American Mineralogist, Volume 71, pages 895-899, 1986

Myrmekite replacing in progrademetamorphism

JorrN R. Asnwonrn Department of GeologicalSciences, Aston University, Aston Triangle, Birmingham B47ET, United Kingdom

Ansrnlcr In pelites from the central Menderes massif, Turkey, myrmekite partly or completely replacesalbite in rocks that had developed, at a slightly earlier stageof , two plagioclasescoexisting acrossthe peristerite gap. The proportion in the myr- mekite is consistentwith the Al and Si immobility theory of myrmekite origin. This requires exchange of Ca for Na with the rest of the rock (l + x)NaAlSi'Oa I xc-a2+ - Na,-.Ca,Al,*.Si3-,O8 + 4xSiO, + 2xNa*. Myrmekite produced by progradereplacement of albite has the same properties, and develops under the same controls, as myrmekite producedby retrogradereaction from K-. Poikiloblastic oligoclasecontaining bleb- by quartz may often representcoarsened myrmekite. Myrmekite is important as an indi- cator of Al and Si immobility in a variety of petrological environments.

IurnooucrroN over to myrmekite, the proportion of quartz in the inter- Myrmekite is the commonestexample of a symplectite, growth is a function of the plagioclasecomposition: the i.e., an intergrowth of minerals the mutual boundaries of relationship is identical for exsolution (Phillips and Ran- which are approximately perpendicular to a reaction in- som, 1968) and for replacement,as shown by Ashworth terface that moved as the intergrowth formed. The con- (1972). Some of the Si segregatesfrom the Al by counter- stitution of myrmekite is simple and comprises plagio- diffusion within the reaction interface. Carstens (1967) clase intergrown with a smaller amount of quartz, the pointed out that this myrmekite-forming interface is in- latter having a rodlike or wormlike morphology. As a basis coherent,i.e., there is no crystallographiccontinuity across for understandingsymplectites in general,it is important it. The formation of the integrowth is analogousto duplex tounderstandtheoriginofmyrmekiteinitsvariouspetro- cell growth in metals(Yund and McCallister, 1970). logical settings.The study of myrmekite hasa long history, Both of the recognizedmetamorphic myrmekite-form- reviewed by Phillips (1974). Myrmekite resemblessome ing processes(by replacement of and exsolution from textures that arise from crystallization of melts (rod eu- K-feldspar, respectively)are retrograde.The replacement tectics),andingraniticrocksHibbard(1979)hasproposed processis usually part of a hydration reactionthat pro- a melt-grown origin. On the other hand, definitely mag- ducesmica in addition to the plagioclaseand quartz (Phil- matic quartz-feldsparintergrowths (graphic and grano- lips et al., 1972; Ashworth, 1972). K-feldspar may be phyric textures) are usually distinguishable from myr- completelyreplaced(Ashworth,l972,Fig. lb). Ithasbeen mekite (Barker, 1970). It is also found in metamorphic suggestedthat the former presenceof K-feldspar can be rocks (Phillips, 1980a),including metasedimentaryones inferred in polymetamorphic roc(s by the detection of in which a magmatic origin is ruled out (Nold, 1984). coarsenedmyrmekite (Ashworth, 1979). Ashworth (1976) has inferred that replacementof K-feld- Myrmekite is reported by Cooper (1972) in metamor- sparby myrmekite is part of a reaction that also consumes phic rocks without K-feldspar, from relatively low grades residual melt in migmatites. (associatedwith the crest of the peristerite gap); Cooper Myrmekite is usually found adjacent to K-feldspar; (1972) and.Shelley (1973) interpreted these occurrences Phillips (1974) has incorporated this spatial association in terms ofprograde reactions. Cooper (1972) proposed into his definition of mymekite. The intergrowth often a reaction betweenclinozoisite and albite, producing an- has a convex boundary toward the K-feldspar, suggesting orthite component and excesssilica, but this does not that it formed from that mineral. Both exsolution and explain the intimate (symplectic) intergrowth. Shelley replacementwill produce