American Mineralogist, Volume 70, pages822-828, 1985

The origin of sector twinning in harmotome

Mzunrro Axrzuxr lnstitute of Mineralogy, Petrology, anil Economic Geology Faculty of Scierrce,Tohoku Uniaersity, Sendai 980, Japan

Abstrrct Harmotome from Strontian, Scotland and Korsniis, Finland was studied from the stand- point of atomic ordering produced on the surfaceduring growth. The (010)face has several ihombic growth hillocks consistingof four equivalentvicinal faces,and thin sectionspardlel to (010)show correspondingfourling texture. The thin section parallel to (001)shows some domains in which the extinction is inclined slightly to the baxis, suggestingtriclinic sym- metry. Deviation from orthorhombic to monoclinic symmetry can be explained by ordering of barium, as observedby structurerefinement. Deviation from monoclinic to triclinic symmetry is attributed to AfSi ordering which was not shown by structure refinement.Both atomic orderings are produced on the side face of a growth step during Sxowth. If the two side faces are symmetricalwith resp€ctto mirror planesin pseudo-orthorhombicharmotome, the two correspondingdomains will be twinned.The spacegroup of harmotomeshould thus be Pl.

InFoduction showed{l0l}, {010}, {111} and {001} forms.The surface internal textures of harmotome from both The minerals of the harmotome- series features and localitiesare similar. All photographsreported in the pres- (Ba- ssi- 11032' l2H2O) have been described rCa-o.rAl- pap€r obtained from harmotome crystals from as orthorhombic, monoclinic and triclinic. Kalb and ent were and Lehtinen (1967) showed that the Klotsch (19,14)suggested that phillipsite is orthorhombic, Strontian. Sahama has one large growth hillock with four and Steinfink (1962) refined the crystal structure in the Korsnds crystal on the (010)face. The (010)faces of the present spacegroup B2mb. Conversely,Sadanaga et al. (1961)re- vicinal faces have severalrhombic growth hillocks, suggest- fined the structure of harmotome in P21lm. Grnf (196a) specimens growth mechanism. The (001) face is divided suggestedthat the orthorhombic symmetry exhibited by ing a spiral elongatedin the direction parallel to some membersof this seriesis merely a geometricaverage into several blocks, each block has one spiral growth pattern. produced by twinning a monoclinic structure.Subsequent- the b'axis. and are slightly irregular. The (101)and ly, Rinaldi et al. (1974)refined the structuresof both mem- The block boundaries (111) are finely striated parallel to the c-axis' bers in P2rlm. Langeman(1886) reported harmotome with faces features were observed by differential inter- triclinic optics. After surface optical microscopy, the surface was Single crystals invariably consist of complex sectoral ference reflection glassslide and a thin sectionwas made. Fig- twins, and Deer et al. (1962)suggest that the twinning may cementedto growth featureson the (010) face and simulatesingle crystal forms, such as tetragonaland rhom- ures 1 and 2 show corresponding internal texture bic dodecahedra.Sahama and Lehtinen (1967) described Figure 3 ropresents the transmitted light. Figure 4 is a sectoral twinning in harmotome from Finland. Akizuki viewed in cross-polarized photograph of the lower rectangle of (1981a,b, 1984) has interpreted sector twinning in some high-magnification minerals in terms of atomic ordering of such cations as Figure 3. of which are indicated by AVSi and Fe3*/Al produced during growth. The surfaoe Growth hillocks, the centers the internal fourling sec- and internal texturesof harmotome suggestthat the secto- arrows in Figure 1, correlatewith optical orientations of the rial twinning may originate by the samemechanism. tors shown in Figure 3. The domainsr showing white and black contrast are indicated Optical observations by lettersX and Y. The four domains correspondingto the a twin relation. A high index face Crystal facesand internal texturesare describedin terms growth hillock are in slightly inclined to (010) consists of closely-spacedkinks ofan orthorhombiccell. The {11U and {001} {101},{010}, correlate principally with the internal tex- forms in the pseudo-orthorhombicorientation correspond (Fig. 2), which to the {100}, {010}, {110}, and {001} forms in the mono- clinic orientation. Harmotome crystals from Strontian, 1 A sector corresponding to a vicinal face on a hillock is called Scotland and Korsnds, Finland were used for the study. a domain to distinguish it from the sector corresponCing to a face Both crystals were several millimeters in diameter and in the macrocrystal. 000H@x/8 5/0708-0822$02.00 822 AKIZUKI: SECTOR TWINNING IN HARMOTOME 823

Fig. 1. Differential interferencereflection photomicrograph showing growth featureon the (010)face of harmotomefrom Strontian, Scotland.Vicinal faceson the growth hillock consistof two kinds of stepsas shown by two white lines.Two summitsof growth hillocks are indicatedby arrows; the pseudo-orthorhombicsetting is used.

