American Mineralogist, Volume 71, pages 1426-1433, 1986

Crystal chemistry of two coexisting K-richterites from St. Marcel(Val d'Aosta, Italy)

ANrNrnA.r-n Morrl.NlDipartimento di Scienze della Terra, Sezione Mineralogico-Cristallografica, Universiti di Roma "La Sapienza,"

Citti Universitaria, 00185 Roma, Italy

Wrr-r-rnvr L. GnrrpruMineralogisk-Geologisk Museum, Sars' Gate I, N-0562 Oslo 5, Norway

Ansrnrcr

Two coexisting F-bearing K-richterites from the blueschist-facies metamorphic man-ganese ore deposit St. Marcel (Val d'Aosta, Italy) have been studied by combined ernrnand n anall'tical techniques. Despite their diferent colors (pink and blue) and cell dimen-sions, these amphiboles have formulae summing up to near l6 cations pfu on the basis of22O + 2(OH + F) and are chemically almost identical. The major differences are in theF/OH ratio and the Ca content, and significant minor differences occur in Fe and Ti.

The rn spectra in the OH-stretching region are notably different, probably as a result ofthe different F contents. However, both amphiboles have a prominent absorption peak at3731 cm ', typical of the K-OH-(MgMgMg) configuration, plus weak bands at 3672 and3653 cm ', typical of configurations having the A site empty. Moreover, pink K-richteritehas a strong absorption band at 3715 cm-' that was assigned to the Sr-OH-(MgMgMg)configuration, balancing the A sites that are empty. Site assignments based on eupa andIR compare well with those determined by crystal-structure refinement. However, they alsoallow detection of empty A sites in the structure.

INrnooucttoN

Although amphiboles are widespread in the Earth's crust,K-richerite may be considered a rare mineral. Neverthe-less, it has one of the widest experimentally determinedP,Z stability fields among the amphiboles (Kushiro andErlank, 1970; Gilbert and Briggs, 1974;Hariyaetal., 1974;Gilbert et al., 1982). These experimental data are consis-tent with the occurrence of K-richterite, which ranges fromhyperpotassic volcanic or subvolcanic rocks (Carmichael,1967;Velde, 1967;Thy,l982) to the upper mantle. Themost common occurrence of K-richterite is in ultramaficxenoliths from kimberlites (Erlank and Finger, 1970; Aoki,I 9 7 5 ; Dawson and Smith, | 9 7 7 ; Erlark and Rickard, I 9 7 7 ;Jones et al., L982), where it is believed to have formedby infiltration metasomatism in the mantle, prior to en-trainment of the xenoliths in the kimberlite.

K-F-richterite is even rarer and was defined as a naturalspecies only recently, in contact-metamorphic ejecta fromthe Monte Somma volcano (Della Ventura et al., 1983),although F-bearing K-richterites have been known for sometime (Olsen etal., 1973; Delaney et al., 1980; Jones et al.,1982). This apparent rarity may be due to a lack of Fanalyses.

Nothing is known about the mutual relationships be-tween F- and OH-bearing K-richterites, nor between theseand richterite. All four endmembers have been synthe-sized (Huebner and Papike, 1970), but the thermody-

namic properties of the solid solutions have not beenstudied completely.

F-poor and F-rich K-richterites occur in contact and inapparent equilibrium within the St. Marcel manganesedeposit, Val d'Aosta, Italy. Only richterite was previouslyknown from this deposit (Rondolino, 1936; Martin-Ver-nizzi, 1982); these K-bearing amphiboles are the productsof blueschist-facies metamorphism and represent the firstoccurrence in such a geologic environment. However, moreinteresting is the apparent equilibrium relationship ob-served between the phases, as this implies the possibilityof a compositional gap between phases difering solely intheir volatile contents.

Pnrnor,ocrc SETTTNG

A description of the Praborna orebody in the St. Marcelvalley has recently been presented (Griffin and Mottana,1982). The present sample was collected from the dump;thus the geologic relationships are not known. The chipavailable for study (14 x l0 x 5 mm) consists mostly ofgranular braunite intergrown with quartz, as well as a clearquartz zone containing a cluster of pink and blue prisms.The splinter is presumably a part of the main ore, crossedby one ofthe numerous quartz veins containing both high-Pand low-P minerals (Griffin and Mottana, 1982; Martin-Yernizzi,1982\.

The individual prisms are pink next to quartz and blue

0003-004x/8 6 / | | | 2-t 426502.00 t426

Table 1. Unit-cell parameters and physicalproperties of St. Marcel K-richterites

VM-9 pink VM-9 blue

9.9951 (3)1 7.971 0(8)5.2715(11

104.743(2)91 5.70(1 0)

3.0403.0s(2)1.606(3)1.623(3)1.635(3)

80(3)18(2)

Note: Unit-cell parameters from L. Ungaretti (pers.comm.). Values in parentheses refer to estimatedstandard deviation of the last dioit.

in the center ofthe cluster. The blue phase is never ob-served in contact with quartz. The elongate crystals (1.0-1.5 x 0.5 mm) are in parallel alignment,with mutual con-tact occurring along prism faces. They are either pink orblue; no intermediate shades or zonations in color havebeen observed.

The aggregate is coarse but not tough; individual grainsup to 0.5 mm in size can be extracted with gentle needlepressure.

