o . s b ! o . o a o g . o o 9 o . g g 2 o . g g o

N a + 0 . o 2 4 6 . 0 8 6 o . t o 4 o . g 9 r o . o B 3 o . g g r o . o 5 a

K + 0 . 9 5 s q . B g o 6 . 9 2 o 9 . 8 8 7 s . g 2 o 9 . 9 3 5 0 . 9 6 6

( r v ) 4 . s o g 4 . 0 0 2 3 . 9 8 9 4 . O 9 9 3 . 9 9 4 3 . 9 9 6 4 . 6 L O

( v r ) s . 9 8 2 0 . 9 7 7 r . 0 2 5 6 . 9 ' 7 8 I . O I 2 I . 0 2 8 0 . 9 6 4

T o r a l 4 . 9 9 0 4 . 9 7 9 5 . o r 4 4 . 9 8 7 5 . 0 0 6 5 . A 2 4 4 . 9 7 4

11yy =gi4+1a13+.r14q2+.rps2+. 1y11 =g+1p6+ag62+

6 5 . 6

1 8 . 4

o . 0 6

o . o 4

0 . 0 6

r 5 . 9

o . 7 2

L g O . I

2 . 9 9 8

0 . 9 9 9

9 . 6 0 r

0 . 0 9 0

0 . 0 0 3

0 . 0 6 5

6 . 9 3 8

3 . 9 9 8

L . 6 0 6

5 . 0 0 4

gowths (Fig. 3). These sanidine microlites are clear andshow no detectable sign of alteration. They mostly areenclosed in optically continuous micropoikilitic quartz,especially near the maryins of relict quartz and feldspar(Fig. a). Acicular and spherulitic sanidine and quartz in-tergrowths have been noted previously in the Rose domegranite (Franks et al., l97l), and other fine-grained sam-ples ofthe granite, not considered in this study, containaltered mafic minerals. Na-rich feldsparhas not been foundin the fine-grained matrix.

To the best ofour knowledge, naturally occurring hol-low alkali feldspars have not been previously described.The morphology ofthe hollow sanidine is further evidencethat Rose dome granite was partially melted, and that thefine-grained matrix is the recrystallized partial melt. Thesimple fact that hollow sanidines are found in the Rosedome granite, and that their morphology is indicative ofgrowth from a supersaturated melt, is additional evidencethat the granite was partially melted by the peridotite.

Optical studies

Measurements of 2V* were made for six hollow andequant matrix feldspar grains using a Bausch and Lomb

5-axis universal stage. The measurements were correctedfollowing the procedures of Emmons (1948). Corrected2V, values range from 37 to 54, but only one of themeasurements exceeds 44'. Orientation procedures andmeasurement of 2V* were handicapped by the low bire-fringence and dark, first-order gray interference colorsshown by the microlites cut nearly perpendicular to theacute bisectrix of the optic angle (Fig. a). Consequenflyerrors could be as large as +3" for measurements of I 4.2V, values for the grains analyzed using the microprobeare given in Table l.

The hollow feldspar microlites have nearly euhedralshapes that range from equant to elongate when vieweddown the acute bisectrix (Figs. 3 and 4). The orientationofthe optic plane is about perpendicular to the long di-mension of the elongate grains (Fig. 5). The optical datafor the elongate hollow grains accord either with ortho-clase crystals that are tabular parallel to {010} or withhigh sanidine that is elongate parallel to the b crystallo-graphic axis or the trace of {001 } (Smith, 1974; Su et al.,1984). X-ray diffraction and 2V studies of relict primaryK-feldspar in Rose dome granite (Franks et al., l97l),however, are consistent with the thermal conversion of

Z . BxO

SMITH AND FRANKS: Mg-RICH HOLLOW SANIDINE 63

Y ' b

O.O25 mm

Fig. 5. Sketch ofhollow sanidine (Fig. a) showing approxi-mate orientation of optical axes and inferred b-axis. Sketch notcorrected for universal stage rotation. BxO, obtuse bisectrix.2V : 40+4'. C, D, E, and F mark locations of microprobeanalyses in Table l.

primary microcline to high sanidine after entrapment inthe mica peridotite. The data from the relict feldsparscombined with the textural evidence of partial melting(Fig. l) and rapid crystallization (Figs. 3 and 4) supportthe inference that the hollow microlites are htgh sanidine,and that they commonly are elongate parallel to the traceof {00 I } in sections cut nearly perpendicular to the acutebisectrix or a crystallographic axis (Figs. 4 and 5). In con-trast, the lathlike K-feldspar microlites (Figs. 2 and 3) arelength fast, do not allow measurement of 2V*, and accordwith elongation parallel to a.

