the Scanned PDF - Mineralogical Society of America
Transcrição
the Scanned PDF - Mineralogical Society of America
American Mineralogist, Volume 66, pages 127-141, 1981 Exsolutionrelationshipsin a clinopyroxeneof averagecomposition Cao.orMno.uoMgo.rrSirOu: X-ray diffraction and analytical electron microscopyr WBrqnv A. GoRDoN,' DoNALD R. PEAcoR,Pnrup E. BRowN, Enrc J. EsseNe Departmentof GeologicalSciences Universityof Michigan,Ann Arbor, Michigan 48109 AND LAWRENCE F. ALLARD Department of Materials and Metallurgical Engineering University of Michigan, Ann Arbor, Michigan 48109 Abstract A pyroxeneof averagecompositionCaoa.Mno from Balmat, New York is exs2Si2Ou "rMg6refinementswere performed in part to solvedto coexistingP2r/c and,A]c phases.Structure limit the compositionsof the individual phases,as were plots of cell parametersand average Ml-O and M2-O bond distancesrs. composition.Site occupanciesshowedthat the phases are manganoandiopside and kanoite. Transmissionelectron micrographsrevealedthat the phasesare lamellar, share(001) as the interphaseboundary, and are 2000A wide. X-ray energy dispersiveanalysisof individual lamellae yielded CaaurMno*MgorrSirOufor the A/c phaseand Cas,2Mn,orMgo,.SirOufor the F2r/c phase.Dark-field imaging revealedthe presenceof antiphasedomain boundariestn the F2r/c phase,indicating the existenceof a A/cP2t/c ttansition for the Ca-poor phasein the Ca-Mn-Mg pyroxene system.In situ heatttg experimentsyield a temperatureof the A/c-F2,/c inversion of 330+20oC.Observationof antiphaseboundary (APB) positionsbefore and after heating through the inversion temperature showedthat the APB positionswere essentiallyunchanged,indicating stabilization of APB's by the concentrationof Ca or somedefect at the APB's. Lattice fringe imagesacross the lamellar boundary (001) indicate that the interfaceis semi- or completelycoherent.The featuresof this exsolutiontexture are consistentwith initial exsolutionby spinodal decomposition or nucleation and subsequentslow cooling in a regionally metamorphosedphase. Introduction X-ray diffraction photographsconfirmed that rl did In the investigationof a magnesianrhodonite from indeed consist of two pyroxenes.We report results Balmat Mine No. 4, New York (Peacoret al., 1978), obtained primarily through single-crystalX-ray difa pyroxenecoexistingwith the rhodonite was qualita- fraction and further characterization of the two tively analyzed and determined to have major Mn phasesby scanningtransmissionelectronmicroscopy analysis. and Mg with minor Ca. This is near the ideal compo- in conjunctionwith X-ray energy-dispersive provided generosity two samples, by the Our of sition of kanoite, MnMgSirOu (Kobayashi, 1977). primarily David Dill the mine staff, Dr. of consist of Quantitative wavelength dispersiveanalyseson the pink rhodonite, buff-colalternating bands of bright electronmicroprobewere obtainedfor this pyroxene (Table l). A pale blue schiller observedin the Mn- ored pyroxene,and qufitz. They were found by minpyroxeneindicatedthe presenceof a periodic micro- ers at BaLnat and were savedonly becauseof their structure,and preliminary single-crystaland powder unique and colorful appearance,and are from a complexly interlayeredsequenceof metamorphosed,impure, siliceouscarbonatesand evaporitestypical of rContribution No. 361 from the Mineralogical Laboratory, Dethe southwestpart of the Grenville Province (Engel, partmentof GeologicalSciences,The University of Michigan. 2Presentaddress:ExxoN Co., U.S.A., P,O. Box 1600,Midland, 1956).The najority of the units consist of calcitic Texas79702. and dolomitic marbles with variable amounts of 0003-004x/8 1/0 102-0r27$02.00 127 128 GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSII/PS Table l. Electron microprobeanalysisof Balmat clinopyroxene oxidewt.% si02 52.09 l,,toiar atio/2Si 2.00 AI 03 0.25 2 0.01 l 4 n 0 2 1. 0 6 0.69 l'490 | 4 .21 o.a? Feo 0.28 0 .0 t Cao 10 . 4 1 0.43 sum 98.30 l' S t a n d a r d s usedwere synthetic tephroite for lln, Irving clinop y r o x e n ef o r C a a n d S i , I n g a m e l l ' sa l B a n d i n ef o r A l a n d F e , a n d l ' l a r j a l a h t i o l i v i n e f o r M g . D r i f t , a t o m i c n u m b e r ,f l u o r e s c e n c e , a b s o r p t i o n , a n d b a c k g r o u n dc o r r e c t i o n s w e r e a p p li e d t o t h e d a t a , u s i n g t h e p r o g r a nE M P A DVR I I ( R u c k l r d g ea n d G a s p a r r i n j ,1 9 6 9 ) diopside, tremolite, phlogopite, calcite, anhydrite, and primary talc. Scattered Mn-rich pods have yieldedthe amphiboletirodite (Brown et al.,1979)as well as the assemblagesMn-pyroxene-Mg-rhodonite-talc-quartz and Mn-pyroxene-Mg-rhodonitemanganoan calcite-quartz-anhydrite. The Balmat ore deposithas been regionally metamorphosed,the peak metamorphicevent there being estimatedat P : 6.5+.5 kbar and T : 625+25oC,as calculated from sphaleritebarometry (Brown et al., 1978),twofeldspar, iron-titanium oxide, and calcite-dolomite thermometry(Bohlen and Essene,1977;Brownet al., 1978),and from the presenceof sillimanite in the adjacent metapelites. Determinationof lattice parameters Cleavagefragments of the pyroxene were examined by Weissenbergand precessiontechniques.Two reciprocal lattices were observed in the diffraction photographs;characteraationof the extinction rules combinedwith identification of the phasesas pyroxenesled to the assignmentof spacegroupsF2,/ c and A/c. The class(b) (h+k:2n+l) reflectionsof the primitive phase were sharp, with no diffusenessor streaking. A powder diffractometer pattern obtained with quartz as an internal standard, CuKa radiation, and graphite monochromator,showed only single unresolvedpeaks,and a combinationof single-crystaland powder-diffractiontechniqueswas required to determine accuratelattice parameters.Least-squaresrefinementof the averagelattice parametersfor the two phaseswascarriedout with the program Lclse, written by CharlesW. Burnham. Estimatesof the lattice parametersfor the individual phaseswere then ob- precession,and Weissenbergtained by Weissenberg, diffractometertechniques.Thus, preciseabsolutevaluescould not be obtained,but differencesin the valuesof correspondingparameterswere appted to the averagevalues as determined with the powder diffractometer,resulting in the final valuesa : 9.78(2), b : 8.93(9),c : 5.32Q)4, F : 108.60(7)'for the primitive phase, all:da : 9.76(2), b : 8.93(9),c : 5.27Q)4, B : 106.44(4)ofor the C-centeredphase. The magnitudesof the b axesof the two phasesare virtually identical, their a-axis magnitudesare only slightly different,but a significantdifferenceexistsin their valuesof c. The c axesdiverge by an angle of 2.16(2)",and a and b of the