ture noted in Figure 4. In both areas showing black and the domains of the growth hillock, whose corresponding white contrast in Figure 4, the extinction directions are surfaceshave growth stepsparallel to eachother, show the symmetrically inclined to [0O1], although the extinction sameoptical orientation, with the optic axial planesof the direction in the areaon which kinks are extremelydense is {111} sectorsinclined symmetricallywith respectto (100) parallel to [001], suggestinga optically orthorhombic and (001),as well as to the optic axial planesof the fourling structure. domains of the hillock shown in Figure 3. Their extinction In Figures I and 2, growth featuresare different on each directions are inclined at 2" to the b-axis on the (001)sec- sideofthe line a'-a' and a"-a" respectively,and in Figure 3, tion. Thus, the eight {ll1} sectorsare triclinic, and are in a the internal texture varies at b-b. Frequently,different in- twin relation with respect to the (100), (010) and (001) ternal textures overlap in the section. Figure 2 indicates planes.In transmitted light, the {111} sectorshows only a that the area with kinks (left side)is partially mantled by few broad lamellaecorresponding to the striations on the the area with straight steps parallel to [001]. The super- surface. structure is thin comparedto the substructure,and as the The (001)face (with spiral patterns shown in Figure 5) surfaceis polished slightly, the one to one correlation be- was very slightly polished,and a thin sectionwas prepared. tweenFigures 2 atd,3 is not alwaysclear. The fine growth The section shows broad, long domains correspondingto steps parallel to (110) occur between the [001] steps or the blocks on the surface,though the pattern correspond- close to them. Corresponding [001] straight lamellae, ing to the spiral is not seenin cross-polarizedligbt. The which were observedin the (010)thin section,show extinc- contrastis too weak for a clear photograph in the horizon- tion inclined in the samedirection as that of the substruc- tal section.On a universal stage,the (001) section shows ture with the stepsparallel to (101),though the lamellaeare two kinds of domain whose optic axial planes are tilted not clear in Figure 3. Although the extinction anglesbe- symmetrically with respect to (100). For the two areas tween the [001] lamellaeand the substructureare slightly whosegrowth directionsare mutually opposed(Fig. 5), the diferent, no twin relation is seenin the lamellae. optic axial planes are symmetrical.For the area in which The four {111} sectorscorresponding to macrocrystal the steps are normal or nearly normal to the b-axis, the facesare observedat the crystal rim (Fig. 3). Two sectors section shows fine, short lamellae (about 0.2 x 0.03 mm) arelight and the other two aredark. The {111} sectorsand parallel to the b-.axis,although c6rrespondingpatterns are AKIZT]KI: SECTORTWINNING IN HARMOTOME

Fig. 2. Differential interferencereflection photomicrographshowing growth kinks and [001] striations on the (010)face of harmo- tome. not observedon the surface;the optic axial planesof the domain) during growth. The differencein extinction angles lamellae are inclined symmetricallywith respectto the b' betweenthe two domainsis 4o on the (001)section, that is' axis,suggesting triclinic symmetry. the extinction angle is 2" to the b-axis,suggesting triclinic The (001)thin section,obtained from closeto the crystal symmetry.The optical orientations and 2V values,which same section,are slightly different center,consists of {010}, {111} and {110} sectorswith were measuredon the sectors,i.e., 2V,: 85' in growth bands. Figure 6 shows the {010} sector with two betweenthe {010} and {110} kinds of domains (white and black) crossing fine growth {010} sector, and 2v,:84" in the {110} sector of the bands in the (001)section. The optic axial planes of both Strontian specimen. domains are tilted and mutually symmetricalwith respect Discussion to (010); thus the white and black domains correspondto thosein Figure 3. Figure 6 also showsthat featureson the Harmotome from Andreasberg, Hart4 Germany is (010) surface vary during growth, e.9., the white domain pseudo-orthorhombic,and the crystal structure has been marked by a star extendson the other vicinal face (black refined in the monoclinic spacegronp P2tlm (Sadanagaet AKIZUKI: SECTOR TWINNING IN HARMOTOME 825

Fig. 3. Cross-polarizedphotomicrograph of harmotome, showing the internal texture just below the crystal surfacein the (010) section.Two arrows show centersof fourlings,which correspondto the summitsof the growth hillocks shown by arrows in Fig. l. The texturesin the two rectanglescorrespond to the surfacefeatures shown in Figs. I and 2. The positionsof letters a' and a" correspondto a' and a" in Figs. I and 2. Optical vibration directionsX and Y are representedin the two kinds of domarns.

