LN Canadian Mineralogist Vol. 30, pp. 1n-136 (1993) A CRYSTAL.CHEMICAL INVESTIGATION OFALPINE GADOLINITE FRANCESCO DEMARTIN Istinto di Chimica Strumtristica Inorganica" Universit degli Studi, ViaG.Venezian 21, I-20133 Milan'Italy TULLIOPILATI Centro CNR perIo Studio delle Relazioni fra Strunura e Reattivit Chimica, viaGolgi 19, I-20133 Milan'Italy VALERIADIELLA Centro CNR di Sndio per la Stratigrafiae PetrografiadelleAIpi Centrali,via Bonicelli 23, I-20133Milan' Italy PAOLO GENTILE AND CARLO M. GRAMACCIOLI Dipanimentodi Scienze dellaTerra, Ilniversit degli Studi,via Bonicelli 23, I-20133Milnn' Italy ABSTRACI Gadolinite-(Y)specimens from variouslocalitiesin the Alps havebeen examined by electron microprobe and single-crystal X-ray diffraction. tn generat, dysprosium is the most abundant rare-earth, although a few samples containapproximately equal ulnount, of Dy undYb-, andin oneinstance, Gdpredominates.Incontrasttomanynon-Alpineoccunences, mostof these specimens showonly lirnited amounts of the lighter REE.Thereis an almostconstant presence of calcium (up to 4 wt7o. CaO' and.possibly twice thai amount for morequestiorible samples;; iron is often markedly deficientwith respect to the tleqretlcal formula,andin arleast one case (Glogstafelberg), the material should moreproperly be ialled hingganite-(Y) (4.0wrTo FeO).In some specimens, a silnificant substituiion of S fJi Be (up to about 4.ZwtUo UrOll canbe deduced-from crystal-structure data, on the basis of linear inteipolation of the measured Be-O''6ond lengfhswith reip".t to other gadolinite-group minerals.This substitution is more exteisive for specimens high in Ca andlow in Fe, andwhich therefore grade towarddatolite.No evidence for replacement of Si by B hasbeen iound. Minor amounts of thorium (up to 0.4 wtTo ThO2)commonly arepresent' anduranium (0.3 wtVo UO) was found in one specimen. As for xenotime and monazite, the behaviorbf Y is not uniquelydetermined by the ionic radius,some specimens beingespecially emiched in this element with respect to the middle-heavy rareearths (up to 4 I .5 wt%o Y 2O) . Keywords:gadolinite, hingganite, rare earths,yttrium, beryllium, boron, pegmatite, fissure, Alps, crystal-structure analysis' electron-microprobe analysis. SOMMAIRE Nousavons caract6ris6 plusieurs fchantillons de gadolinite-(Y) provenant de localit6s alpines par microsonde dlectronique et par diffraction X sur cristal unique.C'est en g6n6ral le dysprosium qui est la plus lbol9anre desterresrares, quoique certains 6chantillons contiennent une proporrion6quiialente Oe Oy et de Yb, et quoiquele Gd pr6domine dals y1-{es 6chantillons' Contrairement i plusieurs "*..pi". provenant d'ailleun, li plupartdesdchantillons ne.contiennent que de faibles quantit6s des terres iaresl6gdris. p calciumestpidsent dans presque tousleicas, en quantitds allantjusqu'd 4Vo de CaO,etpouvant atteindre le double dececidans certains 6chantillons moins bien caract6ris6s. Le fei est fortement ddfcitair,e pqq rypport ela formule id6ale; dffi"^ffi;i.s |1n 6iei"L$'i"1i;;6;F;ifiil;a;64;rii6fre o€G appeto nineganite-ffi @.MoF{'par pids). Dans cerains 6chanrillons, une proponion importante duBeest remptac6e par le B (usqu'l environ 4.2Vo deB2O3par poids), d'aprbs les donn6es obtenues sur la structure crlstaitine de cetteespbce, d la lumibred'une interpolation deslongueurs Be-O mesur6es pour certains de nos &hantillons, par rappor A d'autres min6raux du groupede la gadolinite.Cette substitution est plus r6pandrre pour les echantillons richesen Ca et pauwes en Fe, et doncceux qui montrent une tendance versla composition de la datolite.Nous ne trouvons aucun indice d'un remplacement du Si par le B. De faiblesquanritds deTh (usqu'i 0.47o deThO2par poids)sontassez courantes, et1.3vo de UO2 esipr6sent dans un de nos 6chantillons. Comme dansle cas du x6notime et de la monazite, le comportement de I'yttrium nesemble pas ddpendre uniquement du rayonionique; certains 6chantillons sont fonement enrichis en Y par rapport auxterres fares moyennes et lourdes cusqu'h 41.57o YzO:). (rraduit parla R6daction) Mots-cl6s:gadolinite, terres rares, ynrium, b6ryllium,bore, pegmatite granitique, fissure, Alpes,6bauche dela structure cristalline' analyse i la microsonde 6lectronique.
Transcript
LN
Canadian MineralogistVol. 30, pp. 1n-136 (1993)
A CRYSTAL.CHEMICAL INVESTIGATIONOF ALPINE GADOLINITE
FRANCESCO DEMARTINIstinto di Chimica Strumtristica Inorganica" Universit degli Studi, Via G. Venezian 21, I-20133 Milan' Italy
TULLIOPILATICentro CNR per Io Studio delle Relazioni fra Strunura e Reattivit Chimica, via Golgi 19, I-20133 Milan' Italy
VALERIADIELLACentro CNR di Sndio per la Stratigrafia e Petrografia delle AIpi
Centrali, via Bonicelli 23, I-20133 Milan' Italy
PAOLO GENTILE AND CARLO M. GRAMACCIOLIDipanimento di Scienze dellaTerra, Ilniversit degli Studi, via Bonicelli 23, I-20133 Milnn' Italy
ABSTRACI
Gadolinite-(Y) specimens from various localities in the Alps have been examined by electron microprobe and single-crystal
X-ray diffraction. tn generat, dysprosium is the most abundant rare-earth, although a few samples contain approximately equal
ulnount, of Dy undYb-, andin oneinstance, Gdpredominates.Incontrasttomanynon-Alpineoccunences, mostof these specimens
show only lirnited amounts of the lighter REE. There is an almost constant presence of calcium (up to 4 wt7o. CaO' and.possibly
twice thai amount for more questiorible samples;; iron is often markedly deficient with respect to the tleqretlcal formula, and in
ar least one case (Glogstafelberg), the material should more properly be ialled hingganite-(Y) (4.0 wrTo FeO). In some specimens,
a silnificant substituiion of S fJi Be (up to about 4 .Z wtUo UrOll canbe deduced-from crystal-structure data, on the basis of linear
inteipolation of the measured Be-O''6ond lengfhs with reip".t to other gadolinite-group minerals. This substitution is more
exteisive for specimens high in Ca and low in Fe, and which therefore grade toward datolite. No evidence for replacement of Si
by B has been iound. Minor amounts of thorium (up to 0.4 wtTo ThO2)commonly are present' and uranium (0.3 wtVo UO) was
found in one specimen. As for xenotime and monazite, the behaviorbf Y is not uniquely determined by the ionic radius, some
specimens being especially emiched in this element with respect to the middle-heavy rare earths (up to 4 I .5 wt%o Y 2O) .