X-na,y AND PHYSTCAL DATA

The amphiboles were identified from their unit-cell pa-rameters. Those listed in Table I were obtained by therer routine on an automated single-crystal diffractometer,a preparatory step toward the refinement of the crystalstructure (courtesy of L. Ungaretti). Procedure and in-strumentation are as described by Ungaretti et al. (1983).

Like other richterites, the St. Marcel amphiboles haveparticularly large a parameters (Cameron and Papike,1979). The a parameter also shows the largest differencebetween the two grains, but the values for b and c agreewithin two standard deviations; the larger I/ of the bluevariety is mainly a result of its larger a dimension.

When compared with synthetic F- and OH-K-richteriteendmembers (Huebner and Papike, 1970; Cameron et al.,1983), these richterites show a and V intermediate be-tween the two endmembers (Fig. l), such that the F/(OH +F) ratio in the pink and blue variety would be >0.5 and<0.5, respectively, provided that no Na is present at theA site.

Optical data were obtained with the spindle stage, Car-gille liquids, and white light on grains different from thoseused for single-crystal X-ray and eura work, from thesame batches used for rn spectroscopy. The refractive in-dices of blue richterites are systematically higher thanthose of pink richterite, as is the extinction angle,but2Y*is smaller (Table l). Density was measured on the samegrains using the floatation method in Clerici solution; thepink richterite is denser than the blue one (Table l).

1427

a (A )

1 0 . 0 4 9 1 0 0 5 0

1 0 0 0 0

9 . 9 4 4

K - O H - R | 2 s s o 7 s K - F - R

Fig. l. Variation of the a unit-cell parameter in syntheticrichterite endmembers (Huebner and Papike, 1970; Cameron etal., 1983) and in natural K-richterites from St. Marcel (P, pink;B, blue: present work) and Monte Somma (MS: Della Venturaet al., I 983). The lighter circles for B and P represent the F/(F +OH) contents derived from the chemical analyses.

CHevrrcar, coMPosrrloN

The same grains used for structure refinement were ana-lyzed by eps electron microprobe. Methods, standards,and precision are given in Griffin and Mottana (1982). Inaddition, K and Sr were analyzed, using orthoclase andcelestite as standards. F was measured by wos micro-probe, using a fluorapatite standard. The pink grain waslost prior to the quantitative determination of F. Anothergrain from the batch used for optics was analyzed usingthe blue amphibole as a secondary standard. The analysesare given in Table 2. If Mn and Fe are expressed as MnrO,and FerOr, as is probably the case for minerals in equi-librium with braunite + quartz, and if the amount of HrOis calculated to balance the formulae, totals in the range99.4-99.7 wto/o are obtained.

Both analyses show high SiO, and MgO contents, butcontain little AlrO, and MnO (despite their coming froma manganese ore) and only traces of FerO, and TiOr. Thedifferences in FerO, and TiOr, although apparently minor,are analytically significant, whereas the variations in SiO,and MgO are not (cf. Grifrn and Mottana,1982). Alkalisare present in nearly equal amounts; KrO always exceedsNa,O and is particularly high for an amphibole, althougheven higher values are known from amphiboles in potassiclavas (Carmichael, 1967). The original richterite fromLingban, Sweden (Michaelson, 1883), also contained KrOin excess of NarO.

Among richterites with measured F, the blue grain isthe amphibole richest in I!O. Prider's (1939) "magno-phortite," although reported by Deer et al. (1963) as aK-richterite with 1.29 wto/o F, is in fact a K-fluor-mag-nesio-katophorite according to the IMA classification(kake, 1978).

The formulae calculated on the anhydrous basis of 23oxygens (Robinson et al., 1982) give sums very close tol6 cations, with a slight excess for the blue grain (probably

MOTTANA AND GRIFFIN: TWO COEXISTING K-RICHTERITES

a (a)b (A)c (A)B f )v(A1D(g cm 3)D."""q

B̂l

2V, (")Zlrc

10.0220(411 7.9810(9)5 2750(21

1 04.690(3)91 9.s1(13)

3.0243.03(2)1 .615(3)1.630(3)1.640(3)

76(3)20(2)

r428

^\, Fl o

J

8o

o 1

I K - F - R |

r M S

r o P' o

o

o ' r,s

r t\'o o

o

K R i

o

oB

o

@

o . G 1 . 0

K / ( K . N a A )

Fig. 2. Compositional variation of richterites with known Fcontent (data after Charles, 1977:' Delaney et a1., 1980; Jones etal., 1982; and Della Ventura et al., 1983; MS); (P, pink; B, blue:present work)

due to overestimation of silica) and negligible deficiencyfor the pink one. We therefore normalized cations to 16,i.e., all sites filled. The deviations in stoichiometry are sosmall that we consider these richterites to have completeSi occupancy of the T position. This is in marked contrastto the richterites from upper-mantle rocks, in which de-ficiencies at the tetrahedral position are often so large asto remain unfilled even after allocation of all available Aland Fe3* (Dawson and Smith, 1977; Jones et al., 1982).

In the St. Marcel richterites, the C group cations arealmost entirely Mg, Mn being the second major cationpresent. Note, however, that minor Ca is required to fill

MOTTANA AND GRIFFIN: TWO COEXISTING K-RICHTERITES

I

o- 0 E

the five C sites, a procedure that, although unusual, isconsidered to be "reasonable" by Robinson et al. (1982)for alkali amphiboles. Allocation can continue by fillingthe B group almost equally with divalent (Ca, Sr) andmonovalent (Na) cations, leaving only a little Na to enterthe A site, together with the dominant K.

