American Mineralogist, Volume 62, pages 107-l 14, 1977

Electron diffraction study of phasetransformations in coppersulfides

ANonswPurNIs Departmentof Mineralogyand Petrology,Uniuersity of Cambridge DowningPlace, Cambridge C82 3EW, England

Abstract

The order-disordertransformation behavior of chalcociteand djurleite,and digeniteand anilitehas beenstudied by direct observationsof the transformationsin a transmission electronmicroscope. Chalcocite and djurleiterepresent low-temperature superstructures of the high-temperaturedisordered chalcocite subcell, and there need be no compositional differencebetween them. The djurleite superstructureis the more stable form at room temperature.The intermediatestructure in digeniteforms through modulatedstructures ratherthan throughtwinning. A metastable6a digenitetype superstructure,isochemical with anilite, forms on heating.This 6a superstructureis metastablerelative to the formation of anilite.

Introduction cannotbe directlymeasured in this way,the courseof a transformationcan be followeddynamically, and For compositionsnear CurS a numberof phasesare subtle changeswhich would be very difficult to ob- known to existat room temperature.These may be serveby X-ray methodscan clearly be seen. dividedinto two broadcategories based on thenature of the sulfur close-packingin the structure:(1) I: The transformationbehavior of chalcociteand chalcociteand djurleitewith structuresbased on hex- djurleite agonal close-packingof sulfur atoms: (2) digenite- like structuresand anilitewith sulfur atomsin ap- Chalcociteis generallyconsidered to have a com- proximatelycubic close-packing.The low-temper- positionvery closeto CuzS,whereas djurleite, origi- aturephase relations have been summarized by Mori- nallythought to beCur.rrS (Roseboom, 1966; Takeda moto and Gyobu (1971)and Barton (1973).The et at., 1967a)is now consideredto be a solidsolution transformationbehavior of someof thesephases and with a compositionrange from about Cut.*S to the relationshipsbetween them havebeen studied by Cu' nzS(Mathieu and Rickert,1972; Potter, 1974). directobservation of the transformationsin a trans- The phasesand cell dimensionsof chalcociteand missionelectron microscope. djurleite may be summarizedas follows. Between 104' and 435'C chalcociteis hexagonalwith an : Method 3.95,cn = 6.75A and spacegroup P6"/mmc. Above Naturally-occurringcrystals of the phaseswere 435"C it has the cubicclose-packed digenite struc- crushed,after chilling in liquid nitrogen to prevent ture.Below 104'C a pseudo-orthorhombiccell with a deformation.The finestgrains were collected from a : ll.92, b : 27.34, c : 13.44A, spacegroup Ab2m is suspensionin absolutealcohol and mountedon a formed (Buergerand Buerger, 1944).This can be standardcarbon-coated copper grid for observationsdescribedas a superstructureof the hexagonalform in an AEI EM6G transmissionelectron microscope with a : 3en,b : 4y3av,,c : 2cn.More recentlyit operatingat l00kV. Observationswere made firstly has been shown that low-chalcociteis monoclinic on untransformedgrains by usinga smallcondenser (Evans,l97l), but the pseudo-orthorhombiccell will apertureand a defocussedbeam to keepthe illumina- be usedhere for convenience. tion and the heatingeffect of the beam minimal. Djurleite is orthorhombic,diffraction aspect P*t?* : Transformationsin the grains,both on heatingand with celldimensions a :26.90, b 15'72,c: 13.57 cooling,can be observedby focussingand defocus- A. the diffractionaspect is alsocompatible with the singthe beamor by movingthe beamlaterally on and monoclinicspace group P2'ln. Takedaet al. (1967a) off the grain. Although the temperaturein the grain point out that thesecell dimensionsare closelyre- 107 PUTNIS,' PHASE TRANSFORMATIONS IN COPPER SULFIDES

