( a )

i s

302 THE CANADIAN MINERALOGIST

cril ---+

Ftc. 100. Left: powder Raman spectrum of actinolite. Right: Iaser Raman spectrum of a crystal of tremolite[from White (1975) and Blaha & Rosasco (1978), respectively].

800 600 400 ?oo

C 111-l --1,-

t r r l r r lt 200 600 0

principal features of each spectrum are thesame. Each is characterized by a sharp primarymaximum at -650 cm-t and a sharp, second-ary maximum at -510 cm-t, although thelatter is much weaker in crocidolite than inthe remaining three amphiboles. In addition,there are a number of less intense broader peaksat lower frequencies. The narrow symmetricalshape and high relative intensity of the 650cm-t and 510 cm-r peaks characterize themotions involved as torsional oscillations of theOH groups. The narrow, symmetrical peak-shape is indicative of very weak hydrogen bond-ing or none at all. The remaining peaks (below400 cm-') can be assigned to translatory vibra-tions of OH groups, optical (broader, lesssymmetrical and less intense than the oscillatorymodes) and acoustic modes (the lowest frequen-cy modes).

CeltoN DrsrrusurloNs rr{ ArvrpHtsor,ps

The wide variety of cation co-ordinations inthe amphiboles, together with the large struc-tural compliance of some of the sites, result inconsiderable chemical complexity in theseminerals. A knowledge of the site occupanciesand order-disorder relationships in the amphi-boles is essential to a better understanding oftheir crystal chemistry, phase relations, opticaland electrical properties and oxidation-dehy-droxylation mechanisms. 1 Consequently, theamphiboles, more than any other mineral group,have been studied from the viewpoint of site-population sharacterization.

Warren (1930) considered chemical varia-

100). Tremolite is fairly well ordered andproduces a detailed spectrum, whereas actinolitehas significant cation disorder and gives asmeared weak spectrum; only a few weak bandsare obcerved, the bridging oxygen mode at670 cm-' being the only band of appreciableintensity, with weak bands observed at 1O55,1030, 387, 367 and 223 cm-'. It should benoted, however, tlat the laser-Raman techniqueis useful for the characterization of fibrous orasbestiform amphiboles where these are highlyordered (as is often the case).

N eutron-inela.stic scqttering

Infrared absorption studies concerning theOH group in amphiboles have concentrated onthe region of the principal stretching bandaround -3650 sm*!. There are other vibrationsof the OH anion, considered as a rigid groupbound in the structure, and these carry informa-lion conceming the bonding of the bH at ionto other ions in the structure. However, thesevibrations occur below lO@ cm-r, where thevare complicated by the extensive vibrational andlattice modes of the structure. Because of thedifferences in neutron-scatt€ring cross-sections,the most intense peaks in neutron-inelastic-scattering spectra are those resulting from themotion of hydrogen-bearing group6. Conse-quently, oscillatory and translatory modes ofthe OH anion can be examined using thistechnique.

Naumann et al. (1966) have reported neutron-inelastic-scattering spectra for four amphiboles;the spectra are shown in Figure l0l. The

( b )

I

| | | I t r I l I

ro@

I

o r o c ) c | ( 'F F F F Nr o G ) $ t o l F

(.?'I tlI t l

\ l l Anthophylrte

o6lo|

I

oj

sl

|o$(o

IActlnolltr

Tremolhe

Crocldollte6 | lo F

l l\ l

THE CRYSTAL CHEMISTRY OF THE AMPHIBOLES 303

o.:J

ct.:tl(U

trozUJFz

ENERGY GAIN (meV)ooO O O O O T O C ) ^C\l r l0 C)Ol- - ltlt

0 50 100 150 200CHANNEL NUMBER

Ftc. l0l. Neutron inelastic scattering spectra formiscellaneous amphiboles [from Naumann et al,(1e66)1.

TABLE 58. NORMAL CATION SITE.ASSIGNMENTSIN MP}I.IBOLES

C a t i o n A M ( 4 ) M ( l ) , M ( 2 ) , M ( 3 ) T ( l ) , T ( 2 )

s iAIFe3+l ' l

Fe2+MnMgL iCaNaK

tions in the monoclinic amphiboles in termsof cation occupancies in the amphibole structure.Hc showed that the structures of several mono-ctinic amphiboles are very similar to that oftremolite with "certain replacements of struc-turally equivalent ions". Consideration of theamphibole struclure, together with observedchemical variations in natural amphiboles andionic radii of the substituting cations, ledWarren ( 1930) to group certain elementstogether into four subgrottps, similar to thegeneral formula given in the previous sectionon nomenclature. He also showed that certaincations may occttr in more than one subgroup,suggesting that this was a major factor frustrat-ing previous attempts to systematize amphibolecompositions, According to current usage, theCa group of Warren (1930) is split up into theA and B gror.rps (see Nomenclature section);apart from this, the general formula and cationgroupings used today are virtually the same, asindicated in Table 58. Whittaker (1949) showeda nonrandom distribution of group-C cationsover the M(l), M(2) and M(3) sites in magnesio-riebeckite(3), and Papike & Clark (1967) showeda nonrandom distribution of group-C cationsovcr the T( l) and T(2) sites in potassian titanianmagnesio-hastingsite(24). In the last fifteenyears, considerable effort has gone into thecharacterization of cation ordering in amphi-boles using the experimental techniques sum-marized above. What follows is a summary ofthe results from a wide variety of techniques;this gives a general view of cation sitc-occupan-cies in amphibolcs.

Aluminum

As shown by Warren (1930), Al maY beboth a C-group and a T-grottp cation, i.e., Al

304 THE CANADIAN MINERALOGIST

may occur in both octahedral and tetrahedralco-ordination. When in tetrahedral co-ordination,Al shows considerable ordering; because of thesimilarity in X-ray soattering power of Si andAl, relative ordering of these cation speciescannot be directly distinguished by X-raydiffraction. However, the considerable disparityin size between Si and t"Al (0.26 and 0.39 A,respectively) allows ordering of Al over thetetrahedral sites in the amphibole structure tobe examined by comparison of mean bond-lengths at the various sites (see previous sectionon ordering of tetrahedral Al). In monoclinic(C2/ m) amphiboles, Al shows the site orderingT(l)>>T(2); subsilicic amphiboles definitelyhave Al at the T(2) site, but the presence ofAl at T(2) in amphiboles where Si26.0 atomsp.f.u is somewhat questionable (Ungaretti 1980).ln the orthorhombic (Pwna) amphiboles, Alshows the site ordering TIB>TIA>T2B)T2A;as indisated in Figure 41, the T2B site showsconsiderable Al occupancy in gedrite.

