American Mineralogist, Volume 69, pages 531-540, I9E4

Magnesium staurolite and green chromian staurolite from Fiordland, New Zealand

C. M. Weno

Department of Geology,University of Otago, Dunedin, New Zealand

Abstract

Staurolite with up to 65% MgO and magnesium as the dominant divalent cation isrecorded from an incompletely reconstituted metatroctolite from Fiordland, New Zealand.It occurs in complex assemblages including pyrope, hornblende, spinel and corundum, andis estimated to have crystallized at about 12 kbar and 750"C. The a cell edge of thisstaurolite (7.E914) is markedly larger than that of other natural staurolites, which may beattributed to substantial substitution of Mg for Al in the "kyanite" layer of the staurolitestructure.

A green staurolite with2Vo Cr2O3 is also recorded from a mafic metatuffrom Fiordland.

Introduction

In their thorough review of staurolite crystal chemistry,Griffen and Ribbe (1973) noted that staurolite exhibits"remarkable chemical constancy for a mineral so chemi-cally and structurally complex". This relative constancyis at least in part due to the rather narrow range ofmetamorphic conditions and bulk rock compositions inwhich staurolite has generally been observed-typicallyin moderately high grade pelites. Several features of itschemistry hint at the wide range of staurolite composi-tions possible in diferent bulk compositions metamor-phosed under appropriate conditions. The affinity ofstaurolite for zinc, with ZnO contents commonly up to2Vo (e.9., Guidotti, 1970) and occasionally 7.5Vo (Juur-inen, 1956), is well known, and Griffen (1981) has recentlysynthesized pure Zn-staurolite. Blue cobalt staurolite or"lusakite" with 8.5Vo CoO associated with cordierite-anthophyllite gneiss has been described from Zambia(Skerl and Bannister, 1934). Gibson (1978) has describedan occurrence of staurolite in sheets of amphibolite andhornblendite in a metamorphosed layered basic intrusionin central Fiordland, New Zealand. The staurolite isunusually low in iron, compensated by higher magnesiumand aluminum contents. Hellman and Green (1979) haveexperimentally produced magnesium staurolite (Mg >Fe) from tholeiitic compositions at high pressure, andSchreyer and Seifert (1969) synthesized iron-free Mg-staurolite.

During investigation of a complex polymetamorphicterrane in the Dusky Sound area of Fiordland, 20-40 kmsouthwest of the localities described by Gibson (1978),staurolite has been observed in a wide variety of rocktypes. ln addition to typical low pressure pelites withandalusite and higher pressure pelites with kyanite andgarnet, staurolite host rocks include metagabbro, meta-

troctolite, probable mafic metatuffs, and a metabasaltdike. These metabasites are geologically independent ofeach other, and of the body described by Gibson (1978).In many of these rocks the staurolite is present in veryminor quantities (<0.I modal percent), is fine grained andoften paler than usual. It seems likely that staurolite ismore widespread in metabasites than the present meagerrecord would suggest.

Many staurolites from the Fiordland metabasites havecompositions differing only subtly or not at all from thoseof pelitic staurolites. However this paper presents data ontwo extreme varieties found in metabasites of the DuskySound area, being the first records of natural staurolitewith magnesium as the dominant divalent cation, and ofgreen, chromian staurolite.

Nomenclature

"Magnesium staurolite" is used in this paper to denotestaurolite with magnesium cations in excess of all othersexcept aluminum and silicon (where all iron is combinedas Fe2*). "Magnesian", "chromian" etc. imply unusual-ly large, though still subordinate magnesium, chromiumetc. contents. Magnesium staurolite should not necessari-ly be regarded as a new end-member mineral speciesanalogous to iron staurolite in which Fe is the dominant(divalent) cation, since Fe and Mg may occupy differentsites in the structure (Smith, 1968; Griffen and Ribbe,1973:' and see below).

"Lusakite" (Skerl and Bannister, 1934) is a cobaltstaurolite with Co > total Fe > Mg, the only silicatemineral with a substantial cobalt content. It has not in thepast been accepted as a valid mineral species, perhapsbecause of a misconception of the actual composition;thus Fleischer (1980) quotes the formula as (Fe2*,Mg,Co)zAlqSLOB(OH), implying that Fe > Mg > Co.

0003-004)v84l0506-053 l $02. 00 531

532 WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE

Recently Cech et al. (1981) have reasserted the claim of"lusakite" to be a valid species on the grounds that cobalt isthe dominant cation. However, it must be presumed thatthe crystal chemistry of Co in staurolite is as complex asthat of Fe and Mg. In the absence of definitive data on thedistribution of the Co, the non-commital term "cobaltstaurolite" is more appropriate.

Analytical methods

Mineral analyses were made with the University ofOtago's lnol rxa-51 electron microprobe, an automatedcrystal spectrometer machine, using a variety of naturaland synthetic oxides and silicates as standards. The datawere corrected on-line by the method of Bence and Albee(1968). Six 5-second counts were integrated for eachelement analysis. Greater precision was obtained forsome trace elements (Zn,Mn in Table I staurolites; V, Cuin Table 3 staurolites) by integrating ten l0-second countson peak and background. Vanadium analyses were cor-rected for interference from TiKB. Tabulated analyses arethe means of 2-4 similar analyses, usually on. the samegrain.

Magnesium staurolite

Magnesium staurolite was found in specimen OU48255, a partially metamorphosed ultrabasic rock whichwas probably a hornblende olivine gabbronorite (Streck-eisen, 1976) prior to metamorphism. For the sake ofbrevity, and to emphasize the role of igneous olivine andplagioclase as the major reactants responsible for themetamorphic assemblages in the rock, it is hereafterreferred to as a metatroctolite. The specimen is a pebblecolleeted from float in a small stream draining amphibo-lite-facies metasediments and dioritic orthogneiss northof Dusky Sound. It is tentatively inferred that it wasderived from a xenolith in the orthogneiss.