4Dl Electron diffraction patternsshowed that twinning wascommon in bothphases, and to illustratethis and to identifythe phasepresent, grains were sought in an orientation such that the diffraction patternscon- tained the a* and c* axesof the hexagonalsubcell, i.e.,the beam was parallel to the [010]direction in the grain. Such reciprocal lattice sections are used ab throughoutthis investigation. Interpretationof electrondiffraction patterns. Both low-chalcocite and djurleite are oo2 oF simple super- structuresof the high-chalcocitehexagonal subcell, andthis relationship gives rise to threepossible orien- tationsof the low-temperatureorthorhombic forms o9o. .ry ogo . ry relative to the subcell.These three orientationsare relatedby 60' rotationsabout the hexagonalc axis. Fig. I a,b. The top direct lattice illustrations show two of the Figuresla and b showtwo of theseorientations for three possible orientations of the chalcocite superstructure(dashed line) relative to the high-chalcocitesubcell (full line) The arrow on the chalcocitesuperstructure together with a*c* te- the left is the direction of the electron beam which leads to the ciprocal lattice sectionsobtained in each case.The formation of the a*c* diffraction pattern below each diagram. In third orientationof the superstructureproduces a these diffraction patterns the large spots are the subcell reflections diffractionpattern as for Figure lb. Figures2a andb while the smaller spots are due to the superlatticeorientation show the similar above. situationfor the djurleitesuper- structurewith the two differenta*c* reciprocallattice lated to the hexagonalchalcocite subcell with 4 : sections. 4c6,b : 4en, c : 2y4a1,.Roseboom (1966) and The diffraction patterns from untransformed Potter(1974) reported that djurleitedecomposes to grainsin variousorientations are consistent with the hexagonalchalcocite and high-temperaturedigenite aboveinterpretation of differenttwin orientationsin above93oC, whereas Cook (1972)found that djur- both chalcociteand djurleite.The twinningin djur- leitetransformed to a tetragonalform above93oC. leite is the sameas that describedby Takeda et al. Thistetragonal form, firstdescribed by Djurle(1958) (1967b). has beenencountered for compositionsCu,.r.S (Ja- Transformationbehauior. Crystal grains which nosi,1964) and CurS(Skinner et al.,1964;Serebrya- showed only a single phase were selectedfor the naya, 1966)and has sulfur atomsin approximately heatingexperiments. Increasing the temperatureby cubicclose-packing. beam heatingcauses the spontaneousdisappearance As chalcociteand djurleite are often finely inter- of the superstructurespots in both chalcociteand grown (Evans, l97l), and transformationsfrom djurleite, leaving the high-chalcocitesubcell reflec- chalcociteto djurleite have been reported(Cook el tions(Fig. 3a).On slowcooling, spots appear at the al., 1970),the relationshipbetween these two phases reciprocallattice points (lz VzO), indicating the for- needs further investigation.Transmission electron microscopyprovides a methodof studyingsuch inter- growthsand observingtransformations in thesema- terials. Results Crystalsof chalcocitefrom Carn Brae,St. Just, a b. Penzance,and St.Ives in Cornwall,and Bristol Mine,

Connecticut,were examinedin the electronmicro- @. scope.In all samplesobservations made on a large numberof untransformedgrains indicated that both chalcociteand djurleite werepresent and that djur- leitewas often the more abundantphase. Bulk com- &o z'o &o 2bo positionalanalysis of the startingmaterial was not Fig. 2 a,b The equivalentsituation to Fig. I for the djurleite made. superstructure. PT.]TNIS:PHASE TRANSFORMATIONSIN COPPERSULFIDES r09

t T increasinq --r \t-

(b)The Fig.3. Thesequenceoftransformationsobservedinanc*c*diffractionpattern.(a)Thehigh-temperaturechalcocitesubcell orientations of the intermediate 21,6X lca superstructure. (c,d) Twinned orientations of the chalcocite superstructure. (e,f) Twinned cycle, with equal djurleite superstructure.The key to these diffraction patterns is in Figs. 1,2 The arrows show a cooling and reheating possibility of forming the chalcocite or djurleite ruperstructure and the reversible chalcocite to djurleite transformation'

