RENDICON11 SOCIET}, lTALlANA DI MIl"ERALOG1A E PE,TROLOGIA. 1986. Vol. 41 (2). pp. 181.192

Polysomatism and the classification of minerals

GIOVANNI FERRARISDipartimento di $cienze della Terra dcll'Universita, Via S. Massimo 24, 10123 Torino

MARCELLO MELLINI *C.N.R., e.s. Geologia Slrutturak e DinamiC1l dell'Appennioo, Via S. Maria '3, ':6100 Piu

STEFANO MERLINODipanimento di Scierw: della Tern dell'Uni\oersitl, Via S. Mana ':3, ,:(,100 Piu

ABSTRACT. - The C'OOCC'Pts of polysoJIllltism andpolysomatk series take into account ocllh (hemistl)'and structure of minerals. Therefore they $('Cm wdlworth ronsidering for the purpose of I:'mphll$izing,for instano:e, ~tageneses and solid·statl:' transfor·madons of the minerals. The following new tum·pIes of polysomalic series are given.

S~,it!s oj tht! inophitt! . The S..X series is basedon S (scrpenline) and X [MgoSi.o.l:OH),.·H,o]modules. Known memben arc serpentine, C1Irlo­sturanite and defective stru(1U1l':S obsC'rved as la·mellae inlergrown within carlooturanite itself. Theseries acrounts for minerals which can be definedas H.O-rkh and Si·poor serpentine. The stabilityof the manlx:rs is discussed.

Smes oj the IIntiforilt - The S,:,S;,.T seriesil based on S (serpentine) and T (tak) modules(+ and·- indiOlle the opposire po1arities of thephyllosilicale la}>er). PoI)'SOfflC'S of the series .reserpentine itself .nd the amigorite minenls withdifferent 11 plIrameren and Si/Mg rarios.

St,i~s oj Iht sebiljilrzikitt - Schafarzikite, veniliai·te, apuanire and Ihdr def«rive stru(:\ures ate shownro be member of a polysomatic series Sc..U. basedon Se (schafarzikite) and U ("Fe:·'·Fe:' "'SM­0,. S.] modules. Intergtowth of different polysomesexplains the non stoichiometry revealed by someanalyses.

Serits of tM 6 X 9 A slrueturts - The seriesA..B. is based on A Oawsonite) .nd B (AISiO.oH)modules. By roupling modules, also throughtranslations and rotations, seven so alled 6 X 9 AStlUCtull':S are obtained: sunassite, mad.llite, pum­pellyite, julgoldite, ardennite (twO types) andlawsonite.

Kty words: poly!Ol1l3tism, dassi6cation of mi­ncra.ls, ioophite, antigorite, schafarzikite.

(*) Now at: Dipartimento di Scienze della Tern,Univenitl di Perugi., Piazu Univenitl, 06100Perugia (lu,ly).

Ru.ssur-"TO. _ I rooceui di poIi50marismo t dlserie poli50m.rica rengono CUIlIO ddk caratreri·stidle chimkhc e strutrurali dei minerali e vrngonoritenuti urili per la d.ssificazione di questi quandoli si voglia raggrupp.re al fine di evidenz.i.me,per csempio, relazioni patagenetidle e trasformazioniallo stato solido. Si riporrano i seguenti nuoviesempi di serie polisomatkhe.

Suie ddl'inofiu - La serie S..X si rostruisceaccoppi.ndo moduli S (serpentioo) e X [MgoSi.o.(OH),•. H,()] e oomprende il serpentioo, la c:arla­sruranite e struUure difettive osservlte come lamellenella c:arlosturanite stessa. La se·tie rendC' como diminerali definibili come serpentini ricchi in 11,() epoveri in Si. Viene discussa 11 stabiIiti dei poIi.soma ooti.

Suit tltll'ffflliforitt La serie S.:.S;.·T si 00­

snuisce acroppiando moduli S (serpentino) e T(talco) (+ e - indicaoo le polaritl opposte megli strati fillosiliatici possono IS5Umc:re). Tale seneromprende il serpentino e I'anrigorite pettSata comeun insieme di termini caranerizzati <la peculiarivalori dd parametro a I:' dd rappotto Si/Mg.

Suit! dd/il sehajllrtikiu - Si mootla rome schafat·zikite, vetsiliaile, apuanite e le loro strutture difet·tive siano i membri di una serie polisomaticaSC..U. rostituit. da moduli Se (schafarzikite) I:'moduli U ["Fe:- '"Fe:' '''Sb:' 0,. So]. Le ooncre-­scite spiegaoo la oon st«hiomerri. tTOY.1lI in al·cune .n.lisi.

Strie tltlle struUurt 6 X 9 A • Questa serieA.a. l: basata sulla a.xnbinuiooe di moduli A(lawsonite) e B (AlSiChOH). Combinando taU m0­

duli, anche tramite tasl.azioni e l'OIu~i, si lfIOSuacbe • tale serie poiisom.atica appllrtmgono sursassite,macfaIIite, pumpellyite, julgoldite, udennitl:' (d~

tipi strutturali) e lawsonite.

Pil,oit ehiiltJe: polisomatismo, dusiflCll.zione <...·iminer.li, inofite, amigorite, schaf.rzikite.