The igneous mineralogy of the specimen has beenlargely reconstituted by extension of irregular metamor-phic coronas between olivine or orthopyroxene and pla-gioclase to encompass most of the rock. Staurolite ispatchily distributed in trace quantities as two types,designated I and 2, which are distinguished primarily onthe basis of their local parageneses, and secondarily onthe basis of their compositions (Table 1). Type I stauro-lite, with Mg about 45, occurs with hornblende andpleonaste spinel (tcorundum, -rclinozoisite, +anor-thite). Type 2 staurolite, with Mg about 55, occurs withgarnet and more magnesian hornblende. Both types ofstaurolite are exceptionally magnesian, with about 5Voand, 6%o MgO respectively, compared with the normalrange of l-3Vo.

Petrography

Specimen OU 48255 is heterogeneous on a scale of afew millimeters, partly because of initial (igneous) hetero-geneity on this scale, and partly because of the variable

Table l. Microprobe analyses of staurolite from metatroctoliteou 48255

Type 1

L 2 3 4 5

sio2Alz0eGzos@

!rcll9or@ZnOriq

26.88 27.40 27.3953.98 53.99 54.370.00 0.01 0.010.07 0.03 0. I5

o. ls 0.19 0.155.05 5.11 5.22

11.86 rt.o? 10.940.05 0.06 0.000.07 0.14 0.10gErd tt3d 9135

27.L9 27.75 27.4256.16 54.77 5{.550.0I 0.02 0.020.02 0.04 0.02

0.04 0.07 0.035.61 5.27 5.488.30 8.74 9.310. I3 0.02 0.03

.0.34 0. I7 0.23qrgc vtTt rfud

Caticn8 per 46 oojEens

siAJ.GCa

ltrrEFe7nTi

7.406 7.52L 7.488L7.526 L7.467 L7.5L60.000 0.003 0.0010.020 0.010 0.043

0.034 0.044 0.0342.073 2.092 2.L272,733 2.5t2 2.5020.009 0.010 0.0000.015 0.029 0.021

2'T zTT 2TIT

7.374 7.538 7.44817.9'18 17.531 17.4990.001 0.005 0.0040.006 0.011 0.006

0.009 0.015 0.0072.269 2.537 2.6221.882 1.985 2.1150.026 0.004 0.0050.068 0.034 0.048

z''r 2'3F. ZTJT

w, 43.1 45.L 45.9 54,7 56.1

Ihit celt dincnsicre, r:rp. Z 6t 4 7.891(3)b 16.617(s)c 5.658(2)

* 100 !,tgl(!t +rb)

nature and extent of the metamorphic reactions whichhave incompletely reconstituted the rock. Olivine (Fozs),orthopyroxene (En6e), clinopyroxene (WoasEnesFsT, 5VoAl2O3), plagioclase (Anzs_82) and poikilitic hornblende arepresent as igneous relics in minor amounts, generally nomore than a few percent each, though locally dominant.The olivine is surrounded by extensive coalescing coro-nas consisting of orthopyroxene (locally with symplecticintergrowths of magnetite), followed by pale green horn-blende then a symplectic intergrowth of fine vermiculargreen spinel (pleonaste) within coarse irregular horn-blende. In more thoroughly reconstituted patches olivineand igneous pyroxenes are lacking, metamorphic ortho-pyroxene and hornblende are dominant, with prominentspinel both in symplectite with hornblende and lesscommonly in a fine granular mosaic. Dolomite and mag-netite may also be present. Relict igneous plagioclase isrimmed by anorthite (Ansa_ee) often accompanied byclinozoisite (and zoisite?). Hornblende-spinel enclosesplagioclase but spinel may be absent within a few hundredmicrons of the plagioclase, very thin plates of corundum(<300 x 3-5 pm) or fine prisms of type I staurolite beingpresent instead (Fig. 1a). The corundum, staurolite and

WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE 533

Garnet was produced by a further reaction affectingsubstantial domains interpenetrating with the garnet-freedomains. Garnet clearly replaces hornblende-spinel sym-plectites (Fig. lb); additional orthopyroxene appears tobe consumed also. Extensive aggregates of equigranularpolygonal hornblende generally lie adjacent to the garnet.This hornblende is significantly more magnesian than thatremote from garnet (see below and Table 2); it has clearlyrecrystallized in response to growth ofthe garnet. Chlor-ite is an infrequent associate of this assemblage but is notfound in garnet-free domains. Small irregular inclusionsof relict spinel or hornblende are common but not ubiqui-tous in garnet. Semi-radiating sprays of platy corundummay be present instead, and rarely, granular corundumoccurs with hornblende adjacent to garnet. A few subhe-dral staurolite prisms (<200 x 50 pm), designated type 2also occur as inclusions in garnet (Fig. lc) or as separateadjacent grains.

Table 2. Microprobe analyses of minerals in metatroctolite,ou 48255

L 2 3 { 5 6 7 8 9bbl EL aIrI gBt gnt got cbl opn eo

/, / | -17 4.92 O.l7 {0.69,40.33 /aO.O2 27.97 53.54 39.15l?-: I t L.5l 6. t .33 23.12 22.92 22.1L 22-6t 3. I5 3I.15

o.o5 0.L 0 0.o2 0.02 c.03 0.o2 0.o3 o.o7l . o 2 3 . 2 0

sio2u293rn%

'fo2%

ruhotlgoco

xi2ot2oezorm

7.96 6.57 t9.Ar l7.s rA.77o.Ll 0.o5 0.Io 0.57 0.51

l{.oo I7.o5 L.62 l{.5A 12.30ur.l6 12.03 0.o€ a.o6 s.71

I.63 l.9A O O.O2 0.O3o.ot o.o5 0 0.ol oo.ol o o.o3 0.ol o.02

o.o7

r7.o5 5.33 lt.a2 o.aao.a6 0 0 .31 0 . Io

lrr.20 30.28 28.06 0.1s7JU O.O1 0 .20 23 .34

o 0 .o3 0 0 .02o o 0 .o l oo .o l o .o t 0 .o3 0

Fig. l. Metamorphic textures in metatroctolite OU 48255.See Fig. 2 for mineral abbreviations. Scale bar 0.2 mm in eachphotomicrograph. (a) Reaction zone between orthopyroxene(heavily outlined at right) and plagioclase. Spinel, staurolite andcorundum occur as fine inclusions in coarse hornblende. planepolarized light. (b) Two textural variants of hornblende-spinelsymplectite (center) partially replaced by garnet (center andright). Garnet and granoblastic hornblende to left. Crossedpolarizers. (c) Garnet (right-center to upper left) with stumpystaurolite inclusions at upper left and slender corunduminclusions right-center. Crossed polarizers.

spinel grains may show a crude zonation away fromplagioclase towards orthopyroxene in the above order oroccur mixed in pairs, all as dispersed grains within a"background" of hornblende.