Nous avons caract6ris6 plusieurs fchantillons de gadolinite-(Y) provenant de localit6s alpines par microsonde dlectronique et
par diffraction X sur cristal unique. C'est en g6n6ral le dysprosium qui est la plus lbol9anre des terres rares, quoique certains
6chantillons contiennent une proporrion 6quiialente Oe Oy et de Yb, et quoique le Gd pr6domine dals y1-{es 6chantillons'
Contrairement i plusieurs "*..pi".
provenant d'ailleun, li plupart des dchantillons ne.contiennent que de faibles quantit6s des
terres iares l6gdris. p calcium est pidsent dans presque tous leicas, en quantitds allantjusqu'd 4Vo de CaO, etpouvant atteindre
le double de ceci dans certains 6chantillons moins bien caract6ris6s. Le fei est fortement ddfcitair,e pqq rypport ela formule id6ale;dffi"^ffi;i.s |1n 6iei"L$'i"1i;;6;F;ifiil;a;64;rii6fre o€G appeto nineganite-ffi @.MoF{'par pids). Dans cerains
6chanrillons, une proponion importante du Be est remptac6e par le B (usqu'l environ 4.2Vo deB2O3par poids), d'aprbs les donn6es
obtenues sur la structure crlstaitine de cette espbce, d la lumibre d'une interpolation des longueurs Be-O mesur6es pour certains
de nos &hantillons, par rappor A d'autres min6raux du groupe de la gadolinite. Cette substitution est plus r6pandrre pour les
echantillons riches en Ca et pauwes en Fe, et donc ceux qui montrent une tendance vers la composition de la datolite. Nous ne
trouvons aucun indice d'un remplacement du Si par le B. De faibles quanritds de Th (usqu'i 0.47o de ThO2 par poids) sont assez
courantes, et1.3vo de UO2 esipr6sent dans un de nos 6chantillons. Comme dans le cas du x6notime et de la monazite, le
comportement de I'yttrium ne semble pas ddpendre uniquement du rayon ionique; certains 6chantillons sont fonement enrichis en
Y par rapport aux terres fares moyennes et lourdes cusqu'h 41.57o YzO:). (rraduit par la R6daction)
Mots-cl6s: gadolinite, terres rares, ynrium, b6ryllium, bore, pegmatite granitique, fissure, Alpes, 6bauche de la structure cristalline'
analyse i la microsonde 6lectronique.
t28 THE CANADIAN MINERALOGIST
INTRODUcnoN
Whereas gadolinite has been known for nearly twocenturies, its discovery as a fissure mineral in the Alpinereglon took place more recently (parker et at. li40).Because of its geologically "young,, age, gadolinitefrom the Alps is very different from that of most otheroccurences; it commonly is referred to as .,noble,'gadolinite, as the crystals are sharp and transparent, theircolor is generally pale green or blue-green, reminiscentof beryl or some bluish varieties of titanite, with whichsuch crystals of gadolinite may easily be confused.
As is the case for monazite and xenotime" there arealso occurrences in granitic pegmatites. Most of thesehave been discovered in Val Yigezzo (Ossola) (Mattioli1977, 1978, Turconi 1982, Albertini 1988); in theeastem Alps, occurences are known at the so-called"Plattenbri,iche''in the Rauris Valley (Meixner 1976)and in the Markogel pegmatite near Villach (G.Niedermayr, pers. comm., 1990).
Most of the scientific work on Alpine gadolinite hasbeen concerned with its mere identification. In a fewcases, optical and morphological data and X-ray powderpatterns are given; however, no structure refinement hasbeen carried out so far, and accurate values for theunit-cell parameters are lacking in the literature. Simi-larly, recent results of quantitative chemical analyses,including the distribution of the rare-earth elemenrs(REQ, ue not available.
The unusual perfection and "freshness" of the crys-tals, which are almost unique in nature, make th;mideally suited for chemical analysis and a study by X-raydiffraction. Advanced metamictization and leachingaccompanied by oxidation, commonly observed innon-Alpine gadolinite, might lead to grossly erroneousconclusions about the crystal chemistry of this mineral.For instance, a recent unpublished investigation carriedout by our group on the glassy core of crystals fromBaveno (which are much older than Alpine specimens,though found in the same vicinity) showed that there isan almost complete series of compositions between"true" gadolinite and an amorphous mass of hvdratediron oxide and silica.
Also, crystallographic data of good quality are rareon this species (Miyawaki et al. 1984). They can beimportant in view ofthe existence ofextensive solid-so-lution with other minerals of the same goup, e.g.hingganite (Y,Yb)2BqSi2O8(OH)2 (Semenov er a/.1963, Voloshin et al. 1983, Yakubovich et al. 1983\ oreven datolire CaBSiO4(OH), calcium-gadolinite,homilite Ca2FeBrSi2O,o (Miyawaki et al. 1985), etc.These solid solutions commonly involve compositionalvariation involving light elements such as Be or B, andpreselce of water. For this reason, electron-micronrobedata must be supplemented by other data in order toestablish the chemical composition exactlv.
