Structural control of the chlorine content of OH-bearing silicates (micas and amphiboles)

0016-7037/85/33.00

+ .oo

Structural control of the chlorine content of OH-bearing silicates (micas and amphiboles) M. VOLFINGER*, J.-L. ROBERT’@,D. VIELZEUF** and A. M. R. NEIVA*** lCentre de Recherche sur la Synthese et Chimie des MinCraux, GIS CNRSBRGM, I A, rue de la Rrollerie, 45045 OrEans Cedex, France l*D&wtement de Giologie, LA. IO et I.O.P.G., 5, rue Kessler, 63038 Clermont-Ferrand, France ***DepaLFtment of Mineralogy and Geology, University of Coimbra, 3000 Coimbra. Portugal (Received April 12, 1983; accepted in revised form September 18, 1984)

Abstract-Experimental studies of the incorporation of chlorine in trioctahedral biotite-like micas, belonging to the series phlogopite-annite, phlogopite-KCo&Si,,dOHh and phlogopi~KNi,AlSip,o(OH~, wereperfomml at 600°C and 2 khan, with a duration of two weeks. The results confirm, for the incorporation of an anion in a crystal structure, the fundamental role of the dimension of the anion site, as has been established for cations in previous works. In biotites, the dimension of (OH-Cl) site is mainly controlled by the rotation angle a of the tetrahedra around a direction approximately parallel to E. The experiments were performed using hydrothermal solutions with KCI z 0.5 M; under these conditions, the quantity of incorporated chlorine does not exceed 1300 ppm in the most receptive mica (annite) and is twenty times less in the less receptive ones (phlogopite, for example). These results are applied 10 natural biotites in porphyry copper deposits, metamorphic rocks, and m&c rocks. We conclude that most natural biotites which have a chlorine content of 1000 ppm or more crystallized in equilibrium with a fluid phase with chloride contents of several molar (minimum 3 M). The consideration of micas applies in the same way lo amphiboles. A clear correlation between the Cl content and XF, is observed which can be interpreted in terms of local structure of the minerals. The structural factors which favour the fixation of chlorine, a large anion, are the same which favour the fixation of large alkali cations (replacement of Na by K). This explains the observed correlations between Cl and K in natural amphiboles. INTRODIJCIION

K( Mg,_,M~)(Si~Al)OldOHh,

A SUITABLEmethod of studying the incorporation of trace ions in minerals is ion-exchange between the

mineral and a hydrothermal solution. Recently, a model has been proposed which allows the interpretation of distribution coefficients of trace cations between hydrothermal solutions and silicates, based on local crystal site properties (VOLFINGERand ROBERT, 1980). This model has been established from experiments involving feldspars, feldspathoids, micas and the alkali cations, Li, Na, K. Rb and Cs. The relation is independent of the silicate structural family. It depends only on the adaptation of the trace cation to the structural cavity it enters, as measured by the difference between the size of the fixed trace cation and the size of that cavity. The same model applies to the exchange of alkaline earth elements between hydrothermal solutions and felspars (VOLFINGER, 1980) and to alkali cation exchange involving homblendes (ROBERT, I98 I ). Hence it is interesting to investigate the same model for the fixation of anions. Among the silicates containing anions (micas, amphiboles, tourmalines, feldspathoids, . . .) one must choose those whose structure is the best known, in particular those for which it is possible to calculate most of the interatomic distances and angles from their composition, using some reliable models. The trioctahedml mica family satisfies this requirement and three series of biotite-type micas were used in the present work. The formula of these micas is 31

with M*+ = Fg’, Co*‘,

Or Ni2+. The first part of this paper mainly deals with experimental results concerning the exchange of OHin the micas indicated above with Cl-. The chlorine anion is used in most hydrothermal experiments involving mineral synthesis and cation exchange reactions. The influence of anions, including chlorine, on the distribution of cations is known (IIHAMA, 1965). but data concerning the distribution of chlorine itself between silicates and solutions are sparse. Preliminary data and interpretation of the incorporation of traces of chlorine in trioctahedral micas, K(Mg, M’+)(Si,AI)O,,,(OHh, synthesized in Cl-bearing hydrothermal solutions (600°C, 2 kbars) are given in ROBERT and VOLFINGER (1975, 1976) and integrated

into a general work concerning the crystal chemistry of trioctahedral micas (ROBERT, 1981). These results show significant variations in the degree of replacement of OH- by Cl- with the composition of the octahedral sites of the mica. The highest Cl content is observed in the ferrous end-member (annite). Experimental results published by MUNOZ and SWENSON ( I98 I) show the same behavior of chlorine. These results can be interpreted in terms of the differences between the size of the anionic site and the size of the Cl anion. A similar structural model explains the partioning of trace cations between silicates and hydrothermal solutions (VOLFINGER and ROBERT, 1980).

After extending the crystal chemical model to anions in synthetic trioctahedral micas, the behavior of chlorine in OH-hearing natural silicates (micas and amphiboles) will he discussed. Such behavior is of geochemical interest as described by MUNOZ and SWENSON (I 98 1). Data on natural clinoamphiboles may he interpreted with the same model of structural control of the chlorine content because the model is based only on site dimensions and is independent of the mineralogical family. EXPERIMENTS,

RESULTS

AND INTERPRETATION Experimenral

method

The introduction of Cl- in a mica crystallized in equilibrium with a hydrothermal KCI solution can be represented by the exchange reaction: mica - OH + KCI S mica - Cl + KOH To a first approximation, the partition coefficient D corresponding to this equilibrium is written using the concentration of the chemical species concerned:

(1)

CA,, and CAraH are the atomic concentrations ot’ CI and OH in the mica solid solution (gram-atoms X lO-3/g). Mnc, and MKo” are the molarities of KCI and KOH in the hydrothermal solution. Micas were crystallized at 6OO“C. 2 kbars from gels 01 appropriate compositions prepared by the method of HAMILTON and HENDERSON(1968). In the phlogopite-annite series iron is divalent. Ferrous iron can be obtained in two different ways: (a) by using the solid buffer method with double capsules (which limits the quantities of solid and solution that can be studied) or (b) by using a starting gel containing Fe3+ and Fe0 in appropriate proportions. This second method was used by IIYAMA(pers. commun., 1973) in his study of phase relations in the assemblage blotite + garnet + cordierite at 800°C. 5 kbars. We have used this method in the present work with an initial atomic ratio Fe’/ Fe” + Fe” = 0.4 in the gel, which produces a single mica phase at 600°C. 2 kbars. However some experiments were done with annite using NNO and QFM buffers. Table I and Fig. 1b show that they give similar results. In the two other series (Mg-Ni). (Mg-Co), the transition cation remains divalent under the experimental conditions. The gel (80 to 100 mg) was placed in a gold tube with a KCI solution (sometimes KCI-KOH). The experiments were performed in cold-seal pressure vessels at 600” and 2 kbars for different durations. Chlorine as well as water are supplied to the mica by the hydrothermal solution during the crystallization. KCI solutions are chosen in order to prevent any exchange of alkali ions between mica and fluid. Results are

_, _ ._ _ _E!!+r_.