myrmekite from K-feldspar if (1973, p. 336) found a lack of textural indication that Al and Si are immobile relative to other cations (Ash- clinozoisitehad reacted.He also observedgreat variations worth, 1972).Exsolution in a closed system("schwantke in the quartz proportion in the myrmekite, which grades hypothesis") and metasomatic replacement ("Becke hy- into normal poikiloblastic .Noting that both pothesis") are thus endmembersof a continuum of mixed Schwantkeand Becke hypothesesin their classicalform reactions in which myrmekite is produced, and in this require K-feldspar as a precursor to myrmekite, Shelley sensemynnekite is polygenetic(Phillips, 1980b).If Al (1973)favoredanalternativeinterpretation,inwhichpre- and Si are quantitatively retained as K-feldspar is made existingquartz was supposedto recrystallizewith an elon-

0003404x/86/0708-0895$02.00 89s 896 ASHWORTH: MYRMEKITE REPLACING ALBITE

up to staurolite + kyanite grade (Ashworth and Evirgen, 1985).The reasonfor development of assemblagesof rel- atively high metamorphic grade compared with other areas (in which the peristeritegap closesin the garnet zole; e.9., Cooper, 1972) appearsto be that arro was low, in the presenceof graphite (Ashworth and Evirgen, 1985). The specimensstudied here come from the part of the stau- rolite + kyanite zone mapped by Ashworth and Evirgen (1985, Fig. 1). No granitic intrusions are found nearby. Temperature estimates in the range 520-600"C for this zone (Ashworth and Evirgen, 1985) are obtained from garnet-biotite geothermometry using the calibration of Hodgesand Spear(1982).

Pr,acrocr,.qsE TEXTURESAND coMposrrroNs In many pelites from this area,at $ades approximating the crest ofthe peristerite gap, oligoclasecontains ovoid blebs of quartz with dimensions 10-50 pm (Fig. la), dis- tinctly smaller than the quartz grains in the surrounding matrix (which are > 100 pm in size). Other oligoclase Fig. l. Myrmekitesin pelitesfrom thecentral Menoeres mas- grainscontain groupsof small flakesof mica, mostly mus- sif. (aHc) Transmittedlight, crossedpolars; (d) back-scattered covite. Some of the quartz inclusions are elongate (Fig. electronsErrr image. (a) Plagioclasewith quartzblebs (top left and la); where clearly rodlike, they occur as subparallel swarms top right), showingconvex (center). boundarytoward albite At so that the intergrowh has the internal characteristicsof the top left, the quartz blebs are clearly elongate,so the inter- myrmekite (Fig. lb). To further characteize the plagio- growthis classifiedas myrmekite. (b) Clear example ofmyrmekite (center),in contact with relict albite (in extinction). Nonmyr- clase-quartzintergrowh, three rocks were studied in de- mekitic oligoclaseoccurs at the right, and at the top left where tail. Theseare pelites,with the assemblagequartz-plagio- it is overgrownby the myrmekite.(c) Confusedcontrast due to clase-biotite- muscovite- chlorite- garnet-staurolite - kyanite. overlapofquartz and plagioclasewithin the thin section.Dark They are homogeneousschists with no migmatization or specksare thought to be graphite,on the basisofthe reflected- vermng. light propertiesofthe few that arelarge enough to be identified. Two of these specimenscontain small amounts of al- (d) nsE-modesnv photographof the samearea as in (c),showing bite, in addition to oligoclase,which occursas two textural surfaceoutcrop of quartz (darker gray) and oligoclase(lighter varieties (with and without quartz rods). There are gra- The bright mineralis Black with $ay). biotite. areas bright edges dations, with optical continuity, between the two kinds areholes in the polishedsurface. This is the typeofimage used ofoligoclase, but both have sharp boundaries against al- in the quantitativework to detemine the quartz proportion. bite (Fig. lb). Where there is any definite curvature to the boundary between albite and quartz-bearing oligoclase, gation parallel to the growth direction of the enclosing this is convex toward the albite (Fig. la). The coarseness feldspar. Becauseof the absenceof K-feldspar, Phillips of quartz