When Al is a C-group cation, very strongsite-ordering occurs. However, because of thesimilarity in X-ray-scattering power of Mg andAl, relative ordering of these cation speciescannot be directly distinguished by X-raydiffraction. Whittaker (1949) suggested thattrivalent cations should be ordered at the M(2)site in magnesio-riebeckite(3). Refinement of thestructure of glaucophane(26) (Papike & Clark1968) showed that the mean bondJength atthe M(2) site is much shorter tfian the meanbond-lengths at the M(l) and M(3) sites,suggesting that Al is very strongly or even com-pletely ordered in the M(2) site in glaucophane.Later studies have confirmed that trivalentcations tend to be ordered at the M(2) site, thevariation in size of the M(2) octahedron gene-rally being compatible with this premise(Hawthorne 1978a). Results that are pertinentto this argument for dAl-ordering are those forglaucophane(26), holmquistitel31], gedrite[32],gedrite[33], holmquistite[47], tschermakite[48]and ferro-tschermakite(54). However, these re-sults are not sufficiently accurate to tell whetheror not this ordering is complete. Bancroft &Burns (1969) have asigned weak peaks in theinfrared absorption spectra of glaucophane tocation arrangements involving Al at the M(1)or M(3) sites. Because of the problems associatedwith the derivation of site occupancim frominfrared absorption spectra (Hawthorne 1983),this cannot be considered as good evidence foroccupancy of M(l) or M(3) sites. In support ofthis view, Strens (1974) reported that of fifteensamples of glaucophane examiued by this

method, none show minor bands. In a detailedcrystallographic analysis of twenty-six blueamphiboles, Ungaretti et al. (1978) used meanbond-length and charge-balance arguments toshow complete order of Al at M(2). Thus therethere is no direct evidence of Al occurring atthe M(l) and M(3) sites, the current evidencefavoring complete order at the M(2) site (exceptperhaps for oxy-amphiboles, in which drasticdisordering of cations may accompany oxidationby dehydroxylation).

Ferric iron

Fes* is generally considered as a C-groupcation; there is little evidence to prove otherwise(Phillips 1963), and Leake (1968) has showmthat amphibole compositions deficient in Si*Alalso tend to be classified as poor on groundsother than those connected wilh the summationof T-group and C-group cations. However, inthe clinopyroxenes, the presence of Fe8+ intetrahedral co-ordination has been shown(Huckenholz et al, 1969, Hafner & Huckenholz1971, Virgo 1972b), and the similarity of thepyroxene and amphibole structure-types suggeststhat tetrahedrally co-ordinated Fe3+ could oscurin amphiboles. Hawthorne & Grundy (1977b)provided some X-ray and Mijssbauer evidencefor Fe'' occupancy of the T(2) site in zinciantirodite(s7) but indicated that this evidence isof marginal significance.

Whittaker (1949) proposed that Fe'* isstrongly ordered at the M(2) site in magnesio-riebeckite(3). Later studies assumed that trivalentcations (including Fe'*) were ordered at theM(2) site on the basis of the mean bondJengthat M(2) being shorter than those at M(l) andM(3). Surprisingly enough, it is only recentlythat amphiboles sufficiently high in Fe3+ (potas-sian ferri-taramite(59) and potassium-arfvedson-ite(67), see Appendix 83) have been examinedto provide more definite size-evidense of thisordering. As ordering patterns based on meanbondJength are not very accurate, whether ornot Fe3* is generally completely ordered atM(2) is not certain. Subsilicic titanian magnesianhastingsite(s8) apparently has Feu* at the M(l)or M(3) sites (or both; see Appendix B3), butthe refinement results for this amphibole aresomewhat peculiar. Detailed analysis of meanbond-lengths and charge reguirements indicatescomplete Fe'+ order at M(2) in some cases(Ungaretti et al. 1978) and the occurrence ofsmall amounts of Fe8+ at the M(1) or M(3)sites (or both) in other cases (Ungaretti et al.l98l). It is not clear whether this represents the

THE CRYSTAL CHEMISTRY OF THE AMPHIBOLES 305

equilibrium-crystallization configuration or isthe result of postequilibration oxidation bydehydroxylation.

Numerous amphiboles have been examinedby Mdssbauer spectroscopy, a technique thatis sensitive to iron in its various valence-states.Burns & Greaves (1971) examined a series ofactinolite samples and found that in some cases,the half-width of the doublet assigned to Fe3+wa$ significantly larger than the doubletsassigned to Fe2+ at the various sites. Theysuggested that this was due to the presence ofseveral closely overlapping doublets resultingfrom the presence of Fes* at the M(l) or M(3)sites (or both) in the structure. However, thelarger half-widths of the Fes* doublet may bean artifact of the fitting procedure (Hawthorne1983) or the results of variation in next-nearest-neighbor occupancies (Dowty & Lindsley 1973,Dollase & Gustafson 1978). Thus in normalamphiboles, this cannot be considered as def-inite proof of Fee* occupancy of the M(l) orM(3) sites. Miissbauer spectra of oxidizedamphiboles (Ernst & Wai 1970) show a distincrincrease in half-width of the Fes+ doublet overthe doublet in the corresponding unoxidizedamphibole; in some sases, visible nesolution ofmore than one Feer doublet occurs (Ernst &Wai 1970, Fig. 6). Infrared absorption investiga-tion of the same sample (C-4980) showed twominor peaks at 3625 cm-l and 3654 cm-l. Asthese were present in both natural and heat-treated material, they were assigned toconfigurations involving Fe8+, suggesting thepresence of Fe"* at M(l) or M(3) (or both) inboth unheated and healed material; however,note that a peak at 3654 cm-r is not compatiblewith a configuration involving Fe3", accordingto Table 55. Several studies of natural amphi-boles (Burns & Prentice 1968, Bancroft & Burns1969, Burns & Greaves t97l) by infraredabsorption spectroscopy have demonstrated thatFe3* oscurs at the M(1) and M(3) sites. Strens(7974) has suggested that these authors haveoverestimated the intensity of those peaksassigned to Fe3*. Burns & Greaves (1971) haveindicated that oxidation occurs during samplepreparation for infrared absorption spectroscopy;thus the small amounts of Fe3+ possibly observedin these studies are due in some part to suchsample preparation. This is supported by theresults of Strens (1974) for numerous alkaliamphiboles, where no peaks attributable to Fe8+were observed in the infrared absorption spectraand where the samples were prepared by amethod whdre no significant oxidation occurred.