__ _-. ___ __.. _.

_. , _.

Cnn_PM.6 j+__--------3

I

/’

b

/‘,

Ann-RI __ ._

A LKOH O.Oli’! I _ -----

___ _,_

,-~I (KOH

Annltc

11’

0.W)

-I------_________,_

,(_

Ann-Phl

_

_

.6 [KOHOJH) _ ----

1 m

--I-

I 30

a

I

I

LO

50

days d FIG. I. Incorporation of chlorine by two micas of the annite-phlogopite series, as a function of duration of synthesis. Mica compositions: XF. = 0.6 and X,, = I. CA,: atomic concentration of Cl in the micas (gram-atoms x IO-‘/g). MKcl and MKoH: molarities of KC1 and KOH in the starting solutions. a: CAI/ kfKn vs. time. For a mica at fixed MK~I, CA, is lower when MKoH is greater. b:

C*,

VS.time.

&Ctl~KOH

on the same curve for a given mica; after IO days, there is no significant variation in CA! MKCIIMhOH points

39

Cl in micas and amphiboles given in Table 1 and Fig. 1. They show that fit&en days is sufficient to reach a reproducible, stable state which permits a comparison of behavior of chlorine in the different micas. Chlorine contents were determined with the radioactive isotope’%Cl.The isotope is introduced in KCl of the initial solution; hence it probably is in isotopic equilibrium with the natural stable isotopea of chlorine (“Cl, “Cl) in the final products. The isotope Wl is a pure fl- emitter with an energy of 0.7 14 MeV and a half-life of 3.1 X IO’ years. The radioactivity of the samples was measured with a GeigerMiiller counter. The low energy of the jY particles requires great care in the preparation of standards and samples. For the same counting geometry, the measurement must be done on the same weight of solid for the sample as well as the standard. The detection limit of Cl by this method and the accuracy of the measurement depend on the specific radioactivity of the solution and on the characteristics of the measuring apparatus. In the experiments, most of chlorine remains in the solutions, and the radioactivity must be kept within acceptable safety limits. In a general way, the radioactivity of a solution used for one run is 0.02 PCi. The order of magnitude of the radioactivity of the mica crystallixed in equilibrium with this solution is about 100 to 5000 times lower depending on the type of mica and on the chemical composition of the starting solution. The detection limit of the analytical procedure is 1 ppm Cl in the mica. Near the detection limit, low radioactivity of the sample requires long counting times (500 mn or more) and the uncertainties of the measurements can reach 25%. Table 1 gives the relative uncertainties calculated for all the experiments. The weak radioactivity of the solid phase, due to the low incorporation rate of chlorine in the mica does not permit us to perform reversal experiments. Because the main subject of the paper is OH--Cl- exchange between mica and fluid, one must be sure that all chlorine of the solid phase is really in the mica structure and not adsorbed on the particles’ surface. In all previous studies performed at CRSCM concerning the distribution of trace cations between silicates and hydrothermal solutions, adsorption was taken into account. The measurement of the quantity of adsorbed ions is greatly facilitated by using the radioactive tracers At the conclusion of each experiment, the solid phase was washed under room conditions with a solution of 2 M KC]. The radioactive % was then assumed to be removed from the surface of particles and mplaced by non-radioactive chlorine. A&r washing, the remaining radioactivity measures the amount of chlorine completely fixed in the structure of the mica. An example: Mg-phloogopite (95.6 mg) crystallixed in a solution of KCl (0.45 M) incorporated a bulk amount of 7.8 X IO-* rmok of Cl. Of this amount, more than half is adsorbed on the surfaces of particles. The amount of Cl fixed in the mica structure, measured after washing, was 3.55 X IO-’ rmole of Cl. In our experiments, the amount of chlorine incorporated in the mica phase is small compared to the large chlorine content of the solution. Therefore, changes in the chlorine content of the solution are negligible during the experiments. For the same reason, the amount of OH- in the mica phase can be calculated from the weight of the final solid, (i.e., CA1-OH = constant) and D’ t &cl

I

M~cl . MKOH

The molarity (&On) of the hydrothermal solution is unknown under the experimental P, P conditions used; it

’ Although annite always contains a significant proportion of Fe’+ (nl5%) (WON= et a/.. 1971). we use the ratio Fe/ Fe + Mg.

depends on the KOH content of the starting solution, on the solubility of the mica in water and on the reaction of KC1 with H20. For runs performed under similar conditions (amount of solution, M&, Mxon is pmsumed to be constant when no KOH is added to the starting solution and, from Eq. (2), the partitioning of chlorine between mica and fluid can be characterized by the ratio: D’IMuou = G&MU-,

.

For some runs, a small quantity of KOH was added to the starting solution. In these cases it was assumed that final MKoH is proportional to initial Mkon, and a D’ is defined

Experimental results and interpretation The crystal chemical model of the distribution of trace cations which we wish to apply to chlorine in this paper, states that the fixation of a trace element in a silicate structure is controlled by the respective sizes of the trace ion and the structural cavity it enters. In practice, this means that for a given composition of solution, silicates will incorporate more of a trace ion the closer the two sizes are together. Table 1 gives the initial composition of the mica gel, the initial composition of the solution, the final chlorine content of the mica (a&r removal of adsorbed chlorine), its corresponding atomic concentration CA,, duration of runs, C,,/ MKO ratio and D” coefficient. F~RUR 2 shows the variation of the chlorine content of mica &a function of composition as measured by M’+/M’+ + Mg, (X$), with M’+ = Fe*‘, Co2’ and Ni’+.’ Table I and Fig 2 show that the amount of chlorine incorporated in the micas depends on the cation composition of the mica. In the Mg-Ni series, no variation of the amount of chlorine fixed in the micas is detectable because it remains at a very low level. This low level is also observed in some parts of the two other series. In the Mg-Co series, the amount of chlorine incorporated in the mica remains very low and quite constant from the Mg end-member (Xc0 = 0) to a mica having Xc0 z 0.6; for Xc0 > 0.6, the amount of chlorine incorporated by the mica increases with increasing Co content. The phlogopite-annite series behaves like the Mg-Co series. The amount of chlorine in mica is constant between the phlogopite end-member (XF. = 0) and mica having XF, P 0.4. In this interval, the amount of chlorine in mica is extremely low. For higher values of ,I’,, the amount of chlorine in mica increases gradually to a maximum in annite (XF, = I). Annite contains approximately 40 times more chlorine than phlogopite under the same experimental conditions (PHfl, T, composition of the solution).