97.55 97.30 100.26 lOO.92 lqr.?3 9!t .6r A5.25 IOO.20 98.05

f,iEEt Ftule

si 6.293 6.3!17 o.q!g 5.962 5.975 5.98,1 s.3Ao I .9I9 2.999rI ' 2-920 2-4 3.942 3.992 a.qr2 3.949 5.125 0.133 2.A13?i O.OO7 O.Ori O O.q)2 O.qr3 O.OO3 0.00,1 O.d)l O.0O4

tr"3+ o.(to o.r&

Fc2+ o.! ta9 o.?s3 o.a53 2.1a5 2.327 2.L32 o.a5? o., t4,a o.o5?h o.ollt o.dt? 0.oo5 0.070 0.075 0.050 0 0.oo9 0.oo7r|9 2.97t 3.C;2O 1.136 3.tal 2-7L7 2.720 A.5ar 1.498 O.Ol7Ca 1.456 1.436 O.qt5 0.638 O.9II l . l9 l O.OO2 0.009 1.916

h o.t50 0.5|i!i o o.(Xr o.qr9 0 0.oI2 0 0.oo{x o . o o 6 0 . q x r o o . d t 2 0 0 0 0 0e o.f l ) l o o.qr l 0.ool o.oo2 o.d) l 0.001 0.001 0b o.fir3 -

l5.l6s L!t.6a7 6.q!2 16.0{3 15.023 l5.O3A 20.060 t.O13 8.qr0

2 3 2 1 4 2 1 2 1 2 1 2 5 6 1 2

5 3 . 9 5 6 . 1 9 l . O 7 7 . t

. Iffi disfui.tueal bebpq Pel9s @ril FeO @ditrg to otoichi@tryirr sptel @til cliwisite. Atl i@ a F& ia ot)et mireple .

7, hotrtk .b b sqrwl od qpa 7 ot@lita2, haribl4r& dj@t to g@tX, spiwl di@x to hotrtZenib 1 qil qp 7 st@lita1, g@t w 5, g@t aili@t to tupe 2 ot@Lite

to)

,e 75.a s;2.2 55.A 59.2

IryEotE 52.4,r .drne 36.0gtBaolE 1O.5slaasaftle L.2

a 5 . l a . 63A.5 3{.915.r 19.5

r . 3 l . o

6, g@t ?ia 7, ehldrte8, Et@rphia I, cliMoisita

5t4 WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE

Mineral chemistry

The patchiness of the two groups of assemblages (i.e.,those with and without garnet) on a millimeter scaleimplies a very limited range of diffusion and equilibration.This impression is supported by the mineral chemistry.The equilibration of garnet with type 2 staurolite but nottype I is demonstrated by the consistently lower Mn andFe contents of type 2 staurolites (Table l) and their higherMg noted above. Other variations in staurolite chemistryprobably reflect local variation in the chemical potentialsof some components. Thus staurolite 4, with significantlyhigher Al and lower (Mg+Fe) contents than the otherstaurolites, is an inclusion in garnet which (atypically)impinges on plagioclase, the primary source of the alumi-na in the metamorphic system. Staurolite 3, with asignificant trace of calcium, is in contact with clinozoi-site. (The CaO content was consistent in multiple analy-ses and is not an artifact ofthe edge effect.) The titaniumcontent is variable and low (cf. Grifen and Ribbe, 1973)reflecting the low bulk rock titanium content-primaryhornblende with less than lVo TiO2 is the only significantsource of titanium.

Hornblende analyses I and2 (Table 2) are analogous totypes I and 2 staurolite. The lower Mn content and higherMg ratio of hornblende 2again relate to equilibration withgarnet. The sodium content of hornblende, though mod-est, is only slightly lower than that of the very smallquantity of surviving igneous plagioclase (-Aneo) andconsiderably higher than that of metamorphic plagioclase(-Anss), hence the majority of the sodium in the systemis in hornblende. The higher sodium content of horn-blende 2 relative to hornblende I is thus interpreted as aconsequence of the reduced modal hornblende after crys-tallization of garnet.

Garnet is zoned and pyrope-dominant (Table 2). Garnet5 enclosing staurolite has a lower Mg than the others,such that the usual relationship that garnet has a slightlylower Mg than coexisting staurolite (Albee, 1972) isretained. The higher almandine component in this garnetsuggests that staurolite may be restricted to domainsslightly richer in iron. The marked enrichment in thegrossular component from core to rim (garnets 4-6, Table2) is perhaps a reflection of increasing pressure duringcrystallization.

A feature of the clinozoisite composition is the appar-ent substitution of (Fe2+ + Mg + Mn) for about4Vo of theCa atoms.

Re ac tion re lations hip s

Two sequential groups of metamorphic assemblagesare established on the basis of the petrography andmineral chemistry, garnet and chlorite being restricted tothe second group. Suggested reactions are as follows:

orthopyroxene * plagioclase * H2O--+ hornblende(l) + (spinel, staurolite, corundum)

+ anorthite (-+clinozoisite) (1)

and subsequently

hornblende(l) + spinel f orthopyroxene--+ hornblende(2) + garnet tchlorite (2a)

and hornblende(l) + spinel + orthopyroxene--+ hornblende(2) + garnet + (corundum,staurolite).