Detailed crystal-structure refinements by

TABLE 59. rd vnLuEs FoR FEz+ ORDERING ATM(I) AND l'1(3) FOR THE AMPHIBOLES0F BoccHro ET AL. (1978) ANDUNGARETTT ET AL. ( le8t)

i nc l ud i nqFe3+

0. 460.480 . 5 10 .460 .430 .450.430 .400 . 5 60 .480 . 3 90 .580 . 5 40 . 5 20 .350 . 3 70 .40

0 . 5 50 . 6 00 . 6 5

wi thoutFe3+

A4A5B3B6c lc3D6D7D9D t 8E2E9E t 3E l4F3GHH2

0 . 7 30 .890 .91o,nt

n 7 1

1 . 2 32 , 3 6

(70)(7t )(72)

Va' lues are calculated us ing both the observedFez+ s i te-occupancies,^and- inc luding a l l Fe3+at each site in the Fez+ site-occupancy

Bocchio et al. (1978) and Ungaretti et al, (1981)show the presence of small amounts of Fes*at the M(l) and M(3) sites (see Appendix B3).Table 59 lists the Kd values (for Fe") for M(1)*M(3) ordering. Each of these suites of amphi-boles were taken from a sm,all region over whichthe ambient conditions probably were fairlyuniform. For the three amphiboles of Bocchio eral. (1978), the Kd values vary considerably. How-ever, if Kd is calculated assigning all Fe at M(1)and M(3) as Fe'*, the v.alues are fairly similar.For the amphiboles of Ungaretti et al, (1981),Ko values are fairly similar for all amphibolesexcept those with Fe3* at M(1) and M(3).Considering all Fe as Fe2+ at these sites, thecalculated Ko values for the aberrant amphibolesnow fall into the same fairly restricted rangeas the rest of the amphibole suite. Both of theseresults suggest that all the Fe at M(l) and M(3)in these amphiboles was Fe2' at the time ofcation equilibration, when the Kd values wereset, and that those amphiboles with Fe"* at M(1)or M(3) (or both) underwent postequilibrationoxidation sufficiently mildly or at suffioientlylow temperatures that there was no accompany-ing disorder among the cations.

Titanium

Titanium may have the valence states 3 +

306 THE CANADIAN MINERALOGIST

and 4+ in an oxide environment, but untilrecently was thought to be in the 4* state inminerals (ber et al. 1963, Hartrnan 1969).More recent work has shown the presence ofconsiderable Tie+ in lunar (Burns et al. 1976)and meteoritic (Dowty & Clark 1973a, b, Mao& Bell 1974, Mason 1974) pyroxenes, and inmelanite-schorlomite garnets (Burns 1972,Huggins et al. 1977, .Srhwartz et a.l. l98O).Prewitt et al. (1972) synthesized NaTis+Si:Oo,demonstrating that octahedrally co-ordinatedTis* does o'ccur in pyroxenes. The similarity ofthe pyroxene and amphibole structure sugge$tsthe possibility of Ti"* occurring in the amphi-bole structure. However, there is no convincingevidence concerning the valence state of Ti inamphiboles, although the wider range of struc-tural environments in amphiboles suggests thatmixed Ti valences are a distinct possibility. Evenif Ti were always trivalent upon crystallization,subsequent partial oxidation-dehydroxylationcould give rise to cation disorder and theformation of mixed valences of Ti.

Ti is generally considered as a C-group cation,although some authors do assign Tin* to theT group where Si*Al is less than 8.0. Thereis no evidence of tetrahedrally co-ordinated Tiin amphiboles. In most X-ray studies, Ti hasbeen assigned to the M(2) site together with allthe other trivalent cations in the formula unit.However, there is no direct evidence for thisassignment of Ti. The only direct evidence forTi ordering in amphiboles is the neutronrefinement of potassian oxy-kaersutite(40) byKitamura et al. (1975). They showed that Ti isstrongly ordered at the M(l) site. Whether thisis an equilibrium-crystallization feature orwhether it is associated with postcrystallizationdisorder during oxidation-dehydroxylation is notknown. However, the cation disordering thatis known to occur on high-temperature oxida-tion-dehydroxylation (Ernst & Wai 1970) andthe ordering behavior of trivalent cations sushas Al and Fes+ (see above) suggest that theordering of Ti at M(l) in oxy-kaersutite is apostcrystallization phenomenon accompanyingoxidation-d ehydroxylation.

Schwartz & Irving (1978) have examinedkaersutite using the general technique of Burns(1972); they derived Ti3+/CIi'++Tia*) ratios of0.04. 0.04 and 0.09 for three samples of kaersut-ites with TiO, in the range 5.1-5.8 wt. Vo. Asindicated by the authors, this is not convincingevidence for Tis* in amphiboles because of thelow values and tlte uncertainty of the method;however, it seems to be the only way to examinethis problem at the moment.

The possibility also exists that mixed valencqsof titanium may occur as a result of stereo-chemical criteria. For exxample, bond valenceconsiderations suggest that local occupancy ofT(2) by Al'+ might be coupled with the occu-pancy of M(2) by Tia*, whereas Si occupancyof T(2) might be associated with Ti"* in M(2).In view of the numerous possibilities for mixedvalences and site ordering of titanium in amphi-boles, it is perhaps fortunate that titanium isgenerally a minor constituent in mct varietiesof amphibole.

Ferrous iron

Fe" is generally assigned to the B and Cgroups of the amphibole formula and canoccupy the M(1), M(2), M(3) or M(4) sites. Asthe character of the ordering patterns exhibitedby Fe2* changes with the type of amphibole,it is convenient to consider each principalamphibole group separately.