STlWClWRAL CONSIDERATIONS ON THE PHLOCOPITR-ANNITE SERIES A structural model for OH-Cl exchange in mica To interpret these data, consider OH- (or Cl-) in a trioctahedral mica structure. OH- (conventionally O(4)) is bonded to three (M*+)” cations; it lies in the same plane as tlte apical oxygens o(3) of the tetrabedra and in the middle of the hexagonal ring defined by six TO, tetrabedra. The anion (OH- or Cl-), bonded to 3M’+ is in a cavity limited by the six apical oxygens O(3) indicated above and by six basal oxygens, o(l), O(2) of the same hexagonal ring of tetrahedra (Fig. 3).

M.

Starting solution (molarirks) ‘kc1

%OH

Voltinger

n

41

Cl in Duration (days)

0.76

0.012

2

26.1

'Se

0.56

0.096

5

4.4

Annie '5,

-

0

Ann.-Phlog. series

7.35

9.67

0.116

1.24

2.21

0.212

3.14

0.314

0.10

10

6.0

0.59

0.012

IC

37.8

0.57

0.097

36

6.7

f

I.1

0.60

0.012

36

55.4

* 3.3

1.07

0.088

9

15.5

* 2.0

* 2.3 t

I.0

1.91

0.10

9

22.4

* 2.9

4.14

0.097

9

47.8

f 6.2

0.5,

20'

157

*

,I

1.82 10.6 1.88 15.6 4.37 6.31 13.5

44.3

18.0 3.30 26.0 4.08

0.216 0.320 0.312 0.359

3.30

0.330

3.26

0.303

8.20

L.IS

1.55

0.275

20**

84

t 6

23.6

0.53

55

99

* 7

27.9

I.15

15.2

0.7 0.6

0.44 0.50 0.53 0.53 0.53

0 0 0 0 0

15 I5 15 15 15

91.4 89.1 72.2 46.8 24.5

208 178 176 88.3 46.2

0.5 0.4 0.3 0.2 0.1 0

0.54 0.54 0.45 0.54 0.60 0.46

0 0 0 0 0 0

15 15 15 15 15 15

32 16 14 II 12 14

* t * * * *

5 3 3 3 3 3

o 5 0:5 0.5 0.5 0.5

o 0 0 0 0

15 15 15 15 15

149 115 74 38 14.5

* f f f *

10 8 6 5 3

0.5

0

I5

4.18 3.79

XPe

0.8

Phlog. Co.phlog.series

y 0.9 0.8 0.7 0.6

Ni.phlog. (lki - I)

c

chlorine

l

: atomic conerntrarion of At with external oxygen buffer

NNO.

with

1.2

* 0.9

3.9

I 0.9

l*

t

0.58

5.4

lolid

ppm

Am.-Phlog. - 0.6)

final

in micas

324 * 13 316 t 13 256 t 13 166 * II 87 f 6

11 * 3

9.03 4.51 3.95 3.10 3.38 3.9

16.7 8.35 8.78 5.74 5.63 8.5

42.0 32.4 20.9 10.7 4.09

84 64.8 41.8 21.4 8.2

3.10

6.2

(uate/mg).

QPH.

2b gives the distances OH-q 1, 2) long, OH-q 1, 2) From the published data on the crystal structures of phlogopite and annite, it is possible to calculate short, OH-O(3), Cl-O(3). Cl-q I, 2) long, Cl-o(l, 2) short and Cl-O(3) calculated from the atomic coorthe distances between OH and the oxygens limiting the cavity. In all trioctahedral micas, the ring of dinates given in Table 2a. To understand the structural control of the incortetrahedra is more or less regular, depending on the poration of chlorine in the mica structure, consider rotation of the tetrahedra about the direction T-O(3), the distances between the theoretical position of approximately parallel to P. This rotation of the chlorine and the positions of the surrounding oxygens. tetrahedra is required to produce a good fit between The factor limiting the possible incorporation of the tetrahedral and octahedral sheets (DONNAY ef al.. chlorine is purely geometrical: for chlorine to be 1964; HAZEN and WONES, 1972; GUIDOTTI ef ni.. accommodated, the dimension of the chlorine cavity 1975). It transforms the hexagonal symmetry of the must be such that the shortest distance, dcI.o B r0 ring of tetrahedra into a ditrigonal symmetry and + ro. The ionic radii used here are from SHANNON creates two sets of distances OH-0( 1, 2) similar to and PREWITT (1969, 1970). The choice of an ionic the distances K-0( I, 2): three long distances and radius requires knowledge of the ion’s coordination. three short ones. The rotation of tetrahedra around Oxygens 0( I, 2) are bridging and therefore can be the T-O(3) bond does not change the distance considered as essentially twofold coordinated, with OH-O(3). an ionic radius rg = I .35 A (SHANNON and PREWJIT, Tables 2a gives the atomic coordinates of OH-, in phlogopite and annite, from HAZEN and BURNHAM 1969, 1970). Oxygen O(3) is bonded to three octahedrally coordinated M*+ cations and one tetrahe(1973) and the atomic coordinates of Cl- in the same drally coordinated cation; it can be considered as two micas, calculated from the coordinates of OHfourfold coordinated, with an ionic radius rg = 1.38 and from the difference between the ionic radii. Table

41

Cl in micas and amphiboles

CD--_O-fp-, 0

.I

.2

.3

.

.4

---xtltJ

.S

.b

.7 1$ xt42+ ---i)

.9

1

FIG.2. Incorporation of chlorine in three mica series as a function of Xw, and variation of u angle as a function of XF. (dashed curve). CA,: atomic concentration of Cl in micas (as in Fw 1) at 6OO“C,2 kbats MKo: mob&y of KCI (-0.5 Mf Circtcs:annite~te xriu(X~). !Squates K~~~P,~~H~ phlogopitc series (A&). Diamonds ~i~(Si~lP,dOH~~ srics (X& a angk tfotation of tetrabcdra):from TAKEDAand MOROSIN(1975).