two lattices are parallel. We therefore predicted an intergrowth of these phaseswith (001) as the common plane of least strain, a relation verified by direct indexing of the lamellar interface. Crystal structure refinements Refinementsofihe crystalstructuresof the coexisting pyroxeneswere carried out in part to determine the nature of the individual site occupancies,which in turn restricts the compositionsof the individual phases.Theserefinementsalso provide data on the generalnature of pyroxenestructuresas a function of composition,and specifically for two compositions which appearto have formed under equilibrium conditions. The measurementof intensity data was difficult becausethe class (a) diffraction maxima from the separatelattices are poorly resolved.Intensities for thesediffraction maxima could not be measuredutilizing an automatedsystem,and were determinedby graphically recording intensity profiles and integrating the areas using a planimeter. Even this method failed to provide resolution for diffraction maxima with a relatively large value of the index h, and the data sets were significantly smaller than is usually the casefor pyroxenes.Class(b) diffractions for the primitive phasewere measuredwith the Supper-Pace automated system, which employs background counts on each side of a continuous peak scan.The class(b) and (a) diffraction maxima were placedon the samescaleby comparing class(b) diffraction intensitiesasdeterminedwith both methods. A crystal approximately 0.16 x 0.08 x 0.39 mm was selectedand mounted for rotation about the b axis of the Weissenberg-geometrydiffractometer. MoKa radiation monochromatedwith a flat graphite crystal was used. For the C-centeredphase,374 ntensities were measured,while for the primitive r29 GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS phase we determined the intensities of 982 diffractions, of which 386 were class(a). Theseinclude 139 unobservedintensitiesfor the class(a) diffractions of the primitive phase,27 for class(b) of the primitive phase, and 2l for the C-centeredphase. Intensities were determinedfor reflectionshaving sind lessthan 0.5.All data were correctedfor Lorentz, polarization, and absorptioneffects(p : 38 .--'), by a modified versionof the programABsRpwritten by C. W. Burnham. The structure was refined using the program RFINE2(Finger and Prince, 1975).Form factorswere those of Doyle and Turner (1968). The weighting schemeof Cruickshank(1965,p. lla) was used,and all reflections with individual discrepancy factors greater than 0.5 were rejected. For the primitive phase,separatescalefactors were used for the class (a) and (b) diffractions. Form factorswere initially chosenfor Ml and M2 of both phases,basedonly on generalpyroxenecrystal-chemical principles in relation to the average composition as determined by electron microprobe analysis.For the C-centeredphaseMl was assumed to be occupiedby Mg and Mn, and M2 by Ca and Mn. For the primitive phaseMn and Mg were both assignedto Ml and M2. Theseare only approximations to true site occupancieswhere three different cations may occupy a given site. This is especially significantfor M2 of the primitive phase,which must contain someCa in addition to Mn and Mg. Refinementproceededwithout complicationthrough the refinement of anisotropic temperature factors. The final R values as computed including all data, both observedand unobserved,are 5.7Vofor the primitive phaseand 5.4Vofor the C-centeredphase. Structurefactorsare listed in Table 2.'Atom coordinatesand temperaturefactors are listed in Table 3, thermal ellipsoid data in Table 4, selected interatomic distancesin Table 5, and site occupanciesin Table 6. The distanceswere calculatedwith the program oRFFE(Busing et al., 1964} utilizing the full variance-covariancematrix of atomic parameters and standarderrors of lattice parameters. Discussionof refinementresults The site occupancies(Table 6) are consistentwith phasesnear diopside and kanoite in composition. The generalnature of pyroxene structure variations as a function of compositionis well known; Clark et al. (1969) determined the details of the diopside structure, and Ghose (written communication) has 3To receivea copy of Table 2, order Document AM-81-147 from the BusinessOffice,Mineralogical Societyof America, 2000 Florida Avenue,NW, Washington,DC 20009.Pleaseremit $1.00 in advancefor the microfiche. Table il. Atom coordinatesand anisotropictemperaturefactors for A/c m:rnganoandiopside (above) and F2r/c kanoite (below) Atomxyz B lI M1 0 M2 0 Si 0l 0 . 2 1 r 4 ()r 0 . 4 0 7 5 )( t 3/4 0.7635(3) o . 4 r 2( 3 r) 0.8ss7(8) 02 03 0.382e(3) 0.r358(2) 0.r489(2) MI M2 0 . 2 5 0 5 )( l 0.2534(o) SiA S'iB 0tA 0 . 0 4 r 3 ()1 0 . 5 4 6 6I () 0.8674(3) 0 . 6 5 4 42() 0 . 0 2 3 5 )( i 0 . 3 4 i l( 2 ) 0lB 024 028 0.3726(3) 0 . r 1 8 (r 3 ) 03A 038 0.623e(3) 0.r034(2) 0.6038(2) 0.0e2e(2) 0. t 0 7 2 ( 1 ) 0.2510(3) 0.482 r (3) 0 . 8 3 8 82() ( 4) 0 .3 3 8 3 0.8382(4) 0.500e(4) 0.eer6(4) 0 . 2 5 e 03() 0 . 7 r 6( 3 r) 3/4 r3(2) 28(r) r 8 ()r 822 rr ( 3 ) 5 r( 2 ) 24(2) r8(2) o.67rs(8) 3r(3) 38(4) 2 8 ( 4) 0 . e 9 e 4 ( 7 ) 2 1( 2 ) 46(4) 0 . 2 3 s(72 ) 0 . 2 3 3 2) ( 1 0 .2 7 1(03) 0.2408(3) 0 .r 6 5 2 ( 9 ) 0.r367(e) 0.3255(9) 0.358e(e) 0.s707(6) 0 . 4 e 0( 5r ) 1 7( 2 ) 1 3 ( )r 2 3 ( )l r 7 ()r r 6 ()r 1e(2) 2s(3) 3 r( 3 ) 1e(2) 2 1( 2 ) 23(2) 3 r( r) re(2) 21(2) 23(4) 22(4) 26(4) 38(4) 4 r( 3 ) 44(3) 62(12) 8 6( 8 ) 812 B l3 823 0 6(3) 0 0 -4(2) -4(5) - 4 ( 6) r64(r7) 0(2) -t (2) 3(2) 12(2) il(4) 2 7( 4 ) 80(r6) -4(2) 7(4) -r6(6) 2(2) 6 ( r) 4 ( r) 4(1) e6(s) I (0) 3(0) - 2 ( t) rr ( 2 ) -6(2) e6(5) -t (t) il (2) 16(5) -6(2) -4(2) - 7( 2 ) re(5) r6(4) e(4) r2 ( 3 ) ro(5) -r2(s) -r7(5) r7(4) 4(2) 5(3) 34(4) e5(8) 112(20) 72(7) ro5(4) r38(r3) r6(r3) 124(12) r4s(14) r30(ro) i l 2 ( 10 ) 0 - 2 ( t) 2(2) I (2) -8(2) e(5) 130 GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS Mgor.SirOufor the C2/c phase, and MnoerMg,o, SirO.for the P2,/c phase,ascalculatedbasedonly on occupancyfactors,which were artifically constrained as describedabove.The relative volumes of the two A n g l ew i t h phaseswere determinedfrom preliminary transmis90 sion electronmicrographs,which confirmed that the r r( r 5 ) 2 r ( r 5 ) phaseswere indeed lamellar in nature. The computed averageweightedcomposition,consideringrel30(5) 6 0 (5 ) ative volumes,did not, however,agreesatisfactorily 90 with the observedaveragemicroprobe composition. 80(6) r05(e) The compositionsare plotted in Figure 1. Part of this Ie(8) discrepancyis related to the limiting assumptions 7 7( B ) made about cation distributions for the two phases. r 4 0 ( 3)7 l27(38) We thereforeconcludedthat although the site occu85(e) panciescorrectlyindicated that the phaseswere near 86(7) diopsideand kanoite in composition,details regard7(7) ing site occupanciesremainedto be determined. 