The iron-magnesium-manganese amphibolegroup is characterized by considerable orderingof Fe'*. It was first shown by Ghose & Hellner(1959) and Ghose (1961) using X-ray diffractionthat in the structure of the cummingtonite -grunerite amphiboles, Fet* shows a nonrandomdistribution over the four M sites, being prefer-entially ordered at the M(4) site. This orderingpattern has since been confirmed by additionalX-ray diffraction (Finger 1969a) and extensiveMiissbauer work (Bancroft et al. 1967a,b, 1968,Hafner & Ghose 1971, Buckley & Wilkins l97l);Ghose & Weidner (1972) have shown that thedegree of ordering is to some extent a functionof equilibration temperature. The Mtissbauerspectra of cummingtonite - grunerite amphibolesshow incomplete resolution (Appendix F), withthe doublets from Fe'* at M(l), M(2) and M(3)showing virtually exact overlap; thus Miissbauerspectroscopy provides no information con@rn-ing the ordering of Fe2* over the M(1), M(2)and M(3) sites. However, the structure refine-ments of cummingtonite(2l) and grunerite(22)both show the Fe'* site-occupancy M(3)>M(1)>M(2). The ordering of Fe2+ over the M(1),M(2) and M(3) sites has also been examined byBancroft et al. (1967a) using a combination ofMiissbauer and infrared absorption spectroscopy.Miissbauer spectroscopy gives the Fe'* cationratios at the .M(I)*M(2)+M(3) sites and theM(4) site, whereas the infrared absorptionmethod gives the Fes* cation ratios at theM(l)*M(3) and M(2)+M(4) sites; thus therelative ordering of Fe'* between the M(l)*M(3) and M(2) site.s may be derived from the

THE CRYSTAL CHEMISTRY OF THE AMPHIBOLES

combination of these two methods. The exper-imental results of Bancroft et al, (1967a) alwaysindicate the Fe3+ site-occupancy M(l), M(3)>M(2); these authors stated that "in some gru,nerites, Fe'* becomes enriched in the M(2)position", but provided no experimental evidencefor this. Similar experiments by Buckley &Wilkins (1971) atso showed Fe2* to be prefer-entially ordered at the M(l)*M(3) sites relariveto the M(2) site. The strong absorption band at-1 tram in the B absorption spestrum of amphi-boles of the cummingtonite- grunerite series(Fig. 88) is compatible with the high Fe'+ occu-pancy of the M(4) site throughout this series.As with most other amphi'boles, Fet* orderingis affected by blocking of sites by other cationsin the structure. Ca occupies the M(4) site incumminglonite - grunerite amphiboles, thus af-fecting the relative ordering of Fez* betweenthe M(4) and M(l)*M(2)+M(3) sites @ancroftet al, 1967a); similar behavior has been recog-nized in clinopyroxenes (Hafner 1975). Thepreferential ordering of trivalent cations at theM(2) site may similarly affect Fe2* orderingover the M(l), M(2) and M(3) sites.

The Fe2* site-preference is completely differ-ent in the amphiboles of the tirodite - danne-morite series, although data are only availablefor the tirodite end of the series. Using acombination of M6ssbauer and infrared absorp-tion spectroscopy, Bancroft et al. (1967a)Efound the Fe2* site-occupancy M(l)+M(2)+M(3)>M(4) for tirodite{7}. The Fe2* cationordering over the M(1), M(2) and M(3) sitesis distinctly different from that exhibited byamphiboles of the cummingtonite - gruneriteseries; tirodite shows strong ordering of Fe2+at the M(2) site, and the overall Fe'* site-occupancy M(2)>M(4)>M(1)+M(3). A com-bined X-ray-diffraction and Miissbauer-spec-troscopy investigation of zincian tirodite(S7)(Hawthorne & Grundy 1977b) showed the Fez+site-occupancy M(l) + M(2) + M(3) > M(4), inagreement with previous infrared absorptionand Miissbauer results. On the basis of bond-length considerations and total scattering-powersat each site, Fe2+ seems to be strongly orderedin the M(2) site in zincian tirodite(57); althoughthis conclusion is not as well established as therelative Fe2+ ordering pattern over the M(4)and M(l)*M(2)+M(3) sites, the overall patternof ordering M(2)>M(4)>M(1), M(3) is com-

*There is a misprint in Table 4, page 1021 of thisreference; the quoted Fe2+ occupancy of M(2) in theunit formula of "Mn cummingtonite" should read0.57. NOT 0.07.

patible with the results af Bancroft et al.(1967a) on tirodite{7}. Papike et al. (1969) haverefined the structure of tirodite(28) and foundthat (Fe*Mn) is strongly ordered at the M(4)site. On the basis of the results of Bancroft et al.(1967a), they assumed that Mn is stronglyordered in the M(4) site, and assigned cationsite-populations at M(l), M(2) and. M(3) on thebasis of mean bondJengths. Both the sitepopulations given by Papike et al. (L969) andderived from the curves of Hawthorne (1978a)and given in Appendix B3 show the Fef+ site-occupancy M(2)>M(4)>M(I), M(3) in agree-ment with Bancroft et al. (1967a) and Hawthorne& Grundy (1977b).

Fe2+ ordering patterns in the ofihorhombicFe-Mg-Mn amphiboles differ significantly fromthose of the monoclinic Fe-Mg-Mn amphiboles.Refinement of the structure of anthophyllite[23](Finger l97Db) showed the Fe2* site-occupancyM4>>MI-M2-M1, in agreement with pre-vious results by Miissbauer and infrared absorp-tion spectro$copy (Bancroft et al. 1966). Morerecent Mijssbauer results on anthophyllite[23](Seifert & Virgo 1974, 1975, Seifert 1977) showexceUent agre€ment with the X-ray results(Appendix C3). This distribution is also sup-ported by the polarized absorption spectra ofanthophyllite[23] (Fig. 87), where the barycentreenergy of -700O cm-' for the I p,m(B) and2.4 pm(a) bands is compatible with the<Mzt-O> distance in anthophyllite[23], ac-cording to the correlations of Faye (1972). Thereis as yet no information on the relative orderingof Fe2' over the Ml, M2 and M3 sites inanthophyllite with significant Fe2* occupancyof these sites. Significant differences occurbetween Fe'* site-ordering in anthophyllite andgedrite. Crystal-structure refinement of twosamples of gedrite (Papike & Ross 1970) showan Fe" site-ordering M4)M1-M3)M2;however, the ordering is not as strongly devel-oped as in anthophyllite. Gedrite samples havebeen examined by Miissbauer (Seifert 1977) andinfrared absorption methods (Burns & Lawl97O), but the peak overlap is such that thespectra cannot be adequately resolved; thus noinformation on cation ordering in gedrite hasbeen derived using these methods. The polar-ized-absorption spectra of gedrite[32] (Fig. 87)supports the Fet* cation-ordering arrangementderived from the X-ray study of Papike & Ross(te70).