A. Oxygen o(4) = OH is bonded to three M2* octahedral cations; it is in fourfold coordination (3M’+, 1Ii+). Chlorine, replacing hydroxyl, moves towards potassium and remains bonded to 3M”. In phiogopite, the distance OH-K is 4.01 A; the distance Cl-K is 3.58 A. In annite, these distances are, respectively, 3.97 A and 3.52 A. Chlorine in trioctahedral micas thus can interact with potassium, and we consider ~hkxin~ as being in fourtbld coordination (3M2+, iK+). The only tabulated value of the ionic radius of chlorine is ret = 1.8 1 A, for Cl in sixfold coordination. To obtain a value of r& from rz, we decwsed the ionic radius of Cl” in the same proportion as the radius of oxygen decmases from sixfold to fourfold coordination. Such a ~tion gives r6 = 1.78 A. With the values of the ionic radii of oxygen and chlorine taken in the correct coordinations, the shortest distance& u&o. (contact-distance) are rQ + rq,. 2) = 3.13 A ti ra + ra3) = 3.16 A. From Table Zb, it is clear that the distances ClO(l, 2) short = 3.09 A and C&O(3) = 3.08 A in Phlogopite are too short to accommodate chlorine

without great deformation of the local structure. In contrast, in annite, all the Cl-0 distances are greater than the required minimum vahnz and chlorine can enter, without large defbrmation. In annite, the chlo rine site is larger and more regular than in phlogopite. The set ofdistances given in Table 2b was calculated from published crystal data on the end-members, phlogopite and annite. Detailed crystal structures of intermediate micas of the ~l~pit~nni~ solid solution have not been determined, but it is possible to predict the change in these atruuuma with changing composition using some simpk and reliabk structural models (DONNAY i?f d., 1964, HAZBN and WONES, 1972; GuIDoTn ef al., 1975; TAKEDAand MOROSIN, 1975). In this seria the main structural modification due to Fe”‘-?+@+ exchange is the variation of the angle Q, the rotation of mtmhedm about the direction T-O(3) roughly parallel to C. From the models cited above, the variation of a permits the fit of the tetrabedralsbeettothewtakdml sheet. In effect, thesizeoftbetetraWm remains c.onstant, whereas ~e~of~~~ from the pb@qkte end-member (r& = 0.720 A) to the am&e end-

M.

Volfinger n ol.

A

b

FIG. 3. Diagram of the (OH, Cl) site and nature of the interactions between Cl and the surrounding oxygens. Above, projection on the u-c plane; below, projection in the u-6 plane.

member (r$+ = 0.770 A). For phlogopite and annite (Table 2a and 2b), HAZEN and BURNHAM (1973) found a values of 7.5’ and 1.5’ respectively. From their geometrical model of the effects of cation substitution on crystal structure, TAKEDA and MOR~SIN (1975) give u values of -6.5” and -I* for the same micas. All the data from the literature show that annite, with a low a angle, is very close to the ideal mica structure, with large and regular potassium and hydroxyl sites. In contrast, phlogopite, with a higher a angle has smaller and more distorted potassium and hydroxyl sites. In the series phlogopite-annite, a does not decrease linearly with increasing Fe/Fe + Mg ratio. TAKEDA and MOR~SIN (1975) have

shown that (Y decreases greatly from ~6.5” at XF, = 0 to -2.5” at X,, z 0.4, between XF, = 0.4 and X,, = 1,thedecrease of (Yis less, from ~2.5” to z I”. The variation of (Y with X,., from TAKEDA and MOR~SIN (1975), is reproduced in Fig. 2. Interpretation of experimental results /or the phlogopite-annite series In Fig. 2, it is clear that the chlorine content of mica C&I, in the phlogopite-annite series, is related to (Y. For XF, = 0 (phlogopite end-member), a is large and the distance OH-Q 1, 2) is too small to allow accommodation of a significant amount of