182 G. FERRARlS, M. MELLINI, S. MERLlNQ

Introduction

The purpose of any classification is 10

establish a simple although comprehensivescheme, in order to collect, compare anddistinguish individual objects and to predictunknown phenomena to some extent. Sucha scheme must rely on few basic featuresof the objects to be classified. Whereas peopleusually agr~ on the aim of a classification,disagreement often arises just where basicfeatures are concerned. In fact, individualaptitudes and immediate purpose largelydetermine the particular choice. For instance,different approaches such as X-ray diffractionor optical microsoopy or field experienceoften lead to quite different classificationof minerals. This is understandable and,maybe, fruitful as stated by LIEBAU (1985),who maintains that <le the best classificationwhich can be chosen is the one that is bestable to serve the particular purpose underconsideration _.

&th very broad and very specific clas­sification schemes have been proposed forminerals. Some authors (e.g., STRUNZ, 1938;ZoLTAI, 1960j POVARENNYKH, 1972; LtMADE FARIA, 1983; LIEBAu, 1985) preferred tocover very large topics and proposed quitebroad classification schemes. Although theimportance of these effons can hardly beoveremphasized, these comprehensive sche­mes may Ixrome partially inadequate, orcumbersome, when small arrays are to bedealt with. In the latter case a decreasinggenerality could be hopefully balanced by adeeper insight into the relevant phenomena.In fact, very specific classification schemeshave been proposed for well defined classesof compounds (e.g., MEIER (1968) forzeolites; GoTTARDI and GALL1 (1985) fornatural zeolitesj SMITH and RINALDI (1962)for framework silicates based on four- andeight-membered rings].

Most of (he quoted approaches assumethe chemical nature of the minerals justas a boundary condition, useful to definewhich objects are to be considered. In otherwords, structural crystallography may do­minate over crystal chemistry, as chemistrydoes not directly enter within the set ofbasic rules. As a result, almost geometricschemes may be produced (LIMA DE FARIA,1983).

The concepts of polysomatism and poly­somatic series (THoMPsoN, 1978) simul­taneously take into account both structureand chemistry. Through these concepts,fearures such as solid-state transformationsand paragenetic assemblages can quitestraightforwardly be included within theclassification scheme. Polysomatism, there­fore, seems to us well worth consideringwhen we are interested not only in theclassification, but also in the behaviour ofminerals, even though it suffers from lackof generality, from its rather specializednature and limitations in the number ofclassified objects.

A few examples of polysomatic series willbe shown in this paper, just to stress howthe approach works. The list is by no meansintended to be exhaustive, as the examplesonly refer to our recent work.

PoIY80mC8 and polY80malie series

The basic ideas that underlie the poly­somatic theory may be found scatteredthroughout a literature dating back to whenpeople started recognizing the occurrence ofcommon structural and chemical modulesthat, differently combined, produce a wholefamily of closely related structures (e.g.,MAGNELI, 1953; WARSHAW and Roy, 1961;KOHN and ECKART, 1965; ROTH andWADSLEY, 1965). THoMPsoN (1970 and1978), eventually, put forward a com·prehensive formulation of the theory,exploited its implications in terms of com­positional space, chemographic relationshipsand possible paragenetic assemblages, anddiscussed several examples of polysomaticseries.

The basic building units of a polysomaticseries are one-dimensional beam modules, ortwo-dimensional layer modules, whose struc­ture and chemistry are usually idealized.By the combination of two. or perhaps more,modules, in different ratios and/or withdifferent combination rules, several structuresare produced which define the members ofthe polysomatic series. These series are col­linear in composition, namely the idealizedcompositions of the members linearly rangebetween the two cnd-member compositions.Polysomes (from the Greek = several bodies)

POLYSOMATlSM AND THE CLASSIFICATION OF MINERALS 183

are akin to polytypes, in that some kind ofmixed-layer polytypism can be found wilhinthem. A funher improvement was attainedon introducing exchange vectors (THOMPSON,1981) which afford an efficient, and easilymastered, approach to isomorphic substitu.tions making allowance for compositionsother than the end members. Typicalexamples of polysomatic series are bio­pyriboles (THOMPSON, 1978, 1981; VEBLEN,1981), humites (THOMPSON, 1978), serpen­tjne and chlorite (THoMPsoN, 1978),pyroxenoids (KOTO et al., 1976).

The inophite I)olysomatic 8e,.ie8

Crystal-chemistry

Inophite is the name of a polysomaticsenes which includes the serpentine-likemineral carlosturanite Mg~ISiI202s(OH~4'

H~ and the related structures which occur

•....

re_ In faet. the structure of carlosturanite(fig. 1) just consists of the same octahedralsheet which occurs in 1; 1 uioctahedralsilicates, with modifications in the tetrahedralsheet; rows of silicon vacancies in thetetrahedral centres, compensated by fourhydrogen atoms, split the tetrahedral sheetinto parallel hydrogen bonded strips. Thesilicate strip consists of triple chains whichare generated by connection of four-repeatsingle-crankshaft chains. Depending on theabundancy and distribution of tetrahedralvacancies, many structures are possible, withdifferent chain.multiplicity and chemistry (inparticular, they have inversely correlatedsilicon and water contents).