The Fe2* distribution in holmquistite is verydifferent from that exhibited by the otherFe-MyMn amphiboles. Crystal-structure refine-ment (Irusteta & Whittaker 1975) shows the

308 THE CANADIAN MINERALOGIST

TABLE 60. #T SIrc.OCCUPANCIES FROI4 CRYSTAI.-STRuCTURE REFINEMENTS

!l(3) ltl(4) ordering patten

site populations in calcic amphiboles, but thishas tended to concentrate on the simpler chem-ical species. Amphiboles of the tremolile - ferro-actinolite series were first examined usingMdssbauer spectroscopy by Bancroft et al.(1976b); the resolution achieved in this studywas inadequate, but improved resolutionin a later study (Greaves et al, 197l)led to the derivation of complete Fe"*site-occupancies, with additional results beingpresented by Burns & Greaves (1971). Sitepopulations for actinolite have also been derivedby infrared absorption sp€ctroscopy (Burns &Strens 1966, Wilkins 1970) and a comparisonof results of both methods is given by Burnset al. (191O) and Burns & Greaves (1971). Theresults of Wilkins (1970) all show the site occu-pancy M(1)}M(3)>M(2), including an actin-olite also examined by Burns & Strens (1966),who derived the Fes* site-preference M(2))M(1)+M(3). The infrared absorption resultsof Burns & Greaves (1971) generally show theFe!* site-preference M(2)>M(l)+M(3), butthis is at variance with their own Miissbauerdata, which show the Fe'* site-occupancyM(l)+M(3)>M(2)u; these authors concludedthat the Miissbauer results are more reliable.Goldman & Rossman (1977a) have providedspectral evidence that Fe" does occur at theM(4) site in amphiboles of the tremolite - ferro-actinolite series. The polarized-absorption spec-tra of a tremolite and an actinolite show verystrong bands at 1030 nm(F) and 2474 nm(a);the large band-intensity, together with thesimilarity of the spectra to others from Fe'*in large, very distorted co-ordination polyhedra(Burns 1970a, Goldman & Rossman 1976,1977b), suggests that these bands are due to theM(4) occupancy of Fe"*. In addition, the ob-served energy of the barycentre is compatiblewith the correlations of Faye (1972). The be-havior of the 1030 nm band in more iron-richactinolite (Burns l970a, Fig. 5.13) is alsocompatible with this assignment. The chemicalanalysis of this actinolite (Burns & Greaves 1971,Table l, number 5) indicates significant C-groupcation occupancy (> 5.15 atoms p.f.u.) of theM(4) site. Comparison of the average Fe'*site-occupancies at tlte four M sites [M(1)-M(2)-M(3)-0.48, M(4)-0.061 is in line withthe more equal band-intensities of Fe2* at M(4)and at M(1), M(2) and M(3) expected from

*There is an error in Table 5 of this reference; theFez+ populations for the M(2) and M(3) columnsare interchanged. However, the site occupancies aregiven correctly in their Table 6.

M(2)

0 .29 0 .760.s9 0 .99

- 0 .09

0 . s

(21) 0 .29 0 .05(22) 0.85 0.77(28) 0 .03 0 . 16(57) - 0 .1 7( 6 1 ) 0 . 2 5t23 l 0 .03 0 .02t3l I 0.40t32 l o . l2 (0 .04)r33r 0 .33 (0 .09)

lr(4)11(4)r,r(2)r'r(2)r'1(3)l'l(4)M ( 3 )t4(4)M ( 4 )

11(r)n(1)l'r( 1 )r(r )til( 1 )i l (1 )r'1( l )M ( t )i lo )r'r( r )M ( l )r'1( 3)M(3)M(3)|' l( l)ir(3)r'r(3)H( 3)t 't{: i

> x ( 1 ) - M ( 3 ) ' M ( 2 )M ( 3 ) > M ( 1 ) > M ( 2 )r ' r ( 4 ) > x ( r ) > x ( 3 )r ' 1 ( 4 ) > x ( r ) - M ( 3 )l,r(r) >>r,r(2) - r '1(4)M ( r ) - M ( 2 ) - M ( 3 )r ' r ( r ) >>M(2) - r '1 (4 )r ' { ( l ) - x ( 3 ) > M ( 2 )r,r(3) > M(1) >>!r(2)

3)

2 \2 \

l )

( 0 . 1 r )0 .46

(0 .20)( 0 . 1 7 )

o.is0.35o. : ,

o .os( 0 . 3 r )0 . 1 80.07

( 0 . r 5 )( 0 . r 5 )( 0 . r 6 )( 0 . 2 r )

0 .520 . 1 80 . l 3

-0.34

o. : ,

(24)( 37)*( 3s)(3e)(42)(44)(45)(46)a48l(4e)(50)(54)

(60)(61 )(70)*r( 71 )**(tzr*"(73)

( 5 1 )(59)

(64)

(oz)*( 68)*( 6 e )

0.350 . 6 10.230.320,29I .000.460.490.520 . 70 . 60 . 6 r

(0 .6e)0 .360.300 . t 70 . 1 90 . 1 70.32

0.200.850.73

1 . 0 00.93' r .00

0.930.59

:values glven in parentheses for t ie M(2) si te are predoninant ly Fe$consJderable lh ls lncluded ln the si te-occupancles giveni where anesl lnate of l ln sl te-occuponcies has been mde, indivldual est imtedFez' s l te-occupancles are given r l t i the combined Fe+l ' ln occupancy 1n

Jffilft"lin*r, srsnrncant Fe3+ at tre r,r(r) and/or M(3) sites

Fe2* site-ordering M3 > M I >> M2 - M4 - O.0,and this is confirmed by Mdssbauer and in-frared absorption spectroscopy (Wilkins et c/.1970, Law & Whittaker 1981) (see AppendicesC3, F and G).

The calcic amphibole group shows an ex-tremely complex variation in Fee+ ordering. Theresults of crystal-structure refinements aresummarized in Table 60, where it can be seenthat no exact overall pattern is apparent. How-ever, the predominant Fe'* ordsring pattern ap-pears to be M(1)>M(3)>M(2)=M(4) althoughthe Fez* content of the M(2) site is obviouslystrongly affected by the trivalent-cation conte.ntof the M(2) site. Ferro-tschermakite(54) showsM(3)>M(I)>>M(2)>M(4), and this is compat-ible with the Miissbauer spectrum (Hawthorne1973)" although it should be noted that theother structure showing the Fe2* site-occu-pancy M(3)>M(l), glaucophane(26), is alsocharacterized by a particularly small mean sizeof the M(2) constituent cation. There has besnsome spectroscopic work on the derivation of

0.03 0 .510.560 . 1 0 0 . 4 20.39 0 .65

Calcic group

0.350.580.240.32^ a ^0.480.27

0.78(0 .84) 0 .070.450.240.22 0 .030.28 0 .030.29 0 .040 . 4 5 0 . 1 7Sodic calclc qmup