Cl in micas and amphiboles

43

(OH-Cl) site remains quite distorted even at Xc, = I. In contrast, the replacement of Mg by Ni in the Mg-Ni series increases the tetrahedral rotation angle x Y 2 o (-9” in the Ni-end-member, HAZEN and WON% O(l.2) long 0.325 -0.269 0.166 1972), because the ionic radius of Ni2+ in octahedral O(l,2) short 0.518 0 0.166 coordination (0.700 A) is smaller than that of M$ Phlogopife O(3) 0.630 0.166 0.390 OH 0.133 0 0.401 in the same coordination (0.720 A). In this series, Cl 0.116 0 0.356 the size of the (OH-Cl) site decreases with increasing O(l,2) long 0.303 -0.254 0.167 XNi = Ni/Ni + Mg. The chlorine content of the Mg0(1,2) short 0.543 0 0.168 Amite O(3) 0.629 0.167 0.389 phlogopite end-member is very low. With increasing OH 0.124 0 0.393 Cl 0.108 0 0.348 XNi, CAt should be even lower, but these values of CAt are so low that no significant variation could be measured as shown by Fig. 2. Table 2b. Distances between OH or Cl and the surrounding oxygens in the Qualitatively, the behavior of the Mg-Co series is anionic sites of phlogopite comparable to the phlogopiteannite series. In the and annite (in angstroms). phlogopite-annite series, (r was measured for both O(l.2) long O(I,Z) short O(3) end-members by single crystal X-rays diffraction 3.39 3.07 PhlogopiteOH 3.69 (HAZEN and BURNHAM, 1973) but, for the Co and cl 3.42 3.09 3.08 Ni end-members, (Ywas calculated from a structural 3.51 3.14 &mire z 3.57 model (HAZEN and WON& 1972). The incorporation 3.30 3.24 3.17 of Cl depends on the (OH-Cl) site dimension; in a given mica, this site dimension depends on a! but also on the dimensions and distortions of the surchlorine (CA, is very low). Between XF, = 0 and XF, rounding polyhedra. Within a mica series a is a = 0.4, a decreases but remains too large to allow an valuable guide to characterize the (OH, Cl) site dimension. The phlogopite-annite series is a good increased accommodation of chlorine (C*,.C~remains very low). For XF, > 0.4, o gradually decreases which representation of natural biotites, and in the following corresponds to a continued increase of the small OH- examples, we refer to this series. Q(1, 2) distance. Between Xrc = 0.4 and XF, = 1.0, the chlorine content of mica (C’,,n) increases reguGEOCHEMICAL CONSIDERATIONS larly, indicating an increasing accommodation of the site for Cl. The value XF, - 0.4 is noteworthy. It In nature, (OH-Cl)bearing minerals often crystallize corresponds to discontinuities in the variations of in equilibrium with an aqueous fluid phase whose both (Yand CA,_c,with XF,. composition is unknown. From the previous considerations, it is possible to predict what kind of mica Interpretation of results in rhe iv&-Co and Mg-Ni is the most able to accommodate chlorine. At equimica series librium, for a given composition of the fluid phase, The Mg-Co mica series behaves qualitatively like micas with low CYangle will contain more chlorine; the phlogopite-annite series with respect to OH-Cl micas with small tetrahedral cations (e.g., high Si/AI substitution. The variation of a in the Mg-Co mica ratio) and/or large octahedral cations (e.g., high Fe2+/ series has not been determined in detail; (Yis known Mg ratio). Figure 4 is a general summary; natural only for the end-member, smaller than the a angle tetrahedrally or octahedrally coordinated elements of phlogopite. Between the Mg end-member (phlog- are Si, Al, Ti, Fe and Mg In addition, some octahedral opite) and the Co end-member, the angle a! decreases sites are generally vacant. In practice, compositional and the (OH-Cl) site becomes larger just as in the variations can conveniently be represented by (Si/ phlogopite-annite series. The main difference between Si + AI)“’ and XF, = Fe/Fe + M 2+ + AI” + Ti. the behavior of the two series lies in the difference From Fig. 4, the chlorine content of mica must between the ionic radii of Co& thib,+pin]= 0.735 A increase with each parameter, for a given Cl content and Fe&. = 0.770 A. To obtain the minimum size of the fluid phase with which mica is in equilibrium. of a necessary for the mica structure to accommodate Keeping in mind the structural control of OH-Cl measurable amounts of Cl, the mica requires Co/Co substitution in mica, due to its cationic composition, + Mg > 0.6. Just as in the phlogopite-annite series, it should be possible to estimate the Cl content of a discontinuity is observed in the chlorine content of the fluid phase which was in equilibrium with micas mica in the Mg-Co phlogopite series as a function of either from laboratory experiments or from nature. Xc,, but at Xc., - 0.6 (>XFe - 0.4). Similarly, the As a first example, consider biotites of the phlogmaximum chlorine content of mica in the Mg-Co opite-annite series investigated in this work. All these phlogopite series, at the Co-end-member, is lower micas were crystallized in a solution of KCI - 0.5 than that of the annite end-member, because the M, at 600°C and 2 Kbars. For biotites with XF, Table 2a.

Coordinates of 0(1,2) (long X-O bonds),0(1,2) (short K-O bonds),0(3),OH and Cl in phlogopitc and annite, after Hazen and Burnham 119731

M. Vollinger er u/

tetrahedral

batholith crystallized in a pure KC1 solution. as in the experiments, they could not have crystallized in equilibrium with such a Cl-poor solution because o( decreases Al their Cl content is approximately 40 to 20 times TI . greater than the Cl content of our synthetic phlogopite, ln this case, we must consider that biotites of Guichon Creek batholith crystallized in the presence of a chloride-rich fluid whose concentration was several molar. This conclusion is the same as that of Hot.LAND( 1972) who concluded that highly concentrated saline chlorine solutions are required to fix appreciable amounts of chlorine in biotites. It can be seen from Table 3. that the Cl content of Bethsaida biotites is about twice as low as the Cl octahedral content of Bethleem and Skeena ones. It was shown cations above that the critical parameter cv is the same for FIG. 4. Diagrammatic repfe3entation of the ability of all micas of the three Guichon Creek intrusive phases. trioctahedral micas to accommodatechlorine,as a qualitative Hence, if temperature is constant, which is a reasonfunction of their composition. The same dim applies to able assumption (JOHAN, pers. commun.), the decrease amphiboles. in Cl content of biotites from phase 2 to phase 3 is due to a decrease in the chlorine content of the coexisting fluid phase. = 0.3,0.6 and 1.O, the Cl contents at these conditions If the chlorine-bearing species in the fluid phase are, respectively: 11 ppm, 33 ppm and 335 ppm. include not only KCI, but also HCI, the Cl content Biotites with 0.3 < XF, < 0.6 are reprrscntative of of biotites can be higher, In such an acidic environthe majority of natural ones. ment (MUNOZ and SWENSON. 1981), annite incorIn a second example, we consider biotites from a porates about 10 times more chlorine than under the well-characterized porphyry copper deposit, the Guiconditions of our experiments. chon Creek batholith (Canada) (JOHAN et al., 1980). Biotites of the early intrusive phase (Guichon and From geological and petrological studies, thm sucChataway) contain about twice as much chlorine cessive intrusive phases are recognized. All three (1360 ppm) as those of the later phase 2. According phases contain biotites with quite constant composito JOHAN et al. ( 1980) no fluid phase was evolved tion: during the crystallization of these two intrusions. In this case, considering the high chloride concentrations of solutions evolved from intrusive phases 2 and 3, X (Si~.~~Al~.~~)0~o(OH, Cl).. intrusions of phase 1 may have been chloride-rich In order to compare the Cl contents of these silicate melts. Because of the strong preference of Cl biotites with the results of our experimental work, for the aqueous phase (KILNIC and BURNHAM, 1972). on the phlogopite-annite series, we must estimate the such a melt could have produced the highly concencritical structural parameter (Yof the natural biotites. trated chloride brines of phases 2 and 3. A quantitative evaluation of this suggestion requires experiments on Starting from a phlogopite composition (a f 7.59, the partitioning of chlorine between silicate melts, a decrease in Si/Si + Al (0.750 in phlogopite; 0.706 in the present biotites) results in an increase of a. In biotites and highly concentrated hydrothermal solucontrast, an increase in X,, (0 in phlogopite; 0.33 in tions. As a third example, we consider zones of contact the present biotites) results in a decrease of a. Using these values, and the experimental data of ROBERT metamorphism in green-phyllites and metagraywackes ( 198 I), the estimated a angle for these natural biotites intruded by granitic magmas of AlijbSanhns, Northis that of phlogopite (-7.5“). It is then possible to em Portugal (NEIVA, 198 I). compare the natural data to our experimental ones for phlogopite. It is clear from Table 3, which gives Table 3. Characteristics and occurthe chamcteristics and occurrences of the natural rences of biotitcs of the Guichon Creek batholith. biotites, that only the Cl content of biotite varies in the different intrusive phases of the Guichon Creek BETHLEUSW CHATAWAY BETWAILU Localities GUICHON batholith. Since a free fluid phase evolved from the SKEENA two late intrusive phases (JOHAN and MCMILLAN, Si/Si+Al 0.703 0.712 0.705 1980), we can compare the Cl contents of their 5. 0.336 0.330 0.324 biotites to our experimental data. Cl (ppm) I360 625 380 In our experiments (PHI = 2 Khan, T = 600°C), 3rd 2nd 1st Intrusive (late,uith evolved fluid) ptlll*e (early) the Cl content of phlogopite grown in a 0.5 M KC1 solution is 15 ppm. If biotites of the Guichon Creek Fell’ .-*