Besides these well-known schemes oi substitution(Ito l967,Ito & Hafner 1974, Miyawaki et al. l9g4),
other types of substitution may not be excluded a priort.Among these, for instance, the presence of additionalberyllium replacing silicon has been postulated by someauthors, on the grounds of infrared absorption spectra(Aleksandrova et al. 1966); similarly, the presence offour OH groups substituting for SiOa, as in hydrogarneqis another possibility; furthermore, significant amountsofP and F [in view ofthe isostructural relationship toherderite CaBePOa@,Ott) and its groupl or of Bi [as inminasgeraisite (Y,Bi)2CaBqSirOlsl also may be pre-sent.
CHEMICAL ANALYSIS
Electron-microprobe analyses were performed onpolished grain-mounts, using the wavelength-dispersionARL-SEMQ instrument of the Italian National Re-search Council (C.N.R.) at the Centro di Studi per laStratigrafia e la Petrografia delle Alpi Centrali, Milan.
To determine the content of Si, Ca, Fe and U, a seriesofnatural and synthetic standards was employed. Fortherare earths, Y and Th, synthetic lithium metaborateglasses were prepared as specified in our contributionson Alpine monazite and xenotime (Mannucci et al. 1986,Demarttn et al. l99l a, b). The accelerating potential was20 kV, the sample current (on brass) 0.01 pA, and theMAGIC IV correction procedure was applied (Colby1968, with modifications). In all the samples hereexamined, the concentrations of Tb, Ho, Mg, Sc, P, S,Sr, 84 Na, F, and Al are below the limits of detection ofour instrument (about 0.1 wt%o);forTm and Lu, owingto interference from Dy, the limit is higher (about 0.4wtVo).
Besides the interferences among the REd the analyti-cal line of dysprosium @yfct) is too close to a line ofiron (FeKa), so that on one side the background was toohigh, leading to unacceptably low values for the DyrO.content. This interference has been accounted for bymeasuring the background ofDylcr on one side only.
Our data pertain to material from fwelve localities(Table 1). The variation in oxide percentages reportedincludes both statistical uncertainty in counting andactual compositional variation in the sample, since formost specimens the reported values are the average ofseven analyses, performed on different points. For thespecimens from Hopffeldboden and Triolet, owing tolack of homogeneity of the samples and also to theimpossibility ofrepeating the measurements in optimumconditions, the results are less accurate than for the otherspecimens. For example, the Y content is surely too high,and this also explains the unusually high value for theMelSi ratio for the specimen from Hopffeldboden (seebelow).
Unfortunately, t}re amounts of some lighter elementscould not be determined by our microprobe analysis;among these, boron and beryllium are particularlyimportant for gadolinite. This lack of data can be
A CRYSTAL-CIIEMICAL INVESTIGATION OF ALPINE GADOLINITE r29
TATI,E 1. ANAIITICA! DATE ON AIPIT'E GAI'OTJIISITE
se!,re (1) (2) (3) ({) (5) (6)
L a 2 O 3 0 . 2 ( r )
G z O : - 0 . 3 ( 1 ) 0 . 9 ( ? ) 0 . 4 ( 3 )
P r 2 o 3 O . 2 l 2 l
N d 2 o 3 0 . 3 ( 3 ) O . 1 1 2 ) 1 . 4 ( 9 ) 0 . 9 ( 5 ) 0 . 1 ( 1 ) O . 2 l 4 l
m z o : 0 . 4 ( 4 ) 0 . ? ( 2 ) 1 . 0 ( 5 ) X . 2 ( 6 ) 0 . 3 ( 2 ) 0 . 2 ( 3 )
d z o 3 1 . 4 ( 6 ) 1 . 8 ( 6 ) 1 . 8 ( 6 ) 3 . s ( 1 2 ) 1 . 4 ( s ) r . x ( 1 4 )
D y z o : 4 . 1 ( 1 ) 4 . 0 ( 1 ) 3 . 9 ( 4 ) 6 . 6 ( 6 ) 4 . s ( 5 ) 4 . 0 ( 9 )
(11) Trtolet, Mont Bl,anc, fmmftssm (PalsMom 1990) i (12) A\E
veglta, slmplol, lmm flss@: rc AlbErdnl (1980). $ This ftgre
ls not rallable, o*ing to ovesdmdon of Y. t Ratlo (atonlc) of
the metals ln ttre Y psldou to sl. tt nado (atonlc) of tbe !!! to
Y. g Rado (atomlc) of Ca to Y.
obviated using results from crystal-structure refinement(see below), which indicate common partial substitutionof B for Be. The presence of a realistic amount of boronin the Glogstafelberg sample was confirmed by asemiquantitative microprobe analysis performed with aXL SEM equipped with an EDAX energy-dispersionsDectromeEr.
Even for the most boron-rich specimens, the atomicratro Melsievaluated from our data (see the last entriesin Table l),where Me indicates Y plus the other metalssubstituting for it, remains close to the theoretical valueof 1.0, showing lack of extensive substituuon of B or Befor silicon; a similar conclusion is reached by consider-ing the values of the Si-O bond distances (see below) orthe mean-squarc arnplitudes obtained from crystal-structure refinement. Our results seem to contradict theassertions of some authors about gadolinite from othersources lsee forinstance: Oftedal (1964), Chdst (1965)]'but are in agreement with the findings of Miyawaki aral. (1984); moreover, in our case, because of the lowtemperature of origin, the possibility of such substitutionin the Alpine environment (especially for the fissurespecimens) is low. Full occupancy of the tetrahedralposition by Si may be a general rule for all specimens ofgadolinite.
The values for beryllium and water reported in TableI have been deduced from crystal-chemical considera-tions. For BeO (plus boron), a l:1 atomic ratio withresoect to Si has been assumed, because there is noevidence for appreciable substitution of Si by lighterelements.
Whereas the presence of Fe3+ was found in somenatural and synthetic members of the gadolinite seriesNakai 1938, lto l967,Ito & Hafner 1974), here theaverage Fe-O distances in all the samples we haveexamined by X-ray diffraction rangefrom2.ll 4to2.199A (see below). These values are in the usual range forFe2*, and this is confirmed by the absence of evidentoxidation in the samples. For these reasons, we haveconsidered iron to be exclusively in the +2 state.