Elongation parallel to a is characteristic ofthe rapidlycrystallized plagioclase and alkali feldspar studied by Lof-gren(197 4,1980), Bryan (1972), andFenn (1977), wheth-er or not the crystals are hollow. The shapes of hollowmicrolites in the Rose dome granite, then, differ signifi-cantly not only from the lathlike microlites in the granite,but also from the shapes of hollow feldspars studied byother workers. The hollow plagioclase studied by Bryan(1972) and most of the alkali feldspar grown by Fenn(1977) were tabular parallel to {010}. However, with in-creasing potassium content, the alkali feldspar grown byFenn became less elongate parallel to a, less lathlike, andblocky due to ". . . a reduction in growth rate anisotropyalong c and a . . . ." (Fenn, 1977 ,p. 152). Hence, althougltthe hollow sanidine in the Rose dome granite differs sig-nificantly in morphology from previously reported hollowfeldspars, their blocky, equant to elongate habit is anal-agous to Fenn's observations and consistent with theirpotassic compositions.

Feldspar compositions

Analytical procedures

Analyses were performed using an ARL-AMx electron micro-probe operating at an acceleration potential of I 5 kV and samplecurrents of 12-14 nA. the electron beam was focused to less than3 pm in diameter. Standards were analyzed before and after eachunknown. Counting times of ten seconds were employed. Fivepoints were averaged for Si, Al, K, and Na, ten points for Ca,Mg, and Fe. The data were corrected for background; dead-timecorrections were not required at the low sample currents used.

The data were reduced using the method of Bence and Albee(1969) and the correction factors ofAlbee and Ray (1970).

Microcline (K Al, Sl), anorthite (Ca), and osumilite (Fe, Mg)provided by the Smithsonian Institution (Jarosewich et al., 1980)and Amelia atbite (Na) obtained from W. H. Taylor were usedas standards. Na was also checked against the microcline stan-dard. The osumilite used as the Fe and Mg standard has Si andAl contents very similar to K-feldspar and 4.00 wt.o/o KrO. Theonly other major components of this standard are 6.380/o FeOand 5.8390 MgO. This standard was chosen for Fe and Mg toreduce possible errors associated with the correction procedures.Although much of the Rose dome granite is altered, the fine-grained sanidine-quartz matrix is very fresh. Many ofthe sanidinernicrolites are also sufrciently large to allow quantitative electronprobe microanalyses (Fig. 4).

The analyzed hollow sanidines were fresh, showing no signsofalteration. The unaltered state ofthese feldspars, and the qual-ity ofthe sanidine analyses can be judged using the sum oftheoxide weight percents, which are all close to one hundred, andthe structural formulae, all of which have close to five cationsper eight oxygens, or a total of one octahedrally coordinatedcation and four tetrahedrally coordinated cations per eight oxy-gens (Table 1). Further, the results ofthe analyses given in TableI are consistent between different hollow sanidines and within asingle hollow sanidine. Also, our analysis ofa relict potassiumfeldspar shows no compositional peculiarities (H, Table l).

Feldspar compositions and tetrahedral substitutions

The compositions of three hollow sanidines and a relictK-feldspar are given in Table 1. The average structuralformula for the seven analyses of the hollow feldspars is(&.r,oNao.orrCao.oor)ut(Mgo.o14FEo.os3Alo.82eSir.orr)ttOr. The

structural formula indicates that the hollow sanidines arerich in Mg, Fe, and Si, and are depleted in Al relative to

stoichiometric K-feldsPar.Because the sum of tetrahedrally coordinated cations

per eight oxygens of all feldspars, whether stoichiometricor not, must be four, the data also suggest that Fe and Mg