These structures form a polysomatic series,as all of them are made of two chemicallydistinct modules, S and X in fig. 1. The Smodule is a (lOO) section of the flat-layerserpentine structure with Mg3Si:tO~OH)4

composition. The X module corresponds to

•I

b_

Fi~. J. - ,Sm,Jclure of carlosturnnitcMEI.LlNt Cl al. (1985)].

(serpentine) and X (sce text) [ After

as fault lamellae within carlosturanite itself(CoMPAGNONI et al., 1985; MELLlNI et aI.,1985), This has been the first polysomaticseries to be: vOted and accepted as such bythe International Mineralogical Association(IMA).

Carlosturanite is an asbestiform silicate,strongly reminding chrysotile in aspect andproperties. The similarity of properties isdue to the close structural and chemicalrelationships existing between carlosturaniteand the idealized flat-layer serpentine structu-

an hypothetical hydro-silicate which can bederived from serpentine by substitution of[(OH).·H20]~- for [ShOd~- and hasMg.si~a(OH)H - H~ composition. Thecomplete series would be X, SX, ... , S~(carlosturanite itself), ...• S (serpentine);namely the different polysomes are structu­rally and chemically intermediate betweenthe hypothetical hydro-sorosilicate X and thelayer silicate serpentine S. The ideal chemicalformula of the general Sn,X polysome issimply [Mg3SbO~(OH)4 ]n,·MgeSi203(OH)14·

IS; G. fERRARIS, M. MELlINI, S. MERLlNQ

be assessed. Carlosturanite IS likely to bea real phase indeed, namely a stable as­semblage, within the MgO-SiO:!-H:=O system,as indicated by its abundance in the holotypelocality and by its widespread occurrence.In particular, textural analysis at the opticaland electron microscopy scale, indicateddiopside + chrysotile + carlosturanite asa stable paragcnctic assemblage. On thisbasis, CoMPACNONl et al. (1985) suggestedPH,O"" 2 khar and T".. 250-JOO~C as likelycondirions for the formation of carlosturanilein the light of the available data on theserpe:ntinite metamorphism (TRoMMsDORFF,1982). Eventually, C1rlosturanite is replacedby brucite and chrysotilc, according to thereaction:

Mg~ISil:!0:!8(OHl:t~ 0 H~O --0

carlosturanite

H:O. As regards t~ unit ttUS, the band cparameters are common to serpenline.whereas a is [(m + 2) /2) a. (0. = a ser·pentine) and the cell is primitive when mis even. Viceversa. a = (m + 2) Q. and theceU is c.centcrcd when m is odd; for carlo­sturanite (m = 5) a = 36.70, b = 9.41,c = 7,291 A, P= IOU',

Several faults actually affect the Q perio­dicity of carlosturarute and can be inter·prered as chain multiplicity faults. Usuallythey are isolated lamelJae, continuous within(010) and scattered rhroughout the carlo­sturanite matrix. Only other three periocli.cities besides the 18 A onc (al2 of carlo­sturanite) have been observed in the latticeimages. They are the 16 A- and the 21 A­perioclicities, and the 24 A-half-periodicityof the S~X, SuX and S1X polysomes, respecti­vely. Among them, saX and SlX are com­mon, whereas S~X is rare. 6Mg,S;,OdOH), +

ChrysolilcJMg(OHh + JH,O,

brucite

di

,-- -----....b

s "-~,_--_-..:----6 ..~X-:;:.-::..--

-~

Fig 2. _ ChemIcal relationships among X, SX, ... , S .polysomes of the inophite polysomalic serieswllhm tne Mg·Si0.-l-LO sysleffi (di == diopside, b == bruclle).

Ph4U rel4Jionships

The .whole compositional range of theinophite polysomatic series defines a tie­line which joins the serpentine compositionto the X hydro-sorosilicate, within theMgO-SiO:!-H:!O system (fig. 2). Most pro­bably, carlosturanite is an important phasefor metamorphosed maphics and ultra·maphics. In fact, after the first descriptionof the mineral, new findings were re·ported, from Taberg, Sweden (MELLlNIand ZUSSMAN, 1986) and from sevet1lllocalities within the Lanzo Massif, WesternAlps, si:rpentinites (unpublished data). Onthe other hand, no structure other thanSr,X (carlosturanite) has yet been found asdiscrete crystals, but only as faults. On thebasis of these observations the stability, orinstability, of the different polysomes can

Polysomes other thAn S~X are probablymetastable, formed as growth defects, underpoT conditions not far from the C1rlosturanitestability field. Quite interesting, the corn·positions of the observed faults just dusteraround the carlosturanitc composition (fig. 2),and two of them are less hydrated structures.A partiAlly similar behaviour has been repor·ted for other polysomatic series too, such asbiopyribotes. In this series, no discrete crystalother than jimthompsonite or chesterite wasfound, whereas many different periodic oraperiodic structures were observed as me·tastable phases by HRTEM (VEBLEN andBus ECK, 1979).