0 . i 5 0 . 0 40.850 . 7 1 ( 0 . 8 5 ) -

Alkal i qrcup

0.290.670.88

o.io(o.eg) l0.48(0 .64) -0.80

I't(M(u{tl(

'M(l',l(M (ultit(I't(M(M(M II't(M (M(l.t(M(M (

> lit(>>r,,|(

" I,t(

r '1(3) , r 'r(1) >>M(2) - 14(4M(r ) > M(3) >>M(2) - r , r (4M(r ) > x (3) >>M(2) > M(4l' l(1) >>14(l) " I ' l(2) - lr l(3I' l( l) > r' l(3) ' l t(2)

"r{(ar4( r ) > M(3) " r { (2 ) - M(a

M(3) > I ' l ( l ) "M(2) - x (a

THE CRYSTAL CHEMISTRY OF T}IE AMPHIBOLES 309

[wo groups of co-ordination polyhedra withvery different asymmetry. There is little spectralinformation on the more complex calcic amphi-boles. Hiiggstriim er al. (1969) examined a seriesof calcic amphiboles but did not obtain adequateresolution of doublets to derive site populations.Burns & Greaves (1971) examined a pargasiteusing Mdssbauer spectroscopy; the site popula-tions ind'icate the Feu* site-occupancy M(l)>M (3 ) > M(2) >> M(4). The optical-absorption spec-tra of a pargasite (Goldman & Rossman 1977a)show the Fe'* site-occupancy to be M(l)*M(2)+M(3)>M(4). Semet (1973) showed, usingincompletely resolved Miissbauer spectra, thatFe"*-bearing synthetic magnesio-hastingsite hasthe Fe'* site-occupancy M(l)f M(3)>M(2).Khristoforov et al. (197i have examined aseries of calcic amphiboles of unspecified com-position; their results, given in diagrammaticform, indicate the Fe2* site-occupancy M(3)>>M(l). However, the spectra presented by themcannot be considered reliable (off-resonancecounts from 16000 to 38000, maximumabsorption up to 957a, half-widths up to- 1.0 mm/s) .

The only published data on Fe2* site-occu-pancies in sodic-calcic amphiboles are fromcrystal-structure refinements, the relevant databeing summarized in Table 60. Fluor-rich-terite(35) shows the site preference M(2)>M(l)>M(3)>M(4); as discussed later. the preferenceof Fe'' for the M(2) site in fluor-richteriteappears to result from the antipathy of Fe"for F-co-ordinated sites in silicate minerals. Theremaining amphiboles show the Fe" site-occu-pancy M(3)>M(l )>M(2)> M(4) ; however,both contain considerable Mn at the M(l), M(2)and M(3) sites, and tle exact site-populationsarc not known with certainty. For potassianferri-taramite(59), mean bond-length criteriasuggest Mn to be strongly ordered at M(3) withthe resulting Fe2* site-occupancy M(l)-M(3),but this assignment is quite speculative. Virgo(1972a) reported the Miissbauer spectrum ofsynthetic ferro-richterite, synthesized by Charles(1972, 1974); the spectrum could not be ade-quately resolved, but indicated the presence ofoctahedrally co-ordinated Fe"* in an amphibolenominally containing only Fez'. Ungaretti et al,(1981) have presented refinements of severalsamples of winchite, barroisite and magnesio-taramite. All these show the Fe" site-occupan-cy pattern M(3)>M(l)>M(4)>M(2).

Considerable work has been done on Fe"'site-occupancies in the alkali amphibole group;Table 60 summarizes some of the eadier resultsof crystal-structure refinements. Glaucophane

(26) shows the Fe'* site-ordering pattern M(3)>M(1)>>M(2)>>M(4); this pattern has been con-firmed in several samples of glaucophane byM6ssbauer sp€ctroscopy (Bancroft & Burns1969, Ernst & Wai 1970; see Appendix F)and in ferro-glaucophane(69) by X-ray andM,iissbauer data. Structure refinement of fluor-riebeckite(68) shows the Fe" site-occupancyM(l )> M(3)>>M(2)-M( ); however, this seemsto be caused by the ordering of Li at the M(3)site and is probably not representative of Fef*in Li-free riebeckite. Ernst & Wai (1970) exam-ined several samples of riebeckite by Miissbauersp€ctroscopy and derived the Fe2+ orderingpattern M(3)>M(l), as in glaucophane. Con-versely, a riebeckite from Laytonville, Califor-nia, (19), examined by Banoroft & Burns (1969)by Miissbauer and infrared spectroscopicmethods, shows the ordering pattern M(l)>M(3). Ungaretti et al. (1978) examined a rie-beckite of almost identical composition fromthe same locality and derived the Fe'* orderingpattern M(3)>M(l) from site-occupancy refine-ment of single-crystal X-ray data (see Fig. 102).Several samples of magnesio-riebeckite examinedby Miissbauer and infrared spectroscopicmethods (Bancroft & Burns 1969, Ernst & Wai1970. Burns & Prentice 1968; see Appendix F)show the Fe" site-preference M(l)>M(3), with

,

.a

O

.ooa.n

,l '

7n":rr

Fe"* M (3) sile'occuponcy -..*

Frc. 102. The distribution of Fe3* over the M(l)and M(3) sites in glaucophane - ferro-glaucophane- magnesio.riebeckite - riebeckite amphiboles ( o )and nyboites (l ) From Ungaretti et al. (1978)1.The * indicates riebeckite of igneous origin thatcontains significant Li,

ocooJaooo=.t=

+o

l!

3 1 0 THE CANADIAN MINERALOGIST

crossites showing Fd* enrichments into bothM(1) @rnst & Wai 1970) and M(3) @ancroft& Burns 1969). An unusually extensive study byUngaretti et al. (197&) showed that the Fer+site-ordering pattern in amphiboles of theglaucophane - ferro-glaucophane - magnesio-riebeckite - riebeckite series is extremely sys-tematic (Fig. 102), with the pattern M(3)>M(t)throughout the whole compositional rangeexamined, which excluded magnesio-riebeskite.The regularity of Figure 102 is so striking thatit questions the previous spectral results onmagnesio-riebeckite; additional work on magne-sio-riebeckite is required.

Magnesium

The behavior of Mg in crystals is generallyantipathetic to that of Feu*, and in thermodyna-mic treatments Kretz 1959, 1963, Mueller 1962,1967, 1969, Saxena 1968a,b, l969a,b), order-ing of these two species is usually consideredas an exchange reaction. Thus the ordering andsite occupancies of Mg will only be discussedwhere they are of particular crystal-chemicalinterest.