mationof a 2a X lc hexagonalsuperlattice (Fig. 3b). the regionfrom which diffractedbeams are collected This diffractedintensity appears continuously and is the samein eachcase. (The selectedarea aperture changesas a function of temperature,characteristic usedcollects diffracted intensity from a circulararea of a second-ordertransformation. On furthercooling approximately1000 A in diameter')Commonly the the superstructurereflections corresponding to either djurleitesuperstructure forms from the chalcocitesu- chalcociteor djurleiteappear, irrespective of the su- perstructureas the graincools and transformsback perstructureof the original untransformedgrain. to the chalcocitesuperstructure with slight heating. This transformationoccurs as a distinctdiscontin- This transformationbehavior is illustratedin Figure uity. Repeatingthe heatingand cooling cycleshows 3, which showsthe sequenceof a*c* diffractionpat- that the endproduct of the transformationsequence ternsseen in a singlegrain cycledthrough the trans- has an approximatelyequal likelihood of beingthe formations. chalcociteor the djurleite superstructure,although The transformationcycle could be repeatedmany il0 PUTNIS: PHASE TRANSFORMATIONS IN COPPER SULFIDES

times, and has been observedin many grains with ens with decreasingtemperature, although different various orientations.In eachcase the result has been investigators disagree on the estimated deviations the same, with the chalcocite and djurleite super- from stoichiometry. Roseboom (1966) detected no structuresforming with approximately equal proba- compositional deviation at l80oC, but Wagner and bility as high-chalcociteis cooled. The transforma- Wagner (1957) found that at 400oC chalcocite co- tion from the chalcocite to the djurleite existingwith digeniteof compositionCu,.rS can have superstructureoften observed on cooling suggests a Cu:S ratio of lessthan 1.93. that the latter is the more stable form at room tem- Although the presentstudy does not contribute to perature.The tetragonalform (Janosi, 1964;Skinner the question of the compositional variations which ql., et 1964) was not encountered in this study, may be possible in the ordered forms, nor to the evidently becausethe transformation to the tetra- likelihood that djurleite may be more stable at a gonal form is sluggish(Cook, 1972). slightly copper-deficientcomposition, it does con- At thesetemperatures there is no evidencefor any clude that the chalcociteand djurleitesuperstructures bulk compositional changesin the grains due to sul- are differentordering schemesof high-chalcociteand fur loss, and the reversibilityof the transformations that they neednot be associatedwith a differencein indicatesan isochemicalDrocess. composition. This conclusionimplies a free-energydifference be- Discussion tween the two superstructures,and henceone form The transformation sequenceclearly implies that has the lower free energyat room temperature.The the chalcocite and djurleite superstructurescan exist electron diffraction evidencesuggests that the djur- at the same chemical composition. This conclusion leite superstructuremay be the more stable.This is requiresa reexaminationof the relationshipbetween supported by the fact that, in a sequenceof poly- these two phases,and the compositional variation morphic transformations involving superlatticefor- possiblein both the disorderedhigh-chalcocite struc- mation with a decreasein temperature,the largest ture and the orderedstructures formed on cooling.At superlatticewould have the lowest symmetry and lou' temperaturesthe maximum copper deficiencyof would generally be the most stable form. On this an ordered structure based on the high-chalcocite basis, as the true symmetry of chalcocite is mono- structureis about Cu,.rrS(Potter, 1974)and the com- clinic, djurleite should be monoclinic as well. Elec- positional range extends to stoichiometric CurS. tron diffraction patternsof twinned crystalsdo in fact Within this field a number of phases, all super- show a relativedisplacement of superstructurespots structuresof a high-chalcocitesubcell, are known to from each twin member, indicating a symmetry less exist. The two most common phasesare djurleite, than orthorhombic. This is compatible with the re- with a composition range from about Cur.rrS to sults of Takeda et al. (1967a) and the monoclinic Cu,.rrS,and chalcocite,which is apparentlyrestricted spacegroup P2r/n. Cook et al. (1970)have reported to compositions very close to CurS. Cook er c/. that over a two-year period a chalcociteneedle was (1970)report a4a6 X 2c6hexagonal phase with com- converted to djurleite, but assumeda compositional position Cu,S (1.96 > x 2 1.8), and Eliseevet al. changedue to oxidation; my resultssuggest that this (1964) have describeda hexagonalphase with a 3an may occur isochemically. X 2cn superstructure and a composition around The