The antigorite polY80matic series

Chemical analyses show (WHITTAKER andWICKS, 1970; WICKS, 1979; WICKS llnd

POLYSOMATISM AND THE CLASSifiCATION OF MINERALS 185

PLANT, 1979) that antigorite has a Sicontent definitely higher than that requiredby the ideal fonnula M&3Si::O~OH)~. Anti­gorite is based on the fundamental cellof serpentine except for the tl parameterwhich, as for inophile, is a variable « multi­ple .. of tl. - 5.3 A. KUNZE (1961 andprevious papers there quoted), on the basisof an experimental two-dimensional electrondensity map, proposed. a structural modelfor antigorite with tl = 43.3 A, the mostfrequent value of this parameter. The modelis characterized by a [100J sinusoidal mo­dulation of the serpentine layer in such away that tetrahedra belonging to adjacenthalf waves point in opposite directions, asshown in the highly schematic fig. 3. Whilstin crysotile the rolling of the layer compres­ses the octahedral sheet, in antigorite thefolding (not shown in fig. 3) is opposite andcompression is higher for the already smallertetrahedral sheet. This sheet can thereforeaccomodate an extra [010J row of disilicategroups in each wave, in such a way thatone out of two inversion sites has anoctahedral row sandwiched between letra-

inversion lines, SPINNLER (1985) describesanrigorite in terms of a polysomalic seriesbased on three modules. To avoid too slricta dependence upon a poorly known slrUcturemodel, we propose a two-module polysomaticseries to describe antigorite. Our modulesare S, with the serpentine composition asfor inophite, and T with the talc compositionM~i.OloCOHh to take into account theextra tetrahedra. These modules are (100)slabs one octahedron thick, i.e. V3/2 theoctahedral edge. Noting with S· and S- themodules with an opposite polarity of thetetrahedral sheet, the whole series can bewritten S';'5;;;. T or [Mg;ISi10~(OHNJ...+...·[Mg;ISi~OlO(OHhJ. According to HRTEMimages (YADA, 1979) and KUNZE'S (1961)model, the condition m = m' is not necessary;i.e., the tWO half waves can be of differentlength. In each wave Ihe number of [010Joctahedral rows per cen is m + m'+ 1 andthe Si/Mg ratio is given by [2(m+m'+2)]/[3(m+m'+l)], i.e. Si/Mg > 2/3. By com­parison, the moth inophile member has aSi/Mg ratio given by [2(m +1)JI(J(m +2)],Le. Si/Mg < 2/3; in both cases when the

Fig. J. - Schematic repT('$('ntation of Ihe antigorile slructufe (a::: ofJ.J A mc:mbcr) slittd into S ($Cf­pcntine) and T (talc) modules.

hedra as in laic (T module in fig. 3). Detailsof the structure, particularly at the waveinversions, are slill controversial (UEHARAand SmRozu, 1985), but the number andthe posilion of the extra tetrahedra ac­counting for the excess Si seem beyonddispute. According to this model, thevariability of the tl parameter can be con­nected wilh differenl lengths of the strucluralwave and requires a variable Si excess whichdoes nOI contrast the available chemical data(WHITTAKER and WICKS, 1970).

In order to lake into account the con­clusions of KUNZE (1961) concerning theconnections of the tetrahedrn a.t the wave

number m of S modules tends to be infinitethe ideal serpentine composition is obtainedwith Si/Mg = 2/3.

In the sample studied by KUNZE (1961)there are 16 octahedral rows across a andthe average width of each row is 43.3/16 =2.71 = 3.14 v3/2 A, which correspondsto the value expected for an ideal brudlelayer (tlb = edge of the oclahedron =3.14 A). According 10 the recenl study byUEHARA and SHtROZU (1985) the differentvalues measured for the tl parameler ofantigorite are multiples of 2.7 = tlbV3/2 A.more than of tl./2 = 2.6 A. Members withm +m' odd and even thow P and C monocli.

186 G. FERRARIS, M. MELLlNl, S. MERUNO

41

cl

TABLE 1

CrySllIl<MMUIII dtlttl /0' minmlls0/ llx scbs/llrlikitt po/YJomtllic 1"i('l

Fig. 4. _ Schafarzikite (4), veniliaite (b) andapuanile (cl structures shown as memben of apolysomalk series generated by Se (schafarzikite)and U (sec: lex!) modules.

_".,u,.. "":' "''';-',. ..~ ..~ .._, "'-.....m •• to "":' ",.:, ",,:' 1Il..~....., .~ ..~ ",m --," 'I••:' " ..;- ",.j "'''~~..l. ... I .•' 1/.... -.-parent structure of the whole group (MEL­LlNI el al., 1979). Its cryslal structure (6g.4 a) is ba~ upon chains of corner-sharingSh" -0 pyramids which are connected withchains of edge·sharing Fe:'06 octahedra(ZEMANN, 1951; FISCHER and PERTLIK,1975). The crystal struclure of versiliaite isformally derived from schafarzikite (fig. 4 b),by substituting every fourth Sb:l~ cation inthe pyramidal chain by a Fe3~ cation, and

I•

Se

S.

The 8Chafarzikite polYllomatic seriell

Schafarzikite, versiliaite and apuanhedefine a group of closely related minerals.Whereas schafarzikitc is strictly an iron andantimony oxide, versiliaite and apuanite aremore correctly defined as iron and antimonyoxysul6des. Selected data for t~ mineralsare given in table 1.