38(r8) 60(20) The average Ml-O and M2-O bond distances 112(7) were also usedto determine site occupancies.Interatomic distancesas functions of composition were 57(5) examinedfor compositionalrangesof Ml and M2 of e 6 (r 0 ) Table 4. Thermal ellipsoid data for manganoan diopside and kanoite l v l n - D i o pi d se Atom Axi s t4l 1 2 3 0.066(7) 0 076(4) 0 . o e (' t7 ) r' r5' r82( (rr55)) 0 90 90 I 2 3 0 . 0 e 8 () 0 . 12 5 ( 2 ) 0 .r 4 4 ( 2 ) 60(s) r 5 o ( 5) 90 90 90 0 1 2 3 0 .0 8 8( 2) 0.098(3) 0.il2(4) 1e(e) 7 r( e ) e5(5) 74 ( a ) r56(9) r08(9) 0.090(5) 0.il8(8) 0.'r28(7) r3(B) 81(il) 8 1( 7 ) e r( 5 ) r28(40) 38(40) 0.r07(7) o.r3(5) 0.r46(7) .92\32) 1 75 ( 2 4 ) 86(6) 6(r7) 93(33) 9s(8) 0.089(B) 0.r05(5) 0.r42(6) 58(22) 147(22) e 0 ( 4) 72(6) i8(r0) 22(7 ) l'1'l 0 . 0 72 1 4 ) o .o B 4( 4) 0'r05(3) 43(r0) il5(il ) 58(6) r4(r4) 154(r3) 100(s) M2 0.094\2) 0.il4(2) 0.125(2) 42(3) il0(4) s5(3) r12(4) r58(3) e 4( 4 ) 56(3) ee(4) 14 5 ( 3 ) 0 . 0 8 (r 3 ) 0.08e(2) 0.il4(3) 50(r8) r 3 6 (B r ) r04(4) 46(rB) 46(r8) r 0 3 ( 4) 7 r( 4 ) e t( 3 ) re(3) ( 3) 0.083 0.092(3) 0.il4(3) 28(12) 65(13) r03(4) 7 0 (r 3 ) r53(10) r07(5) 7 t( 4 ) r00(7) 2 1\ 4 ) 014 0.089(6) 0 . 0 9 6 () 7 0 .r 3 6 ( 6 ) 36(46) 124(48) r 0 r( 6 ) (48) 'rr24r6 ( 4 5) 7 8 ( 7) 7 4 1)7 86(16) r6(6) 02A 0 089(8) 0.r2s(5) 0 . 13 r( 6 ) 6 4( 8 ) r53(r5) 99(35) 35(B) ' r6 ' 13o(( r26r ) ) 6e(7) 8 9 ( 3 )7 2 r( B ) 03A 0 . 0 8 (75) 0 .r ' r 4 ( 5 ) 0 . r5 0 ( 4 ) 2 6 ( 5) 7B(8) i r2(3) 65(6) 1 2 71 7) 4 7( 6 ) 3e(6) 5t(6) 0lB 0 089(9) 0 .r 0 2 ( 6 ) o . r2 5 ( 7 ) ' r76e9( (2r08)) 028 0 .0 8 8 6( ) 0 . r1 8 ( 7 ) 0.148(6) 30(9) 65(e) r05(5) Si 03 Rmsampli tude A n g l ew ' i t h abc* 90 Anglewith Kanoite siA 5iB 038 0.084(5) 0 . r ' 1( 4 ) e 4 ( r) r 62(8) 1 4 s ( )7 r03(4) 2 11 1z ) 81(19 7 r( e ) 7 4( 8 ) r 4 7( B ) r8(8) lre(5) rrB(6) 42(5) r46(6) r08(9) 97(il) le(e) 66(6) il 0(9) 32(7\ 42(4) 7717\ 5 t( 4 ) refined a pyroxene structure near ideal kanoite in composition.We thereforeshall not describethe details of the structuresbeyondthe parametersgiven in the tables,exceptto point out the well-defined7-fold coordinationof M2 in kanoite. The structurerefinementswere performed in part to limit the compositionsof the individual phases. The refined compositionscorrespondto CaorrMn",n Table 5. Selected cation-oxygen interatomic distances for kanoite and manganoan diopside MnD - io p si d e Kanoi te M r- 0 2 A 028 0rA 0 rB 2.045(4) ( 3) 2.066 2 . 0 7(75 ) 2.087 \4) t4r-02 0t 0t ave. 2x2.056(3) 2x2.068(4) 2x2.14013) 2 .O B B 0rA 2.166(4) 018 2.reo(4) ave. 2..l05 M2-028 2.1??(4) 0 2 A 2 . r 6 1( 4 ) 0 rB 2 . 1 7(74 ) 0 r A 2 . 1 e (73 ) 0 3 A 2 . 4 ? (13 ) o3B 2.7rs(3) uJb ^ 2.495 ^ . F t ^ \ 2.380 )14-UZA t.f,vJ(+J o'rA 1.612(4) 03A r.648(3) o3A r.682(3) .634 S iB - O 2 B l . 5 B B ( 4 ) 0tB r.613(4) 038 r . 6 7 3 ( 3 ) 038 L-62!13) ave. ave. 4) 2 x 2 . 2 7( 1 ? x 2 . 3 0(73 ) 2x2.613(3) 2x2.7BB(3) 4.do3tJ,, ave. ave. M?-02 01 03 03 1.637 si-02 0r 03 03 r.se0(3) r.605(3) r.666(3) 1. 6 7 (73 ) ave. I .635 13l GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS Table 6. RefinedMl and M2 occupanciesfor maganoandiopside and kanoite Mn-Diopsi de Mz K a n oi t e M n0 . 0 6 ( 1) l 4 g0 . 9 4 MI C a0 . 8 7 ( 2 ) M n0 . 1 3 142 l 4 n0 . I 0 ( l ) Ms 0.90 Mn0.86(l) M g0 . 1 4 each phase.Linear variation of bond distanceswith compositionwas assumed. For the P2,/c phase,the range clinoenstatite-synthetic kanoitewas usedfor both Ml-O and M2-O. Data for bond distanceswere obtained from Morimoto et al. (1960)for clinoenstatiteand from Ghose (written communication)for synthetickanoite; these are the only pyroxenesfor which data exist in the Capoor edgeof the Ca-Mn-Mg quadrilateral.For MlO the inferred occupancy of the Ml site correspondedto (MgornMn ,,), which is in excellentagreement with the refined site occupancy.For M2-O the refined bond distance value fell just beyond the range representedby the clinoenstatite-kanoitevalues. However, as some Ca must be present in the P2,/c phase(as it coexistswith a manganoandiopside),the slightly larger value of the M2-O distance (2.30vs.2.2564 for synthetickanoite) is entirely reasonablewhen the presenceof Ca in M2 is considered. For the A/c phase,the compositionalrange diopside-johannsenitewas chosenfor Ml, and diopsideCZ/ckanoite was chosenfor M2. Bond distancedata from Clark et al. (1969)for diopside and Freed and Peacor(1967)for johannsenitewere used.An M2-O interatomic distancefor a theoretical A/c kanoite wasgeneratedfrom the linear regressionequationsof Ribbe and Prunier (1977)for A/c pyroxenesand effective ionic radii of Shannonand Prewitt (1969). For Ml-O the inferred site occupancycorresponded to (MgornMno,,), which agreeswell with the refined value for that site. This method was not entirely applicable to M2 of the C2/c phase,as Ribbe and Prunier (1977) have shown that the M2-O distance in C2/c pyroxenesis not a function of the ionic radius of M2 alone.Rather,it is a function of the ionic radii of both the Ml and M2 sites.Thus, as the composition of the Ml site does not change appreciably in the range diopside-kanoite, the value obtained for the M2 site occupancy of the C2/c phase (CaorrMnoor)does not reflect any appreciableionic substitutionbetweenCa and Mn in M2. The overall results of this technique were in reasonableagreement with thoseof the structurerefinements. For cell parametersthis same basic method was applied, with the compositional ranges clinoenstatite-MnrSirOufor P2,/c and diopside-kanoitefor A/c.This methodhad the samegenerallimitations as the previous procedure. First, for the Ca-poor edgeof the Ca-Mn-Mg quadrilateral,cell parameter data exist only for ctnoenstatite (Stephensonet al., 1966),syntheticMnrSirO. (Syono et al., l97l), znd synthetic kanoite. Second,the compositional range clinoenstatite-MnrSirO.does not account for small amountsof Ca, which are likely to be presentin the P2,/c phase,and which will have a significant effect on the cell parameters.Thus, the value shown in Figure I for the composition of the P2,/c phase from this method is only a rough estimate.For the A/c phasethe estimateof compositionwas more tightly constrained;average Ml-O and M2-O bond distanceswere generatedfrom the observedcell parameters and Shannon and Prewitt's (1969) effective ionic radii in the linear regressionequationsof Ribbe and Prunier (1977) for A/c pyroxenes.From these equations,Ml and M2 ionic radii were computedfor the A/c phase,which in turn allowed a more accurate estimateof composition to be made fot A/c than for Y2r/c.The