On these grounds, and considering also the depletionof iron with respect to the theoretical end-member(see below), gadolinite from the Alps has been assumedto obey the general chemical formula: Y.+rCa2fe;,Be2-282.Si2O1 oHz,*zn+,, or (l-y)Y 2Q'2yCaO'(1-l)F eo.2SiOz.(2-22)BeO'zBzOr'("r+fz)H2O. Here' the pa-rameters r, , and e have been determined for each sampleby measuring the contents in iron, calcium, and boron,respectively, Unfortunately, since our determinatioa ofthe boron content is based on crystal-structure refine-ment, our general chemical formula can be used onlywhere the necessary X-ray-diffraction data are avail-able; for this reason, the entries relative to BrO3, BeO'HrO, and the total, are left blank for five samples inTable 1.
Whereas in general the stoichiometric ratios MelSiare satisfactory, the totals are low with respect to thetheoretical values. In any case, the totals in Table I can
130 THE CANADIAN MINERALOGIST
be raised if minor amounts of the less abundant i?EE(which are surely present, although below the detectionlimit) are considered. Another reason for the low valuesof the totals can be incipient metamictization (seebelow).
A general average of the REE distribution for all ourAlpine samples is reported in the frst column in Table2, together with separate averages for fissure andpegmatite samples (columns 2 and 3), the conespondingvalues for selected other (non-Alpine) samples, and theoverall crustal average.
The REE disribution clearly reflects the Oddo-Hark-ins rule (Oddo 1914, Harkins 1917), with rhe elemenmof odd atomic number less abundant than the corre-sponding elements of even atomic number. In thesamples we have examined so far, with two exceptionsonly, dysprosium is the mosr abundant REE. in thesample from Bosco, the content of Yb2O3 (in weight)equals that ofDy2O3, and in another case (Glogstafel-berg), gadolinium prevails. With respect to other occur-rences (Table 3), the geate$t majority of the Alpinesamples show a marked depletion of the lighterREE (Lato Nd) in favor of the other members of the series. Thereis only a minor difference between the specimens fromfissures and those from pegmatites: the heaviest REE areslightly more abundant in the former, and ytrium ismore abundant in the latter. A more sisnificant differ-ence between the two kinds of occurrenie is the contenrof boron, which is lower for the pegmatite specimens.
On the whole, t}re average distribution of REE (plusytrium) is not far from the general average given byAleksandrova et al. (1966)for one type ofoccurrence ingranitic pegmatites (compare column I with column 7in Table 2); with respect to the natural geochemicalabundance, this corresponds to a marked enrichment inthe middle-heavy REE (Dy, Er). The distributio n of RE E
TAILI 2. A COIIPARISON Or lt'ttE Rrf, DISTRIBUTXONIN A'IPINI CAIOITXNIFE, COI.iPAAED TO EEAT IN
EAIOLINITE FROM OTttER L@ATJITIES
( r ) a2 t (3 ) (4 ) ( r ) (6 ) (7 )
L62O3 0.1G 2 O 3 1 . O O . 4 I . 1Pr2oJ O.1N d 2 o 3 1 . 9 1 . 3 1 . 9s d 2 o 3 X . 9 1 . 3 X . so d 2 0 3 4 . 1 3 , 9 4 . 3D y 2 o 3 4 . 2 9 . 5 6 . 6E ! 2 o 3 6 . 2 6 . 9 4 . rY b 2 O 3 5 . 1 5 . a 3 . aY 2 O 3 7 0 . 2 7 O . 9 7 6 . !t / D y 9 a . 6 ? . s 1 1 . 5
Saqlrlss fen 1 to 12 are labelsd @ in Tabls 1. (13) Mtyamfd st a1.(1984), 12.08 wt6 FeO, no BzO3. (1.4) NtlBs€D (19?3) . (15) Seg8lstad &IaBs (1978), about 10 wtA FeO, 0.55 wtg 8203. (16) Voloshtn et al.(1983), 1.31 wtg FeO, uo 8203. (1?) geBerov et al. (1,963), niDeEl not)msd at ttrat tlBs;3,44w|%B2Ay 1.26 wtg FeO. (18) Mtyar€td et al.(X987), 5.65 fiq FeO, tncs of 8203. (19) lrllyawald er at. (1985),X?.03 wt$ f'eo. (20) S@enov et al. (1963), g.g0 rrg FsO. (21) Fott elal. (19?3). (22)Lagere Cibb8 (19?4). (23) F@!dstql. (1980).
in nonmetamict gadolinite from Hundholmen, Norway,reported by Nilssen (1973),looks even more similar rothe average of our data (column 5); rhere is, on the otherhand, a remarkable difference with respect to the othernon-metamict gadolinite from Japan studied by Miy-awaki et al. (see column 4), which is considerably richerin the lighter REE and poorer in yttrium. In our Alpinesamples, the greatest amount of lighter rare-earths hasbeen observed in the fissure sample from Glogstafelberg(5.5 wtVo),andthe next highestamount (4.2 wt%o),inthepegmatite sample from "Strada", Arvogno. These fig-ures correspond to 1 1.8 wt%o a\d 8.9 wt7o, respectively,of the total of the REE oxides (plus Y). The Glogstafel-berg sample also shows remarkably high amounts of Smand Gd (8. I and 12.6 w t%o of the total of the REE oxides),pointing out the unusual character of this occurrence;note that the REE distribution of hingganite-(Y) fromTuva (Semenov et al. 1963) is similar (compare columns8 and 9 in Table 2).
The range ofthe observed values for the Y2O3 content(21.5 to 41.5 wt%o) and the atomic REEX ratio varywidely (Table 1). The Y2O3/Dy2O3 ratio (in rerms ofweight), reported in the last line ofTable 2, in general is
Th€ va1u6, in wtt, aF @tculatsal on the lotal of the REE oxldss -Y2o3. (1) Total avengs of mr data; (2) avemgs ofiii aata forflsgure q-plesr wlth ex@pdon of the epeclnen frcm glogstafelberg,
A CRYSTAL-CTIEMICAL INIVESTIGATION OF ALPINE GADOLINTIE 131
higher than for the crustal average. This parameler isimportant because the ionic radii of Y3* and Dy3* arevirtually the same; accordingly, there is evidence forprocesses of fractionation depending not only upondifferences in ionic radius (see also Mannucci et a/.1986, Demartin et al. l99la). This is in line with ourobservations for xenotime, in particular concerning thesample from Bosco (Demartin et al. l99lb)i it is worthnoting that for the same locality, the Y2O3lDy2O3 ratioin gadolinite is 25.9. Another peculiarity of thegadolinite from Bosco is the virtual absence of all theREE lighter than Gd.