Schafarzikitc was considered to be the

h)

nic cell, respectivdy; the value of tk a para­meter (Urns OUI to be (m +m'+ 1) 2.7 Aand 2(m+m'+l)2.7 A in the two C3M:S,

respectively. Experimental a values nOI

integer multiple of 2.7 A are interpretedas average values from a mixture.

POLYSOMATISM AND THE CLASSIFICATION OF MINERALS 187

Fig. 5. - Connection of pyramidal chains throughsulfur bridges in the schafarzikite polysomatic series:(a) single chain in schafarzikite; (b) double ribbonsin versiliaite; (c) infinite layer in apuanite.

introducing a sulfide anion between pairs ofFe3

+ cations belonging to two adjacent chains(MELLINl and MERLINO, 1979). The excessnegative charges introduced with the sulfideanions are balanced by oxydation of twooctahedral Fe2 ' to Fe:!+ per each sulfideanion.

Similarly, the crystal structure of apua­nite (fig. 4 c) is formally derived assumingthe same substitution every third Sb3 +. Amajor structural difference exists betweenthese two derivative structures. Whereas thepyramidal chains are connected by sulfurbridges to build up double-chain ribbons inversiliaite, they form infinite sheets in thecase of apuanite. A scheme for the differentconnections is given in fig. 5.

Although MELLINl and MERLINO (1979)actually recognized the modular structure ofthese minerals, and described versiliaite asbuilt up by two layers with v'Fe~""Sb~+OI6

and "Fe~' "Fd+ "'Sb~+ 0 16 S2 composi­tions respectively, the polysomatic nature ofthe whole group was not completely realizedat that time. The systematic description ofthe group was instead performed mainly interms of stuffed and substitutional derivativestructures.

Adopting now the point of view of poly­somatism, two fundamental layer modulescan be recognized within the diffetentstructutes. and they are suited to generatethe whole family. These modules are sketchedin 6gs. 4 and consist of (001) structural slabs.The first module, Se hereafter, is just a(001) slab of the unmodified schafarzikitestructure. The slab is one octahedron thickalong [001], namely it has e/2 thickness.As shown below, the Se module is to be

u

S

Fe

~Sb

~ ~!2!!!l!.,1, "., "l, ,(',

~=

.......".. "':'IAl ,1"" ,1''','''',.0,' I ,. ,. ,. ,"'." ••~,,", .. ",1'"~ ,''',,(,,'.',, ',.0, 11 '..., ,. .,,, "'.,. ','-"" .. ",I" ,1"'"I''',JI'',.o,11 '.Il ... "." "... ••...,..",.. ",I"~I ..),l"'.II",.o,IJ ..." ,."" ,..... ",.. ."

-" "':-"'.(.. ',1....".".',''',.0,,)I •. Il '.Il '..., -l_'" "''',''''',1'',0,').... ,. ,. ".to -

Fig. 6. _ Chemical relationships among U, (SeU).(apuanite), Se.U (versiliaite), Sc.U and Se. (.$(;hafar­zikite) polysomes of the schafanikile polysomaticseries within the Fe-Sb-S srstem.

TABLE 2Crysta/.chemical data lor minerals

of the 6 X 9 A structures polysomatic series

axis normal to the layer module, and a andb are glide planes whose poles lie withinthe layer module. The second module, U, istwo octahedra thick and contains all themodifications that ptoduce the derivativestructures: sulfi~e ion insertion, substitutionof "'Fe:l ' for '''Sbs+ and of v'Pe'+ for"Fe~·. It is just the same U layer alreadydescribed for versiliaite, and has 2Pi layergroup symmetry. As a consequence, the

preferred over the two octahedra thick Slayer which was chosen by MELLINI andMERLINO (1979), as this latter unit wouldfail to reproduce tetragonal structures suchas apuanite. The composition of the Se mo­dule is obviously "Pe~+ '''Sb~' Os, namelyschafarzikite itself is the first end-memberof the polysomatic series. Its symmetryis described by the layer group 2Pba(THOMPSON, 1978), where 2 means a diad

J'"''1>101

cba

188 G. FERltAltIS, M. MEI.L1NI, 'S. MEItL/NO

:-.

a) hi

. .-A

A

Fig. 7. - Sursassite (a), pumpeUyite (bl, arden·nilC (c, d) and lawsonite (to) structures shown asmembers of a polysomatic series generated by A(lawsonite) and B (AISio.OH) modules. Exponentsrand t indicate rotation and translation, respective­ly, of the modules (sre text). J)

A

B

A'

B

A

POLYSO....1ATlSM AND THE CLASSIFICATION OF MINEIlALS 189

second end·member is the SO called • 6 Aunknown orthorhombic structure" (MEL­UNl and MEIl.L1NO, 1979), namely it cor·responds ro the structure with the highestpredictable sulfur content. With thesechoices. ,he different structures becomeSchafarzikhe Se::Versiliaitc SC::UApuanire ScUScU =(ScUhUnknown 6 A ortho-

rhombic structure UUnknown 18 A ortho­

rhombic structure(MELLINI et aI., 1981) SC..U

•-'A

a

A'

at

A

Fig. 7 e)

The ideal unit cell content of the ge­neral SC..U. member can be expressed as«'Fei" "'Sb:~ 08)", + (·'Fe~~ "'Fe~'

'''Sb~- 0 .. $.!). and the c periodiciry is.... (3 m + 6 n) A. Orthorhombic Pbamstructures occur when the polysomatic for­mula contains an even number of Se layersbetween subsequeOl U layers. Otherwise,tetragonal P4dmbc struCtures are obtained.