are tetrahedrally coordinated in the hollow sanidine (Ta-

ble l). Tetrahedral trivalent Fe is not uncommon in alkalifeldspars, and feldspars of the composition KFe3*SirOthave been synthesized (Faust, 1936; Wones and Apple-man, 1963). Tetrahedrally coordinated divalent Fe andMg occur in terrestrial and lunar plagioclase (Bryan, 1974;Longhi et al.,1973 Beatty and Albee, 1980; Crawford,1972),atdthe partitioning of these two elements betweenplagioclase and liquid has been studied experimentally(Murphy, 1977; Longhi et al., 197 6). These substitutionsare attributed to feldspar components of the formulaCaMgSirO, and CaFe2*SirO.. Plagioclase of both thesecompositions has been synthesized (Sclar and Kastelic,1979; Sclar and Stead, 1980).

Although it may be argued that the iron in the hollowsanidine is trivalent, there is no ambiguity as to the va-lence state of magnesium. Therefore, the following dis-cussion focuses mainly on the structural implications ofMg substitution in sanidine. Ferrous iron in sanidine,however, must occupy sites structurally similar to thoseoccupied by Mg.

Ca is nearly absent in the sanidine. The atomic Mg/Caratio is greater than l0 (Table l). Tetrahedral Mg and

SMITH AND FRANKS: Mg.RICH HOLLOW SANIDINE

Fe2* in these feldspars cannot, therefore, be attributed tothe magnesian and ferroan anorthite components dis-cussed above. Rather, the chemical analyses and the cal-culated sanidine structural formulae can be rationalizedby postulating a substitution of the feldspar component(K,Na)r(Mg,Fe2+)Si?Or6 into the hollow sanidine. Thiscomponent requires that an octahedral site in a [Sioo8]ostructural unit be occupied by an alkali cation (K*) inorder to satis$, the excess negative charge left unbalancedby the monovalent octahedral alkali cation in a[(Mg,Fe,+)SirOr]2- structural unit.

Feldspars of the composition (I!Na)r(Mg,Fe)SirO,u havenot been synthesized, but magnesian and ferroan ana-logues of leucite and kalsilite have (Roedder, l95la,b,1952). The synthesis of these compounds, moreover, doc-uments the substitution of Mgz+ and Sio+, or Fe2+ andSia+, for two tetrahedral Al3+ in potassic framework alu-minosilicates.

High- and low-temperature polymorphs of KrMgSirO,,are isomorphous with high- and low-temperature leucitepolymorphs, and the high- and low-temperature poly-morphs of KrMgSirO, are isomorphous with kalsilite andkaliophilite (Roedder, l95la, l95lb). KrFe2+SirO,, andKrFe2+SirO, are isomorphous with their magnesian coun-terparts (Roedder, 1952). Furtherrnore, there is completesolid solution between KrMgSirO,, and leucite (Schairer,1948). The isomorphism and solid solution shown bythese compounds accords with limited solid solution ofKr(Mg,Fe,+)Si,O,. in the disordered high-sanidine struc-ture. The substitution ofMgr+ and Sia+ fortwo tetrahedralAl3+ cations may be thought of in terms of the reaction:

2 {KA lS i3O8}+MgO+S iO ,+ Alzor + {KMgSirOs}- + {KSi408}*.

The average KrMgSirO,u content ofthese hollow sanidinesis 6.8 wt.0/0. The mean KrFeSirO,u content is I1.2 wt.o/oif all iron is assumed to be ferrous. The Rose dome san-idines contain more than three silicon cations per eightoxygens (Table l), and as far as we know are the mostsiliceous feldspars analyzed to date.

DiscussionTextures and morphology:i mp I i cat ions for pet ro genes i s

The best known occlurences ofhollow feldspars are thecalcic plagioclases in the glassy rims of mid-ocean ridgebasalts (Bryan, 1972, 1974) and the lunar mare basalts(Crawford, 1972). These plagioclases, which tend to besector zoned, grew from supercooled melts that result fromthe extrusion of anhydrous basalts into cold, deep-sea andlunar environments. Hollow plagioclase feldspars are notfound in either subaerially erupted magmas or magmaserupted at shallow water depths (Bryan, 1974). Hollowplagioclases have been grown from the melt by Lofgren(1974, 1980) at linear cooling rates of 2{ per hour andin isothermal drop experiments resulting in supercoolingsof 40 to 200'C.