Among the elements subsdruting for Y and the REE,the most important is Ca generally present in quantitiesof the order of l-2uttVo CaO. There are also specimenscontaining up to 4 wtVo CaO (Val Nalps, Glogstafel-berg); a still higher figure (nearly 8 wt%o) has beenobtained for gadolinite from Triolet, but the analyticaldata are not as reliable, owing to the extremely small sizeof our crystal and to the possible presence of impurities.Unfortunately, the poor quality of the specimen pre-cluded an investigation by X-ray diffraction.
Small quantities of the actinides also are present'Thorium could be detected in four cases (about 0'3 wtToThO); in one case only (Monte Bassetta), a similaramount of UO2 also has been detected. From these data,no substantial difference between the specimens fromfissures and pegmatites can be observed.
In nearly all our specimens, the Fe content showssubstantial departure with respect to the ideal formula ofgadolinite (see below, Table 5). In one case, at least(Glogstafelberg), the occupancy of the Fe position isdecidedly less than 507o; therefore, the material shouldbe more properly ascribed to hingganite-(Y1, whosepresence in the Alps has not been documented so far. Itis interesting to note that this mineral is associated withmonazite-(Nd), another very rare REE mineral. For thisreason, t}le very peculiar occulrence of Alpine REEminerals, first mentioned by Graeser & Schwander(19S?), appears to be even more unusual. The depletionof Fe is in agreement with the results of refinement ofthe site occupancy (see below, Table 5); this rules outthe possibility for calcium to be present in the Fe site, asin minasgeraisite (Foord et dL l 986), since otherwise theinferred occupancy would be greater than the corre-sponding value determined by microprobe analysis.
Although X-ray data of good quality can be obtainedfrom singlecrystalsof many of ourAlpinesamples, evenwithout heating, signs of incipient metamictization seemto be indicated by the width of the X-ray-diffractionpeaks of some samples, even if these are perfectlytransparent and show sharp crystal faces. This spread hasbeen accounted for by selecting an appropriate scan-width in our collection of X-ray data (see the second linein Table 4).
Since the lowest totals for the compositions reportedin Table 1 correspond to the highest degrees of peakbroadening, these effects may, at least in part, both be
EAEIE it. DEIIAILB OF A-RAi DHrA-co:;l,Eellol AND RIFINEUEN!ot gllRocllrBB otr !!P!NE OAIOLIIfIIIE
o . 0 3 2 0 . 0 3 4 0 . o 2 2 0 . 0 2 a 0 . 0 2 0 0 . 0 3 0
D va lue ln ths
r e r q n t t n g s c n @ " s o . o s o 0 . o 4 0 o ' o 4 o o . o 3 o 0 . 0 4 0 o . 0 2 5 0 . 0 4 0
r c l & r . 1 r 8 1 . 2 5 4 1 . 3 8 ? 1 . 0 a ? 0 . 9 a 2 0 . 9 9 0 1 . 1 5 7
llha ""an
*!dth f,or aach leflectlon (degr6€6) !s equal to tro.3stano'
s n - r r r l r " l - t l e " l I / r l F o l t , ' a . - 1 x 1 j r o l - r l r c i t 2 1 * l r o l 2 ) r / 2 .
e cor1n1 | ro l -l I r" I ) 2/ (Nolu..u"tro.r-Nu.115se6 ) I 1/2.
tnet ghtt ng 6ch@s, Fl / (r lpo) ) 2, o(ro)-t l ( r ) + ( 9r l2 t t I 2 / 2No\P -
due to incipient metamictization. The reduction ofanalytical totals may be due to density variations: in fact,these phenomena already appear at the beginning ofthemetamictization process. In this respect, a classicalexample is provided by zircon (Holland & Goufried1955); another good example is shown by the mineralsof the ekanite group (Diella & Mannucci 1986' andreferences therein); for gadolinite, the unit-cell volumedecreases by about3%o after heating (Ueda 1957). Somesamples (no. 7) showing incipient metamictization donot seem to contain appreciable amounts of U and Th.However. such an occurrence seems to be well knownin this mineral (Ewing 1975).
CnYsral-SrnucruRE ANALYS$
Wherever possible, transparent fragments of singlecrystals of gadolinite were mounted on a NONIUSCAD-4 diflractometer, using MoKcr radiation (1,0 .7 107 3 A; . tntensity data for the structure refinementswere obtained from seven specimens; in addition tothese, for three samples, only unit-cell data could beobtained, from a least-squares fit of25 reflections with20 ranging from 30o to 38o; these are compared in Table3 with those of related minerals. The minor variationsare undoubtedly due to compositional differences.
For the boron-rich samples, the value of c tends to besmall; this is consistent with the differences in unit-cellparameters between gadolinite and either datolite orhomilite. The value of c seems to be particularlysensitive also with respect to the i?EE distribution. Forinstance, for the specimen from Japan (which is richerin lighter REE elements than our Alpine samples), thevalue of c is large; following this trend, it is still largerfor gadolinite-(Ce) from Norway (Segalstad & Larsenl97a). In any case, the volume of the unit cell of
aaap1€ (2,Y { ( 1 ) 2 . 3 0 6 1 2 1Y - O ( l ) ' 2 . 3 7 3 1 2 1r - o ( 2 ) 2 . 1 7 2 ( 2 1v - o ( 3 ) 2 . 6 A L ( 2 1Y - o ( 3 ) ' 2 . 4 5 9 1 2 1r - o ( 4 , 2 . 3 6 4 1 2 'Y - o ( s , 2 . 4 9 0 1 2 1y - o ( 5 ) , 2 , 4 O 6 ( 2 tY4(av€rage) 2.424t occupancy 1.1.7B . V . € @ 2 . A O
3 r - o ( 1 ) 1 . 6 0 4 ( 2 )s i - o ( 2 , 1 . 6 3 6 ( 2 )s t - o ( 3 ) 1 . 6 3 3 ( 2 )s 1 - o ( 4 ) 1 . 6 4 2 1 2 'Si-O(aY6!a9€) L.629
B e - o ( 2 ) 1 . 6 2 9 ( 3 )8 6 - 0 ( 3 1 1 . 6 4 r ( 3 )B € { ( 4 } X . 6 1 4 ( 4 )B s { ( s ) 1 . 6 0 4 ( 3 )Bd(averag€l 7.621B occutscy 0.16B . V . a @ 2 . 0 6Forul chuge& 2.f6
F e - O ( 2 1 x 2 2 . 2 e f ( 2 ,F 6 { ( 4 ) x 2 2 . 2 2 3 ( 2 1re-o(s)x2 2.O3sl2lF6-O(aver. I 2.190!e occupancya 0.66Ps occupscy- O.70B . V . s @ 1 . 9 6
The atom are l8boled ac@rding to lliyawau el al. (1984) i. tlF nmberilg of ths samplsa ls ths@o s l|r Table 1. " FrcE crlBtal-structure analysls;" D frcE results of electmn-nlcrcprcbeenalyass. Bond-valence au h v.u. (Brcm 1981). . Se ten.