The blue grain has F(OH + F) : 0.44 and shouldtherefore be considered an F-bearing K-richterite; the pinkone has F/(OH + D : 0.62, thus meeting the requirementfor being classified a K-F-richterite (Fig. 2). As none ofthe kimberlitic K-richterites exceeds 0.5 in F/(OH + Dratio, this is only the second known example of a K-F-richterite, after the Monte Somma type occurrence (DellaVentura et al., 1983), and the only known high-P occur-rence.

All the allocations suggested so far have been madeaccording to standard, "no(nal" or "reasonable" crystal-chemical considerations (Hawhorne,1982a, p. 69; Rob-inson et al., 1982, p. 7), the only addition being that ofcombining Sr and Ca, as seems reasonable on the basisof valence and ionic radius. However, the formulae givenin Table 2 do not consider the possibility that local va-cancies may be present. Thus, although stoichiometricallycorrect, these formulae may be subject to changes in thecase that other, independent methods of establishing siteoccupancy yield different values (Hawthorne, 1983b,I 983c).

Ixrn,lnnn sPEcrRoscoPY

In the decade 1965-1974, infrared spectroscopy foundextensive application in studies aimed at assigning thevarious cations present in the amphibole formulae to sites

Table 2. Microprobe analyses and calculated formulae of St. Marcel K-richterites

Analysis (wt%) Atoms oer formula unit

1 ' 21

sio,Alr03

Tio,FerO"MnrO.MgoCaOarh

NaroK"OFHrO

Total

56.74 57.140.59 0.390 06 0.000.00 0 .100.67 0.53

22.89 23.036.94 5.310.63 0 934 10 4.654.88 5 .161.401 1.00(0.81) (1.1s)99.71 99.43

7.953 8.0240.0478.000 8.0240.050 0.0650.006 0.0000.000 0.01 10.071 0.0574.779 4.8180.094 0 025s.000 4.9760.948 0.7720.051 0.0761.001 1 .1522.000 2.0000 . 1 1 3 0 1 1 00.872 0.9220.985 1.032

1 5.985 16.032

0.620 0.444(0.380) (0.556)

7.960 8.0050.0408.000 8.0050.058 0.0650.006 0.0000.000 0.0110.071 0.0574.783 4 8060.082 0.0565.000 4.9950.961 0.7410.051 0 0750.988 1 .1842.000 2.0000j27 0.0780 873 0.9221.000 1 .000

16.000 16.000

0.620 0 444(0.380) (0.5s6)

b l

AIVI

2TAlvl

TiFeMnMgCac>cCa"SrNaB

>BNa^K

>CAT

FOH

Notei Samples are (1) VM-g pink (four spots) and (2) VM-g blue (three spots). Values in parenthe-ses are calculated.

. On 23-oxygen basis. -'On 16-cation basis. t wD-Er,rp analysis on a different grain (error + 0.20).

f r e q u e n c y ( c m - t )

3 7 0 0 3 6 0 0

Fig. 3. The hydroxyl-stretching region of the St. MarcelK-richterites (P, pink; B, blue) and of the Monte Somma typesample (MS). Note the variation of the transmission scale foreach sample.

in the structure. However, after Strens' (1974) review, themethod fell partly into neglect, because the many prob-lems involved had become apparent. Nevertheless, the rnspectrum gives a variety of qualitative and semiquanti-tative information, so that the technique retains its use-fulness (Hawthorne, I 982b).

The rn spectmm of amphibole consists of two mainregions: a low-frequency region (1200-200 cm-r) wherethe vibrations typical of the structure occur; and a high-frequency region (3800-3600 cm '), characteristic ofthestretching (valence) vibrations of hydroxyl. The hydroxyl-bending region is less important as it is difficult to separatefrom the lattice vibrations; in addition, it also may overlapwith the lattice vibrations of adsorbed water, which te-naciously adheres to the amphibole fibers (Strens, 1974).Therefore, efforts toward the quantitative determinationof site populations have been concentrated essentially onthe high-energy region.

The rn spectra ofblue and pink richterites were recordedon a Perkin-Elmer model 983,{ grating spectrometer. Theyare given in Figures 3 and 4, together with a new spectrumof the K-fluorrichterite from the type locality, rerun on

1429

Fig.4. tn. *n"1,'ll,un,.. tourll,o,', ".,,i.t}*r"

Sommatype K-F-richterite (MS), the blue richterite (B), and the pink (P)richterite from St. Marcel. Note the changes in the transmissionscale, but the similarity of the frequency position.

the original disk supplied by Della Ventura et al. (1983).To avoid overcrowding ofthe figures, only a few oftheobserved peaks are indicated, but all are listed in Table 3'

The hydroxyl-stretching region

It is well known that in the amphibole structure OHis located at the O(3) site (Hawthorne, 1982a), where F-(Warren, 1929) and presumably Cl- (kake, 1968) are alsolocated (Fie. 3). The same site may also be occupied byO2-, either competing with OH-, as in oxy- and Ti-bearingamphiboles (kake, 1968; Kitamura etal., 1975), or withF- as in the amphiboles of meteorites (Olsen etal.,1973).The latter are particularly interesting, as they are (Na)-F-richterites.