Alkali feldspars grown from supercooled melts form

tabular crystals at small degrees of supercooling, and skel-etal, dendritic, and spherulitic forms at progressivelygreater supercooling (Fenn, 1973, 1977; Swanson,1977;Lofgren and Gooley, 1977). Fenn (1977) grew hollow al-kali feldspars, having glass-filled cores, from supercooledhydrous alkali feldspar melts. Long (1978), in a study ofthe partitioning of Ba, Sr, and Rb between alkali feldsparsand supercooled hydrous granitic melts, grew sector-zonedsanidine that appear to have small hollow cores (Long,1978, Fig. l0).

The hollow feldspars in these previous studies crystal-lized from supersaturated melts. Although there are somemorphological and chemical differences between the feld-spars described in these previous studies and the hollowsanidine in the Rose dome granite, the gross similaritiessuggest that the latter also crystallized from a supersatu-rated melt.

Supersaturation of these other melts was thermally in-duced. Rapid cooling, however, probably does not ac-count for hollow sanidine in Rose dome granite. The gran-ite occurs as xenoliths in a mica peridotite that intrudedthe Pennsylvanian Stanton Limestone and Weston Shale,which probably were buried by 900 to 1200 m ofyoungerrocks (Frank et al., l97l). Moreover, a considerable vol-ume of hornfels was generated from the Weston Shale(Franks et al., I 97 I ). It is probably a better approximationto infer nearly isothermal formation and crystallizationof the molten matrix within the granite rather than a dropin temperature comparable to that experienced by mid-ocean ridge basalts and lunar basalts, even though loss ofwater from the melt may have been an efficient heat trans-port mechanism.

Supersaturation of hydrous feldspathic melts can beachieved with or without a drop in temperature throughthe loss of volatiles. The feldspar growth textures de-scribed here can be attributed to rapid loss of volatilesfrom the mica periodotite-granite system during emplace-ment at shallow crustal levels, perhaps at depths in theregion of900 to 1200 m (Franks et al., l97l). The presenceof hornfels and metachert fragments in the outer partsof the partially melted granite xenoliths indicates that thematrix was molten during final emplacement of the micaperidotite. Moreover, the presence of hydrous and K-richphases in buchite-like Weston hornfels and Stanton Lime-stone in contact with the peridotite indicates the transferof water from the peridotite to the country rock duringintrusion (Franks et al., 197l).

Mg in sanidine: comparison with other feldsparsThe MgO content of the sanidine microlites,0.44 to

0.53 wt.o/0, is similar to that of hollow sector zoned pla-gioclase in the glassy rims of submarine pillow basalts,0.244.54 wt.o/0, reported by Bryan (1974) but it is un-usually high compared to other alkali feldspars. For ex-ample, 3 I alkali feldspars analyzed by De Pi ei et al. (197 7)all have MgO less than 0.04 wt.0/0. Eight sanidines ana-lyzed by Basu and Vitaliano (1976) all have less than 0.03wt.o/o MgO. Eighteen anorthoclase megacrysts analyzed by

SMITH AND FRANKS: Mg-RICH HOLLOW SANIDINE 65

Mason et al. (1982) using an ion microprobe all havebetween 2 and 52 parts per million Mg by weight. Thealbite and K-feldspar phases ofseven perthites from fourpegmatites analyzed using an ion microprobe all have lessthan I I parts per million Mg by weight (Mason, 1982).Data on Mg is also absent in the recent paper of Scheareret al. (1985) on the major and trace element compositionsof potassium feldspars from three pegmatites in the BlackHills.

The normally insignificant amount of MgO in sanidine,and other alkali feldspars as well, also is reflected in theabsence of data on Mg in the sanidine/glass partition coef-ficients published by Leeman and Phelps (1981) and Ma-hood and Hildreth (1983). The unusual amounts of Mgin the matrix sanidine, however, doubtless reflect par-titioning of Mg between the interstitial ganitic melt inthe xenoliths and the mica peridotite.