resultsof this methodare shown in Figure l. The compositionsdetermined for each phasecorresponded to Ca"rrMno,Mg,oSirOufor the / En l\l9SiO3 \ Rh lvlnSiO! Fig. 1. Plot of Balmat C2/c and P21lc clinopyroxene compositions in the Ca-Mn-Mg quadrilateral. The points labeled A and B are average compositions for the Balmat clinopyroxene as determined from electron microprobe analyses. A corresponds to the average composition of the sample analyzed using srru; B is included to show the range of average pyroxene compositions of the Balmat samples. I and A represent compositions determined by structure refinements and plots of cell parameter v.t composition, respectively. The points labeled X, Y, and Z represent average compositions determined from the results of each method. r32 GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS C2/c phase, and MnosrMg,,rSirOufot the P2,/c phase. The averagepyroxene composition determined by this method, consideringrelative volumes, corresponded to Cao3oMfurrMg,orsirou, in only fair agreementwith the observed average microprobe composition,although in reasonableagreementwith the averagerefinement composition of Ca"rrMno., Mg, ooSirOu. Nonetheless,the resultsof theseattempts clearly underscoredthe necessityfor somemeans of determining the lamellar compositionsdirectly, which led to analysisof this material by techniques of analyticalelectronmicroscopy.Fortunately, at this stage of the study appropriate instrumentation becameavailableto us. Analytical electronmicroscopy The electron microscopewas a JEoL JEM l00cxASID outfitted with a Princeton Gamma-Tech energy-dispersivesolid-statedetectorwith an analytical resolution of 146 eV at 5.9 keV. This instrument combinesthe imaging and diffraction capabilitiesof the conventional(fixed beam) transmissionelectron microscope(crnu) with the analytical capabilities offered by a scanning electron beam instrument (sreu). It has the capability of 2.0A line resolution in the transmissionmode, using a side-entry goniometer and the tilted beam technique for lattice fringe imagery.Incident beam diametersof 50A and lesscan routinely be obtained.The actual spatial resolution for analysis,however, may be considerably increasedin relatively thick samples.A thick nonbeam-defining Mo aperture positioned above the specimenwas used to eliminate most of the hard Xrays generatedby the beam-definingaperture in the condensersystem.A graphite specimenholder was usedto further minimize X-rays generatedin the vicinity of the sample.NonethelessCuKa was detected in all spectra,but was subsequentlydetermined to arise from the specimenrod beneath the graphite holder. A liquid-nitrogen-cooledanti-contamination devicewasemployedat all timesto minimize organic contaminantsformed at the beam-specimeninterface. All electronmicroscopywas performed at 100 rectly observableon the optical level, but could easily be detectedwith a reflected light source by the schiller. To ensurethat the lamellar boundary [predicted to be (001)l would be as nearly parallel to the incident electron beam as possible,only those crystals showingcenteredor near-centeredoptic normal figures were selected.3-mm-diameterbrass washers were then attachedto the crystals with a non-acetone-solubleepoxy. The crystals were subsequently removedfrom the slide by dissolving the glasswax with acetone,and thinned by ion bombardment using standard conditions (7 kV, l00pa Ar ion beam current, and inclination of the sample to the beams of -15"). From conventional transmission electron micrographs,the lamellae were seento be approximately 20004 wide. The interphaseboundary was directly of loindexedas (001)within error of measurement 2". This was accomplishedthrough determination of the direction, luvwl, of the lamellae boundary for each of severaldifferent orientations. Further confirmation of this index was obtained by measuring the anglebetweenthe lamellar boundary and (ll0) cleavagesteps seen in transmissionimages, which was76o45'.Robinsonet al. (1971)studiedthe orientations of exsolutionlamellae in clinopyroxenesand showedthat the interphaseboundary of lamellae exsolvedon a plane that is apparently(001)may not be parallel to (001);that is, the interfaceand (001) may in fact differ by an angle of as much as l0o. Therefore, the interphaseboundary is actually someplane of irrational index that representsthe plane of dimensionalbestfit betweenthe two latticesat the time of exsolution. Our index determination, therefore, while consisientwith the interphaseboundary being (001),may by analogybe up to lo-2o from the (001) plane. Quantitativeanalysis The Cliff-Lorimer (1972)simple ratio method for thin foils wasusedto determinethe individual pyroxene compositions.This method is extremely convenient if the sampleis suffi.cientlythin that atomic kv. number, absorption,and fluorescence(ZAF) effects Specialthin sectionsof the Balmat sampleswere are minimal and may effectively be neglected.The made, using a glass wax that was acetone-soluble, ratio of characteristicX-ray intensities of two elerather than a conventionalthin-sectionepoxy. Single ments is thereforerelated to the concentrationratio crystalsof pyroxeneexhibiting schiller were then se- of those two elementsby the expression(C"/C) : lectedoptically.The optical orientation of this pyrox- k"b(I,/Ib). We have taken the ratio of each element ene appearsin generalto be the sameas that of the presentin the unknown (Mn, Mg, and Ca) to Si, ascommonclinopyroxene,i.e. the optic normal is paral- sumingthat only Si occupiesthe tetrahedralsite (this lel to the D axis. Generally the lamellaewere not di- was also assumedfor all of the standardsused).The 133 GORDON ET AL: CLINOPYROXENE EXSOLUTION REI'.ATIONSHIPS Nuclear Data 6600 computer systeminterfaced with the detectorincluded a program which calculatedthe area under the curve for a chosenpeak, so that the final intensity value corresponded to peak-background.Upper and lower energylevelsfor eachpeak were set manually. These computed areaswere the actual valuesusedin the intensity ratio calculations. The simple ratio method has been used successfully by other researchers on geologicalmaterials(Lorimer and Champness,1973;Copley et al., 1974;Nobugai et al.,1978). Somepotentialproblemsof chemicalanalysiswere contamination buildup and mass loss due to volatilization, both associatedwith the stationary spot mode used for analysis.To check for errors from theseeffects,ratios of characteristicX-ray intensities were taken from the same spot every 200 seconds over a l5-minute period. This was done on the unknown and on eachof nine standards.No detectable errors were found from either effect, as intensity