(2b)

Rigorous chemographic analysis of these reactions in amulticomponent silica-deficient system would be difficultif not impossible. There are no suitable phases fromwhich to project the important minerals onto a 3- or 4-component system. In practice, simple ACF diagramswere found to be useful. This involves the assumption,among others, that silica is sufficiently mobile that spatialvariation in silica activity is not a significant factorlimiting the assemblages. In this way, the essential fea-tures of the two groups of reactions are shown in Figure2, where the somewhat variable mineral compositions arerepresented by points. The "difficulties" caused by silicaare relatively minor, for example consideration of thesilica balance of reaction I suggests that a silica-poorphase is required as a reactant. Presumably this is magne-tite, to account for the lower Mg ratio of (hornblende +spinel) relative to orthopyroxene, and the significantquantity of Fe3+-bearing clinozoisite produced. ReactionI thus becomes

orthopyroxene + plagioclase + magnetite + H2O-+ hornblende + (spinel, staurolite, corundum)

* anorthite + clinozoisite

M e tamorp hic conditions

The formation of pyrope garnet and magnesium stauro-lite implies a high pressure of metamorphism. Schreyerand Seifert (1969) synthesized Mg-staurolite in an iron-free system at high pressures and temperatures, but could

Fig. 2. ACF diagrams illustrating the assemblages andreaction relationships in OU 48255. A is molecular Al2O3-Na2O-KzO, C is CaO, and F is FeO + MgO. Mineral abbreviations: an -

anorthite; plag - plagioclase; czo - clinozoisite; dol - dolomite;hbl - hornblende; mt - magnetite; hyp - hypersthene; spl - spinel;stl - staurolite; co - corundum; gnt - garnet; chl - chlorite; (A)represents the first group of reactions, basically the breakdownofhypersthene + plagioclase; and (B) rgpresents the subsequentreactions, basically the breakdown of hornblende (l) + spinel,but note that hornblende (2) persists with staurolite andcorundum.

WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE 535

not find a stability field below 12 kbar and 800'C. Incontrast Fe-staurolite is stable to pressures as low as Ikbar (Richardson, 1968), implying that the Mg content ofstaurolite in Mg-Fe systems will be strongly pressuredependent. This is corroborated by the experimentalproduction of staurolite with Mg 54-57 in model tholeiitecompositions at about 750" and 25 kbar water pressure byHellman and Green (1979). Under these conditions thestaurolite Mg approaches that of the bulk rock, 70.

More precise assessment of the metamorphic condi-tions affecting this complex natural system, which in-cludes mixed volatiles as indicated by the presence ofboth hydrous and carbonate minerals, is a difficult prob-lem. The limited range of equilibration and the metastablepersistence of substantial volumes of igneous relics andearly metamorphic assemblages, suggestive of fluid-defi-cient conditions, are further complicating factors. Theuncertain origin of the metatroctolite specimen precludesthe use of assemblages in adjacent rocks to assist indetermination of the metamorphic conditions.

In a pioneering study Obata and Thompson (1981)analyzed reaction topologies in the model mafic-ultramaf-ic system CaO-MgO-Al2O3-SiO2-H2O, calibratedagainst the limited experimental data. Their reaction 26(hornblende * orthopyroxene + spinel + anorthite :garnet + chlorite) is closely analogous to reaction 2a.This reaction is "water conserving" hence essentiallyindependent of a(HzO), and is predicted to occur at 11-13kbar and 700-800"C. The effect ofiron on the system, andthe changed topology with the addition of phases notconsidered by Obata and Thompson such as clinozoisiteand staurolite, are undoubtedly important. Neverthelessthese figures are a useful starting point for an estimate ofthe P-? conditions responsible for the formation ofmagnesium staurolite in the metatroctolite.

Some indication of the temperature of metamorphism isgiven by Fe-Mg partitioning between adjoining clinopyr-oxene and orthopyroxene rims. Clinopyroxene, which isconsidered to be entirely of relict igneous origin, showsextensive exsolution of fine hornblende lamellae (cf.Yamaguchi et al., 1978). A thin rim (-30 g.m) is generallyfree of these lamellae and markedly depleted in Al2O3(ZVocf . 5%). The rim is interpreted to have metastably equili-brated with the immediately adjacent phases. Two-pyrox-ene temperatures for these materials calculated afterWells (1977) are reasonably consistent at 760'C (130).Although this estimate is compatible with that derivedfrom the phase relations of Obata and Thompson (1981),it must be treated with caution since it is unclear when thepyroxene rims were last able to equilibrate relative to thetiming of the metamorphic reactions.

Schreyer (1967) suggested that the lack of naturalmagnesium staurolite was due to the absence of veryaluminous, magnesian bulk compositions metamor-phosed at sufficiently high pressure. In a sense thepresent exceptional staurolite reinforces this view. It isinvolved in only local equilibrium with other aluminous

phases, while a substantial proportion of relict pyroxeneand olivine remains. If the rock had been fully reconsti-tuted the staurolite is unlikely to have survived. Formagnesium staurolite to form in more typical basic com-positions, presumably much higher pressures are re-quired, tending towards the 24-26 kbar used by Hellmanand Green (1979) to crystallize magnesium staurolite inmodel tholeiite compositions.

S taurolit e cry s t al c he mis try

Idealized formulae such as (Fe,Mg)+AlraSi7.5O44(OH)4,(Fe,Mg)d|z 33Si8O44(OH)+, or (Fe,Mg)4AlrESi8O46(OH)2misrepresent the complex crystal chemistry of staurolite(Smith, 1968). Griffen and Ribbe (1973) inferred from astatistical study of staurolite analyses that Fe, Mg and Alare each distributed over both the tetrahedral Fe site andthe octahedral Al sites, with Fe preferring the former andMg and Al the latter. Thus Fe and Mg are not related bythe simple one-for-one substitution scheme implied by theformulae above. The present analyses were normalizedon a 46-oxygen basis, equivalent to assuming 4H performula unit. On this basis the total of Mg + Fe cations isunusually high, up to 4.81, compared with a maximum of4.47 and mean 4.07 for the 18 staurolites with less than lVoZnO analysed by Gritren and Ribbe (1973). This feature isalso manifested by unusually high cation totals, 29.58 to29.82 (cf 29.33 to 29.55, mean 29.43 recorded by Griffenand Ribbe).

Figure 3 explores relations between Mg + Fe and Alcell contents with plots of Mg * Fe against Al, andagainst Al', where AI'is Al + Si-8 (that is, Al cations perunit cell after subtacting sufrcient Al to make up theoccupancy of the Si site to 8). The markedly improvedcorrelation after this adjustment is obvious, and this is

r = o s 0

17-a t73 l7' 18! t63 l7.O 172 t7A

Al (cations per unit cell) Al'

Fie. 3. (A) Mg + Fe against Al, and (B), Mg + Fe against Al'(= Al+Si - 8) for staurolite from metatroctolite OU 48255.Least squares regression lines are shown. In (A) the regression isentirely controlled by the most aluminous staurolite. Slope of theregression line in (B) is -1.59.