free-energy difference between chalcocite and Cu, r,S. A djurleite-like phase with a Cu-rich com- djurleite at room temperatureswould be small, and position (protodjurleite) has also been reported hencetransformations from chalcociteto djurleite in (Mulder, 1973), but is metastablewith respect to bulk samples would be very slow. This could well chalcociteand djurleite. account for the difficulty in synthesisingdjurleite in Clearly, in a systemwhere there are so many pos- short-term experiments(Roseboom, 1966).The ob- sible ordering arrangements of disordered high- servation of the chalcocite to djurleite transformation chalcocite,low-temperature phase relationshipsare in the electronmicroscope may be possiblebecause of complicated by the narrow free-energy range over the slightly elevated temperature in the grain and which they must exist and thereforeby the kineticsof enhancednucleation in the very thin specimensused. the transformations between such ordered phases. Ordering experiments in high-chalcocite have been In contrast, disorderedhigh-chalcocite is generally carried out by Mulder (1973) who found that on consideredto have a very narrow compositionalfield cooling, a metastable djurleite-like phase, termed near CurS (Roseboom, 1966).The stabilitv field wid- protodjurleite, formed before transformation to a PUTNIS PHASE TRANSFORMATIONS IN COPPER SULFIDES mixture of low-chalcociteand djurleite. Pro- Results todjurleitemay be relatedto the 2a X I c intermediate Digenitefrom Butte,Montana, has been used in phasereported here. this investigation.Untransformed grains showed the Theproblem of describingthe behavior of a system 5a type superstructureconsistent with that reported in which a numberof possibleordered phases exists for naturaldigenite. On heating,the superstructure can only be resolvedin termsof an analysisof the spots disappear,leaving the high-digenitesubcell mechanismand kineticsof the transformationsin- (Fig. 4a).The orderingbehavior observed on recool- volved.A detailedanalysis along theselines must ing takes place in stages,as demonstratedby the await the determinationof the structuresand hence evblution of the diffraction patterns. Figure 4a-d the natureof the orderingprocesses in the djurleite showsthe seriesof diffractionpatterns obtained with solid solution. decreasingtemPerature. The onsetof orderingis markedby the appearance II: The transformationbehaviour of digenite and of satellitereflections about the main subcellreflec- anilite tions (Fig. 4b). In reciprocalspace the positionof Both digeniteand anilitehave structures based on these satellitescan be describedby the reciprocal a cubicclose-packed network of sulfuratoms, with lattice vectors(/ufufu), and indicatesthe formation the cationsordered in the tetrahedralsites. Above of a structuremodulated along the Il I l] directionsof about 75oC the cationsare disorderedand a solid the high-temperaturecell. These satellites are at first solutionfield of high-digeniteexists, with a broad compositionalrange increasingwith temperature (Roseboom,1966; Rau, 1966).The structureis face- centeredcubic with a" : 5.56A. Morimoto and Gyobu (1971)distinguish natural digenite(sensu stricto) which hasa stabilityfield in the Cu-Fe-S systemwith Fe between0.4 and 1.6 atomicpercent, and the digenite-typesolid solution in the systemCu-S which is metastablerelative to aniliteat room temperatures(Morimoto and Koto, 1970).Digenite (sensu stricto) quenched from above 75'C has a cubicsuperstructure of period 5c" with the characteristicthat only reflectionswith indices (l0m t l, l\n L. l, l) arepresent where m and n are integers.This has beeninterpreted (Donnay et al., 1958;Morimoto and Kullerud, 1966)in terms of four-componentmultiple twins of a rhombohedral cell with the four triad axesparallel to the triads of thecube. Over a periodof timethe missing reflections appear,although the relationshipbetween this modi- fied 5a type structureand the twinninghypothesis has not beenmade clear. Digenitecompositions synthesized in theCu-S sys- .oo2 tem form similar superstructuralphases on cooling below75'C with periodicitiesvarying from 5c"to 64" .1Tl dependingon composition(Morimoto and Koto, 1970).At room temperaturethese break down to .ooo .220/ aniliteand djurleite. Anilite is orthorhombic,space group Pnma with a :7.89, b :7.84, c = ll.0l A. As thehigh-temper- of ,Ot diffractionpu,,".n, showingthe trans- aturedigenite subcell is only slightlydistorted (Koto Fig. 4. A series tf formation sequencein naturaldigenite. (a) The high-temperature '/u) and Morimoto, 1970)anilite can be consideredas a digenitestructure. (b) The satellitesat \'/,'lu (c) Development superstructureof high-digenitewith a - b : \,Qa., of 5a superstructurespots along [lll]. (d) The full 5a super- c: 2e., and will be referredto in this way. structure n2 PUTNIS: PHASE TRANSFORMATIONS IN COPPER SULFIDES