ratios remainedconstant.We point out, however,that this was determined only for the materials used in this study. Care should be taken with other minerals (most notably feldsparsand carbonates)to check for this effect,as masslossmay lead to seriouserrors in compositiondeterminationsfor some phases.As an additional checkfor beam damage,micro-micro diffraction was performed on the individual pyroxene lamellae(1.e.diffraction from areasof approximately 1000x 1000A;Mardinly, personalcommunication). If specimendamage was occurring, the diffraction pattem should have gradually deteriorateddue to alteration of the specimenin the beam. The resulting diffraction patternswere observedin the microscope over a period of severalminutes,and were found not to change.Therefore, we concluded that sample damagefrom the beam was negligible. Analysisdependson the measurementof k, a constant for each element ratio which includes corrections for the efficiencyof X-ray detectionby the detector (Cliff and Lorimer, 1972). To determine k values,X-ray intensity pairs for CalSi, Mn/Si, and Mg/Si were measuredfrom nine homogeneouswellanalyzed standards routinely used in microprobe analysesat the University of Michigan. The standards used and k values are listed in Table 7. This procedureyielded three intensity ratios for Mn/Si, four for Mg/Si, and six for CalSi. The k valueswere determinedin two ways: (l) by simple calculation from the above expression,using known values of element/Si concentrationin element weight percent and observed X-ray intensities; and (2) by least- squaresrefinementof the straight lines resulting from plots of concentration ratio vs. intensity ratio; the slopesof these lines are k values. There was very good agreementbetweenk values determined from both methods.23 analysesper standardwere made, counting for 200 secondseach.The specimenholder was tilted at an angle of approximately 30o during analysis.The standardswere analyzed as crushed crystalssuspendedon holey carbon films, becauseinsufficientamountsof eachwere availablewith which to prepareion-thinned standards.Care was taken to anzlyzeonly thin edgesof crystal fragmentsin order to eliminateZAF corrections.The plots of concentration vs.intensityratios alsoservedascheckson specimen thickness;beyond a critical thickness,ZAF effects must again be considered.As long as the relationship between concentrationand intensity is linear, the simple ratio method is applicable. Note that k values must be determined for every instrument, as efficiencyof X-ray detection may vary between detectors,with operating conditions, and with data reduction methods. The lamellar compositions were determined by summing the concentrationsof Ca, Mn, and Mg to one for each phase, such that the following expressionscould be used: C.": [k.".'(Ic"/Ist)]Csr (l) C'": [kM"st(IM"/Isi)]Csi Q) C.r: [k',",(I'rlI.')]C", (3) Each equation is the original expressionrewritten to solvefor the concentrationof the octahedral-siteelements. Thus written, the observed intensity ratios and k valuescan be usedto determinethe Si content of eachphase: [kc"s,Gc",/I",) + kM"si(IM"/Isi) + kMesi(I.s/I.')lC"' : (4) I This value of C", was then used in expressionsl, 2, and 3 to solvefor the contentsof Ca, Mn, and Mg in eachphase.Approximately 50 analysesof eachof the Balmat pyroxeneswere made to determineelement,/ Si intensity ratios. Standard statistical analyses yieldedthe final valuesof k to be k."., : 1.27+ 0.09, k'.r' : 1.399i 0.002,and k'r", : 0.47 + 0.04.The resulting errors in the STEMcompositionsare representedby the circles shown in Figure l. The results of this procedure yielded compositions of for the C2/c phase, and Cao.rMnoooMgorrSirOu for the F2'/c phase.The averCao,rMn,orMgoruSirO. age composition,considering relative lamellar vol- t34 GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS Table 7. Standardconcentrationand intensity ratios used in determinationof k values CalSi Conc. I lvln/Si Conc. I Mg/si Conc. I 0.445 0.46 0 . 2 4 40 . 3 4 0.2540.36 2 . 0 0 02 . 8 0 0 . 4 6 50 . 1 9 0 . 4 6 50 . 2 3 0 . 4 7 00 . 2 3 0 . 5 ' l 70 . 2 5 0.490 0.62 0.495 0.51 0.495 0.66 0.675 0.77 0.995 1.27 'Standards u s e dw e r e s y n t h e t i c t e p h r o i t e , B r o k e n H iI I b u s t a mtie , a n d W a b u s h c u m mn ig t o n it e f o r M n / S ;i B r o k e nH i l 1 b u s t a m i t e ,D u c k t o wdni o p s i d e , G o l d i c h d i o p s i d e , A N Uh e d e n b e r g i t eA, N Uw o l l a s t o n i t e , a n d I r v i n g c i i n o p y r o x e n feo r C a l S i ; W a b u schu m m i n g t o n i t e , D u c k t o wdni o p s i d e , G o l d i c hd i o p s i d e , a n d I r v i n g c l i n o p y r o x e n ef o r M g / S i . C o n c e n t r a t i o nr a t i o s f o r t h e s e p l o t s w e r e c a l c u l a t e d i n n r o l ep e r c e n t f r o m s t a n d a r da n a i y s e s . umes,correspondsto CaoruMnorrMgourSirOu, which agreesreasonablywell with the observedaveragemicroprobecomposition.As predictedearlier in considering the refined compositions,the P2,/c phasedoes indeed contain some Ca (approximately six mole percent). The two compositions are analogous to those found in coexistingaugites and pigeonites of the Ca-Mg-Fe pyroxenequadrilateral.The most obvious differencein the Ca-Mg-Mn systemis the relative enrichment of the P2,/c phase in Mn. This is probably due to the marked preferenceof Mn for the larger M2 site,as noted by Ghoseet al. (1975)in a study of site preferencesof transition metal ions in syntheticpyroxenes.They found that, based on site occupancyrefinements,Mn showed a much greater affinity for M2 than did Fe. Another major differencebetweenthe two quadrilateralsis the apparent smaller width of the solvus in the Ca-Mn-Mg system. A probable explanation for this lies with the similarity in ionic sizebetweenCa and Mn; thus the impetus for unmixing would not be as great as that expectedfor Ca vs.Mg and Fe, where cation sizesare more disparate.The results of all methods used to determinelamellar compositionsare shownin Figure 1. The compositionsdetermined with srnu clearly give the best agreementwith the observedaverage microprobecompositions. Antiphasedomains The scaleof antiphasedomains(APD's) in pigeonites has been used by severalresearchersto indicate relative cooling rates (Bailey et al., 1970;Christie el al., l97l; Brown et al., 1972;Ghoseet al., 1972;LalTy et al., 1972,1975; Brown and Wechsler,1973; Nord et al., 1973).At high temperatures,pigeonitespossess C2/c symmetry,in which two chains of linked SiOo tetrahedra are identical. With decreasing temperature, the coordinationof the cationsin the M2 sites changesin concert with shifts in the chains. The