gadolinite decreases with increasing content of theheavier REE elements, as expected from the lanthanidecontraction. As for hingganite, if the position of the Featom is not fully occupied, the unit-cell parameter ,tends to be larger, in line with the increase of the Fe-Odistances (see below, and Table 5).
For the crystals of best quality (irregular fragmentswhose diameters range from 0.1 to 0.2 mm, untreated byheating or other manipulations), reflections with amaximum value of the Bragg angle 0 of 35. with MoKaradiation have been collected. Detailed data about suchcollections are reported in Table 4. For all crystals,empirical absorption corrections were derived accordingto the 'psi-scan" technique of North et al. (19681,followed by the DIFABS routine (Walker & Stuan1983), as described by Demartin et al. (1992). Aftercorrecting for absorption, the disagreement factorsbetween the Fo values corresponding to symmetricallyequivalent reflections in the same crystal are quitesatisfactory (Table 4).
Scattering factors for neutral atoms, along withcorrections for anomalous dispersion, have been takenfrom Cromer & Waber (1974) and, Cromer (1974),respectively, including the real and the imaginarypart. For &e REE, we have assumed the scattering factorof Y. Because of this assumption, and the partial
replacement of Y by the heavier REE, the occupancyfor yttrium is always greater than unity, even whereappreciable quantities of calcium also are present(Table 5).
The refinement was carried out separately for eachcrystal by full-marix least squares, minimizing thequantity Zw(Fo-F")z, and considering a total of 82variables. These include symmetry-unconstrained positional and anisotropic thermal parameters for all theatoms, plus the scale factor, an overall extinctionparameter and the occupancy for Y and Fe. The atomiccoordinates of the starting model were those of Miyawa-ki et al. (1984). The weighting in the last cycles, the finalvalues of the R index and the corresponding weightedindex rR* are reported in Table 4. Fractional atomiccoordinates, anisotropic thermal parameters and tablesof structure factors are deposited at CISTI, NationalResearch Council of Canada, Ottawa. Ontario KIA 0S2.All the calculations were performed on a PDPlll73computer using the SDP-Plus Structure DeterminationPackage (Frenz et al. 1980). In the final Fourierdifference syntheses, no peaks exceeding 1.1 - 0.6 e/A3have been found. These peaks are always close to theREE position and do not imply presence of additionalatoms.
Interatomic distances are reported in Table 5, together
A CRYSTAL-CTIEMICAL INVESTIGATION OF ALPINE GADOLINTTE 133
with values of the e.s.d. The main features of thestructure are in close agreement with the conclusions ofthe earlier works on tle same subject, especially withthose of Miyawaki et al. (1984), which is taken here asthe main reference. The minor differences we haveobserved are significant but can easily be explained. Forinstance, the average (Y+i?EE")-O distance in our sam-ples is 2.428 A, a smaller value than the conespondingaverage (2.437 A) found by Miyawaki et al. Tttis isclearly related to the difference in the proportion of theREE, our samples being poorerin the lighterREE, whoseionic radius is comparatively large. This result of thelanthanide contraction is evident even within our set ofsamples, the smallest (Y+REE")-O distance (2.420 L)and the largest one (2.438 A) being observed for thesamples from Bosco and from Glogstafelberg, respec-tively (Table 5). Among the specimens for which dataare reported in this table, the former is the poorest in thelighter lanthanides, and the latter is the richest. In somecases, these average distances and especially t}te occu-pancy of Y also are affected by the presence ofsignificant amounts of Ca; the smallest figures for the Yoccupancy correspond in fact to the specimens that arerelatively rich in Ca and Y and poor in the lanthanides.
In general, the Fe-O distances of our samples are veryclose to the values obtained by Miyawaki et al. (1984)athe averages for ou1 specimens can be compared withthe average (2. I 82 A) obtained by these authors. For theGlogstafelberg sample, these distances are slightly butsignificantly larger than for all the others. This differ-ence is probably due to the absence of a large part of theiron atoms, whose presence would counteract the elec-trostatic repulsion between the neighboring atoms ofoxygen.
The average Si-O distances for each one of ourcrystals (see Table 5) do not show remarkabledifferences from the general average of al1 our data(1.629 A); the corresponding values inMiyawak:, et al.(1984, 1985) for gadolinite and homilite (1.633 and1.634 A, respectively) are only slightly larger, andthe difference may not be significant. With respectto the range of average length of the Si-O bond^inwell-documented orthosilicates (1.628-1.655 A),as for instance in the olivine-monticellite group(Brown 1970) or the garnet group (Novak & Gibbs1971), our values are situated at the shorter extreme,but there is substantial agreement, taking thepossible overall influence of the crystal structure intoaccount.
Our data thus confirm the virtual absence ofreplace-ment ofSi by other elements, especially boron. Further-more, since in the case ofthe presence of any substituentin two different structural sites a partition coefficientbetween these sites is implied, the highest amounts ofboron substituting for Be should correspond to thehighest concentrations in the Si site; if these concentra-tions are significant, a considerable shrinking of theSi-O bond-length should result. However, no correla-
tion between the Be-O and the Si-O average bond-lengths has been observed.