The rn method is specific only for OH valence vibra-tions, those for F, Cl, and O being either Raman-activeor occurring in the Nrn region. However, the presence ofcompeting anions is evident in the spectrum from thedecreased intensity ofthe OH absorption bands (provided

a constant amount of amphibole is used). The integratedareal intensities of bands in the 3800-3600-cm ' regionare inversely proportional to the measured F contents and

MOTTANA AND GRIFFIN: TWO COEXISTING K-RICHTERITES

6 0

4 0

2 0

bR

c

E

rd

bR

E

s

6 3

6 2

3 8 0 0

f r e q u e n c y ( c m - r )

I 430 MOTTANA AND GRIFFIN: TWO COEXISTING K-RICHTERITES

36381 1461 0791 043

- 1 0 1 5 s979

9 1 8775s738704ooJ

604-590s

f,4U

507459

-420s381

-3bbs

3731

-3707s36713657

3642

1 1461 079

, 1043- 1 0 1 5 s

978954

-920s-775s

738702ooc

606590548cu/459

-420s380360

3642363711441 084b

Table 3. Observed infrared bands forK-richterites

St. Marcel

VM-9 pink VM-9 blue Monte Somma

pink richterite contains less OH, it is substantially moreordered with respect to the coordinating cations than themore OH-rich blue one.

Broadening of the OH-stretching bands is particularlyevident in the Monte Somma richterite. The broad bandat 37 l0 cm ' is most likely derived from the overlap ofat least two bands. Such a broadening may be related tothe disordered cation arrangement around OH as a resultof the high temperature of formation of the sample (over550"C according to Della Ventura et al., 1983, as com-pared with ca. 400"C for the St. Marcel material; see Grif-fin and Mottana, 1982).

In an amphibole with an empty A site, hydroxyl com-pletes the octahedral coordination of the M(l) and M(3)sites by forming the apex of a flat pseudotrigonal pyramid,the base of which consists of two M(l) sites and one M(3)site. When all the cation sites are occupied by Mg, thecluster has the configuration (MgMgMg)-OH-tr, and therR spectrum consists of a single, strong and sharp peak at3673 cm' (peak A of Burns and Strens, 1966). Whencations other than Mg occupy M(l) and/or M(3), a fre-quency shift takes place that depends upon the change inthe energy of the OH-M bond, in turn affected by theelectronegativity of the cation (Strens, 1974). Thereforethe cluster having the configuration (Fe2*Fe2*Fe'z*)-OH-trhas a characteristic frequency at 3625 cm ' (peak D ofBurns and Strens, 1966), and it is not only shifted to lowerenergybut is also weakerthan the 3673-cm-'peakbecauseofthe differences in bond strength. Other clusters, havingdifferent types of segregation such as (MgMgFe'z*)-OH-!and (MgFe'z*Fe'z*)-OH-!, generate independent peaks at3660 and 3648 cm-' (C and D ofBurns and Strens, 1966)as do those involving Fe3t (see Hawthorne, 1983b).

The undisputed assumption behind this interpretationis that frequency shifts occur in discrete steps, each con-figuration having a characteristic vibrational frequency.Thus, in chemically intermediate members of solid so-lutions, the configurations of the endmembers are all si-multaneously present with their own characteristic fre-quency but with peak intensity proportional to the numberof each configuration versus the total number of config-urations. This model, developed mainly by Strens (1966)and by Burns and Strens (1966) and later improved byBurns and Greaves (1971), Nikitinaet al.(1973), and Law(1976) with regard to various theoretical and frtting as-pects, sufers severe drawbacks when accurate quantita-tive determinations of site occupancy are attempted (see,among others, Strens, 1974; Whittaker,1979; Hawthorne,1983a). However, it proves sufficiently accurate for simplesolid solutions like the tremolite-ferroactinolite series, aswell as for qualitative assignments in amphiboles of otherseries (Burns and Greaves, l97l), providing clusteringdoes not occur (Whittaker, 1979).

In amphiboles with filled or partially occupied A sites,a new type of configuration is present. The hydroxyl nolonger forms the apex of the triangular pyr"amid but bridgesbetween the M( l)M( l)M(3) pseudotriangular base and thecation located at or near the center ofthe large, l2-fold

37103668

37313715

3672

3653

9819579 1 6

738713668

-622

c4Y

512457

363s3463 1 1

-290s272

238s

Note; s : shoulder; b : broad.

thus roughly proportional to the HrO contents. Both urapand rn analyses therefore indicate that the pink amphibolecontains Iess OH than the blue one. The rn analysis doesnot give a quantitative estimate of OH, as the spectra wererecorded in transmission mode rather than absorptionmode, and the recording technique is nonproportional.

Pink richterite shows a sequence ofbands which is muchbetter defined than that ofthe blue richterite. As the grind-ing technique and pellet preparation were identical forboth samples, such a difference is real and may be relatedto different clustering of the OH present in the samples.

Pink richterite shows three sharp transmission minimaat3731,3715, and3672 cm-t, plus two other weak shoul-ders on the lower-energy side of the 3672-cm-' band, at3653-3657 and 3638 cm-', respectively.