chains are no longer related by symmetry, and the spacegroup is P2,/c. In pigeonitesthe temperature of this transformation is controlled by composition (Prewittet al.,l97l). The resultingmicrostructureis composedof regions of the P2t/c structure (APD's) related to each other across antiphase boundaries (APB's) by a translationof (a + b)/2 (Morirnoto and Tokonami, 1969).The structureis thus C-centeredat the boundaries. APD size should be a function mainly of cooling rate through the inversion temperature,as the C2/c-P2,/c inversion in pigeoniteseemsto occur by random nucleationand growth of lhe P2,/c phasein the C2/c phase.However, heterogeneousnucleation may occur and theseAPD's may be larger than if formed by random nucleationalone, and thus would give an erroneousindication of cooling rate (Lally et al., 1975).With a relatively slow cooling rate, large APD's with fairly straight boundaries may be expected, becausethe few random Y2,/c nuclei that form in Ihe A/c phasehave more time to grow, and straightening of boundaries is one way in which APD's can becomeenlarged(Cupschalkand Brown, 1968).Few instancesof APD's with straight boundaries have been reported in pigeonites (Champness and Lorimer, l97l; Lally et al., 1975). Carpenter (1978) describesslowly-cooledterrestrial pigeonites, which exhibit not only large APD's but also relatively straight boundaries.He cites evidencefor the enrichment of some APB's in Ca, as predicted by Morimotoand Tokonami(1969).He alsopresentsdirect evidencefor theseAPB's being good nucleation sitesfor augite,as noted by Lally et al. (1975). As mentionedearlier, APD's on the order of 200A or less(Morimoto and Tokonami, 1969)are revealed in single-crystalX-ray diffraction photographsby the presenceof diffusenessin the class(b) reflectionsof the P2,/c phase.However,the class(b) reflectionsof our F2,/c phasewere quite sharp.This was explained by bright- and dark-field transmissionelectron micrographswhich revealedthat not only were APD's present in the primitive phase, but that they were large,rangingfrom -400 to -3500A in wdith. As no nucleation of other phaseswas seen in either the P2,/c or A/c phases,this coarseAPD sizealone in- GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS 135 dicatesa slow cooling rate. In any case,the large domain sizeallows the APD's to difract independently, consistentwith the sharp quality of the class (b) Y2r/c reflections.Figure 2 showsdark-field transmission electronmicrographsof the BaLnat pyroxenelamellae,showingonly diffraction of tll:eP2,/c phase. The APD texture is readily apparent.The reflections used for imaging were sharp, single class(b) reflections, representedby the circled spot in the accompanying diffraction pattern (inset),which is the a*c* plane. Clearly, from this microtextural evidence,the Ca-poor pyroxene in the Ca-Mn-Mg quadrilateral undergoesa C2/c-Y2,/c inversion analogousto that of pigeonitesin the Ca-Mg-Fe system. The temperatureof APD inversion was qualitatively determined with a special sample heating holder in the srBu. An ion-thinned fragment of pyroxenewas enclosedin a folding Cu grid, which was then placedin the heating holder. Immediately prior to heating,the APD texture was imagedin dark-field mode.Both the APD texture and the sharp class(b) imaging reflection were monitored during heating. The rate of heating and cooling was rapid (approximately 50oC per minute). As the inversion temperature was approached,the APD texture was seento disappear quickly and continuously; concurrently, the imaging reflection was seen to decreasein intensityuntil it disappearedcompletely,as seenin the sequenceshown in Figure 3. The samplewas maintained above the inversion temperature(at temperaturesas high as -360"C) for no more than one to two minutes before cooling. This procedure was performedseveraltimes within a period of about 2030 minutes. Care was taken to determine whether changesin sampleinclination during thermal cycling might be causing the disappearanceof both APB contrastand diffraction intensity. Bright-field observation of extinction contours, which are extremely sensitiveto changesin sampletilt, indicated that the specimen remained stable during temperature changes.The results of these experiments yield a temperature of inversion corresponding to 330 t 20"c. Ca enrichmentof APB's In somecases,enrichmentin Ca of APB's can be detectedby the broadeningof APB's or by the nucleation of a Ca-rich phase at the APB's (Lally et al., 1978).We observedneitherof these 1975;Carpenter, effects. but instead noted the distribution of the APB's which relatesto the distribution of Ca. before and after heatingthrough the inversion temperature. In Figure 4, which representsthese conditions, the Fig. 2. Dark-field transmission electron micrographs of Balmat clinopyroxene taken at 100 kV using h+k odd reflections: (A) shows diffraction pattern (inset) of a*c* plane, with imaging reflection circled; note the relatively large APD's and fairly straight boundaries. (B) and (C) also show "pinch-out" lamellae (arrowed) in both the P2,/c (light) and A/c (dark) lamellae. GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS APB's are generallysimilar, even identical in several areas.This indicatesthat somethingis controlling the distribution of the APB's, giving rise to a "memory" effectafter rapid heating and cooling through the inversiontemperatur€.We concludedthat Ca rrust be concentratedat the APB's to some extent, causing this phenomenon,although it is possible that it is causedby the presenceof defects.As heating and cootng were rapid, the likelihood of appreciablediffusion of Ca away from the APB's while the sample was at a temperatureabove that of the inversion is small. In an attempt to directly confirm the presenceof Ca at the APB's, Ca X-ray imagesof lamellae showing the APD texture were recorded.A window was placedon the emissionspectrumaround the Ca peak in order to permit the detection of CaKa photons only. Theseimages(Fig. 5) qualitatively illustrate the enrichmentof the A/c phasein Ca with respectto n,/c.Withlnthe Y2r/c phase,however,it is not possible to determineif the few Ca X-ray photons detectedcorrelatedirectly to the APB's. The amount of Ca presentat the APB's is simply too small to be detected relative to the inefficiency of the method. However,in light of the "memory" effectseenin the APB orientationsfrom the heating experiments,we concludethat some Ca is probably concentratedat the APB's, and this is very likely the factor controlling the distribution of the APB's. Coherencyof the interphaseboundary Fig. 3. Sequence showing the disappearance of p2r/c h+k odd imaging reflection during heat