In the Japanese material studied by Miyawaki et al.(1984), the average length of the Be-O bond is 1.634(2)A. This value is significantly largerthan the correspond-ing results for all our Alpine specimens; similarly, thereare significant differences even within our set of sam-ples, in contrast with the uniform values for the Si-obond lengths (see Table 5). The most plausible explana-tion for all this involves partial replacement of Be byotherelements. Of these substituents, Si canberuledout,since the Be-O and the Si-O bonds are very similar inlength. Instead, boron is the most likely candidate, inview of the structural relationships between gadoliniteand datolite. For the latter mineral, the average of B-Odistances is 1.480(l) A @ant A Cruickshank 1967, Foitet al. 1973), avalue much shorter than the average value(1.65 A) reportedby Ondik & Smith (1962) fortheBe-Obond length; for lromilite, the corresponding averagedistance is 1.504 A (Miyawaki et al.1985).
Since in datolite and in many members of thegadolinite series one atom [corresponding to O(5)] istotally or partially replaced by an OH group, theshrinking effect due to the substitution ofBe by B couldbecome less clear, because the bonds with OH woulddiffer in length from the corresponding bonds withoxygen. For this reason, a better result can be achievedif the Be-O(5) bond is not considered, and the averageof the remaining three Be-O bond lengths is used forcomparison. This- average amounts to 1.475 A fordatolite. to 1.530 A for homilite, and to 1.648 A for theJapanese gadolinite, respectively (Foit et al. 1973,Miyawaki et al. 1984, 1985), the last value beingpractically identical to the general average of Ondik &Smith (1962). In line with these results, the Be-O(5)distance for the Glogstafelberg sample [1.622(5) A] isremarkably different from the correspondtng distancesfor all the other samples, including Nos. 2, 3 and 6, whichalso are B-rich (1.595 to 1.604. A, respectively). Thisconfirms the possibility of the presence of a significantamount of OH replacing O, thereby supporting ourassumption of a relatively high water content: a similarconclusion can be drawn on the basis of a bond-valencesummation (see below).
By carrying out a linear interpolation with respect tothese end members of known composition, the occu-pancy factor for B (in the Be position) reported in Table5 has been calculated for all our specimens for whichadequate crystal-structure data are available. This indexcorresponds to a B2O3 content ofthe order of l-3 wtVo,with a maximum of 4.2 wtVo for the sample from Beura.These figures are not surprising; according to Oftedal(1964), the B2O3 content of 81 samples of gadolinitefrom 22 different localities ranges from 0.05 to 2.5 wtEo,with an average of 0.2wtVo.In most cases, B2O3 rangesbetween 0.05 and O.5 wt%o; another study by Aleksan-drovaet al. (1966) showed appreciable amounts ofB2O,(up to nearly 5.0 vtt%o) in gadolinite samples from the
t34 TIIE CANADIAN MINERAI-OGIST
USSR. Therefore, our results for the Alpine specimensare in the range of values already reported in theliterature for gadolinite.
On applyingbond-valence sums (Brown 1980), thereis a general, although only qualitative, agreement withthe observed substitutions, as can be expectod for partlydisordered structures. For instance, for the Be site, thissum ranges from 2.02 to 2.15, and the highest valuescorrespond to the highest contents ofboron (see Table5); however, if the bond-valence s is assumed to be aweighted average for Be and B (s = 2 and 3, respec-tively), the values are slightly different (see the next linein Table 5). A similar situation occurs for Fe. where thelowest valence-sum (1.87) is obtained for the samplefrom Glogstafelberg, which is the poorest in Fe. ForO(5), whose bond-valence sums are reported in the lasrline of Table 5, the lowest value ( l.2l ) is also relarive rothe sample from Glogstafelberg, which, according to ourchemical considerations, should be the richest in water(OH). However, t}re corresponding values for O(5) in theother samples never exceed 1.58, even for the OH-poorspecimens. For the Y site, the bond-valence sums rangefrom2.69to2.83; here, the correlation with the chemicalcomposition is less clear.
The partial replacement of Be or Si by other elementscould also be reflected in the refined values of theoccupancy factor; however, attempts in this respect havefailed to give reliable results. Similarly, ifthe occupancyis not refined, the average (apparent) mean-squaredisplacement should show appreciable variation withrespect to the 0normalO value. However, the thermalparameters are difficult to compare and interpret, sincethey vary considerably from one sample to another,showing the combined effects of difference in numberofelectrons, ofvariation in bond lengths, and ofresidualabsorption.
The bond angles around Si range from about 106o to117o, and tlose around Be, from l01o to 116., inagreement with the tendency of oxygen-bonded beryl-lium to form less regular tetrahedra than either silicon orboron.
ACKNoWLEDGEMENTS
This work has been made possible by financialassistance from the Italian National Research Council(C.N.R.). We are grateful to a number of collectors, mostof whombelongto the Gruppo Mineralogico Lombardo,who provided us with most of the necessary material,often sacrificing unique specimens. These are: Messrs.Claudio Albertini, Domenico Preite, and Franco Vanini.Special thanks are due to Mr. Ronald Winkler fromBbckstein and to Dr. Gerhard Niedermayr for the Ausfianspecimens, and to Mr. Alex Kipfer for the samples fromSwitzerland. We are also indebted to Dr. Andrea Valdrdfrom the Analytical Department of Philips (Milan) forhaving performed the qualitative analyses forboron. Dr.Michael Fleischer kindly supplied us with a very useful
list ofreferences. The authors are also grateful to Dr. G.Robinson from the Canadian Museum of Nature as wellas to R.F. Martin for invaluable help and advice.
BnowN, G.E. (1970): The Crystal Chemistry of the Olivines.Ph.D. thesis, Virginia Polytechnic Institute and State Univ.,Blacksburg, Virginia.
BRowN, I.D. (1981): The bond-valence method: an empiricalapproach to chemical structure and bonding. /r, Structureand Bonding in Crystals II @4. O'Keeffe & A. Nawotsky,eds.). Academic Press, New York (1-30).
CelaNcHr, N. (1979): Brevi segnalazioni: gadolinite di Beura(Val d'Ossola). Riv. Mineral. Ital. 1,49-50.
Crnrsr, C.L. (1965): Substitution of boron in silicate crystals.Norsk Geol. Tidsskr. 45" 423428.