rods is typified by Figure la, but in a few ex- (1973,1974) did not regardthese examples as being myr- amples,they are thinner in one part ofthe oligoclasegrain, mekite although their internal featuresare indistinguish- always the part adjacent to albite (Fig. 1b). In previous able from classical myrmekite, except perhaps that the studies,this fining ofquartz toward the interfacehas often quartz rods tend to be coarser(Shelley, 1973). beenobserved in myrmekite adjacentto K-feldspar (Phil- In this paper, -quartzintergrowths are de- lips, 1974,p. 182).Where relatively coarse (Figs. la, 1c), scribed from another area associatedwith the peristerite the quartz rods tend to be discontinuous, being transi- gap and lacking K-feldspar. It is shown that the inter- tional in shapetoward the quartz blebs in poikiloblastic growths have all the features of true myrmekite and can oligoclase. be attributed to the same fundamental control that pro- Plagioclasecompositions were estimated in mole per- ducesmyrmekite from K-feldspar, the only differencebeing cent from electron-microprobeanalyses for Na, Ca, and that the mineral replacedis albite. K. Albite is approximatelyAno ,. Adjacent quartz-freepla- gioclasesare approximately An,._,u.In these two speci- RncroN,ql, SETTTNG mens, with quartz rods are in the range The rocks studied are pelitesfrom the central Menderes An,u-rr. The third specimenlacks albite but has abundant massif, Turkey. As described by Ashworth and Evirgen oligoclasewith and without quartz rods, both having ap- (1985), the sequenceof metamorphic zonesdeveloped is proximately the samecomposition range (An,r-rr). Figure garnet, staurolite, staurolite + kyanite, and kyanite. Pla- 2 illustratesthe relatively calcic nature ofthe quartz-bear- gioclasescoexisting across the peristeritegap are observed ing oligoclasesrelative to the peristerite gap (estimatedin ASHWORTH: MYRMEKITE REPLACING ALBITL, 897

L., o F o o .EE (, t

relicf coexistence of qlb with An.^ ., tJ- to

peristerite gop f_ I Fig. 3. Meansand ranges of the estimatedvolume proportion I I of quartzin two specimens,plotted againstmeans and rangesof plagioclasecomposition measuredin the same areasof myr- o o.os o.1o or5 o.2o dl mekite.The theoreticalcurve refersto Al and Si immobility. Fig. 2. Observedranges of mlrmekitic plagioclasecompo- sitionin threespecimens plotted against temperature, estimated from garnet-biotitegeothermometry by the methodof Hodges For quantitative measurement,in each specimen,five areas and Spear(1982). Garnet-rim compositions being rather con- of 4-7 x lOa pm2 (roughly 0.5-l times the area in Fig. stant,the plotted rangesof temperaturearise from variationsin ld) were selected on the criterion that quartz appeared matrixbiotite composition(cf. Ashworth and Evirgen,1985). evenly distributed. Even so, it is possiblethat small areas of quartz-free oligoclase were accidentally covered or, conversely,thal quartz may be overestimatedby encoun- the same area from specimenslacking oligoclase-quartz tering chanceinclusions of quartz not truly belonging to intergrowths). the intergrowth. The photograpSswsls digttized, and areas of minerals were computed in the manner of Mongkoltip Fnoponrrou oF eUARTZ and Ashworth (1983, p. 645). Inclusions other than quartz polars, In transmitted light between crossed thin-sec- (e.g., biotite, Fig. ld) were subtracted from total areas. tion areas in which quartz and plagioclase are superim- The areal proportion was then taken as the estimate of posed make it impossible to determine their proportions volume proportion of quartz in an intergrowth. The re- (Fig. proportion accurately lc). It is likely that the of sults are summarizedin Figure 3, in which the theoretical quartz (Barker, has been overestimatedin optical studies curve is for quantitative retention of Al and Si in the 1970).A surfaceimaging technique is required. It would reaction be possibleto stain the feldspar,but transmittedJight im- ageswould still be confusedby unwanted detail from be- (l+x)NaAlSi3O8+nca2* low the stained surface. The