Blue richterite shows only two poor, broad minima at3731 and 3671 cm-', plus weak shoulders at 3705-37 10,3657,and3642 cm-1 . All of these bands are much broaderthan the corresponding bands in the pink richterite spec-trum. Broadening of the OH valence bands suggests great-er disorder in the arrangement of the OH in the bluerichterite than in the pink one (decreasing segregation, inthe sense of Law, 1976) with reference to the cations oc-cupying the M(l) and M(3) sites coordinating the O(3)(Oh,F)-bearing site. Apparently, therefore, although the

314

250228

314

277

229

MOTTANA AND GRIFFIN: TWO COEXISTING K-RICHTERITES t43l

coordinated cavity formed by two back-to-back Si.Or.rings of the tetrahedral chains (Hawthorne, 1982a). Con-figurations with a filled A site possess greater energies thanthose with empty ones, because the repulsion arising be-tween the cation at A and the proton of the OH dipoleprojecting into the A-centered cavity leads to an increasein the stabiliry of the O-H bond (Hanisch, 1966). Con-sequently, these configurations will show OH vibrationsshifted to higher energy, i.e., they will display positivefrequency shifts with respect to the 3673 cm ' frequencyof the (MgMgMg)-OH-tr configuration. Rowbotham andFarmer (1973) found a strong peak at 3734 cm-' in asynthetic K-richterite and another at 3728 cm-I in a syn-thetic Na-richterite, which they assigned to the (Mg-MgMg)-OH-K and (MgMgMg)-OH-Na configurations,respectively. In addition, Maresch and Langer (1976) as-signed to the (MgMgMg)-OH-Li configuration a vibrationoccurring at 3708 cm-' in a complex orthorhombic syn-thetic amphibole.

Although criticized by Hawthorne (1983b) with regardto the homogeneity of the amphiboles synthesized, Row-botham and Farmer's assignment has not been disputed.Therefore our 373l-cm-' band can be also assigned to theconfiguration (MgMgMg)-OH-K present in both rich-terites. The intensity of this band shows that such a con-figuration is indeed the most significant in our samples.

Following Burns and Strens (1966) and Rowbothamand Farmer (197 3), the band at 367 2 cm-l is assigned tothe (MgMgMg)-OH-tr configuration. We therefore acceptthat a number of A sites are empty in our pink richterite.Moreover, the shoulders at 3653-3657 and 3638-3642cm I can be assigned to configurations such as (Mg-MgMn)-OH-D and (MgMnMn)-OH-!, respecrively, onthe basis of arguments originally suggested by Burns andStrens (1966). This is a further indication that a numberof the A sites are vacant, at least among those close to thehydroxyls occupying the O(3) site in place of F.

The assignment of the sharp, well-defined peak at 3615cm-r in the pink richterite still requires explanation. Theposition is such as to make it unlikely that it derives fromconfigurations involving either Na or Li. Therefore weassign it to the (MgMgMg)-OH-Sr configuration, on thebasis of the following reasoning.

In Strens' ( I 974) theoretical treatment, frequency shiftsfrom the (MgMgMg)-OH-tr configuration, assumed to bethe standard (3673 cm-l), are a function of the electro-negativity ofthe intervening cation. They are negative inthe case of substitutions at the M(l) and M(3) sites andpositive when the A site is occupied. In addition, there isalso a contribution due to the efective charge in such acase. Work on synthetic amphiboles has shown the fol-lowing shifts: +62 cm-' for K+ (electronegativity 0.8);+56 for Na. (0.9) and +35 for Lit (1.0), as listed above.In the present case, a -142 cm-t shift has been observedin addition to the +59 attributed to K*. This shift cannotbe attributed to any other alkali metal ion, the electro-negativities of which are even lower than K (Rb 0.8; Cs0.7), but it can be related to alkaline earths, the electro-

Table 4. Configurations in pink richterite

Wavenumber Contr ibut ionConfiguration (cm-') (%)

MgMgMg-OH-KMgMgMg-OH-SrMgMgMg-OH-CMgMgMn-OH-!MgMnMn-OH-tr

3731371 53672 l3653 |3638 I

87.35 .1

0 7 (

negativities of which are in fact intermediate between thoseof Na and Li (Ca 1.0; Sr L0; Ba 0.9). As Ba is not presentin our richterites (see Table 2), the choice is restricted toCa and Sr.

Charge and electronegativity being equivalent, theprobability for an ion to enter the A site is inversely pro-portional to size, as the larger the ion, the more stable thestructure will be: K (1.64 A; > Sr (1.44) > Na (1.39) >Ca (1.34) for the 12-fold coordination (Shannon, 1976).Thus Sr is not only favored over Ca, but over Na as well,rnore so as the total local effective charge is increased.

This interpretation explains not only why the *56 cm '

shift for Na is not detected in our spectrum, but also whyconfigurations with empty A sites are observed. They areneeded to balance the total charge at the A site becauseofthe entry ofa divalent cation.

From this chemical interpretation, and taking into ac-count the analy"tical data of Table 2, we suggest the con-figurations given in Table 4 to be present in pink richterite.

The filling up of the A site requires a very minor con-tribution of another configuration, which may be Mg-MgMg-OH-Na or MgMgMg-OH-Ca. If they are present,these bands are mostly hidden below those of the domi-nant contributions, although there appears to be in thespectrum another very weak shoulder at 3657 cm-'. Al-ternatively, the balance of the A site is reached by assum-ing the presence of 0.006 Ti at M(l) (cf. Table 2); such asubstitution is possible in amphiboles, but it has beenshown to be unresolvable in the rR spectrum (Kitamuraet al., 1975).