corrodedgrain boundaries)of any sulfur loss.On slowly recoolingsuch a heat-treatedgrain, satellite reflectionsagain appear about the subcellspots, but their positionis now describedby reciprocal-lattice vectors (/"y"yul (Fig. 5b). Further cooling pro- ducesthe additional6a superlatticereflections along I I I ] (Fig.5c), but thisprocess is ofteninterrupted by the abruptappearance of a superstructureconsistent with that of anilite(Fig. 5d). On reheating,the anilite supercelldisorders without the appearanceof the 6c reflections,demonstrating that the behaviorleading to the formationof the6a modulationalong I I l] is metastable.Rapid cooling from the disorderedform also resultsin the formation of the anilite super- structureby a discontinuousmechanism. This is also consistentwith the interpretationthat the formation of the 6a modulationis metastablebehavior. This sequenceof transformationshas been ob- servedin many differentgrains, and showsthat the orderingbehavior in untreatednatural digenite in- volvesa seriesof continuousevents leading to the Fig. 5. A seriesof l0] diffractionpatterns I showingthe trans- formationof a 5a superlattice.Heat treatment of the formationsequence in heat-treateddigenite. (a) The high-temper- grain ature digenitestructure. (b) The satellitesat (t/"f" /.). (c) Devel- abovethe orderingtemperature results in a opmentof 6a superstructurespots along ulll. (d) The anilite changeof behaviorsuch that, on recooling,the con- superstructure. tinuousordering process leads to a 6a modulation along Ill] which givesway to the formation,by a weakbut becomequite strongprior to the appear- discontinuousevent, of the morestable anilite struc- ance of the additional 5a superlatticereflections ture. along the [11] directions(Fig. 4c). The twinning hypothesisrequires that the superlatticereflections Discussion alongthe four [111]directions originate from differ- My observationson the orderingprocess in dige- ent regionsof the crystal,the individualregions hav- nite indicatethat the twinningmechanism which has ing rhombohedralsymmetry. Dark-field observations beenproposed for the transitionfrom the high-tem- usingthe differentsuperlattice spots did not showthe peratureform to the 5a type structureis not tenable. presenceof any twinning,and if individualdomains A similarconclusion has been drawn for the forma- of rhombohedralsymmetry exist they would have to tion of intermediatebornite (Allais, 1968; Putnis and be on a scaleof lessthan aboutl5 A. In all casesthe Grace,1976) for whicha similartwinning hypothesis intensityat point (/uyuyu)develops simultaneously had beenproposed. alongthe four Ill] directions,inconsistent with an The evolution of the diffraction pattern implies interpretationbased on twinning. Further cooling that severaldiscrete rate-controlling processes are in- producesthe full suiteof 5a superstructurereflections volvedin the mechanismof the transformation.The whichappear after several minutes (Fig. ad). developmentof satellitesabout the subcellreflections This orderingprocess is continuousin the sense may be ascribedto a spinodalclustering mechanism that the appearanceof the reflectionsis gradualand for vacantsites, which at this stagewould be statis- their intensityis a smoothfunction of time andtem- tically distributed.The subsequentdevelopment of perature.The processof disorderingand ordering additionalsuperstructure spots along [1ll] then ac- couldbe repeatedindefinitely with