obvious solution is back- - Na,-"Ca"Alr*,Sir-"O, + 4xSiO, * 2xNa*. (l) scatteredelectron (osn)imagery of polished surfaces,which mvrmekite givesatomic-number contrast.Ashworth (1972) usedthis technique rather crudely, in an electron microprobe. The method has become technically easy with the advent of Since the Ca content of the albite is negligible, the theo- semiconductordetectors, designed for ssr imaging in the retical curve is the same as that for the production of scanning microscope (sevr),together with suitable elec- myrmekite from KAlSirO, and is given by Equation 1 of tronics to manipulate the contrast (Hall and Lloyd, l98l). Ashworth (1972). The mean atomic number Z of sodic plagioclasesis close to that ofquartz; the differencefrom quartz, AZ increases INrnnpnntarroN in more calcic oligoclases.Figure ld showsthat adequate In view ofthe possiblesources oferror outlined above, contrast is obtained betweenquartz and oligoclaseofap- the agteement between observed quartz proportions and proximatelyAn2r composition, where tZ = 0.17;in high- theory is reasonable(Fig. 3). Unfortunately, it is not pos- contrast prints, quartz was distinguished from An,u, in sibleto testfor systematicdependence on plagioclasecom- which case AZ = 0.10, close to the practical limit for position, sincethe rangeof oligoclasecompositions in the discriminating betweenphases by this technique(Hall and rocks is restricted. Nevertheless,the data support the in- Lloyd, 1981,p. 363). terpretation of the plagioclase-quartzintergrowths in terms nsnmicrographs were obtained from the two specimens of Al and Si immobility, exactly as in the case of myr- with the most abundant plagioclase-quafizintergrowths. mekite formed from K-feldspar (Ashworth, 1972). Since 898 ASHWORTH: MYRMEKITE REPLACING ALBITE the textural featuresalso agreewith those of classicalmyr- ofelongation ofquartz. Spatial fluctuations in the quartz mekite, it would be unreasonableto use a different name proportion will arise. In the present area, the common for the presentintergrowths; they are myrmekite formed oligoclasewith ovoid quartz blebs is interpreted as coar- from albite, as distinct from the commoner occurrences senedmyrmekite. of myrmekite formed from K-feldspar. In caseswith K-feldspar as precursor,it is often difficult Cot{clusroNs to distinguish exsolution from replacement myrmekites, The myrmekite described here is clearly of metamor- and the two processescan occur simultaneously (Ash- phic origin, since the rocks are metasedimentarywith no worth, 1972; Phillips, 1980b). In the present case,exso- sign of granitic injection and the inferred temperatures lution can be ruled out; only tiny proportions ofoligoclase are too low for anatexis.This myrmekite, replacingalbite could unmix from the observedalbites, whereasthe myr- as a result of progrademetamorphism, formed under ex- mekite-forming processhas removed albite altogether from actly the samecontrols as myrmekite replacing(or exsolv- some rocks. This is clearly a replacement effect, and a ing from) K-feldspar under retrograde conditions. It is particularly simple one in which one mineral is replaced possiblethat myrmekite may be discoveredreplacing yet by a two-mineral symplectite at approximately constant other minerals that have lower AVSi ratios than the prod- Al and Si, according to Reaction l. uct plagioclase. In these specimens,nonmyrmekitic oligoclaseand al- Where replacementof the reactantmineral is complete, bite are interpreted as representing a relict coexistence myrmekite after albite might be confusedwith myrmekite across the peristerite gap (Fig. 2). The geothermometry after K-feldspar, but they may be distinguishableby oc- indicates a progrademyrmekite-forming reaction (Fig. 2) currence in different metamorphic settings. In pelites, in which albite is replacedas the peristerite gap ceasesto K-feldspar is abundant only when the prograde muscov- operate.However, the overall effect on plagisslaser.m- iIe-quartz dehydration reaction has operated; retrograde position is not simply to homogenize it at