5

I

Al

St

ca tlons

45

CI in micas and amphiboles

Table 6.

Cb~ractcri~tfa of bi&itss from phylSitas and sahistu in Qont6ct Pataaorphfe 6on66 around grWIit66 of Alij&Sanffnn, North Portugal.

Si/Si+Al

0.677

0.695

0.650

0.689

0.656

0.664

0.69

0.71

0.67

0.71

0.70

0.71

0.71

h

0.66

0.48

0.41

0.44

0.43

0.45

0,42

0.46

0.45

0.43

0.4b

0.46

0.45

300

1500

200

Boo

400

900

600

700

IOM

4200

6oo

2100

4MO

Cf(wd

Two zones af contact metamorphism are recognized around each granitic body. The compositions of biotites in these rocks (Si/Si + Al and XEJ and their chlorine contents are given in Table 4. Compared to the previous example (Table 3). biotites of Table 4 have a similar Si content but have a greater xFe. Applying the structural considerations developed previously, a biotite with X,, z 0.45, as in the present example, can accommodate very tittie chlorine (see Fig. 2). it has a high a angle, enlarged by low Si content (Si/Si + Al - 0.70). In Table 4, slight differences in Si/Si + Al and XF, occur between biotites of the inner and the outer zones. The compositional variations are too small, however, to modify a significantly or to explain the observed differences between the C1 content of biotires of the inner and outer zones. The Cl content of biotites of the inner zones is systematically larger (700-2000 ppm) than that of biotites of the outer zones (200-600 ppm). As in the previous example, if these biotites crystallized in ~uiIib~um with a pure KCI-H& bearing solution, their high CI contents can be explained only by equilibrium with a Cl-rich fluid (several M KCI). Using the same assumption, the differences between the Cl contents of biotites of the inner and outer zones correspond to a decrease in the concentration of chloride in the fluid with increasing distance from the intrusive granite. In the same way, biotites of the latest facies (in pegmatites and quartz-veins) have the highest Cl content, as much as 4200 ppm. Such high Cl contents require equilibrium with late highly concentrated chloride fluids (brines?). As in the previous example. high Cl contents of biotites coutd be also explained by the presence of an acidic Cl-bearing species, for example HCI. in the fluid phase. Consequently, the decrease of the concentration of Cl in biotites from inner to outer zones could be due to a decrease of HCI/KCI ratio, A third parameter, temperature, must be considered. In contact me~mo~hism, the temperature of outer zones is less than that of inner zones. From experimental data of MUNOZ and SWENSON(I 98 1)” a decfwse of temperature favours incorporation of chlorine in annitc. If this behavior is valid for all biotites, the differences of Cl content between inner

and outer zones cannot be explained by a temperature effect. As the crystal chemical effect is negligible in the present case, we must conclude that the nature and ~n~nt~tion of Cl bearing species in the fluid phase is of fundamental importance. We take as the last example micas in mafic rocks, peridote xenoliths in kimberlites described by DELANEY et al. (1980). These authors distinguish two textural groups of biotites: primary biotites which are in ~uilib~um with other m&c silicates and secondary ones. which form alterations rims, around garnets and orthopyroxenes. We consider here the primary biotites because of their remarkably constant composition, whatever their occurrence:

X

(Sis.~~Al2.,~~2o(OH, F, C%

which is very close to the phlogopite end-member. The chlorine content of these micas varies from 400 lo 1100 ppm (DELANEY et al., 1980) but a direct comparison with the experimentai results, given in Fig. 2 and Table 1, is not justified because of differences in the T-P conditions of equilibrium. However, some recent experimental data conceming relations between malic magmas and fluids performed by one of us (M.V.) as part of a general study on the metallogeny of chromium ores (JOHANet al., 19841, are available. Data were obtained at iOOO’C, 8 kbars, T-P conditions closer to those at which micas described by Delaney et al. formed. In these experiments, a biotite crystallizes from a basaltic melt when KCI-bearing fluid reacts with the magma. The other phases are liquid (glass), pyroxenes and spinel. The composition of this biotite is

similar to phlogopite, but with a larger (Yangle due to the presence of some octahdd atuminum (see Fig. 3). Cornto the natural biotites of this example, these synthetic biotites thereforc can accommodate less chlorine. The analysis of the experimentally-produced biotites have shown that when the (Na, K) Cl molarity of the starting fluid is 3 M, the Cl content of the biotite is no more than 300

M. Volfinger M a/.

46

ppm. But when the chloride concentration of the fluid is 9 M, the Cl content of the biotite reaches 2700 ppm. If the synthetic and natural biotites are comparable, we may conclude that natural micas of this example equilibrated with a fluid phase having a chloride concentration greater than 3 M. EXTENSION

TO A~PHI~L~

The models of crystal-chemical control of the incorporation of chlorine in trioctahedral micas applies to amphiboles because it takes into consideration the local environment only. which is the same in the two mineralogical families. The anion (OH, F, Cl) is adjacent to three octahedral cation positions and lies in the middle of a pseudo-hexagonal ring of apical oxygens of (Si, Al) tetrahedra. In micas, distortions of the tetrahedral sheet are frequently required to fit it to the octahedral sheet; the higher the distortion, the less the hydroxyt site can accommodate chlorine, as demonstrated in the first part of this paper. Similarly, in amphiboles, distortions of the tetrahedral chains are required to fit the octahedral strip to it. In micas, the tetrahedral rotation angle zy is used to measure the distortion of the tetrahedral sheet; the same angle exists in amphiboles. in amphiboles, the angle 0(5)-0(6)-O(5) (see Fig. 5) is classically used to measure this distortion (PAPIKE ef al.. 1969); we refer to it as 6. The tetrahedral rotation angle cy, in amphiboles. is related to d by: Q = (X - 8)/Z. In most amphiboles. the kinking of the chains, 6 = 0(5)-0(6)-O(5), is in the range 160” to 180”, which corresponds to a rotation angle a in the range 0” to 10’. very similar to that of trioctahedral micas.