The contribution of the Mn segregation to the config-urations involving the empty A site can be calculated fromthe intensity ratios of the C and D to the A peak (Burnsand Strens, 1966; Nikitina et al., 1973). The amount ofMn at the OH-coordinating M(l) + M(3) sites is 0.003atoms pfu at best, out of a total of 0.071 (Table 2). Therest is either in sites coordinated by F or enters the M(4)sites together with Na displaced by Sr entering the A site.

It should be stressed here that the assignments givenabove refer only to the 380/o of the A sites adjacent to OH;nothing can be said for the 620/o for which F occupies theO(3) position.

The region of lattice vibrations

Despite their large number and strong intensity, theabsorption bands in this region add very little to our in-sight of the amphibole crystal chemistry @ig. a). In factthe two examined richterites have spectra identical in all

1432

details, in terms of their frequency, and also in perfectagreement in their intensity when minor variations inchemical composition are also taken into account. On thecontrary, there is a striking difference between the St.Marcel richterite spectra and that of the Monte Sommasample. They have essentially identical frequencies, butdiffer in the sharpness and intensity of the bands. As awhole, the Monte Somma richterite spectrum seems toindicate a disordered structure in comparison with thehighly ordered St. Marcel spectra. This may be the resultof the differenl P.T conditions of formation. The MonteSomma amphibole formed rapidly in a high-Z, low-Pcontact-metamorphic environment, whereas the St. Mar-cel amphiboles formed in a low-Z, high-P regional meta-morphic episode (over several million years?). However,this interpretation cannot be supported by other data, asthere are no spectra of volcanic K-richterites available inthe literature.

Clusn oF THE col,oR

The type K-F-richterite from Monte Somma is color-less in grains about 0.5 mm in size, as are most richterites.The color ofthe two St. Marcel amphiboles may be relatedto their contents of transition metals. Although they havenearly equal amounts of Mn, they differ essentially incontaining Ti (pink) and Fe (blue). It may be questionedwhether these minor amounts are enough to produce theclear colors exhibited by 0.5-mm-thick grains. However,no other chromophores were detected at the rt'rre level,and we are forced to attribute the diflerent colors to themodifying effects of Ti and Fe on the prevailing Mn. ThusK-F-richterite presumably is pink owing to the effects ofTi and Mn3*. This interpretation is supported by the colorshown by other Mn3*-bearing minerals present at St. Mar-cel (pyroxenes, epidotes and phengites) although these allshow violet to red, rather than pink hues. The color ofthe blue K-(OHlrichterite probably is due to electrontransfer between Mn and Fe, although this interpretationis admittedly tentative. There are at St. Marcel no otherblue minerals to test it, nor are blue colors in mineralsusually attributed to such a mechanism. In most cases itis related to Fe2+-Ti4+-Fe3+-Ti3+ charge transfer (e.g., kya-nite, sapphire), but in amphiboles it has been demonstrat-ed to be due to the Fe2+-Fe3+-Fe3+-Fe2+ charge transfer (cf.Smith and Strens, 1976). We have no optical spectra tosupport our suggestion of the possibility of a Mn2+-Fe3+-Mn3+-Fe2+ charge transfer. However, Fe in the blue grainprobably is too little to cause the color by itself, there isno Ti to allow a transfer similar to that in kyanite, andthe properties of Mn and Fe are very similar. These pointssuggest that the proposed mechanism is likely.

CoNcr-usroNs

The crystal chemical study of the pink and blueK-richterites of St. Marcel has produced no clue as to howminerals with such a small difference in cation compo-sition, but with a large difference in anion substitution,can coexist in apparent chemical equilibrium. The causeof the color is apparently related to the presence of trace

MOTTANA AND GRIFFIN: TWO COEXISTING K-RICHTERITES

amounts (<0.010 atoms pfu) of Fe in the blue richteriteand Ti in the pink one, in both cases in the presence ofMn3*.

The rn study completes the knowledge of the crystalchemistry in a way unexpected and undetected either viaa careful er"rpe or xno investigation. The presence of emp-ty A sites is revealed by the occurrence ofm bands typicaloftremolite-type amphiboles. As these cannot be detectedby microprobe analysis and were not detected during crys-tal-structure refinement (L. Ungaretti, pers. comm.), wesuggest that domains with empty A sites occur to balancelocally the presence of Sr at the A site in place of a mono-valent cation. In fact, in addition to the well-known 373 I -

cm-I band typical of the (MgVtgMg)-OH-K clusters, thespectra show a band at 3715 cm-| that can be explainedas due to the (MgMgMg)-OH-Sr configuration. This re-quires the simultaneous presence of empty A sites tomaintain the local bond-valence requirements, as well asthe overall neutrality of the crystal structure. Thus, rnspectroscopy, although still fraught with difficulties, is ableto give additional information about aspects of crystalchemistry, even in the case of complex structures such asamphiboles. Some of this information is not apparenteither in a conventional evaluation ofthe analysis nor ina structural refinement. This additional information doesnot conflict with that given by the above methods. rnspectroscopy can discriminate among unit-cells with emp-ty vs. full A sites.