Mg-partitioning between feldspars and melts

Estimation of the partition coefficients between the Rosedome matrix sanidine and the mica peridotite offers in-sight into the nature of the interaction between the graniticpartial melt and the mica peridotite, as well as into thepossible importance of Mg-rich metasomatic fluids de-rived from the mica peridotite melt. Although the distri-bution of quartz and feldspar is highly variable in therecrystallized partial granitic melts, we estimate a quartzto sanidine ratio of about four to one where the sanidineis enclosed in quaftz. Minor apatite is the only other phasein the fine-grained matrix we studied. Sanidine is the onlymatrix phase that contains appreciable Mg in our samples.The sanidine/whole-matrix Mg partition coefficient is,therefore, estimated to be about 5. The sanidine/granite-melt Mg partition coefficient must be greater than 5, es-pecially because the melt probably was water saturated(Franks et al., l97l).

In contrast, a number of studies indicate that feldspar/melt Mg partition coefficients generally are less than one.The Mg mineral/matrix partition coefrcient of 12 ho-mogenous alkali feldspars determined by De Pieri andQuareni (1978) range from 0.00 to 0.33. These values arebased on chemical analyses offeldspar and glass separatesand the higher values may be in excess ofthe actual par-tition coefrcients because of impurities and inclusions inthe feldspars. Ion microprobe analyses of an anorthoclasemegacryst from Mt. Erebus, Antarctica, and a glass in-clusion within the megacryst yield an anorthoclase/glassMg partition coefrcient of 0.008 (Mason et al., 1982).Using electron microprobe data Longhi et al. (1976) dem-onstrated that the Mg plagioclase/liquid partition coeffi-cients for low-Ti lunar basalts and terrestrial basalts areabout 0.041.

The large feldspar/melt Mg partition coefficient esti-mated for the hollow sanidine could reflect rapid growthfrom the melt, consistent with the hollow character of themicrolites. Schiffrnan and Lofgren (1981), for example,documented an increase in iron concentration in andesinefrom the range 0.66-0.69 to the range 1.56-2.55 wt.o/o as

the cooling rates of a basaltic melt were changed from0.5"C per hour to 218'C per hour. They attributed thisincrease of Fe content to kinetic effects of Fe entrapmentat the plagioclase-liquid interface. However, even thoughthe Fe plagioclase/liquid partition coemcient increasedsympathetically with cooling rate in these experiments, itnever exeeded unity, and rapid growth does not appearto be the sole factor responsible for the htgh Mg contentof the Rose dome sanidine.

Mineral/liquid partition coefficients are also a functionof both crystal structure (Philpotts, 1978; Takahashi andIrvine, 1981) and melt structure (Watson, 1976, 1977Hart and Davis, 1978; Ryerson and Hess, 1978; Mahoodand Hildreth, 1983). Relative values of partition coeffi-cients of two isovalent elements that occupy the samecrystallographic site primarily are controlled by the crystalstructure @hilpotts, 1978). The absolute value of a min-erallliquid partition coefrcient, however, is primarily afunction of liquid structure. In general, the mineraVmeltpartition coefrcient of a network-modifuing cation, suchas Mg, increases sympathetically with melt polymeriza-tion. This effect is clearly illustrated in the partitioning ofnetwork-modi$ing cations among minerals in equilibri-um with a strongly depolymerized Fe-rich melt and ahighly polymerized Si-rich melt. The mineral/melt par-tition coefficients are much larger for the Si-rich melt thanfor the equilibrium Fe-rich melt (Watson , 197 6; Ryersonand Hess, 1978). The nonequlibrium partitioning of ele-ments between the silica-undersaturated ultramafic micaperidotite and the granitic melt may approximate theequilibrium distribution of elements between low-tem-perature Fe- and Si-rich immiscible liquids.

Although the sanidine/granite-melt Mg partition coef-ficient is greater than one, the sanidine/mica peridotiteMg partition coefficient must be much less than one. TheMgO content of one sample of noncalcareous Rose domemica peridotite is 22.31 weight percent (Franks et al.,1971). Using this value as an order of magnitude estimateof the Mg content of the mafic alkaline melt yields asanidine/mica-peridotite Mg partition coefficient of 0.022.This value is about three times that of the anorthoclase/glass partition coefrcient (0.008) determined by Mason etal. (l 982), about one halfofthe plagioclase/basaltic-liquidMg partition coefficient (0.041) of Longhi et al. (1976),and within the range of the alkali-feldspar/matrix Mg par-tition coefrcients (0.0H.33) of De Pieri and Quareni( I 978). The similarity of the hollow-sanidine/mica-peri-dotite Mg partition coefficient to other feldspar/melt pairssuggests that the distribution of elements between thegranite partial melt and the mica-peridotite melt enclosingthe granite xenoliths achieved metastable equilibrium, eventhough no peridotite melt permeated deeply into the xeno-liths studied by us.