(A) ( B )

og

o=

oo' E 4 Sf

o

=@o

()4+o

\.

r = o 9 e

WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE

strong support for the common assumption that it is Alwhich makes up the Si site to essentially full occupancy(Smith, 1968; see also Griffen et a1., 1982). The extent ofthe adjustment is considerable because of the variableand low Si content ofthe unit cells. The low Si content ispresumably a manifestation of the low silica activity ofthe rock, indicated by the absence of quartz and theabundance of spinel with minor corundum.

The very good negative correlation of Mg + Fe againstAl' (despite considerable variation of Md supports Grif-fen and Ribbe's (1973) deduction of an important(Mg,Fe)=Al substitution which they inferred to takeplace mainly in the octahedral sites. The slope of theregression line is close to -1.5, the figure to be expectedfor the simple substitution 3(Mg,Fe)=2A1.

Cell dimensions

Griffen and Ribbe (1973) plotted staurolite unit celldimensions against iron content, and demonstrated that dwas insensitive to iron content while b, and to a lesserextent c, both increase with increasing iron. These resultswere interpreted in terms of the structure (Smith, 1968),which has a monolayer with the approximate compositionFe2Als 7O2(OH)2 (the "iron" layer, see Fie. 4) lyingparallel to (010), alternating with slabs of Al2SiO5 compo-sition and kyanite structure. Since Fe2* is the largestsignificant cation in staurolite and assuming that all ormost of the cation substitution takes place in the ironlayer, increasing Fe content causes expansion of thethickness of the iron layer thus increasing b. ln the adirection the Fe tetrahedra are flanked by only slightlyoccupied U octahedra which are readily deformed, hencea does not increase. The Fe tetrahedra are linked by Al(3)octahedra to form azig-zag chain parallel to c, hence theexpansion of c with increasing iron content. Because of

Fig. 4. The "iron" layer of staurolite, cf. Dickson and Smith(1976) and Ribbe (1980). (Note however the distortion of theseauthors' diagrams due to assumption that c/a is 0.5 rather than0.70) The unit cell boundaries are shown by the dotted lines.

the indirectness of this linkage, and the geometric integri-ty of the adjoining kyanite layers, the increase in c issmall.

Griffen et al. (1982) have recently refined this analysisby plotting cell dimensions against mean ionic radius ofthe cations inferred to occupy the iron layer, to allow forthe effect of deviations from simple Fe,Mg chemistry. AllFe, Mg and Zn are assumed to be in the iron layer,together with Al in excess of that in the kyanite layer, andcations are distributed between tetrahedral and octahe-dral sites according to defined rules. The cell dimensionsvary according to essentially the same pattern with im-proved correlation coefficients, and the interpretationwas reafrrmed.

Unit cell dimensions of a single crystal of type 2staurolite, determined with the use of a Gandolfi X-raycamera, are presented in Table l. Because of the smallcrystal size and slight contamination with garnet only 14lines were available for the least squares refinement of thedata. An orthorhombic cell was assumed.

Cell edges b and c of this staurolite lie close to theregression lines of cell edges against mean ionic radiuscalculated by Griffen et al. (1982), and within the field ofother staurolites.l However, at 7.8914, a is substantiallylarger than in other natural staurolites (range 7.865-7.8794), and this is particularly notable since the varia-tion observed in other staurolites is not significantlycorrelated with mean ionic radius in the iron layer. Thissuggests that a is varying in response to changes in sitesin the kyanite layer.

The outstanding chemical features of this staurolitewhich might account for the abnormal a cell edge are thehigh Mg content and the high cation total. Figure 5 showsa plotted against Mg cations and against cation total forthe magnesium staurolite and the 19 staurolites used byGritren et al. (1982). There are good positive correlationsin both cases, and these remain significant even if themagnesium staurolite is excluded from the regressions.On this basis it is suggested that significant substitution ofMg for Al occurs in the kyanite layer of staurolite, as wasmodelled by Smith (1968) in his site refinements. Thiseffect would appear to be stronger when the cation total ishigher, perhaps because ofthe lower number ofvacanciesand the resulting "competition" for sites.

The most likely site for divalent ions to substitute forA13+ in the kyanite layer is Al(2), because as noted byGriffen and Ribbe (1973) this site is coordinated to theunderbonded O(3) and half-hydroxyl O(lA) and O(lB).Al(2) sites are in pairs lying on either side of the iron

I It is assumed that the composition of the X-rayed stauroliteis the mean of analyses 5 and 6 of Table l, because it was pickedfrom a plagioclase-free part ofthe specimen. It was necessary toadd Mg as well as Al to the Fe site to make up the occupancy to 4cations since the Al content of the iron layer is insufficient. Theresultine mean radius is 0.6034.

,.%:,,,

l

I

0X r t

X G o

1.O t.6 2.O 2-6Mg cations per unit cell

'-nX s

X GRo

20.t

537

Griffen and Ribbe (1973) note that Fe-Mg staurolitessynthesized by Richardson (l%6) and Fe-free Mg-stauro-lite synthesized by Schreyer and Seifert (1969) have a celledges consistently about 0.0154 larger than the thenknown natural staurolites, whereas c is indistinguishableand b is perhaps a little smaller in the synthetic stauro-lites. These features are matched by the natural magne-sium staurolite. In the light of the foregoing discussion itis suggested that to a considerable extent divalent ions aresubstituting for Al in the kyanite layer of the syntheticstaurolites. Since the large a is also shown by the pure Fe-staurolite, it would seem that at least in the syntheticminerals both Fe and Mg can substitute for Al. Thesynthetic staurolites were all prepared at high pressure(-20 kbar), and the natural staurolite is also inferred tohave formed at high pressure (-12 kbar). It seemsreasonable to conclude that substitution of divalent ionsfor Al in the kyanite layer is promoted by high pressure ,and that consequently the a cell edge of staurolite may bea crude pressure indicator. A larger cell edge might beexpected to cause a lower density, which would beincompatible with higher pressure . However the negativeeffect of the cell edge on density is greatly exceeded bythe positive effect ofthe larger cation total (Fig. 5b).