AcxNowr,nocMENTS

These critical samples were kindly donated by Dott. V. Mattioli(Milan) as a part of his continuing support of our investigationofthe mineralogy of St. Marcel. Prof. L. Ungaretti (Pavia) carriedout crystal structure refinements and Dotts. G. Della Venturaand G. C. Parodi ran the IR spectra. Critical readings by Prof.Dr. K. Langer (Berlin) and Dr. W. V. Maresch (Bochum) greatlyimproved the manuscript. Investigations at St. Marcel are fi-nanced by M.P.I. (Italy) via the project "Crystal Chemistry andPetrogenesis," and by the Norwegian Research Council for Sci-ence and the Humanities (NAVF1. A grant-in-aid from the C.N.R.,C. S. Mineralogia e Petrologia delle Formazioni Ignee, Roma,permitted A.M. to present this work at the 3rd E.U.G. Meetingin Strasbourg (April a, 1985). Detailed reviews by F. C. Haw-thorne, P. McMillan, and G. R. Rossman greatly improved themanuscnpt.

RnrrnsNcts

Aoki, Ken-Ichiro. (1975) Origin ofphlogopite and potassic rich-terite bearing peridotite xenoliths from South Africa. Contri-butions to Mineralogy and Petrology, 53, 145-156.

Burns, R.G., and Greaves, C.J. (1971) Correlation of infraredand Mossbauer site population measurements of actinolites.American Mineralogist, 56, 2010-2033.

Burns, R.G., and Strens, R.G.J. (1966) Infrared study of thehydroxyl bonds in clinoamphiboles. Science, 153, 890-892.

Cameron, Maryellen, and Papike, J.J. (1979) Amphibole crystalchemistry: A review. Fortschritte der Mineralogie, 57,28-67.

Cameron, Maryellen, Sueno, Shigeho, Papike, J.J., and Prewitt,C.T. (1983) High temperature crystal chemistry of K and Nafluo-richterites. American Mineralogist, 68, 924-9 43.

Carmichael, I.S.E. (1967) The mineralogy and petrology of thevolcanic rocks from the Leucite Hills, Wyoming. Contributionsto Mineralogy and Petrology, 15,24-66.

Charles, R.W. (1977) The phase equilibria of intermediate com-

MOTTANA AND GRIFFIN: TWO COEXISTING K-RICHTERITES t433

posit ions on the pseudobinary Na.CaMg,SirOrr(OH)r-NarCaFerSirOrr(OH)r. Amerian Journal of Science, 277, 594-625.

Dawson, J.B., and Smith, J.V. (1977) The MARID (mica-am-phibole-rutile-ilmenite-diopside) suite of xenoliths in kimber-lite. Geochimica et Cosmochimica Acta, 41,309-323.

Deer, W.A., Howie, R.A., and Zussman, J. (1963) Rock-formingminerals. Volume 2. Chain silicates. Longmans, Green andCo., London, 379 p.

Delaney, J.S., Smith, J.V., Carswell, D.A., and Dawson, J.B.(1980) Chemistry of micas from kimberlites and xenoliths. II.Primary- and secondary-textured micas from peridotite xeno-liths. Geochimica et Cosmochimica Acta, 44,857-872.

Della Ventura, Giancarlo, Maras, Adriana, and Parodi, G.C.(1983) Potassium-fluorrichterite from Monte Somma (Cam-pania, Italy). Periodico di Mineralogia, Roma, 52,617-630.

Erlank, A.J., and Finger, L.W. (1970) The occurrence of potassicrichterite in a mica nodule from the Wesselton kimberlite,South Africa. Carnegie Institution of Washington Year Book68,320-324.

Erlank, A.J., and Rickard, R.S. (1977) Potassic richterite bearingperidotites from kimberlite and the evidence they provide forupper mantle metasomatism. (abs.) 2nd International Confer-ence on Kimberlites, Santa Fe, Extended Abstracts (unpaged).

Gilbert, M.C., and Briggs, D.F. (1974) Comparison of the sta-bilities of OH- and F-potassic richterites-A preliminary re-port. (abs.) EOS, 55,480-481.

Gilbert, M.C., Helz, R.T., Popp, R.K., and Spear, F.S. (1982)Experimental studies of amphibole stability. Mineralogical So-ciety of America Reviews in Mineralogy, 98,229-353.

Griffin, W.L., and Mottana, Annibale. (1982) Crystal chemistryof clinopyroxenes from the St. Marcel manganese deposit, Vald'Aosta, Italy. American Mineralogist, 67, 568-586.

Hanisch, M.K. (1966) Messung des Ultrarot-Pleochroismus vonMineralen. VL Der Pleochroismus der OH-Streckfrequenz inRiebeckit. Neues Jahrbuch fiir Mineralogie Monatshefte, 109-t t2 .

Hariya, Yu, Oba, Takanobu, and Terada, Shoichi. (1974) Sta-bility relation of some hydro-silicate minerals at high pressure.4th International Conference on high pressure, Proceedings,Kyoto,206-210.

Hawthorne, F.C. (1982a) Crystal chemistry of the amphiboles.Mineralogical Society of America Reviews in Mineralogy, 9,A.,t-t02.

- (1982b) Amphibole spectroscopy. Mineralogical Societyof America Reviews in Mineralogy, 9A, 103-139.

-1t983a) The crystal chemistry of the amphiboles. Cana-dian Mineralogist, 21, 173-480.

- (1983b) Quantitative characterization of site-occupanciesin minerals. American Mineralogist, 68, 287 -306.

- 1t 983c) Characterization of the average structure of nat-ural and synthetic amphiboles. Periodico di Mineralogia, Roma,52, 543-58 l .

Huebner, J.S., and Papike, J.J. (1970) Synthesis and crystal chem-istry of sodium potassium richterite (Na,K)NaCa.MgrSirOrr(OH)r: A model for amphiboles. American Miner-alogist, 55, 1973-1992.