- & Wasrn, J.T. (1974): Intemational Tables for X-RayC ry s tal l o g r ap hy 4, T able 2.2.b. T\e Kl,noch Press, Birmin-gham, U.K. (present distributor: Kluwer Academic Publish-ers, Dordrecht, Holland).
DEMARTTN, F., Gnavaccror-r, C.M. & Pu-en, T. (1992): Theimportance of accurate crystal-structwe determination ofuranium minerals. II. Soddyite (UO2)2(SiO4)(2HrO. ActaCrystallogr. C48,I-4.
-, Prant, T., Drplm, V., Dor.rzsllr, S., GENrr-E, P. &Gneuacctot-t, C.M. (1991b): The chemical composition ofxenotime from fissures and pegmatites in the Alps. Can.Mineral. 29.69-75.
& Gnarr,raccroLl C.M.(1991a): Alpine monazite: further data, Can. Mineral. 29,6t-67.
DELLA, V. & MANNUccr, G. (1986): A uranium-rich ekanite,(Ths.76,Ue.21)(Ca2.6 1,Fee.ea,Mnn.6 1 )Si7.eO26, from Pitiglia-no, Ialy. Rend. Soc. Ital. Mineral. Petrogr.4l,3-6.
A CRYSTAL.CHEMICAL INVESTIGATION OF ALPINE GADOLINITE 135
EwrNc, R.C. (1975): The crystal chemistry of complex niobiumand tantalum oxides. IV. The metamict state: discussions.Am- M ineral. 60. 7 28-7 33.
Forr, F.F., Jn., Pstt-t-tps, M.W. & Gtsss, G.V. (1973): Arefinement of the crystal structure of datolite,CaBSiO4(OH). Am Mineral. 58, 909-914.
Foono,8.E., GANES, R.V., CRocK, J.G., SnrvoNs, W.8., JR.& Bensose, C.P. (1986): Minasgeraisite, anew member ofthe gadolinite group from Mina.s Gerais, Brazil. Am.Mineral. Tl,603-607 .
Fnsrz, B.A. & Assocnres (1980): SDP Plus Version 1.0.Enraf-Nonius. Delft , The Netherlands.
GRAESER, S. & SCHWANDER, H. (1987): Gasparite-(Ce) andmonazite-(Nd): two new minerals to the monazite groupfrom the Alpl Schweiz Mineral. Petrogr. Min. 67, 103-I 13 .
IIARKTNS, W.D. (1917): The evolution of the elements and thestability of complex atoms..L Arn. Chem. Soc,39,856-879.
Hou-eNo, H.D. & GorrFRrED, D. (1955): The effectof nuclearradiation on the structure of zircon. Acta Crystallogr. 8,291-300.
Iro, J. (1967): Synthesis of calciogadolinite. Am. MineraL 52,1523-15n.
- & HAFNER, S.S. (1974): Synthesis and study of gadoli-nites. Am. Mineral.59, 7m-708.
Kprsn, A. (1979): Die Mineral-Paragenesen im Stollenprofildes Furkabasistunnels, Fenster Bedretto (Ronco, TD.Schwe iz St rahle r 5, 45-92.
KoNrRUs, K. (1965): Die Funde der Beryllium-MneralienPhenakit, Milarit u. Gadolinit in der Ostalpen. Der Auf-schluss 16,7U75.
Lacen, G.A. &GrBBs, G.V. (1974): Arefinementof thecrystalsffucnrre of herderite, CaBePoa(oH). Am Mineral. 59,9t9-925.
Merrwuccr, G., Dmm, V., GRAMAccloLI, C.M. & Pu.an, T.(1986): A comparative study of some pegmatitic and fissuremonazite from the Alps. Car. Mineral, 24,4W74.
Marnon, V. (1977): Euxenite e tanteuxenite in ValleYigezzo.Riv. M irw ral. hal. 8. 9L98.
- (198): Gadolinite in ValleYigezzo. Riv. Mineral. Ital.9"7-9.
MsDo{E& H. (1976): Gadolinit und andere Berylliummineraleaus den Plattengueisbriichen der Rauris (Salzburg), miteiner zusammen-fassenden bersicht ber die alpinen Beryl-liumminerale. D er Aufschluss TI, 309-314.
Mrvawerr, R., N.a.rat, I. & Necnssue, K. (1984): A refine-
ment of the crystal structure of gadolirtrte. Am- Mineral,69,948-953.
OKAMoro. A. & Isose, T. ( I 987): Thefirst occunences of hingganite, hellandite and wodgtnite inJapan. Kobutsugaku hsshi,18, 17-30 lin Japanese, absr.in Am. Mineral. 75,432 (1990)1.
NAKAI, T. (1938): On calciogadolinite, a new variety ofgadolinite found in Tadachi, Nagano, lapan. Bull. Chem.Soc. Japan 13, 591-594.
NrrssEN, B. (1973): Gadolinite from Hundholmen, Tys[ord,North Norway. Norsk Geol. Tidsskr. 53,343-348.
NonrH. A.C., Pnn-ltps, D.C. & Scor-r MArnEws, F. (1968): Asemi-empirical method of absorption cofiectior,. Acta Crys'tallogr. A24,351-359.
Novar, G.A. & Gtsss, G.V. (1971): The crystal chemistry ofthe silicate gamets. Am. Mineral. 56,791-825.
Onno, G. (1914): Die Molekulantruktur der radioaktivenAtome. Z. anorgan. allgem" Chem. n, 253-268.
O.V. (1983): Hingganite-(Yb), a new mineral from amazo-nite pegmatites of the Kola Peninsula. Dokl. Akad NaukSSSR 270, I 188-l 192 [in Russ.; for abstr., see Am. Mineral.69,811 (1984)1.
Walrnn, N. & SruARr, D. (1983): An empirical method forcorrecting diffractometer data for absorption effects. ActaCrystallogr. A39, 158-166.
Wntcen, R. (1987): Berylliummineralien aus der Umgebungvon Bckstein. l.apis l2(2), ll-14.
YAKUBovrcH, O.V., MATvENKo, E.N., VoLosH[.i, A.V. &Srvrolov, M.A. (1983): The crystal structure of hingganite-(Yb), (Y0.5rTR036Can.13)Fq.665BelSiO+l(OH). Sov. Phys.Cry stallo gr. 28, 269-27 l.
Received October 27, 1991, revised manuscript arceptedApril22,1992.