In C-centred amphiboles of the cummingtonite- grunerite series, Mg appears to be virtuallyexcluded from the M(4) site. In studies wherethe derived Fee+ owupancy of M(4) is less than1.0, the balance may be assigned to the smallamount of Ca generally present (see AppendixB3). Any significant M(4) occupancy by Mgoccurs only in tirodite P2r/ m(27), and isthought to be the cause of this particularstructural variant (Papike et al. 1968, 1969).The orthorhombis Fe-Mg-Mn amphiboles,anthophyllite[23] and gedrite[32] and [33] showconsiderable occupancy of M4 by Mg, andagain it is this factor that is thought to beresponsible for the existence of the orthorhombicstructure-type. Other amphiboles with Mg atM(4) are the synthetic protoamphibole[20] andsynthetic (Li,tr) (Li,Mg)sM&SieO,,*,(OH)"-,grown by Maresch & Langer (1976). Both ofthese amphiboles have Mg and Li apparentlymiscible at the M(4) site; this is not surprisingin terms of the ionic radi'i and single-chargedifference between these two cations, particu-larly as protoamphibole is an orthorhombicstructural variant with a small M4 site and theother amphibole is of unusual composition tosay the least. Considerably more enigmatic isthe existence of the synthetic monoclinic amphi-bole NaMgNaMgsSirOz(OH)o (e,g., Witte et al.1969), in which the M(4) site is presumablyoccupied by Mg and Na in equal proportions;

with regard to the normal crystal-chemicalbehavior of cations in crystals, this would seemless likely than miscibility between tremoliteand niagnesio-cummingtonite, unless its struc-ture is not a C2/ m structure but a subgroupderivative of th,is.

Manganese

In principle, site occupancies of Mn in amphi-boles should be diffic.ult to derive. Mn and Fehave very similar X-ray-scattering factors andcannot be distinguished by this method; theyhave sufficiently similar electronegativity andmass that they cannot be distinguished by theinfrared absorption method. However, Fe issensitive to the Miissbauer effect and Mn isnot; hence, the distribution of Mn in amphibolescan be derived by a combination of thesemethods. This was first done by Bancroft et al.(1967a) for tirodite{7}, using a combination ofMtissbauer and infrared absorption spectros-copy; they showed that Mn is strongly orderedin the M(4) site, showing the site preferenceM(4)>M(l)+M(2)+M(3). A similar result wasobtained for zincian tirodite(57) using a com-bination of crystal-structure refinement andMiissbauer spectroscopy; these authors alsorefined the Mn site-occupancies of the M(l),M(2) and M(3) sites, but the results were notstatistically significant. Examination of meanbondJengths in tirodite(28) suggests that Mnavoids the M(2) site, which is in accord withthe results for zincian tirodite(57). In the calcic,sodic-calcic and alkali amphiboles, there islittle or no information on the distribution ofMn. In principle, the problem is susceptible to acombination of X-ray and Mdssbauer methods,but in prbctice the difficulties assosiat€d withderiving accurate Fe2* site-occupancies from theincompletely resolved spectra have so far pre-cluded the use of this method. When the sum ofC-group cations exceeds 5.O, some studies haveassigned the excess cations to the B group in theorder Mn;(Fe,Mg); there is no direct proofof this, although it is in line with the behaviorof Mn in the Fe*Mg-Mn amphiboles. Somestudies (Hawthorne 1976, Hawthorne & Grundy1978) have attempted to derive Mn site-prefer-ence in sodic-calcic and alkali amphiboles onthe basis of mean bondJengths at the M(l),M(2) and M(3) sites; the preference M(3))M(l)-M(2) was observed in both studies, butthese results are highly speculative. Bershovet al. (1966) and Manoogian (1968a, b) haveshown by electron-spin-resonance studies thatMn in tremolite is (predominantly) divalent.

THE CRYSTAL CHEMISTRY OF TTTE AMPHIBOLES 3 l l

Both studies concluded that Mne+ occupied 'oone

site" leither M(4) or M(1,2,3) togethed andassigned it to M(4); however, an argument canbe put forward for reversing this assignment,so the situation is still not clear. Goldman(1977) exarnined a Mn-bearing tremolite usingpolarized electronic-absorption spectroscopy; heassigned bands to d-d transitions in Mn3* andperhapo Mnz*. Thus the Mn situation is atpresent rather confused.

Lithium

Li occurs in significant quantities only inspecific types of amphibole, although it is capa-ble of occupying a wide variety of sites inamphiboles. In the Fe-Mg-Mn amphiboles, onlyholmquistite and clinoholmquistite show signif-icant I-i. Crystal-structure refinements (Whitta-ker 1969, Litvin et al. 1,973a, Irusteta &Whittaker 1975) show Li to be virtually com-pletely ordered at the M(4) site. These resultsagree with the infrared absorption results ofWilkins et al. (1970), which show that nosignificant Li occurs at the Ml*M3 sites inholmquistite. Li also occurs in protoamphi-bole[20], where it oscurs both at the M4 site(together with Mg) and at the A site. However,Gibbs (1969) did not observe any significantelectron-density at the A site and suggested thatLi was positionally disordered about the cavitysurrounding the A site. A similar distributionof Li with significant occupancy of the M4 andA sites presumably occurs for the syntheticamphibole (Li,E)(Li,Mg)zMgrSieOzr',(OH)r-"synthesized by Maresch & Langer (1976). Li isnot a common constituent of calcic amphiboles.Calcic amphiboles coexisting with holmquistite(von Knorring & Hornung 1961, Wilkins er a/.1970) show only sm.all amounts of Li (-0.16atoms p.f.u.), and there is no evidence to showwhere the Li occurs in the structure. Conversely,Li can be an impofiant constituent of alkaliamphiboles, .and stoichiometric considerations(Sundius 1946b, Borley 1963) suggest that Li isa C-group cation in these. On the basis ofinfrared absorption evidence, Addison & White(1968b) proposed that U occurs at the M(l)*M(3) sites in riebeckite. Subsequent crystal-structure refinement of a fluor-riebeckite(68)containing significant Li (Hawthorne 1978b)showed Li to be almost completely ordered atthe M(3) site. However, it should be stressedthat this ordering patt€rn pertains to a fluor-amphibole [O(3) - 0.6F+0.4OI{; the orderingmay be different in a hydroxy-amphibole, al-though high Li contents are generally associated

with high F contents in alkali amphiboles(Lyons 1976).