FIG. 5. Diagram of the alkali site of amphibole. Only the basal faces of tetrabedra are represented. The heavy lines cormspondto the chains above the b-c plane, the liit lines, to the chains under 6-c plane. 6 measures the kinking of the chains.

It measures the rotation of tetrahedra around the TT( 1Wt I ) or T(ZKX2) in amphiboles, roughly parallel to a* and around T-O(3) in micas, roughly parallel to P. Because the crystal-chemical reasons and the mechanism of distortion of tetrahedral chains of amphiboles are identical to that of the tetrahedral sheet of micas, the ability of amphiboles to incorporate chlorine must be controlled by the same structural (r.e., the same chemical) parameters. and specifically by the parameter Xr,. Natural chlorine-rich amphiboles are rare. They belong to the hastingsite group; clin~mphi~les with high iron content, and especially with high octahedral Fe” contents. Among these amphiboles is chlorpotassium hastingsite (dashkesanite), described first by KRUTO~ f 1936). More recently. high chlorine amphiboles of the same type have been described by JACOBSON( 1975) and by DICK and ROBINSON( 1979) in skams, with the structural formula: O pp,aaldirection,

by SHARMA( 198 I), in scapolite-bearing calcic metamorphic rocks, with the formula:

x f%.~Al I 92)OZttCi and by KAMINENI el al. (1982) nockite:

I

.,OHmd

in an Indian char-

One of the main characteristics of these Cl-rich amphiboles is their high iron content. This otiation is in satisfactory agreement with those made pmvi~usly for micas. Fig. 4, which is a qualitative diagram correlating the chemical composition of trioctahedral micas to their ability to accommodate chlorine, is also applicable to amphiboles. The positive correlation between chlorine content and X,, has been observed in actinolitic amphiboles (VIELZEUF, 1982); it is illustrate in Fig. 6. In these actinolites, derived from the retrogressive breakdown of orthopyroxenes, chlorine content is related to XF, by: Cl (ppm) = 43269X~, - 9762 (correlation coefficient T = 0.9 I f. From this formula, and an estimated molar weight of 867 g (average of 17 analyses, VIELZEUF,1982), one can calculate a relation between the atomic fraction of chlorine, X0 = Cl/(Cl + OH), and XF, in these amphiboles. The relation is: XO = 1.06X& - 0.24. The relation predicts that beyond the value X, = 0.24, the propottion of chlorine contained in amphibole follows almost exactly the iron content fX& (the slope of the regmssion line is very close to If. The value XF, = 0.23 cormsponds to the opening of the (OH, Cl) site of the amphibole to Cl anions. This situation is similar to that observed

47

Cl in micas and amphiboles in trioctahedral micas (Fig. I). Another charirmeristic of high chlorine amphiboles is their high potassium content. A positive correlation between Cl and K in actinolites has been demonstrated by WELZEUF (1982); this correlation is not coincidental. The same parameter which controls the dimensions of the anionic site (OH, F, Cl site) controls the size of the alkali site in micas and amphiboles. In amphiboles, another parameter also controls the dimensions of the alkali site:

a = 90” - [0(7)-o(7)‘-0(7)“1 90” (PAPIKEet al., 1969), which measures the distortion of the parallelogram of four oxygens O(7) around the alkali cation (see Fig. 5). Both parameters a and A, which reflect the adaptation of the tetrahedral chains to the octahedral strips, are controlled by the dimensions of polyhedra, and particularly, by their iron content (XF,). Thus, we consider that the positive correlation between Cl and K is a consequence of the increasing of XFe.

CONCLUDING REMARKS This study confirms, for anion exchange in silicate minerals, the fundamental role of the local structure of the anion site. In the case of micas and amphiboles, the rotation of tetrahedra controls the dimension of the anionic site (OH, F, Cl site). The closer the symmetry of the ring of six tetrahedra (in the sheet for micas, in the double chain in amphiboles) is to ideal hexagonal symmetry, the easier the replacement of OH- by Cl-. In micas, for given P, T and

composition of coexisting hydrothermal solution, chlorine is accommodated relatively more easily in annite, but much less easily in phlogopite whose (OH, F, Cl) site is more distorted. One can predict from our structural model of OH-Cl exchange that under the same conditions, the incorporation of chlorine in muscovite, with its strongly distorted (OH, F, Cl) site, will be even less than in phlogopite. Similarly, pargasitic amphiboles whose (OH, F, Cl) site is relatively distorted, will accommodate less chlorine. Taking into account structural considerations, it has been shown that for an approximately 0.5 M KCI solution in equilibrium with mica, the maximum chlorine content of mica is 300 ppm (i.e., in the most favourable structural case of annite). This implies the existence of highly concentrated chloride solutions (several molar), or other Cl-bearing species (HCI), in equilibrium with natural micas and amphiboles whose chlorine contents are one to several thousands ppm. Throughout this work, chlorine has been considered as an anion, exchangeable with hydroxyl; this assumption is likely under our experimental conditions. However, the possibility of incorporation of chlorine as molecular HCl must be kept in mind, specially in the interlayer sites of micas. The existence of molecular HCI in silicates has been demonstrated in scapolites (DONNAY et al., 1978). The question of the nature of chlorine in micas and amphiboles could certainly be answered by the use of spectroscopic methods on appropriate samples. To obtain relations of interest in geochemistry, quantitative experimental data on the OH--Cl- exchange as a function of the nature and concentration of chloride species in the fluid (concentrations of KCI. NaCl. HCI, etc) are required. For this work, the structural similarity of most natural biotites with the phlogopite end-member, measured in terms of the tetrahedral range

rotation

angle

a (systematically

in the

7”-8”) will help in applying the experimental

results. AcknoH,IedRemenrs-The authors are indebted to F. Dclbove, Z. Johan and L. Le &I for helpful suggestionsabout crystalchemistry and geochemistry of chlorine in silicates. Edrrorinl handlinK: J. M. Ferry

REFERENCES

FIG. 6. Correlation of the Cl content with XF, in a series of seconda~ actinolites from French Pyr&es. The equation of the regression line is: Cl (ppm) = 43269XF. - 9762 (r = 0.91).