The chemical relationships among the diE­fereOl ideal polysomes are depicted in fig. 6.On considering the actual chemical data,

they plot along the tie-line, but are shiftedfrom the expected positions. For instance,the chemical analyses for apuanite are foundbetween the SC:!U and the (ScUh points.Similarly, versili:lite plots between SC~U,

that is versiliaitc itself, and Sc~, that isschafarzikite. Namely, the crystal has bulkchemical properties intermediate betweentwo adjacent ideal compositions. The expla­nation for this non-stoichiometric behaviourwas given by MELLlNI et a1. (I 981), whofound intermixed occurrence of differentstructures.

The 6 X 9 A structures

This group includes a number of structures,so called because of the approximate valuesof their axial translations (MooRE et al.,1985). Members of the group are lawso­nite, sursassite (and the isostructuraI mac­faIlite), pumpeIlyite (and the isostrocturaljulgoldite), ardennite, orientite, ruizite, santa­fdte, bermanite. According to MooRE et al.(1985), they can be described in terms ofcommon ~[M~~TO"hJ sheets occurringin all of them. Different intusheet marerialis sandwiched between sheets and gives riseto different tetrahedral polymerization. Foristance, Si~OT gtOUps occur in lawsonite,SiO.. and SbOT groups in sursassite andpumpellyite, SiO. and SisOIO groups inardennite and orientite, Si.Olo groups inroizite. A few of these structures (table 2)will now be considered and their polysomaticnature will be stressed. Furthermore, theexamples will show that either slightly dif­ferent or highly different structures can beproduced starting from equal amounts ofthe same, alternating fondamenraI layer­modules, just by different choice of stackingvectors.

Let us refer ro sursassite as the startingmember for the whole description. Itscrystal structure (MELLlNl et al., 1984)consists of edge-sharing octahedral chains,cross-linked by corner-sharing SiO. andSi~01 tetrahedral groups (fig. 7 a). FollowingMooll.E et al. (1985), the fundamental6 X 9 A sheet, named A in fig. 7 aJ canbe easily recognized. Two adjacent A mo­dules sandwich the B module of fig. 7 a,havin~ Ma'SiO~H composition. For the

190 G. FEIlllAR1S, M. MELLlNI, S. MER LINO

sake of generality, we extend now the Amodule to embed also the out-of-the-walllarge cations, such as calcium or divalentmanganese, and water molecules (Xl. Itschemical composition becomes X:lM~OHr.

(SjO~):. Therefore, sursassite and isostructu·ral madallite can be easily described asconsisting of alternating A and B modules{6.6 and 2.7 A thick, respectively}, AB beingtheir polysomatic formula.

Pumpellyitc and juga/dire are polyrypicvariants of the sursassite structural type. Themost important difference is that, in the firstcase, similar tetrahedral groups face eachot~r on the two sides of ~ octahedralchains (6g. 7 b), and, in the second case,different groups face each other (6g. 7 al.The difference is due to the different stackingvectors of the same fundamental modules.In particular, whereas each A layer of sursas·site directly overlaps a corresponding A layerin pumpelIyife, no direct overlap occurs forthe B layers. In fact, every second B layerof pumpellyite is shifted by 1/2 (7+ 1J)with respect to sursassite. The polysomaticformulae can be expressed as AB for sursas·site, and as ABAB' for pumpellyite, with tindicating the 1/2 (~+1t) vector. Similarly,more complex polyrypes might be anal~;

for instance, the ... t. t p • _. 36 A polytypereported by MELLlNI et al. (1984) wouldbecome ABABAE'AB'.

The polysomatic analysis can be extendedto other structures which do not otherwisereveal their close chemical and structuralrelationships with sursassite and pumpellyi­te. For instance. ardennite. Mn~' (Ale(OH)e(As04XSi04):(SiJOlo)], exhibits major struc·tural modi6cations, as Si04 and Si~,o

groups are now presenl. However, an ap­propriate slicing of its structure immediatelyreveals that ardennite too consists of alter­nating A and B modules, in equal amounls.and is a polymorphic modification of sursas·site and pumpellyite (6g. 7 C). Whereasadjace:nt A layers are identical in sursassiteand pumpelJyite, they are correlated by a[100] .twofold axis in ardennite. By speci­fying with the symbols A and Ar this re­lationship, the polysomatic formula for arden­nite comes out to be ABArB. A furtherpossibility is the introduction of BI modulesas those occurring in pumpellyite. Actually,they are possible, and the polysome AEA'B'

(6g. 7 J) just corresponds to a polysomaticvariant for ardennite which was first as~umed

by MELLlNI and MER LINO (1982) as pos­sible structure for orientitc, and then actual­ly found to occur as fault structure in otien­tite cryslals by MELLINI et al. (1986).