Jones, A.P., Smith, J.V., and Dawson, J.B. (1982) Mantle meta-somatism in 14 veined peridotites from Bulfontein mine, SouthAfrica. Journal of Geology, 90, 435-453.

Kitamura, Masao, Tokonami, Masayasu, and Morimoto, Nobuo.(1975) Distribution of titanium atoms in oxy-kaersutite. Con-tributions for Mineralogy and Petrology, 51, 167-172.

Kushiro, Ikuo, and Erlank, A.J. (1970) Stability of potassic rich-terite. Carnegie Institution of Washington Year Book 68,231-233.

Law, A.D. (1976) A model for the investigation of hydroxylspectra of amphiboles. In R.G.J. Strens, Ed. The physics andchemistry of minerals and rocks, 677-686. Wiley, London.

Leake, B.E. (1968) A catalog of analyzed calciferous and sub-

calciferous amphiboles together with their nomenclature andassociated minerals. Geological Society of America Special Pa-p e r 6 8 , 2 1 0 p .

-(1978) Nomenclature of amphiboles. American Miner-alogist, 63, 1023-1053.

Maresch, W.V., and Langer, KJaus. (1976) Synthesis, lattice con-stants and OH-valence vibrations of an orthorhombic amphi-bole with excess OH in the system LirO-MgO-SiOr-HrO. Con-tributions to Mineralogy and Petrology, 56,27-34.

Martin-Vernizzi, Silvana. (1982) La mine de Praborna (Vald'Aosta, Italie): Une s6rie mangan6sifdre m6tamorphis6e dansle facids 6clogite. Thdse docteur 3dme cycle, Universit6 Pierreet Marie Curie, Paris,215 p.

Michaelson, A. (1883) Bidrag til kiinnedom om Amphibolenssammansiittning. Kongl. Svenske Vetenskaps-AkademietsForhandlingar (Ofversigt), 20, 19 5-200.

Nikitina, L.P., Petkevich, E.2., Sverdlova, O.V., and Khristo-forov, K.K. (1973) Determination of occupancy of octahedralposit ions in the structures of amphiboles (Ca,Na,K)r-,(Fe'?*,Fe3*,Mg,AD, [Si,A1].O,, t(OH)z from vibrations of OH .Geokhimiya, 1 66 1-1 668 (transl. Geochemistry International,rr, 1233-1239).

Olsen, E., Huebner, J.S., Douglas, A.A.V., and Plant, A.C. (1973)Meteoritic amphiboles. American Mineralogist, 58, 86 l-872.

Prider, R.T. (1939) Some minerals from the leucite-rich rocks ofthe West Kimberley area, western Australia. MineralogicalMagazine,25, 378-387.

Robinson, Peter, Spear, F.S., Schumacher, J.C., Laird, Jo, Klein,Cornelis, Evans, B.W., and Doolan, B.L. ( I 982) Phase relationsof metamorphic amphiboles: Natural occurrence and theory.Mineralogical Society of America Reviews in Mineralogy, 98,r-227.

Rondolino, funaldo. (1936) Sopra alcuni anfiboli manganesiferidi Praborna (San Marcel-Valle d'Aosta). Periodico di Miner-alogia, Roma, 7, 109-121.

Rowbotham, G., and Farmer, V.C. (1973) The effect of "A" siteoccupancy upon the hydroxyl stretching frequency in clinoam-phiboles. Contributions to Mineralogy and Petrology, 38, 147-r49.

Shannon, R.D. (1976) Revised effective ionic radii and system-atic studies ofinteratomic distances in halides and chalcogen-ides. Acta Crystallographica, 432, 7 5l-7 57 .

Smith, G., and Strens, R.G.J. (1976) Intervalence-transfer ab-sorption in some silicate, oxide and phosphate minerals. InR.G.J. Strens, Ed. The physics and chemistry of minerals androcks, 583-612. Wiley, London.

Strens, R.G.J. (1966) Infrared study of cation ordering and clus-tering in some (Fe, Mg) amphibole solid solutions. ChernicalCommunications, I 5, 5 19-520.

-(1974) The common chain, ribbon, and ring silicates. InV.C. Farmer, Ed. The infrared structure of minerals, 305-330.Mineralogical Society, London.

Thy, Peter. ( I 982) Richterite-arfvedsonite-riebeckite-actinoliteassemblage from MARID dikes associated with ultrapotassicmagmatic activity in central west Greenland. Terra Cognita,2 ,247-249.

Ungaretti, Luciano, Lombardo, Bruno, Domeneghetti, Chiara,and Rossi, Giuseppe. (1983) Crystal-chemical evolution ofam-phiboles from eclogitic rocks ofthe Sesia-Lanzo zone, WesternAlps. Bulletin Min6ralogique, 106, 645472'

Velde, Danielle. (1967) Sur un lamprophyre hyperalcalin potas-sique: La minette de Sisco (ile de Corse). Bulletin de la Soci6t6frangaise de Min6ralogie et Cristallographie, 90, 214-223.

Warren, B.E. (1929) The structure of tremolite. Zeitschrift liirKristallographie, 7 2, 42-57 .

Whittaker, E.J.W. (1979) Clustering of cations in amphiboles.Physics and Chemistry of Minerals, 4, 1-10.

MeNuscnrpr REcEIvED Ocrosen 18, 1985MeNuscnrpr AccEPTED Jurv 8, 1986