Conclusions

Staurolite from the "metatroctolite" greatly extendsthe range of magnesium content known from naturalmaterial, and appears to be the first record of naturalmagnesium staurolite. The rock is very heterogeneousand includes significant volumes of alumina-poor igneousrelics, so that the local systems containing staurolite arevery aluminous. Two phases of metamorphic reactionsare apparent, each incomplete, leaving the rock dominat-ed by complex reaction textures. Metamorphic condi-tions responsible for the crystallization of the magnesiumstaurolite during the second phase, perhaps of higherpressure than the first, are estimated to have been in thevicinity of 12 kbar and 750'C.

WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE

ooEo

oI

(A)^ tle

?.oa

(B)

?.teo(

o

€ ,.""oo

6l

f.af

lotal cations por unit cell

Fig. 5. (A) Staurolite a cell edge against Mg, and (B), aagainst cation total, for the 19 staurolites used by Gritren et al.(19E2), together with thar from OU 4E255. Height of symbolsrepresents +l e.s.d., width is arbitrary. Data points representedby crosses were not included in the regression analyses becauseof major deviations from (Mg,Fe) chemistry. Dual symbol at topright represents magnesium staurolite from OU ,{E255, whichwas included in regressions (l) but excluded from regressions(2). J6 = staurolite 6, Juurinen (1956) (as refined by Grifien andRibbe, 1973), GR l0 = staurolite 10, Gritren and Ribbe, vK =von Knorring et al. (1979). J6 and GR l0 each have about 1.5 Zncations per unit cell. From the data of Griffen (l9El), their a celledges would be expected to increase by about 0.0074 if this Znwas replaced by Fe, putting the points close to the calculatedregression lines. It may be suggested that staurolite vK, from ashear zone in a pegmatitic crystal of amblygonite (LiA(pO4)F),owes its extreme composition (19.3 Al cations per 46 oxygens,2.4 (Fe+Mg+Zn), cation total 29.01) ro substantial substitutionof Li+Al for 2(Fe, Me, Zn).

The a cell edge of the magnesium staurolite is markedlyhigher than that of other natural staurolites but compara-ble to that of synthetic staurolites produced at 20 kbar. It

layer, sharing the short edge O(IAFO(IB). Substitution is inferred that significant substitution of Mg (and perhapsof larger cations in this site would readily increase a. In Fe) for Al takes place in the A(2) site of the kyanite layircontrast c would not be significantly affected, because it in staurolite, this effect being more pronounced at higheris largely controlled by the length ofcontinuous chains of pressures. The a cell edge ofstaurolite may be useful as aAl(l) octahedra parallel to c, as is shown by the unusually crude pressure indicator.short O(IAFO(IBFO(IA) distance (Smith, 1968). If sig- On the basis of their crystallizarion of staurolite innificant Mg were to substitute for Al in the Al(l) sites, c model tholeiite compositions at 24-26 kbar, Hellman andwould be expected to increase. This is not observed, Green (1979) suggested that staurolite may be widespreadalthough it might be that the contraction of c due to the in mafic rocks metamorphosed at high pressures, as forgenerally decreasing Fe with increasing Mg is masking example in subducted lithosphere. While the presentthis effect- Similarly the apparent absence ofany effect on occurrence of staurolite is certainly consistent with thisb from the inferred substitution of considerable Mg for Al suggestion, high pressure is evidently not a necessaryin Al(2) is readily explained by the concomitant substitu- condition for the formation of staurolites in general intion of Al for (Fe,Mg) in the Fe site. metabasites. Spear (19E2) has documented occurrences of

53E WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE

staurolite (with Mg 23-28 in Zn-poor crystals) in amphi-bolites metamorphosed at 5-6 kbar, and authigenic stau-rolite with Mg 19 occurs in a metabasite dike cuttingandalusite-bearing pelites south of Dusky Sound, Fiord-land (Ward, in prep.). The significance of higher pressurewould appear to be principally in promoting a high Mgratio in staurolite, i.e., reducing the partitioning of Fe intostaurolite relative to the more magnesian minerals. Thusat lower pressures staurolite may be restricted to themore iron-rich metabasites.

Chromian staurolite

Staurolite with the unusual pleochroic scheme X clear,slightly bluish green, Y yellowish green, Z muddy green-ish yellow, Z > Y > X, was observed in trace quantities intwo specimens of amphibole-rich rock from north ofDusky Sound. They are from a distinctive lithostratigra-phic unit generally about 20 m thick, dominated by rocksrich in amphiboles (50-99%) with no feldspar and general-ly little or no quartz. Aluminous minerals includinggarnet, chlorite, kyanite and staurolite are major phasesin some streaks, pods and bands within the amphi-bolerich rocks. From the evidence provided by sharplydefined hornblende pseudomorphs after igneous clinopyr-oxene, and the peripheral interbanding on a decimeter-meter scale of the amphibole-rich rocks with metasedi-ments including pelite and metaconglomerate, it isinferred that the unit is probably a metatuff of unusualcomposition.

Petrography

The two rocks with "green" staurolite, OU 48093 andOU 48097, both have the assemblage gedrite-quartz-hornblende-kyanite-chlorite-rutile-phlo gopite-stauro-lite-allanite. Magnesian chlorite, in apparent texturalequilibrium with the other phases, is present as a majorphase in OU 48097, but as only a trace in OU 48093.

Specimen OU 48093 is conspicuously banded withalternating quartz-rich (40:7 0%) and quartz-po or (2-10%)layers 2-10 mm thick. Pale green hornblende is a majorphase only in some of the quartz-poor bands. Thesehornblende-rich bands contain all the staurolite and mostofthe phlogopite and allanite. Staurolite generally occursas a minor associate of kyanite-phologopite (+quartz)pools interstitial to coarser hornblende and gedrite. It isfine-grained, with prisms typically about 50-80 prm long.On a thin section scale its modal abundance is less than0.01%. Petrographic recognition is hindered by its simi-larity to allairite in the same rock, in its pleochroism,relief and birefringence. Some staurolite grains can bedistinguished on the basis of 60'twins, and some allanitesby significant extinction angle, length fast orientation, ora dark radiation halo in adjacent hornblende. In additionto this unusual coloring for staurolite and allanite, somekyanite is also distinctly colored in thin sections, with Zslightly bluish green, Ypaler, X colorless. Gedrite, horn-

blende and phlogopite are normally colored. Except forthe greater abundance of chlorite, OU 48097 is similartexturally to the quartz-poor bands of OU 48093. Stauro-lite is very rare.