The end-members of this A.B. poly­somatic series would have A and B com­positions, respectively. As regards the A end­member, it actually exists and corresponds tolawsonite, Ca[AliOHb(SbOl)]' H20 (fig.7 e). Its polysomatic formula is AA", and,quite interesting, such a structure is alsopresenl as fault structure within the sursas­site crystals (MELLINI et al., 1984). PossibleB structures, which would have AISiO:i)Hcomposition, do not seem to be known,instead. Obviously, other A,nBn polysomesmight exist and their chemical compositionand dool spacing would be obtained by theappropriate linear combination of A and Bmodules.

Conclusions

The fundamental question must now befaced of whether and where the approachis suitable.

A first, important application is connectedwith structural crystallography and topologyof modular structures. For instance, in thecase of ardennite and sursassite-pumpellyite,structures which would appear quite dif.ferent when analyzed in terms of thetetrahedral polymerizations. immediatelyappe9r to be polymorphs and built up bythe same fundamental modules.

A second field of application is due tothose cases where the full understanding ofthe chemical data is quite important. Thispossibility comes out from the fact thatpolysomes just embed within their definitionthe required chemical information. Anexample is the non-sloichiometric behaviourwithin the schafarzikite group.

Bearing in mind the inophite polysomaticseries as a third example, a third importantreason can be found in the study of solidstate transformations, growth defects, sub­solidus phenomena, parageneses and mineralequilibria, namely in mineral reactivity. Infact mineral reactivity strongly depends on

POLYSOMATlSM AND THE CLASSIFICATION OF MINERALS 191

chemistry as well as on Slructure, and boththese features are well taken into accountby the use of the polysomat.ism theory.

As regards the idcnti6cation of basicmodules, it s~ms to us that an effou isrequired to identify the most signi6cant ones,from the point of view of their physicalsoundness and possible crystalchemical use.This means that, based upon an appropriatechoice, more and more structures may byfound to be based upon these modules, justas these fundamental building units cor­respond to the most efficient ways of linkingtogether individual atoms and representenergetically favourable pathways towardscrystal growth.

REFERENCES

CoMPAGNOSI R., FUIlAIIS G., MELLlNI M. (198" •CllnOIturllniu. 11 ntw IIsbtsti/orm rode·/ormingsilitllu from Volt VlITlli/II, /till,. Amer. Minef1ll.,70, 767-772.

FlsQ-lEl. R., PEITLlK F. (197') . Vtr/tintru1f1. dtrKris/lI1s/rulttur dtr Sthtl/tlrVltitt, FtSh,o,. Tscher.Mineral. Petrogc. Min., 22, 2)6.24l.

Gon.....DI G., GALLI E. (198') • Nil/urll1 ZrolitnSpringer-Verlas, Berlin.

KOHN ].A., Ec"An D.W. (196') . Mixtd.lllyerpoly/yptr rtlllud /0 mllgnt/oplumbitt. Amer. Mi­neral., 50, IJ71·IJ80.

KOTO K., MOIlIMOTO N., NUITA H. (1976) - Crys/lll­lographic rtlll/ionlhips 0/ th< pyroxtntr andpyroxtnoids. J. Japan Assoc. Minet:ll. Petrol.Econ. Geol., 71, 248-254.

KUNZE G. (1961) - An/igori/ S/ruleturtlNorttiscMGrundlltl.tn und ihrt prillt/isth< &dtutulll. furdit wtittrt Strptn/int·Fo,schlllll.. Fonschr. Mi·ner., J9, 2Q6.J24.

LIEBAU F. (198') - S/,uc/u,1I1 dxmirtry 0/ li!ictI/tl.Springer.Yerlas, Berlin.

LIMA DE FAIlIA J. (t98J) • A p,oposal /0' il

s/ruCfu,lIl dllssi/iclI/ion 0/ mine,als. Garda deOrta, SCrie rle Geologia, 6, \-14.

MAGNELl A. (19'3) - Structures 0/ the ReOr/ypewith ,ecu,rent dislocations 0/ atoms: c Homolf>­gous. u,its of molybdenum and /"ngIun oxides.Acta Cryslallogr., 6, 49'·500.

MEtEl. W..M. (1968) - ZtOiite s/1UCtu,tl_ In c M~lecolar Sieves., Society of Q1emiQlI Indumy,London, pp. 1G-27.

MELLlNt M., AMOUI.IC M., BAllo:.'NET A., Mu·CUI.IOT G. (1981) - Mit:fOlt,uC/U,tl IInd non­stoichiometry in scbtJ/ilni!eiu.filet mintrllls. Amer.Mineral., 66, 107J.I079.

MELLlNt M., FEUARlS G., CoM'AGNONI R. (198') .Car/osturaniu: HRTEM tvidtnct 0/ a polYl0ma/icsnitl including urptn/int. Amer. Mineral., 70,77J·i61.

1I.ULLlNt M., MULlSO S. (1979) Vt,si!iolitt IIndtlpuani/t: dtrivil/;f}t st,w::tUftS ,dll1td to SCbtJ/flf­zileiu. Amer. Minef1l!., 64, 123}.1242.

MELLlNI M., MULlNO S. (1982) - O,dtf..Jisordtrin lUfIIISSi/t. Absrracu of Papers, l}lh Genef1l!Meeting, IMA '82 (Varaa, Bulgaria), p. J7J.

MELLlNI M., MULINO S., OIlUNDI P. (l9i9) .Versiliai/t ilnd apuani/e, two ntw mintrills /,omthe Apulln Alps, Itllly. Amer. Minera1., 64, 12J(}.12J4.