a Ca content replacive myrmekite is then produced by the reverse of intermediatebetween those coexisting at the gap.The new that reaction: K-feldspar + AlrSiO5 + HrO - muscovite plagioclaseis more calcic than the range spannedby the + qtrarlz + plagioclase, where the quartz and plagioclase relict, coexisting ones (Fig. 2). The specimenwithout al- occurpartly asmyrmekite. K-feldspar is particularly thor- bite, and with compositions around Anro in both myr- oughly removed where this reaction is coupled with crys- mekitic and nonmyrmekitic plagioclase,represents an ad- tallization of residual melt, as in some migmatites (Ash- vanced stage of this evolution. Thus a source of Ca is worth, 1976). Complete replacement of K-feldspar is implied. It is likely that a calcic mineral in the rocks was thereforeexpected mainly in the upper amphibolite facies, consumed.Though there is no epidote-group mineral in whereasthe prograde replacement of albite describedin the present assemblage,it is plausible to speculatethat this paper is characteristicof the greenschist-amphibolite one was formerly present and broke down by prograde faciestransition. dehydration as the myrmekite was forming (it is possible Retrogrademyrmekite is often well-preserved,because that the plagioclasewith muscovite inclusions represents reaction was short-lived: obscure,coarsened varieties re- relict Al-rich domains where this mineral was situated). flect complex metamorphic histories (Ashworth, 1979). Thus some overall reaction can be envisioned that is sim- Progrademyrmekite, on the other hand, is expectedto be ilar to that proposed by Cooper (1972) (though this still coarsenedas a generalrule. The present examples occur does not describe the reaction completely): albite + cli- where the climactic metamorphic temperatures were those nozoisite ' anorthite component + qvartz + NarO + at which the myrmekite was produced. Any further heat- HrO. The production of quartz by an overall reaction of ing would obscure the myrmekite texture. This consid- this kind is not the reason for the occurrenceof myrme- eration explains the apparent rarity of prograde myrme- kite; the quartz in the myrmekite results from local con- kite. Many examplesof poikiloblastic oligoclase,in which servation of Al and Si in the partial reaction affectingthe the inclusions are quartz blebs, may have originated as reactant albite volume. myrmekite, and it may prove possibleto recognizesome It is interestingthat coarseningofquartz rods away from examplesin the light of the present study. the reaction interface is observed here in the prograde Myrmekite forms becauseAl and Si are inherited from case(Fig. lb) as well as being common in the retrograde the reactant mineral. These elementshave also been shown case(Phillips, 1974). This suggeststhat the effect is not to control the developmentof another symplectite(Mong- due to any changein reaction kinetics with changingtem- koltip and Ashworth, 1983).Of course,Al and Si will not perature (which would be expectedto produce opposite invariably be immobile. Not all symplectitescontain Al effectsin the two cases),but may simply be due to coar- and Si, and evenwhere both elementsare present,it seems seningof quartz in already-formed myrmekite while the possible that others control the texture if segregationof reaction was still producing fine quartz rods at the ad- the Al and Si is unimportant [as in the -exso- vancing interface.The simplest interpretation of the coar- lution reaction studied by Boland and Otten (1985)1. sening processis Ostwald ripening, in which larger par- However, immobility ofAl and Si appearsto be common, ticles (of quartz in this case)grow at the expenseof smaller as indicated by the commonnessof myrmekite itself. Im- ones; interfacial free energy will also be reduced by loss mobility here refers to a restricted range of segregationin ASHWORTH: MYRMEKITE REPLACING ALBITE 899 the reaction interface, over distances that are reflected in Carstens,H. (1967) Exsolution in ternary , II. Inter- precipitation in the spacing ofthe