DELANEY J. S., SMITH J. V., CARSWELL D. A. and DAWSON J. B. (I 980) Chemistry of micas from kimberlites and xenoliths-Il. Primary and secondary textured micas born peridotite xenoliths. Geochim. Cosmochim. Acra 44, 857872. DICK L. A. and ROBINSON G. W. (1979) Chlorine-bearing potassian hastingsite from a sphalerite skam in southern Yukon. Can. Mineral. 17.25-26. DONNAYG., DONNAYJ. D. H. and TAKEDA H. (1964) Trioctahedrai one-layer micas-II. Prediction of the st~cture from composition and cell dimensions. Acfa Crysl. 17, 1374-1381. DONNAYG., SHAW C. F., BUTLER 1. S. and O’NEIL J. R. (1978) The presence of HCI in scapohtes. Can. Mine&. 16, 341-345.

48

M. Volfinger ef al

GIJIIXTTI C. V., CHENEY J. T. and CONATOREP. D. (1975)

MUNOZ J. L. and SWENSONA. (1981) Chloride-hydroxyl Interrelationship between Mg/Fe ratio and octahedral Al exchange in biotite and estimation of relative HCI/HF content in biotite. Amer. Mineral. 60. 849-853. activities in hydrothermal fluids. Econ. Geol. 76, I!?I?HAMILTOND. L. and HENDERSONC. M. B. (1968) The 2221. preparation of silicate compositions by a gelling method. NEIVAA. M. R. (I 98 I) Geochemistry of chlorite and biotite Mineral. Mag. 36, 832-838. from contact metamorphism of phyllite by granites. MeHAZENR. M. and WONESD. R. (1972) The effect of cation mdrias e Noticias, Pubi. Mus. Lab. Mineral. Geoi., Untv substitution on the physical properties of trioctahe-dral Coimbru. N” 91-92. 113-134. micas. Amer. Mineral. 57, 103-125. PAPIKE J. J.. Ross M. and CLARK J. R. (1969) CrystalHAZEN R. M. and BURNHAMC. W. (1973) The crystal chemical characterization of clinoamphiboles based on structure of one layer phlogopite and annite. Amer. Minhve new structure refinements. Mineral. Sot. Amer Spec eral. 58, 889-900. Pup. 2. 117-136. HOLLANDH. D. (1972) Granites, solutions and base metal ROBERTJ.-L. (I 98 I) Etudes cnstallcchimiques sur les micas deposits. &on. Geol. 67.28 I-30 I. et Ies amphiboles. Applications a la p&rographie et j la IIYAMA J. T. (1965) Influence da anions sur les equilibtes geochimie. These d’Etat, Universite Paris XI, France. d’ichanges d’ions Na-K darts la feldspaths alcalins a ROBERT J.-L. and VOLFINGERM. (1975) Cristallochimie des micas trioctaedriques. Rapport annuel d’activite, 600°C sous une pression de 1000 ban. Bull. Sot. Franc CRSCM-CNRS, Orleans, France. Mineral. Crist. 88, 6 18-622. ROBERTJ.-L. and VOLF~NGERM. (1976) Echange OH--Cl JACOBSON S. S. (1975) Dashkesanite: ehlotine amphibole dans les micas trioc&hiques. Rapport annuel d’activ%, from St. Paul’s rocks, Equatorial Atlantic Transcaucasia, U.S.S.R. Min. Sci. Invesr.. Co&b. Earth SW Smithsonian CRSCM-CNRS. Orlians. France. SHANNON R. P. and FREWITT C. W. (1969) Effective ionic Inst. 14, 17-20. radii in oxides and fluorides. Acta Cryst. B25, 925-946. JOHAN Z. and MCMILLANW. J. (1980) Possibilites d’indiSHANNONR. P. and PREWI-ITC. W. (1970) Revised values vidualisation d’une phase flu& ci partir d’un bain silicat& of effective ionic radii. Actu Crysf. B26. 1046-1048. consequences geochimiques. Exemple du batholite de Guichon Creek. In Min&uiisationsLi&s aux Granitoides. SHARMAR. S. (198 I) Mineralogy of a scapolite-bearing rock Memoire du B.R.G.M. n099, 26th I.G.C.. Paris. from Rajastan. northwest peninsular India. Lithos 14, 165-172. JOHANZ., LE BEL L. and MCMILLANW. J. (1980) Evolution TAKEDAH. and MOR~SINB. (1975) Comparison of observed geologique et petrologique des complexes granitoides ferand predicted structural parameters of mica at high tiles. In Mi&ralisations Likes aux Granitoides.M&moire temperature. Actu Cryst. B31, 2444-2452. du B.R.G.M. n”99. 26th I.G.C., Paris. VIELZEUFD. (1982) The retrogressive breakdown of orthoJOHAN Z., DUNLOP H., LE BEL L., ROBERT J.-L. and pyroxene in an intermediate chamockite from Saleix VOLFINGERM. (1984) Origin of chromite deposits in (French Pyre&s). Bull. Mineral. 105, 681-690. ophiolitic complexes: evidence for a volatile and sodium VOLFINGER M. (1980) Cristallochimie des distributions rich magmatic liquid and a reducing fluid phase (in prep.). d’elements. Gas de Sr dans le celsian. Rapport annuel KAMINENI D. C.. BONARDI M. and R\O A. f. (1982) d’activite, CRSCM-CNRS. Orleans, France. Halogen-bearing minerals from Airport Hill. VisakhapatVOLFINGERM. and ROBERTJ.-L. (1980) Structural control nam. India. Amer. Mineral. 67, 1001-1004. of the distribution of trace elements between silicates and KILINC 1. A. and BURNHAMC. W. (1972) Partitioning of hydrothermal solutions. Geochim. Cosmochim. Acta 44, chloride between a silicate melt and coexisting aqueous 1455-1461. phase from 2 to 8 kilobars. Econ. Geol. 67, 231-235. WONES D. R., BURNS R. and CARROLLB. (1971) Stability KRUTOVG. A. (1936) Dashkessanite-a new chlorine amand properties of annite. Truns. Amer. Geophys. Union phibole of the hastingsite group. Dokl. Akad. Nuuk. 52. 369-370. S.S.S.R. (Geol. Ser.) 341-373 (in Russian).