of clinopyroxene: (A) 135"C; (B) 260"C; (C) 350"C. Diffraction patterns were taken in dark-field mode, with the electron beam tilted so that the diffracted beam is parallel to the optic axis of the microscope. All micrographs taken at 100 kv. From examinationof the unit-cell parameters,we predictedthat the index of the interphaseboundary would be (001); this was subsequentlyverified by electrondiflraction and transmissionelectron micrographs.This prediction was basedon the fact that rhe greatestdifferencein magnitudesof parameterswas betweenthe c axesof the two pyroxenes.Both phases have the b axis in common, and a and D of the two phasesare parallel. tf both the a and D parameters were identical, the interphaseboundary would be a coherentunstrainedboundary. As the b parameters are equal within the error of observationand the a parametersare nearly equal, we thought that the lamellar interface in the sample might be such a boundary.To investigatethe coherencyof the interface, high-resolution electron microscopy (lattice fringe imaging)was performedusing the (200) reflection. Figure 6 showsa tilted beam lattice fringe image of (200) planescrossingthe lamellar boundary. The accompanying difraction pattern shows the a*c* plane, with the characteristicextinctions apparent. GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS Fig. 4. Dark-field micrographs of APD's of P21/c phase: (A) before heating; (B) after rapid heating (to 350'C) and cooling. Note the extremely similar orientations of APB's in both micrographs (100 kV). GORDON ET AL.: CLINOPYROXENE EXSOLIITION RELATIOI{S.rIIPS closest-packedoxygen anions. From these observations, it appearsthat this interfacemust be a largely coherentboundary.The lattice parametersparallel to the interfaceare so similar, however,that a minimum strain condition would exist if the interface was largely coherentwith dislocationsseparatedby approximately 10004, a situation which would require that the interfacebe defined as semicoherent. Discussionof the exsolutiontexture The Balmat clinopyroxenelamellae are homogeneous phases,share (001), and are approximately 20004 wide. The interphaseboundary, as described above,is largely coherentwith little strain, and the presenceof APD's indicatesthe existenceof a C2/cF),/c inversion for the Ca-poor phase.The Balmat samplesexhibit an exsolution phenomenonformed completelyas a product of regional metamorphism, with a well-definedand simple thermal history. The parent material for this rock was shallow marine sediment.The probable geologic sequenceinvolved heating and crystallizationof sedimentsby regional metamorphism to the peak temperature of 625"C and pressureof 6.5 kbar (Brown et al.,1978);formation of the pyroxene,which waspresumablyhomogeneousat the peak of metamorphism;and subsequent slow cooling from peak metamorphic conditions which permittedthe pyroxeneto unmix continuously to two A/c phases,one Ca-rich and one Ca-poor. Further slow coolingthen allowedthe Ca-poor phase to invert from C2/c to P2,/c symmetry,with the associated formation of relatively large APD's. Another feature of the exsolutiontexture is the "pinching-out" of lamellae (Fig. 7), which occurs in both Fig. 5. (A) Scanning transmission image of clinopyroxene phases.A possiblemechanismfor formation of this lamellae (lighter lamellae arc A/e; (B) scaming transmission Ca feature is describedby Copley et al. (1974); it inX-ray image of same area as in (A), showing A/c lamellae cludes phase nucleation and subsequent lamellar enriched in Ca; 100 kV. growth by the propagationof ledgesalong the interphasesurface.In our sample,however,these ledges may simply be a strain-releasemechanismafter the The centralbeamand the (200)reflectionusedto im- formation of a dislocation,and may not necessarily age the 4.6A periodic fringes are circled. In this im- be the result of heterogeneous nucleation. age,it is possibleto follow thesefringes without inWe cannot assign a specific mechanism of exterruption acrossthe interface, at which point they solution using only the observedmicrostructure,albend slightly. The angle between the fringes from though spinodal decompositionis certainly possible. one phaseto the other is 2.5o,which is equal to the Featuresof the microstructureare consistentwith the angle betweenthe c axes within error of measure- known geologichistory of the sample,especiallythe ment. Viewing the structure in this orientation @* mostly coherentlamellar boundary, the large size of parallel to the incident electron beam) is essentially the APD's, and the apparent Ca-enrichment of equivalent to observing the tetrahedral-octahedral APB's. Of special interest is the small amount of layers of the pyroxene structure, i.e. the planes of strain associatedwith the lamellar boundary, which GORDON ET AL: CLINOPYROXENE EXSOLUTION REI-/ITIONSHIPS Fig. 6. Lattice fringe image ofinterphase boundary (001). Ditrraction pattern (inset) shows the atc* plane; circled are the central beam and the (200) reflection used to image the lattice fringes. Note the slight bending of the (200) planes across the interface' Dark-field micrograph, 100 kV. indicatesthat the a and b parametersof both phases must havebeenas similar at the time of exsolutionas they are now. In general,the exsolutionrelationships exhibited by the Balmat clinopyroxene provide markersin the cooling history of the rock. In sum, techniquesof analytical electron microscopy have provided information complementaryto that determinedby the more conventionaltechniques of electronmicroprobeanalysisand singlecrystal Xray diffraction. The exsolution relationships for clinopyroxenesin the system CaSiOr-MnSiOrMgSiO. are strikingly similar to thoseof the common rock-formingsystemCaSiO.