Calcium

Ca is virtually always assigned to the B groupin the unit formula, and all X-ray studies ofcalcic and sodic-calcic amphiboles appear to becompatible with this assignment, Ca occupyingthe M(4) site in all structures so far examined.It is not certain whether Ca is ever significantlyin excess.of that required to fill the B groupin the unit formula. Practically all compositionsin which Ca exceeds 2.10 atoms p.f.u. areresults of obviously poor analyses, as discussedby Leake (1968); even the composition withfairly well-documented high Ca (2.23 atomsp.f.u.) quoted by him (#801) is classified asinferior on the basis of the sum of C-groupcations (4.58 atoms p.f.u.). Hawthorne (1976)has examined the sums of C-group cations inamphiboles of the eckermannite - arfvedsoniteseries, showing that only lOVo of the analysesare satisfactory according to generally acceptedcriteria (Papike et al. 1974). There are severalpossibilities with regard to this situation: (Dminor occupancy of the M(1,2,3) sites by Cadoes occur; if this is assumed, -7OVo of theamphibole analyses will be classified as satis-factory. (ii) Significant amounts of minor com-ponents (e.g., Li, Cr) were ignored duringanalysis, sufficient to bring the majority of theanalytical results up to the satisfactory level.(iii) Vacancies can occur at the M(1,2,3) sitesin amphiboles; in fact, vacancies at the M(3)site in amphiboles are not disadvantageous withregard to local bond-valence requirements.

In the Fe-Mg-Mn amphiboles, small amountsof Ca are assumed to occupy the M(4) site;whether this occupancy is real or whether theseamphiboles contain incipiently developed .la-mellae of calcic amphibole must await detailedstudies by transmission electron-microscopy.Certainly in coexisting Fe-Mg-Mn and calcicamphibole assemblages, exsolution of calcicamphibole by the Fe-.Mg-Mn amphibole iswidespread (Ross el al. 1968a, 1969, Jaffe et al.1968. Robinson et al. 1969. Robinson & Jaffe1969). In the rare amphibole joesmithite, Caoccupies the two M(4) sites and the A(2) site;this is the only recorded amphibole where theA cavity is occupied by a divalent cation.

Sodium

Na is assigned to the A and B groups of theunit formula. Warren (1929, l93O) proposed

3r2 THE CANADIAN MINERALOGIST

that Na would occupy the M(4) and A sites inthe amphibole structure, and this was confirmedexperimentally by Whittaker (1949) and Heritschet al. (1957). All subsequent crystal-structurework has confirmed these results.

Potassium

K is always assigned to the A group in theamphibole formula unit, and all experimentalwork so far published is in agreement with this(Papike et al. 1969, Cameron et al. 1973b,Hawthorne 1976), T\e positional disorder ofNa and K is discussed in the section on the Asite.

Beryllium

Joesmithite is a remarkable amphibole inwhich Be2* occupies the T(1)B site, apparentlyshowing no disorder with Si over the fourdistinct tetrahedral sites. The occupancy of aT(l)-type site by Be'* is in accord with thecriteria based on local charge-balance proposedby Hawthorne (1978c).

Boron

Kohn & Comeforo (1955) reported the syn-thesis of a calcic fluor-amphibole containingconsiderable B3*. The chemical analysis indi-cates l. l2'B and 0.33d8 p.f.u.; as B3* has notbeen recorded in octahedral co-ordination in anoxide environment" additonal evidence is neededbefore this result can be accepted.

Cobalt

Co2* occupies the M(1), M(2), M(3) and M(4)sites in the structure of "NazHrCogSieO*(OH)r"(Prewitt 1963, Gibbs & prewitt 1968), whichapparently deviates somewhat from its idealcomposition. Similar results are apparent fromthe chemical analysis of fibrous cobalt amphibolesynthesized by Nesterchuk et al. (1968). On thebasis of optical-absorption spectra, Chigarevaet al. (1969) proposed that Co2+ occurs in bothoctahedral and [8]-fold co-ordination in synthe-tic cobalt fluor-richterite.

Nickel

Chemical analysis of synthetic nickel richterite(Fedoseev er c/. 1968, 1970) shows Ni tooccupy the M(l), M(2), M(3) and M(4) sites.Absorption spectra of a nickel fluor-riqhteritesuggest the presence of Nis* in both octahedraland [8]-fold co-ordination (Chigareva et al.1969).

Chromium

On the basis of optical-absorption spectra,

Chigavera et al. (1969) proposed that Cr3*ocsurs in [8]-fold co-ordination in "chromiumfluor-amphibole".

Zittc

Klein & Ito (1968) reported results of chem-ical analyses and physical data for severalzinc-bearing amphiboles from Franklin, NewJersey. In zincian tirodite, the Zn content mayattain 1.22 atoms p.f.u.; a combined X-raystructure refinement and Mdssbauer-spectros-copy study (Hawthorne & Gmndy 1977b) of.zincian tirodite(S7) showed the Zn site occu-pancy M(1) > M(3) >> M(2) > M(4). Chemicalanalyses of zinc-bearing amphiboles of thetremolite - ferro-actinolite series suggest thatZn is strongly ordered at the M(1), M(2) orM(3) sites (or all three) (Klein & Ito 1968); asimilar preference appears to be exhibited byzincian magnesio-riebeckite.

Gennaniunt

Sipovskii et al. (197D and Grebenshchikovet al. (1974 reported synthesis and charac-terization of a fibrous amphibole, NarMguGe"O::(OH)2, where Si is entirely replaced by Ge.As the ionic radius of Ge is slightly larger thanthat of Si, the cell parameters of the germanateamphibole are all slightly larger than those ofthe silicate analogue (Table 6l).

Fecrons ARrectrNc Cerron ORDERTNcIN AMPHIBOLES

The amphibole structures have a wide varietyof cation sites that can acceDt cations in therange 0.26-1.60 A in size and l+ to 4* informal charge. The extreme diversity of chem-istry that this allows, together with the similarityof symmetrically distinct cation polyhedra, en-sures that extensive solid-solution will occur

TABLE 6I. COMPARISON OF CELL PARAMETERS FORGER}4ANATE AND SILICATE

- AMPHIBOLES*NaTMUUSi Uor,

( 0H ), NarMOUGer0r, ( 0H ),

a(8)b(8)c(8)A ( o )

v(83)Space group

e .70 ( I )' r 8 .o r ( r )

5.28(2)' r 0 3 . 0 ( r )

e00(5 )C2/n

2.eg(z)

e .e8 (1 )1 8 . 3 r ( r )

' r 06.5 (2 )e35(ro)C2/n

4 . 0 7 ( 5 )*from

Grebenshchikov et a]. ( '1974)