al., 196l; Rinaldi et al., 1974).However, the oblique optical by the relation betweenbarium and the framework tetra- extinction showsthat the harmotome crystalsexamined in hedra in the two-dimensionalatomic arrangementexposed this work are triclinic. Although a crystal is a three- on the side face of a growth step.Almost all growth steps dimensionalstructure, it is producedby superpositionof a on the (010)face are parallelto (101)or (101).In Figure 7, two-dimensional atomic arrangement. Although atomic both T4(1) and T4(2) coordinate to barium ions on the sites may be equivalent in a three-dimensionalstructure, sameside of the growthstep parallel to (101)in the pseudo- the sites are not always equivalent on a two-dimensional orthorhombic setting,whereas barium and tetrahedra T2(l) growth surface,because the immediatesurroundings of the and T2(2) do not exist on the same growth step. The im- sites are not always the same with respectto the growth mediatesurroundings of the tetrahedraT2 and T4 are dis- plane. Consequently,there is the potential for various de- tinctly diflerent with respect to the barium ion on the greesofatomic ordering to occur during growth. I propose growth step, though both tetrahedraare equivalentin the that the non-orthorhombic characterof harmotomeis pro- three-dimensionalorthorhombic structure. Consequently, ducedon the crystal facesduring growth. The fundamental barium ions are displacedtowards one side of the mirror framework of harmotomeis orthorhombic, but the barium plane during growth, giving rise to monoclinic symmetry.If ions have some flexibility with respectto their position in the growth steps are parallel to (1-01),the situations of the channels(Rinaldi et al., 19741.Harmotome has three tetrahedraT2 and T4 will be reversedfor barium ions on pseudo-mirrorplanes: (100), (010) and (001).Barium de- the way. The growth hillock on the (010) face consistsof viates slightly from the (100) mirror plane, reducing the four vicinal faces{1k1}, and thus the correspondingsectors symmetry from orthorhombic to monoclinic (Gottardi, arein a fourling relation. 1978).This displacementof barium is in a direction defined If the side face of a growth step is normal or parallel to 826 AKIZUKI: SECTOR TWINNING IN HARMOTOME

Fig. 4. An enlargementofthe lower rectangulararca shownin Fig. 3'

Fig. 5. Differential interferencereflection photomicrograph of harmotome, showing a spiral growth feature on the (001) face; the baxis is horizontal. AKIZUKI: SECTOR TWINNING IN HARMOTOME

Fig. 6. Crosspolarized photomicrograph ofharmotome, showingthe internal texture ofthe {010} growth sectorobserved in the (001) section;the (010)face normal to the photographis observedat the crystal rim.

the (100)pseudo-mirror plane, the barium ions will be dis- cations in some zeolites, such as analcime, stilbite, epi- ordered among the two sidesin the channel,and the "1" stilbite, yugawaralite and heulandite,bond to the frame- and "3", and "8" and "9" sites will be statistically work that belong preferentially to Al-populated equivalentin the structure,resulting in orthorhombic sym- tetrahedra. metry. Although some straight stepson the (010) lacn arc The degreeof AfSi ordering may vary becauseof differ- parallel to the (100)mirror plane, the crystal grows by the encesin growth directions. If a tetrahedron is produced fine stepsparallel to (101),which exist betweenthe straight before the large cation site on the side face of the growth stepsor close to them. Therefore,the structure is ordered step is occupied, the tetrahedron is occupied by and monoclinic. rather than aluminum. Alternatively, if the tetrahedron is The monoclinic symmetrymay be reducedto triclinic by formed after the large cation site is occupied, the tetra- AfSi ordering (Akizuki, 1981a).In many zeoliteswith an hedron will be preferentiallyoccupied by aluminum to sat- Al/Si ratio lessthan one, Al/Si ordering may be attributed isfy the local charge balance. If the two tetrahedra are to a kinetic relation between exchangeablecations and mutually equivalent with respectto the cation on the side framework oxygenson the surfaceduring growth. If the face of a step, the Al/Si arrangementwill be disordered cations are isolatedfrom direct contact with the framework betweenthe two tetrahedra. oxygensby water molecules,the AfSi structure is likely to In harmotome,the four tetrahedra(Tl,T2, T3 and T4) be disordered.Such is the casein the stelleritestructure. in coordinated directly to the exchangeablebarium ion are which the ions are completely surrounded by not equivalent on the side face of the step; thereforealu- water molecules (Gall and Alberti, 1975). Exchangeable minum occupancy may be different among them. Fur- 828 AKIZTJKI: SECTOR TWINNING IN HARMOT OME

Kalb and Klotsch (1944),can be explainedby twinning of 'oo monoclinic individuals. The present observationsconfirm | Grn!'s suggestion.The cruciform twins of harmotome are produced by penetration twinning. The fourling sectorial twinnings are a result of crystal growth on hillocks having four vicinal faces.The number of twins correlatesto the number of growth hillocks and kinks of growth steps,and no untwinnedharmotome crystals are found in nature'