the sameresults. cords with further local orderingof vacanciesand If the disorderedstate is maintainedfor a periodof tetrahedralsites. Morimoto and Gyobu (1971)have time and the edgeof the grainunder observation is demonstratedthat the presenceof iron stabilizesthe kept abovethe orderingtemperature, an irreversible structure,and the full developmentof the 5a super- processtakes place in the hottestpart of thegrain. In latticemay be dueto the differentiationof copperand mostcases there is no changein theappearance ofthe iron atomsover specificsites. grain and only rarelyis thereany evidence(such as The irreversibleprocess resulting from heat treat- PUTNIS: PHASE TRANSFORMATIONSIN COPPERSULFIDES ment of the digeniteis probably due to the differen- a widerrange of compositionthan that implied by the tial diffusion of iron atoms acrossa temperature nominalformula, Cu'Sn. gradientin the grain. Sucha lossof iron from the Acknowledgments areabeing heated is consistentwith the formationof Dr. J. D. C. McConnell for helpful the 6a superlatticespots along asan iron-free I would like to thank [11], discussionsthroughout the course ofthis work and ProfessorC. T. compositionequivalent to natural digenitecould Prewittfor commentson the manuscript.Suggestions made, dur- form a 6a superstructure(Morimoto and Koto, ing review, by Dr. R. W. Potter are appreciated.I gratefully 1970).The differentialdiffusion of iron hasalso been acknowledgethe receiptof a grant from the Natural Environment observedin the breakdownof bornite(Grace and ResearchCouncil. Putnis, 1976),in which iron diffusesaway from the References forming structureswith compositions heat source Allais, G. (1968)Etude de I'ordredans les sulfuresde cuivre. alongthe bornite-digenitejoin. Structurede la phasem6tastable de la bornite. Bull Soc.Jr. Transformationsin this heat-treateddigenite well M ineral. Cristallogr., 9l, 600-620. illustratethe general principles of behaviorin a phase Barton,P. B. (1973)Solid solutions in thesystem Cu-Fe-S. Part I: transformation. The high-temperaturedisordered The Cu-S and CuFe-S join. Econ.Geo| , 68, 455465. W. Buerger(1944) Low chalcocite.Am field above about 75oC Buerger,M. J. and N. structurehas a stability Mineral.,29, 55-65. (Morimoto and Koto, 1970),and the anilitesuper- Cook. W. R. (1972)Phase changes in CurS as a functionof structureis apparentlythe stablephase below this temperature. Natl. Bur. Stand. Spec. Pub. 364 , Proceed.5th temperature.The transformationto the anilitestruc- Mater. Sci.Res. Symp.,'103-711. ture appearsto befirst-order. If thetransformation is ' -, L. Shiozawaand F. Augustine(1970) Relationship of sulfideand cadmiumsulfide phases. J Appl' Phys'4l' carriedout irreversibly(in the thermodynamicsense) copper 3058-3063. the high-temperatureform is retained below the Djurle, S. (1958)An X-ray studyon the systemCu-9. Acta Chem. transformationtemperature, and alternativemeta- Scand.12. 1415-1426. stable second-orderprocesses may take place to Donnay,G., J D. H. Donnayand G. Kullerud(1958) Crystal and lower the free energyof the system.In this casethe twin structureof digenite,Cu,S". ,42. Mineral.,43,230-242 Eliseev,E. N., L. E. Rudenko,L. A. Sinev,B. K. Koshurnikovand metastablebehavior involves the formation of the 6a N. I. Solovov(1964) Mineral Sb.,L'uou. Geol. Obshchestuo pri superlatticespots. L'uou.Gos Uniu.,18,385-391. Sincethe sequenceof processesleading to a 6atype Evans,H. T. (1971)Crystal