-FeSiOr-MgSiO.. GORDON ET AL: CLINOPYROXENE EXSOLUTION RELATIONSHIPS References Fig. 7. Ledges (arrowed) along interphase boundary. ,.pinchout" lamella is arrowed in (A). Bright-field rransmission micrographs, 100 kV. Acknowledgments We thank Professor W. C. Bigelow, John Mardinly, and Steve Krause of the Department of Materials and Metallurgical Engineering of The University of Michigan which houses the sTEM and the electron microprobe for continuing operational assistance with the instruments and for many valuable discussions. We also thank Craig Johnson for the final electron microprobe analysis of the pyroxene. We are grateful to St. Joe Minerals Corporation who provided summer support for one of us (pEB), and particularly to Dr. David Dill who saw that the rhodonite- and pyroxenebearing samples were saved. We are also indebted to Dr. Gordon L. Nord, Jr., of the U. S. Geological Survey for his careful review of the manuscript. This research was supported in part by a Sigma Xi Grant-in-Aid of Research and a Geological Society of America Research Grant to Wendy A. Gordon and NSF grant EAR 7g22758 to Donald R. Peacor. Bailey, J. C., Champness,P E., Dunham, A. C., Esson,J., Fyfe, W. S., MacKenzie, W. S., Stumpfl, E. F., and Zussman, J. (1970)Mineralogy and petrology of Apollo ll lunar samples. Proceedingsof the Apollo ll Lunar ScienceConference,169194. Bohlen,S. R. and Essene,E. J. (19'17)Feldsparand oxide thermometry of granulitesin the Adirondack Highlands.Contributions to Mineralogyand Petrology,62,153-169. Brown, G. E. and Wechsler,B. A. (1973)Crystallographyof pigeonitesfrom basaltic vitrophyre 15597.Proceedingsof the Fourth Lunar ScienceConference,887-900. Brown,G. E., Prewitt,C. T., Papike,J. J., and Sueno,S. (19'12)A comparisonofthe structuresoflow and high pigeonite.Journal of GeophysicalResearch,7'I, 5778-5789. Brown, P. E., Essene,E. J., and Kelly, W. C (1978)Sphalerite geobarometryin the Balmat-Edwardsdistrict, New York. American Mineralogist,63, 250-257. Brown, P. E., Essene,E. J., and Peacor,D. R. (1979)Phaserelationsinferredfrom feld data for Ca-Mg-Mn pyroxenesand pyroxenoids.(abstr.) Transactionsof the American Geophysical Union.60. 422423. Busing,W. R., Martin, K. O., and Levy, H. A. (1964)A Fortran crystallographicfunction and error program. U. S. National TechnicalInformation Service,ORNL-TM-306. Carpenter,M. A. (1978)Nucleation of augite at antiphaseboundariesin pigeonite.Physicsand Chemistry of Minerals, 2, 237251. P. E. and Lorimer,G. W (1971)An electronmicroChampness, scopicstudy of a lunar pyroxene.Contributions to Mineralogy and Petrology,33, l7l-183. Christie,J. M., Lally, L. S. Heuer,A H., Fisher,R. M., Griggs,D. T., and Radcliffe, S. V. (1971)Comparativeelectron petrography of Apollo I l, Apollo 12, and terrestrialrocks. Proceedings ofthe SecondLunar ScienceConference,69-80. Clark, J. R., Appleman,D. E., and Papike,J. J. (1969)Crystalchemicalcharacterizationof clinopyroxenesbasedon eight new structurerefinements.MineralogicalSocietyof America Special Paper2, 3l-50. Clitr, G. and Lorimer, G. W. (1972)The quantitative analysisof thin metal foils using EMMA-4-the ratio technique. Proc.eedings of the Fifth European Congresson Electron Microscopy,InstituteofPhysics,Bristol,p. 140-141. Clitr G. and Lorimer, G. W. (1975)The quantitative analysisof thin specimens.Journal of Microscopy, 103,203-207. Copley,P. A., Champness,P. E., and Lorimer, G. W. (1974)EIectron petrographyofexsolution texturesin an iron-rich clinopyroxeneJournalof Petrology,15,4l-57. Cruickshank,D. W. J. (1965)Errors in least-squares methods.In J. S. Rollett, Ed., Computing Methods in Crystallography,p. 112-116.PergamonPress,Oxford. Cupschalk,S. G. and Brown, N. (1968)Kinetics of antiphasedomain growth. Acta Metallurgica, 16,657-666. Doyle, P. A. and Turner, P. S. (1968)RelativisticHartree-Fock Xray and electronscatteringfactors.Acta Crystallographica,A24, 390-397. Engel, A. E. J. (1956)Apropos the Grenville. In J. E. Thomson, Ed., The Grenville Problem,p.74-76. Royal Societyof Canada SpecialPublication L Finger, L. W. and Prince,E. (1975)A systemof Fortran IV comput€r programs for crystal structure computations.U. S. National Bureauof Standards,TechnicalNote 854. GORDON ET AL.: CLINOPYROXENE EXSOLUTION RELATIONSHIPS Freed, R. L. and Peacor,D. R. (1967)Refinementof the crystal structureof johannsenite.American Mineralogist, 52, 709-720. Ghose,S., Ng, G., and Walter, L. S. (1972)Clinopyroxenesfrom Apollo 12 and 14: exsolution,domain sttucture and cation order. Proceedingsof the Third Lunar ScienceConference,507531. Ghose,S., Wan, C., and Okamura, F. P. (1975) Site preference and crystal chemistry of transition metal ions in pyroxenes and olivines.(abstr.)Acta Crystallographica,A'3l, 576. Kobayashi,H. (19'17)Kanoite, (Mn++,Mg)r[Si2O6],a new clinopyroxenein the metamorphicrock from Tatehira, Oshima Peninsula, Hokkaido, Japan.Journal of the Geological Society of Japan,83,53'l-542. Lally, J. S., Heuer,A. H., Nord, Jr., G. L., and Christie,J. M. (1975)Subsolidusreactionsin lunar pyrox€nes:an electronpetrographic study. Contributions to Mineralogy and Petrology, 51,263-281. Lally, J. S., Fisher,R. M., Christie,J. M., Griggs, D. T., Heuer, A. H., Nord, G. L., and Radcliffe, S. V. (1972) Electron petrography ofApollo 14 and 15 rocks. Proceedingsofthe Third Lunar ScienceConference,401422. Lorimer, G. W. and Champness,P. E. (1973)Combined electron rnicroscopyand analysisof an orthopyroxene.American Mineralogist,58, 243-248. Morimoto, N. and Tokonami, M. (1969)Domain structureof pigeoniteand clinoenstatite.American Mineralogist,54,725-740. Morimoto,N., Appleman,D. E., and Evans,Jr., H. T. (1960)The crystal structuresof clinoenstatiteand pigeonite.Zeitschrift fiir Kristallographie, | 44, 120-| 47. Nobugai,K., Tokonami, M., and Morimoto, N. (1978)A study of subsolidusrelationsof the Skaergaardpyroxenesby analytical electronmicroscopy.Contributions to Mineralogy and Petrolo g y , 6 7l,l l - 1 1 7 . l4l Nord, Jr., G. L., Lally, J. S., Heuer, A. H., Christie, J. M', Radcliffe, S. V., Griggs, D. T., and Fisher, R. M. (1973)Petrologic study ofigneous and metaigneousrocks from Apollo 15 and 16 using high voltage transmission electron microscopy. Proceedingsof the Fourth Lunar ScienceConference,953-970. Peacor,D. R. Essene,E. J. Brown, P. E., and Winter, G. A. (1978) The crystal chemistryand petrogenesisof a magnesianrhodonite. American Mineralogist, 63, 1137-1142. Prewitt,C. T., Brown, G. E., and Papike,J. J. (1971)Apollo 12 clinopyroxenes:high temperatureX-ray diffraction studies.Proceedingsof the SecondLunar ScienceConference,59-68. Ribbe, P. H. and Prunier,Jr., A. R. (1977)Stereochemicalsystematics of ordered A/c srlicatepyroxenes'American Mineralogist,62,'llG-720. Robinson,P.,Jatre,H. W., Ross,M., and Klein, Jr., C. (1971)Ori€ntation of exsolution lamellae in clinopyroxenes and clinoamphiboles: consideration of optimal phase boundaries. American Mineralogist, 56, 909-939. Rucklidge,J. C. and Gasparrini, E. L. (1969) Specificationsof a completeprogramfor processingelectronmicroprobedata: eMPADRVIL Departmentof Geology,University of Toronto, Toronto. Shannon,R. D. and Prewitt, C. T. (1969)Effective ionic radii in oxidesand fluorides.Acta Crystallographica,825' 925-946. Stephenson,D. A., Sclar,C. B., and Smith, J. V. (1966)Unit cell volumesof syntheticortho€nstatiteand low clinoenstatite.MineralogicalMagazine,35, 836-846. Syono,Y., Akimoto, S., and Matusi, Y. (1971) High pressure transformationsin zinc silicates.Journal of Solid StateChemistry, 3, 369-380. Manuscriptreceived,February1, 1980; accepted for publication,Seplember15, 1980.