The similarity ofthe MgO content of Rose dome hollowsanidine to the hollow plagioclase grown from relativelyMgO-rich mid-ocean ridge basalts may be due to the in-teraction between the granite and Mg-rich metasomaticfluids derived from the peridotite. Isolated phlogopite

66 SMITH AND FRANKS: Mg.RICH HOLLOI|/ SANIDINE

crystals that gtew within limestone xenoliths in the peri-dotite (Franks et al., l97l) support the idea of K- andMg-rich metasomatic fluids. A metosomatic fluid phasewould have greatly facilitated ion-exchange between thegranitic melt and the peridotite and favored crystallizationof Mg-rich sanidine.

Mg in sanidine: crystal chemical implications

The Mg-, Si-rich sanidine compositions presented hereare the first unambiguous evidence of local charge im-balance within individual four member tetrahedral ringunits in feldspar. Our documentation of the [KSioOr]* and[KMgSirOr]- feldspar components opens the door forspeculation on local charge imbalance in feldspars ofmoreusual composition. Possible charged components in alkalifeldspar are [(Na,K)SioOr]* and [(Na,K)AlrSirOr]-. Thecomponents [CaAlSi3O8]* and [NaAlrSirOr]- may occurin plagioclase. In all cases, equal numbers of positivelyand negatively charged units are required in the crystalstructure if overall charge balance is to be maintained.Betterman and Liebau (1976) synthesized feldspars of theformula LaNaAloSioO,u. The application of the Al-avoid-ance principle, which states that [NOo]'- tetrahedra donot share corners (Smith, 1974) to this feldspar impliesthe existence of [LaAlrSirOr]* and [NaAlrsiror]- com-ponents. The possibility of local charge imbalance be-tween 4-member ring units in feldspars, and, hence, ad-ditional substitutional disorder probably ought to beincluded in atomistic models of the feldspar structures.

Conclusions

l. Mg-rich hollow sanidine in the Rose dome graniteprobably grew from a supersaturated melt.

2. Supersaturation of the granitic partial melt probablywas induced by loss ofvolatiles rather than loss ofheat.

3. The sanidine/granitic melt Mg partition coefficientis greater than five, indicating ion-exchange between thegranitic melt and the mica peridotite that enclosed thegranite xenoliths.

4. Ion exchange between the granitic melt and micaperidotite probably was facilitated by K- and Mg-richmetasomatic fluids.

5. The sanidine compositions indicate the exisitence ofa KrMgSirO,u feldspar component, and document the oc-crurence of local charge imbalance in feldspar.

Acknowledgments

B. M. Barnes has our respect and thanks for keeping the mi-croprobe operational. M. E. Bickford and G. E. Lofgren helpfullyreviewed our original manuscript. Valuable suggestions and com-ments from D. M. Christie are much appreciated.

References

Albee, A. L. and Ray, L. (1968) Correction factors for electronprobe microanalysis of silicates, oxides, carbonates, and sul-fates. Analytical Chemistry, 42, 1408-1414.

Basu, A. and Vitalianio, C. J. (1976) Sanidine from the MesaFalls Tufl Ashton, Idaho. American Mineralogist, 61, 405-408.

Bence, A. E. and Albee, A. L. (1968) Empirical correction factorsfor the electron microanalysis ofsilicates and oxides. Journalof Geology, 76, 382-403.

Betterman, P. and Liebau, F. (1976) Lanthanum<alcium-so-dium aluminosilicates La"Ca, -r.Na"[Alr6sir6O61]. Naturwis-senschaften, 63, 480.

Bickford, M. E., Mose, D. G., Wetherhill, G. W., and Franks, P.C. (1971) Metamorphism of Precambrian granitic xenoliths ina mica peridotite at Rose Dome, Woodson County, Kansas:Part l, Rb-Sr isotopic studies. Geological Society of AmericaBulletin, 82, 2863-2868.