structureof low chalcocite.Nature superstructureis clearly metastablerelative to the Phys.Sci.,232,69-70. (1976) and thermalde- formation of the anilitestructure, the very similar Grace,J. and A. Putnis Cation mobility compositionin bornite.Econ' Geol., Z/' 1058-1059' behaviorof naturaldigenite may alsobe metastable. Janosi,A. (196a)La structuredu sulfurecuivreux quadratiques. The behaviorof naturaldigenite is howeverin some Acta Crystallogr.,17, 3ll-312. way modifiedby the presenceof iron in that the 5a Koto, K and N. Morimoto (1970)Crystal structureof anilite. superstructureis retained at room temperatures. A cta Crystallog r., 826, 9 | 5-924. (1972)Elektrochemisch-ther- Whetherthis structureis stablein the strict senseor Mathieu. H. J. and H. Rickert modynamischeUntersuchungen am SystemKupfer-Schwefel whetherit is a'frozen-in'metastable phase unable to bei TemperaturenI l5-90"C. Z. Phys.Chem' Neue Folge,79, transformto a more stablephase due to kineticcon- 3t 5-330. straintsis as yet unanswered,although the patternof Morimoto,N. andA. Gyobu(1971) Thecomposition and stability behaviorsuggests that the latter may be the case.The of digenite.Am Mineral, 56, 1889-1909. - at formationof modulatedstructures of dif- znd K. Koto (1970)Phase relations of the Cu-S system analogous low temperature:stability of anilite.Am. Mineral.,J5, 106-117. ferent periodicitiesalong the entire digenite-bornite -- npd G. Kullerud (1966) Polymorphismon the join (Morimoto and Kullerud,1966) should also be CuuFeSr-Cu"Sujoin. Z Kristallogr., I 23, 235-254. reconsideredin termsof possiblemetastability. -, K. Koto and Y. Shimazaki(1969) Anilite, cu,s., a new Thecomposition of aniliteis not knownwith preci- mineral.Am. Mineral.,54, 1256-1268. (1973)Protodjurleite, a metastableform of cuprous analyseshave been carried out on two-phase Mulder, B. J. sion,as sulfide.Kristallogr Tech.,8, 825-832. assemblages(Morimoto et al., 1969;Morimoto and Potter, R. W. (1974)Metastable phase relations in the system Koto, 1970).The present work showsthat the6a |ype Cu-S. Geol Soc.Am. Abstractswilh Programs,6'915-916. metastablestructure is isochemicalwith the anilite Putnis,A. and J Grace(1976) The transformationbehaviour of structure,and accordingto Morimoto and Koto bornite.Contrib. Mineral. Petol., 55,3l l-315. Rau,H. (1967)Defect equilibrium in cubichigh temperature cop- (1970) type is limited to a compositionof the 6a per sulfide(digenite). J. Phys.Chem. Solids,28,903-916. Cur.roS.These authors have also noted a transforma- Roseboom,E. H (1966)An investigationof the systemCu-S and tion from the 6q type to anilite, and it is therefore some natural copper sulfidesbetween 25" and 700oC.Econ. possiblethat the anilitesuperstructure may existover Geol.61. 641-672. lt4 PUTNIS: PHASE TRANSFORMATIONS IN CAPPER SI]LFIDES

Sergbryxniyx,N. R. (1956)On the t€tragonalmodification of -, J. D. H. Donnayand D. E. Appleman(1967b). Djurleite chalcocite.Geochem. Inter., 3, 6E7-688. twinning. Z. Kristallogr., I 25, 414422. Skinner, B. J., F. R. Boyd and J. L. England (1964) A high Wagner,J. 8., and C. Wagner(1957). Investigations on cuprous pr€ssurepolymorph ofl chalcocite,Cu2S. Trans. Am. Geophys. suffide-./. Chem.Phys., 26, 1602-16A6. Union,45, l2l-122. Takeda,H., J. D. H. Donnay,E. H. Roseboomand D. E. Apple. man (1967a)The crystallographyof djurleite,Cu,..S. Z. Kristal- Manwcript receioed,January 14, 1976;accepted logr., 125,404413. for publication,August 5, 1976.