Acknowledgments I am gratefulto ProfessorF. C. Hawthorneand ProfessorP. Cernf, Universityof Manitoba"Canada, and ProfessorR. Rinaldi, Universityof Modena,Italy, for critical readingand correctionsto themanuscript. References Akizuki, M. (1984)Origin of optical variationsin grossular- andraditegarnet. American Mineralogist, 69, 328-338. Fig. 7. Tetrahedra (T) and barium ions (circles) in the harmo- Akizuki, M. (1981a)Origin of optical variation in analcime. tome structure projected on (010); water molecules are omitted AmericanMineralogist, 66, 403-409. Four growth steps parallel to (101) in the pseudo-orthorhombic Akizuki, M. (1981b)Origin of optical variationin chabazite. cell and (100) pseudo-mirror planes are shown. The monoclinic Lithos,14,17-21. cell is outlined. Modified from Sadanaga et al. (1961) and Rinaldi eernf, P. (1964)The phillipsite-wellsite-harmotomesymmetry: orthorhombicor monoclinic?Neues Jahrbuch fiir Mineralogie' thermore, the T2{1) and T2(2), and T3(1) and T3(2) sites Monatschefte.129-134. may be nonequivalent with respect to aluminum occu- Deer,M. A., Howie,R. A. and Zussman,J. (1962)Rock-Forming pancy on the side faceof the growth stepsshown in Figure Minerals,Vol.4, Framework . Wiley, New York. Alberti,A. (1975)The crystalstructure of stellerite. 7, becausethe T2(l) and T3(1) tetrahedra are produced Galli, E. and Bulletinde Min6ralogie,98, 1 l-18. before barium ion is incorporated,whereas the T2(2) and Gottardi,G. (1978)Mineralogy and crystalchernistry of zeolites. T3(2) tetrahedraare producedafter. Thus, aluminum occu- In L. B. Sandand F. A. Mumpton,Eds., Natural Zeolites, pancy may be higher at T2\2\ and T3(2) than at T2(1) and Occunence, Properties,Use, p. 31-44. Pergamon Press,New T3(1), and the spacegroup will then reduce to Pl. Iitaka York. (1953)detected noncentrosymmetry in harmotome by the Hoffman, E., Donnay, G. and Donnay, J. D. H. Symmetry and piezoelectriceffect. Although no evidenceof AfSi ordering twinning of phillipsite and harmotome.American Mineralogist' has beenso far observedharmotome, the structuremay be 58,1 105. locally orderedon a very fine scale. Iitaka, Y. (1953)A techniquefor testing the piezoelectricprop€r- 6, 663-6,64.. If the growth steps (Fig. 7) are displacedin the other ties of crystals.Acta Crystallographica, Kalb, G. and Klotsch, H. (1944)Die Symmetriedes Harrrotom direction, symmetrical with respect to the (100) plane, a und Phillipsit under Berticksichtigung der Vizinalerschein- will occur among T-sites,resulting in reversalrelationship ungen.Zeitschrift ftir Kristallographie,105, 315-323. twinning. The mirror and glide planesin a crystal of higher Langeman,L. (1886)Bietrdge zur Kenntnissder Mineralien: Har- symmetrycorrespond to the twin plane in a sectorialcrys- motom, Phillipsit und Desmin. Neues Jahrbuch fiir Mineral- tal of lower symmetry. As the symmetry of an AfSi or- ogie,Geologie und Palaeontologie,2, 83-l4l' deredstructure is triclinic, the twin planesoccur parallel to Rinaldi, R., Pluth, J. J. and Smith, J. V. (1974) Zeolites of the (100),(010) and (00_1)of the pseudo-orthorhombicsetting, phillipsite family. Refinementof the crystal structuresof phil- correspondingto (201),(010) and (001) in the monoclinic lipsite and harmotome.Acta Crystallographica,B3O, 2426-2433. orientation. As the symmetryof the barium arrangementis Sadanaga,R., Marumo, F. and Tak6uchi, Y. (1961)The crystal BarAloSirrO., 'l2HzO. Acta Crystal- P2rfm, the twin planeswith rcspectto the barium ions are structure of harmotome, lographica,14, 1153-1163. (100)and (001).The (010)twinning will be due only to the Sahamq Th. G. and Lehtinen, M. (1967) Harmotome from et al. (1973) reported only the AfSi ordering, Hoffman Korsniis, Finland. Mineralogical Magazine, 36, 444-448. (100)and (001)twinning in so-called"morvenite" twin ag- Steinfink,H. (1962)The crystal sttucture of the zeolite,phillipsite. gregate(their crystal setting differs from the present set- Acta Crystallographica,15, 644-651. ting), but did not observethe (010)twinning. eernf (1964)suggested that the orthorhombic symmetry Manuscript receiaed,July 16, 19E4. of harmotome growth hillocks, which was observed by accepteilfor publication, February 12, 1985.