MELLlSI M., MERLlNO S., PASERO M. (1984) ­X.rIlY and HRTEM study 0/ SU'SilSSitt: Cf']stlll

s/,uc/U,t. stlUleing diso,dtr, IIn~ su,solssi/t·pum­pdlyilt inttrgroWlh. Phys. Cllau. Mintta.ls, 10,99-10'.

t.lELLlNI M., Maw;() S., PASEIlO M. (1986) .X-ray Itnd HRTEM s/,uctu,e IInll1,sis of o,itll­lilt. AmeT. Minel'll.l., 71, 176-187.

MELLINI M., ZUSSM"N J. (1986) - CaNosturanilt(llO/ c p;cro/ite.) from Tllherg, Swedtn. Mineral.Mag., in press.

MOORE P.B., SHEN J., ARAKI T. (1985) - C"s/illchemistry 0/ /ht C:[M,"(J).(T01).} shut: st,uc/u­,Ill prindples and c,ystlll slrw;:/urt 0/ ,uhilt,mru/atliu and oritn/iu. Ama. Minetll.l., 70,17I·18l.

PoV....ENNYtai AS. (l972) • Crys/tJi CMmicll1 diu·si~a/ion 0/ mintrll1s. Plenum Pl'CS5, New York.

Rom RS., WADSLEY AD. (196') - Mul/iplt plHut/ormll/ion ill th< binil'7 slsUm Nb.DrWO.. IV.Th< blocle prindple. Acta Crystallogr., 19,42-47.

S~nTH J.V., RINALDI F. (1962) - F,ilmtworle !/ruc­/urts formed f,om paralltl Jour· and tight.mem­be,ed ,ings. Mineral. Mll8., JJ, 202-212.

SPINNLER G.E. (198') - HRTEM study 0/ an/i·/l,O,;/t pyroxtne·Jt,ptntint rtllctions, and chlo,ite.Ph. D. Thesis, Arizona State University.

STIlUNZ H. (19J8) • Systtmatile und St,ule/u, dt,Sili/eatt. Z. Kcistallogr., 98, 6O-8J.

THOMPSOH J.B. Jr. (1970)· Geonrt/ru:1I1 poJSibili/its/0' tlmpbibolt s/,uctu,es: modtl biopy,iboltl.AIDer. MincnJ., ", 292-Z9J.

TIiOMPSON jB. Jr. (1978) • Biopy,ibolts IInd poly­somlltic str;ts. Amer. Mineral., 6J, 2J9-249.

T"OMPSON J.B. Jr. (1981) • An in/,oduction to themintfillou and ptt,o/OU 0/ the biopy,iboltr.In .. Amphiboles and other hydrous P)'riboles ­Mineralogy •. £d. D.R. Veblen, Rev. Mineral.,Yol. 9A, Mineral. Soc. Amcr.

TIlOMMSOORI'P V. (\982) - Pt/r%gic ilSptCII ofItfptntini/t mt/amorphism. Rend. Soc. hal. Mi·nef1ll. Pellul., J8, 549·"9. .

UEHlr.JlA S., SHtllOZU H. (198') . VtI,ioltion.· inclNmicll1 composition Itnd !tr~tf.:,1I1 proptrtitr0/ tln/i,o,itn Minet1ll. J. (Japan), 12, 299-J18.

VEBLEN D.R. (1981) . Non-cLmicll1 pyriboltl olndpoI,somil/ic ,elUtiolls in bioPJ,ibolts. In c Am­phiboles and Olher hydrous pyriboles • Mine­ralogy., &t. D.R. Veblen, Rev. Miocl'll.l., Vol. 9A,Minef1ll. Soc. Amer.

VEBLf..N D.R., BUSECK PK (1979l. - Chilin·wid/horde, ilnd diso,der in biopyribo/n Amcr. Mi­neral., 64, 687-700.

192 G. FERRAIUS, M. MELLlNl, S. MERLlNO

WARSIIAW C.M., Rov R. (1961) . Cfaui!icafion and<I scheme /or the identification of loyer silieater.Gt:o!. Soc. Amer. Bull., 72, 145.';:.1492.

WHITTAKEIt E.).W., WICKS F.]. (1970) - Chemicaldifferences among the urpentine «polymorphs »:A dircuSiion. Amer. Minera1., 55, 1025-1047.

WICKS F.]. (1979) . Mineralogy, chemistry andcrys/aUography 0/ chrySOlite asbestos. Miner. Ass.Canada _ Short C'.ourse, 4, 35·78.

W1CI(S F.).. PlANT A.G. (1979) Electron-

microprobe and X.Roy.microbeam sJuJin 0/serpentine textures. Canad. Mineral., 17, 785-830.

YADA K. (1979) . Microstructures of chrySOlite andanliRorite by high rerolufion electron microscopy.(anad. Mineral., 17, 679~91.

ZEMANN J. (I9.'H) - Formel und Kristal/strukturder Scha/arzikits, Tscher. Mineral. Petrogr. Mitt.,2, 166-175.

ZOLTAI T. (1960) - Classification 0/ silicates andother minerals with tetrahedral structures. Amcr.Minct1ll., 45, 960-973.