Chemistry

Microprobe analyses of the staurolite from OU 48093show the presence of approximately 27o CrzOr (Table 3).The zinc content is considerable but not unusual, andvanadium and copper are also present as significanttraces. Nickel and cobalt were not detected atthe 0.2Volevel. Griffen and Ribbe (1973) and Ribbe (1980) note thattraces of CrzO3 (<O.lVo) are commonly found in stauro-lite, and Sharma and MacRae (1981) have recently report-ed 0.2-0.8% Cr2O3 in staurolite from gedrite-cordieritegneisses in northern India, but this appears to be the firstrecord of staurolite with sufficient chromium to affect thecolor of the mineral.

Comparison of the analyses of chromian staurolite withthose of non-chromian staurolites from the same strati-graphic unit reveals an antipathetic relationship betweenthe A12O3 and Cr2O3 contents (e.g., analysis 4 in Table 3is almost identical to analysis 3 except for the virtualabsence of Cr2O3 in the former and the proportionallyhigher Al2O3). It is concluded that chromium is substitut-ing for aluminum presumably in the octahedral AI sitessince, although Al is inferred to be a minor substituent inthe tetrahedral Fe site (Smith, 1968; Griffen and Ribbe'1973), Crl+ is unknown in tetrahedral coordination(Burns and Burns, 1975; Bish, 1977; Phillips et al.' 1980).

Controls of chromium content

The range of chromium contents of staurolite and theother phases present in OU 48093 are listed in Table 4.Staurolite is the most chromium-rich mineral present, andthe chromium content of the staurolite-bearing bands isconsiderably greater than that of the staurolite-freebands. That such chromian staurolite has not been report-ed previously is presumably because of the rarity ofchromium-rich rocks sufficiently aluminous to form stau-rolite. It may also have been overlooked because of itscoloring, which mimics common hornblende.

Although the high Cr2O3 content of staurolite relativeto the other minerals present is conspicuous, staurolite'schromium content is seen to be no more than averagewhen recalculated as atomic Cr/Al (Table 4). This sug-gests that the large number of appropriate cation sites isthe main reason for the high chromium content, ratherthan any particular "affinity" of staurolite for Cr. Achromium phase is absent, hence these values need notrepresent saturated Cr contents. A nearby hornblende-cummingtonite rock containing metamorphic Al-chro-mite, OU 48088, has allanite with up to 4.5Vo CrzO: andhornblende with Cr/Al up to 0.10, values about five timeshigher than in OU 48093.

The chromian staurolite is estimated to have formed at

WARD: MAGNESIUM STAUROLITE, CHROMIAN STAUROLITE

Table 3. Microprobe analyses of chromian staurolite. OU 4g093

539

l, 2, 3 Eeparate cqEta.ts of chKnrian Etauolitefttttr Cn {8093

'l EtalrDlite frcn qt {8089

670"C and 8 kbar, from consideration of(l) the anthophyl-lite-forsterite-magnesite-talc assemblage in a nearby podof CO2-metasomatized "metaserpentinite" (Johannes,1969; Greenwood, 1976), and (2) Fe-Mg partitioningbetween garnet and biotite (Ferry and Spear, 1978) andCa partitioning between plagioclase and garnet (Ghent etal., 1979) in adjacent garnet-biotite-kyanite-plagioclase-quartz pelites. The influence of these metamorphic condi-tions on the chromium content of staurolite is difficult toassess. Seifert and Langer (1970) have shown experimen-tally that the saturated Cr2SiO5 content of kyanite in-

Table 4. Chromium content of minerals in OU 48093

creases markedly with pressure but is relatively insensi-tive to temperature. Although the structural differencesbetween staurolite and kyanite are undoubtedly signifi-cant, staurolite may well behave similarly.

Conclusions

Up to 0.45 cations per unit cell of chromium (2 wt.%Cr2O3) substitutes for aluminum in staurolite in a maficmetatuf from Fiordland. It is likely that staurolites withhigher chromium contents will be found in rocks of higherchromium content, perhaps to the extent that Cr3+ wouldbe the dominant cation after Si and Al. The green-dominated pleochroism of chromian staurolite presum-ably results from the superimposition of a chromium-generated green color on the normal yellow pleochroismof staurolite.

Acknowledgments

I am grateful to J. M. Pillidge for the quality of the polishedthin sections he produced, and to Drs. A. F. Cooper, A. Reayand D. S. Coombs for their useful comments on the manuscript.

References

Afbee, A. L. (1972) Metamorphism of pelitic schists: reactionrelations of chloritoid and staurolite. Geological Society ofAmerica Bulletin E3, 3249-3268.

Bence, A. E. and Albee, A. L. (l%8) Empirical correctionfactors for the electron microanalysis of silicates and oxides.Journal of Geology, 76,382-403.

Bish, D. L. (1977) A spectroscopic and X-ray study of thecoordination of Ct'* in chlorites. American Mineralogist, 62,3E5-389.

Burns, V. M. and Burns, R. G. (1975) Mineralogy of chromium.Geochimica et Cosmochmica Acta. 39. 903-910.

Cech, F., Povondra, P. and Vrana, S. (1981) Cobaltoan staurolitefrom Zambia. Bulletin de Min€ralogie, 104, 526-529.

Dickson, B. L. and Smith, G. (1976) Low temperature opticalabsorption and Mrissbauer spectra of staurolite and spinel.Canadian Mineralogist, 14, 206-215.

Ferry, J. M. and Spear, F. S. (197E) Experimental calibration ofthe partitioning of Fe and Mg between biotite and garnet.Contributions to Mineralogy and Petrology, 66, ll3-117.

Fleischer, M. (1980) Glossary of Mineral Species, 1980. Mineral-ogical Record, Tucson, Arizona.