Bryan, W. B. (1972) Morphology ofquench crystals in submarinebasalts. Journal of Geophysical Research, 77 , 5812-5819.

Bryan, W. B. (1974) Fe-Mg relationships in sector-zoned sub-marine basalt plagioclase. Earth and Planetary Science Letters,24 , r57-165.

Crawford, M. L. (1973) Crystallization of plagioclase in marebasalts. Proceedings ofthe Fourth Lunar Science Conference,705-7 17 .

De Pieri, R., De Vecchi, G., Gregnanin, A., and Piccirillo, E. M.(1977) Trachyte and rhyolite feldspars in the Euganean hills(northern ltaly). Memorie degli Instituti di Geologia e Miner-alogia dell'Universita di Padova.

De Pieri, R. and Quareni, S. (1978) Partition coefrcients ofalkaliand alkaline-earth elements between alkali feldspar pheno-crysts and their lava matrix. Mineralogical Magazine, 42, 63-67 .

Emmons, R. C. (1943) The univeral stage. Geological Society ofAmerica Memoir 8.

Faust, G. T. (1936) The fusion relations of iron-orthoclase, witha discussion ofthe evidence ofthe existence ofan iron-ortho-clase molecule in feldspars. American Mineralogist, 21,735-763.

Fenn, P. M. (1973) Nucleation and growth of alkali feldsparsfrom melts in the system NaAlSi3OE-KAlSi3Os-HrO. Ph.D.Thesis, Stanford University, Palo Alto.

Fenn, P. M. (1977) The nucleation and growth of alkali feldsparsfrom hydrous melts. Canadian Mineralogist, 15, 135-161.

Franks, P. C., Bickford, M. E., and Wagner, H. C. (1971) Meta-morphism of Precambrian granitic xenoliths in a mica peri-dotite at Rose Dome, Woodson County, Kansas: part 2, pet-rologic and mineralogic studies. Geological Society of AmericaBulletin, 82, 2869-2890.

Hart, S. R. and Davis, K. E. (1978) Nickel partitioning betweenolivine and silicate melt. Earth and Planetary Science Letters,40.203J19.

Jarosewich, E., Nelen, J. A., and Norberg, J. A. (1980) Referencesamples for electron microprobe analysis. Geostandards News-letter,4, 43-47,257.

Jones, A. P., Smith, J. V., Dawson, J. B., and Hansen, E. C.( I 983) Metamorphism, partial melting, and K-metasomatismof garnet-scapolite-kyanite granulite xenoliths from Lashaine,T anzania. Journal of Geology, 9 1, 1 43-1 65.

Jurewicz, S. R. and Watson, E. B. (1984) Distribution of partialmelt in a felsic system: the importance of surface energy. Con-tributions to Mineralogy and Petrology, 85, 25-29.

Leeman, W. P. and Phelps, D. W. (1981) Partitioning of rareearths and other trace elements between sanidine and coexist-ing volcanic glass. Journal ofGeophysical Research, 86, l0 I 93-1 0 1 9 9 .

Lofgen, G. (1974) An experimental study of plagioclase crystalmorphology: isothermal crystallization. American Journal ofScience, 274,243-273.

Lofgren, G. (1980) Experimental studies on the dynamic crys-tallization of silicate melts. In R. B. Hargraves, Ed., Physics

SMITH AND FRANKS: Mg-RICH HOLLOW SANIDINE 67

of Magmatic Processes, p. 487-551. Princeton University Press,Princeton.

Lofgren, G. E. and Gooley, R. (1977) Simultaneous crystalliza-tion of feldspar intergrowths from the melt. American Min-eralogist, 62, 217 -228.

Long, P. E. (1978) Experimental determination of partition coef-ficients for Rb, Sr, and Ba between alkali feldspar and silicateliquid. Geochimica et Cosmochimica Acta, 42,833-846.

Longhi, J., Walker, D., and Hays, J. F. (1976) Fe and Mg inplagioclase. Proceedings of the Seventh Lunar Science Con-ference, I28l-l300.

Mahood, G. and Hildreth, W. (1983) Large partition coeftcientsfor trace elements in high-silica rhyolites. Geochimica et Cos-mochimica Acta, 47, ll-30.