quartz rods as they form at this interface. granular in alkali feldspar containing calcium solid solution. Contributions to Mineralogy and Petrology, 14, The transport mechanism is a variety of grain-boundary 3r6-320. diffirsion (Yund and McCallister, 1970), and diferent Cooper, A.F. (1972) Progressivemetamorphism of metabasic transport kinetics will operate in a fluid phase (which may rocks from the Haast Schist Group of southern New Zealand. -492. be present in the same rocks where the myrmekite is grow- Journal of Petrology, 13, 457 Hall, M.G., and Lloyd, G.E. (1981) The SEM examination of ing). Nevertheless, where myrmekite (or another Al- and geological samples with a semiconductor back-scattered elec- Si-controlled symplectite) occurs, slow diffusion of Al and tron detector. American Mineralogist, 66, 362-368. Si is a constraint that should be valuable in interpreting Hibbard, M.J. (1979) Myrmekite as a marker betweenpreaque- other features of the rock, such as zoning and reaction ousand postaqueousphase saturation in granitic systems.Geo- 1047-1062. textures. Myrmekite is a clear indicator of restricted dif- logical Society of America Bulletin, 90, Hodges,ICV., and Spear,F.S. (1982) Geothermometry, geoba- fusion range Al of and Si. rometry and the AlrSiO5 triple point at Mt. Moosilauke, New Hampshire. American Mineralogist, 67, | | l8-l 134. Acruowr,BocMENTS Mongkoltip, P., and Ashworth, J.R. (1983) Quantitative esti- Collectingwas guided by Dr. M. M. Evirgen and supportedby mation of an open-systemsymplectite-forming reaction: Re- the British Council, with logistical assistancefrom the MTA In- stricted diftrsion ofAl and Si in coronasaround olivine. Jour- nal of Petrology 24, 635-661. stitute (Turkey) and Etibank. Colleagues at Aston are thanked , Nold, J.L. (1984) Myrmekite in Belt Supergroupmetasedimen- for facilities and in fields (R. advice the of electron optics G. tary rocks-Northeast border zone of the Idaho batholith. Howell, Department of Mechanical and Production Engineering) American Mineralogist, 69, I 050-1 052. and image analysis(Dr. W. G. Collins, Civil Engineering). Phillips, E.R. (1973) Myrmekites from the Haast Schists,New Znaland:.A discussion.American Mineralogist, 58, 802-803. RnrnnrNcns -(1914) Myrmekite-One hundred years later. Lithos, 7, I 8l-1 94. Ashworth, J.R. ( I 97 2) Myrmekites of exsolutionand replacement - (1980a)Myrmekite as a marker betweenpreaqueous and origins. GeologicalMagazine, 109, 4 5-62. postaqueousphase saturation in granitic systems:Discussion. -(1976) Petrogenesisof migmatites in the Huntly-Portsoy Society of America Bulletin, 91,672-674. area, north-east Scotland. Mineralogical Magazine, 40, 661- Geological - ( I On polygenetic m),rmekite. Geological Magazine, 682. 9g0b) -(19'19) Textwal and mineralogical evolution of migma- 1r7.29-36. Phillips, E.R., and Ransom, D.M. (1968) The proportionality of tites. In A.L. Harris et al., Eds. The Caledonidesof the British quartz in myrmekite. American Mineralogist, 53, l 4l l -1 413. Isles-Reviewed, 357-361. GeologicalSociety of London Spe- E.R., Ransom, D.M., and Vernon, R.H. (1972) Myr- cial Publication 8. Phillips, mekite and muscovite developed by retrograde metamorphism Ashworth, J.R., and Evirgen, M.M. (1985) Plagioclaserelations Broken Hill, New SouthWales. Mineralogical Magazine,38, in pelites, central MenderesMassif, Turkey: I. The peristerite at 570-578. gap with coexisting kyanite. Journal of Metamorphic Geology, (1973) Myrmekites from the Haast Schists,New Zea- 3.207-218. Shelley,D. land. American Mineralogist, 58, 332-338. Barker, D.S. (1970) Compositions of , myrmekite, Yund, R.A., and McCallister, R.H. (1970) Kinetics and mech- and graphic . Geological Society of America Bulletin, exsolution. Chemical Geology, 6, 5-30. 81,3339-3350. anism of Boland, J.N., and Otten, M.T. (1985) Symplectitic augite: Evi- dence for discontinuous precipitation as an exsolution mech- anism in Ca-rich clinopyroxene. Journal of Metamorphic Ge- MaNuscnrprREcETvED Jurv 11, 1985 ology,3, 13-20. MeNuscnrpr AccEprEDMancn 18, 1986