Ghent, E. D., Robbins, D. B. and Stout, M. Z. (1979) Geother-mometry, geobarometry, and fluid compositions of metamor-phosed calc-silicates and pelites, Mica Creek, British Colum-bia. American Mineralogist, 64, 874-885.

Gibson, G. M. (1978) Staurolite in amphibolite and hornblenditesheets from the Upper Seaforth River, central Fiordland, NewZealand. Mineralogical Magazine, 42, 153-154.

Greenwood, H. J. (1976) Metamorphism at moderate tempera-tures and pressures. In D. K. Bailey and R. MacDonald, Eds.,The Evolution of the Crystalline Rocks, p. 187-259. AcademicPress, London.

Griffen, D. T. (1981) Synthetic Fe/Zn staurolites and the ionicradius of IvZn2+. American Mineralogist, 66,932-937.

Gritren, D. T., Gosney, T. C. and Phillips, W. R. (1982) Thechemical formula of natural staurolite. American Mineralogist,67,292-297.

sil(DAluOsGzocVzoEop@!rcllp

fto?',ATiJc[n*al

siAIGvCrCaItrll9

tb7,rrTiIbtal

28.07 27.3452.32 52.38I .78 2.010.04 0.08- 0 .05

0.02 0o.Lt 0. I02.79 2.43

28.555r.63r.75

00.153.19

11.160.7r0 .72gtrEd

11.190.900.75

9fr6

2.5890.1840.r57

2r3r

10.701.050.57

tf,7r

2.509o.2L70.120

z!;tf

28.3354.150.04

00 .053.36

LL.270.150 .57

sr$t

Caticn8 lEr 45 o.ygens

7.768 7.6A 7.89717.053 17.308 16.8340.390 0.445 0.3830.009 0.018- 0.014

0.006 0 00.031 0.023 0.035I.149 r.0r4 L.315

7.76317.4880.009

00.012r.372

2.5920.1450.15r

2t'3T

2.5820.0300.Lt7

zTtr

t€ldrt rffiit G,o.ffib.lrlrg blnda fr bed.

l|lrdnnamlcG/aI

.ta:roltt 1.8 - 2.0gdrft 0.2 - 0.6lmblcnde 0.4 - 1.0ktmtt3 0.4 - 1.I

tlrtue L.2$logwlt€ 0.7chlorita 0.5alleite 0.9

0 .02 - 0 .40 . 1 - 0 . 40 .04 - ?

0 . 4 - 0 . 7 '

0 .0250.0300.0440.012

0.0280.0190.025

540

Griffen, D. T. and Ribbe, P. H. (1973) The crystal chemistry ofstaurolite. American Journal of Science, 273A,479495.

Guidotti, C. V. (1970) The mineralogy and petrology of thetransition from the lower to upper sillimanite zone in theOquossoc area, Maine. Journal of Petrology, 11,277:336.

Hellman, P. L. and Green, T. H. (1979) The high pressureexperimental crystallization of staurolite in hydrous maficcompositions. Contributions to Mineralogy and Petrology, 68,369-377.

Johannes, W. (l!}'69) An experimental investigation of the systemMgO-SiO2-H2O-CO2. American Journal of Science, 267,1083-l 104.

Juurinen, A. (1956) Composition and properties of staurolite-Annales Academicae Scientarum Fennicae, Series A, III Geol-ogy, 47, l-53.

Obata, M. and Thompson, A. B. (l9El) Amphibole and chloritein mafic and ultramafic rocks in the lower crust and uppermantle. Contributions to Mineralogy and Petrology, 77,74-81.

Phillips, T. L., Loveless, J. K. and Bailey, S. W. (1980) Cf*coordination in chlorites: a structural study of ten chromianchlorites. American Mineralogist, 65, 1 12-122.

Ribbe, P. H. (19E0) Staurolite, In P. H. Ribbe, Ed., MSAReviews is Mineralogy, v. 6: Orthosilicates. MineralogicalSociety of America, Washington, D. C.

Richardson, S. W. (1966) Staurolite. Carnegie Institution ofWashington Year Book, 65, UE-252.

Richardson, S. W. (1968) Staurolite stability in part of the systemFe-Al-Si-O-H. Journal of Petrology, 9, 467 489.

Schreyer, W. (1967) A reconnaissance study of the system MgO-Al2O3-SiOrH2O at pressures between l0 and 25 kb. CarnegieInstitution of Washington Year Book, 66,382-392.

Schreyer, W. and Seifert, F. (1%9) High pressure phases in the

WARD: MAGNESIUM STAUROLITE. CHROMIAN STAUROLITE

system MgO-Al2O3-SiOrH2O. American Journal of Science,267-1|,407-443.

Seifert, F. and Langer, K. (1970) Stability relations of chromiankyanite at high pressures and temperatures. Contributions toMineralogy and Petrology, 2E, 9-lE.

Sharma, R. S. and McRae, N. D. (1981) Paragenetic relations ingedrite-cordierite-staurolite-biotite-sillimanite-kyanitegneisses at Ajitpura, Rajasthan, India. Contributions to Miner-alogy and Petrology, 78, 48-60.

Skerl, A. C. and Bannister, F. A. (1934) Lusakite, a cobalt-bearing silicate from Northern Rhodesia. Mineralogical Maga-zine, 23 , 598-606.

Smith, J. V. (l%E) The crystal structure of staurolite. AmericanMineralogist, 53, I 139-t 155.

Spear, F. S. (19E2) Phase equilibria ofamphibolites from the PostPond Volcanics, Mt. Cube quadrangle, Vermont. Journal ofPetrology, 23,383426.

Streckeisen, A. (1976) To each plutonic rock its proper name.Earth-Science Reviews, 12, l-33.

von Knorring, O., Sahama, T. G. and Siivola, I. (1979) Zincianstaurolite from Uganda. Mineralogical Magazine, 43, M.

Wells, P. R. A. (1977) Pyroxene thermometry in simple andcomplex systcms. Contributions to Mineralogy and Petrology,62, r29-r39.

Yamaguchi, Y., Akai, J. and Tomita, K. (1978) Chinoamphibolelamellae in diopside of garnet lherzolite from Alpe Arami,Bellinzona, Switzerland. Contributions to Mineralogy andPetrology, 6,263-270.

Manuscript received, December 2E, I9E2;acceptedfor publication, September 19, 19E3.