Mason, R. A. (1982) Trace element distributions between theperthite phases of alkali feldspars from pegmatites. Mineral-ogical Magazine, 45, 101-106.

Mason, R. A., Smith, J. V., Dawson, J. 8., and Treves, S. B.(1982) A reconnaissance of trace elements in anorthoclasemegacrysts. Mineralogical Magazine, 46, 7 -l l.

Mehnert, K. R., Busch, W., and Schneider, G. (1973) Initialmelting at grain boundaries of quartz and feldspar in gneissesand granulites. Neues Jarbuch flir Mineralogie Monatshefte,I 65-1 83 .

Murphy, W. M. (1977) An experimental study of sotdJiquidequilibria in the albite-anorthite-diopside system. M.S. Thesis,University of Oregon, Eugene.

Philpotts, J. A. (1978) The law of constant rejection. Geochimicaet Cosmochimica Acta, 42,909-920.

Roedder, E. W. (1951a) The system KTGMgG-SiO2 part l.American Journal ofScience, 249, 81-130.

Roedder, E. W. (l95lb) The system KrO-MgO-SiO, part 2.American Joumal of Science, 249,22L248.

Roedder, E. W. (1952) A reconnaissance of liquidus relations inthe system KrO.2SiOr-FeO-SiOr. American Journal of Sci-ence, Bowen Volume, 435-456.

Ryerson, F. J. and Hess, P. C. (1978) Implications of liquid-liquid distribution coefficients to mineralliquid partitioning.Geochimica et Cosmochimica Acta, 42,921-932.

Schairer, J. F. (1948) Phase equilibrium relations in the quater-nary system KTGMgO-AlrOr-SiO, (preliminary report). Geo-logical Society of America Meeting, New York, November1948 (not seen; extracted from Roedder, 1951b).

Schifman, P. and Lofgren, G. E. (1982) Dynamic crystallizationstudies on the Grande Ronde pillow basalts, central Washing-ton. Journal ofGeology, 90,49-78.

Sclar, C. B. and Kastelic, R. L. (1979) Iron in anorthite: synthesisand characterization ofthe CaAlrSirOr-CaFeSirO. series. (abstr.)Transactions of the American Geophysical Union (EOS), 60,

Sclar, C. B. and Stead, R. J. (1981) Kinetics of synthesis anddisproportionation reactions of CaMgSiOr. (abstr.) Transac-tions of the American Geophysical Union (EOS), 62,410.

Sheerer, C. K. and Papike, J. J. (1985) Chemistry ofpotassiumfeldspars from three zoned pegmatites, Black Hills, South Da-kota: implications concerning pegmatite evolution. Geochim-ica et Cosmochimica Acta, 49,663473.

Smith, J. v. (1974) Feldspar Minerals, Volume l. Springer-Ver-lag, New York.

Snyder, J. L. (1962) Geologic investigations, Central DavisMountains, Texas. Texas Journal ofScience, 14,197-215.

Su, S. C., Bloss, F. D., Ribbe, P. H., and Stewart, D. B. (1984)Optic axial angle, a precise measure of Al,Si ordering in theT, tetrahedral sites of K-rich alkali feldspars. American Min-eralogist, 69, 440-448.

Swanson, S. E. (1977) Relation of nucleation and crystal-growthrate to the development of granitic textures. American Min-eralogist, 62, 966-97 8.

Takahashi, E. and Irvine, T. N. (1981) Stoichiometric control ofcrystal/liquid single-component partition coeftcients. Geo-chimica et Cosmochimica Acta, 45, 1l8l-l 185.

Watson, E. B. (1976) Two liquid partition coefficients: experi-mental data and geochemical implications. Contributions toMineralogy and Petrology, 56, 119-134.

Watson, E. B. (1977) Partitioning of manganese between for-sterite and silicate liquid. Geochimica et Cosmochimica Acta,4 r . 1363-1374.

Wones, D. R. andAppleman, D. E. (1963) Propertiesof synthetictriclinic KFeSirOr, iron-microcline, with some observationson the iron-microcline:iron-sanidine transition. Journal of Pe-trology, 4, l3l-137.

Manuscript received, December 14, 1984;acceptedfor publicalion, September 16, 1985.