Vous êtes sur la page 1sur 12

American Mineralogist, Volume 86, pages 640651, 2001

Thermodynamics of the amphiboles: Anthophyllite-ferroanthophyllite and the ortho-clino phase loop


BERNARD W. EVANS,*,1 MARK S. GHIORSO,1 HEXIONG YANG,2 AND OLAF MEDENBACH3
2 1 Department of Geological Sciences, Box 351310, University of Washington, Seattle, Washington 98195-1310, U.S. A. Geophysical Laboratory, Carnegie Institution of Washington, 5251 Broad Branch Road, N.W., Washington, D.C. 20015-1305, U.S.A. 3 Institut fr Mineralogie, Ruhr-Universitt Bochum, D-44780 Bochum, Germany

ABSTRACT
Ten new single-crystal X-ray structure refinements of unheated and heat-treated anthophyllite, new measurements of the optical indicatrix of anthophyllite, and previously published data from Mssbauer spectroscopy of heated anthophyllite, show that temperature-dependent long-range order of Fe2+ and Mg on the M-sites of cummingtonite-grunerite and anthophyllite may be considered identical for the purpose of thermodynamic modeling. The difference in solution properties between the monoclinic and orthorhombic series, as expressed in the composition (XFe) dependence of lnKD in natural amphibole pairs, is accomodated through adjustment of an enthalpic term that is independent of order-disorder. End-member thermodynamic properties of cummingtonite and ferroanthophyllite are derived from those already known for anthophyllite and grunerite respectively, using intercrystalline KD data and a fit of the T-XFe phase loop to two critical field constraints: middle amphibolite-facies amphibolites and upper amphibolite-facies metaperidotites. Amphibolites suggest a transition temperature in the system FMSH at 555 C and XFe 0.3, whereas metaperidotites suggest a transition temperature of 650 C at XFe 0.1. LnKD for Fe-Mg exchange between cummingtonite and anthophyllite passes through zero at XFe 0.7, and as a result the T-XFe phase loop shows a minimum at this composition. Extrapolated end-member transition temperatures are estimated to be 800 C (Mg) and 450 C (Fe). At its breakdown to enstatite + quartz + H2O (790 C at 5 kbar), anthophyllite is marginally stable with respect to end-member cummingtonite, and the addition of Ca renders the breakdown reaction metastable. A stability field is possible for end-member ferroanthophyllite. Cummingtoniteanthophyllite phase relations mirror those of the analogous clino- and orthopyroxene.

INTRODUCTION
Accurate representation of calculated phase boundaries for mineral polymorphs in P-T-X phase diagrams requires that molar Gibbs free-energy differences be known with high precision, preferably better than 100 or 200 J/mol. This degree of precision is seldom attained in calorimetric measurements of heats of solution or experimental bracketing of mineral reactions. Thus, constraints of the usual kinds on the free energies of polymorphs from mutually independent sources will generally fail to represent their P-T-X relations correctly. Direct experimental reversal of polymorphic reaction boundaries quickly becomes difficult if not totally impossible as temperature falls much below 1000 C, and extrapolation from hightemperature conditions is not always possible. In these circumstances, the occurrence and compositions of polymorphs in well-defined natural situations may provide the best information with regard to their relative stabilities. This seems to be the case for the phase relations of the orthorhombic and monoclinic ferromagnesian amphiboles. * E-mail: evans@geology.washington.edu
0003-004X/01/0506640$05.00 640

Compositionally the simplest of all amphibole groups, the ferromagnesian amphiboles [(Mg,Fe)7Si8O22(OH)2] occur in a broad range of metamorphic rock types in the form of monoclinic cummingtonite-grunerite (predominantly C2/m, but P21/ m in low-temperature cummingtonite) and orthorhombic anthophyllite (Pnma). In this paper we attempt a quantitative assessment of their mutual stabilities, using what we already know about the thermodynamics of the cummingtonitegrunerite series (Ghiorso et al. 1995; Evans and Ghiorso 1995). Significant variables are pressure, temperature, and composition [mole-fraction Fe/(Fe + Mg) or XFe]. Our goal is a set of thermodynamic properties for anthophyllite-ferroanthophyllite solutions, and an isobaric TXFe phase diagram, the phase loop, for the dimorphs. Aluminous orthoamphiboles (Al-anthophyllite and gedrite) will be considered in a later contribution. Some preliminary reasoning allows us to conclude that both ends (Fe and Mg) of the phase loop possibly occur at temperatures found in the Earths crust. A compilation of Fe/Mg intercrystalline partition data from natural parageneses, most of which probably equilibrated between 500 and 650 C, shows that lnKD for the exchange reaction:

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE

641

grunerite + anthophyllite = cummingtonite + ferro-anthophyllite


Fe7Si8O22(OH)2 Mg7Si8O22(OH)2 Mg7Si8O22(OH)2 Fe7Si8O22(OH)2

(1) falls in the range 0.125 to + 0.025 for Al-free compositions (Evans and Ghiorso 1995, Fig. 3). Assuming some cancellation of non-ideal solution properties between ortho- and clinoamphibole, this range translates into a standard-state Gibbs free-energy of exchange somewhere between +1.0 and 0.2 kJ/M-cation, perhaps near the average value of 0.4 kJ/M-cation. Equation 1 may be rearranged: anthophyllite cummingtonite = ferro-anthophyllite grunerite
Mg7Si8O22(OH)2 Mg7Si8O22(OH)2 Fe7Si8O22(OH)2 Fe7Si8O22(OH)2

(2) to show that the same small standard reaction free-energy is also a measure of the P or T difference between the end-member ortho-to-clino reactions. For example, when the Fe endmember transition (RHS of 2) is at equilibrium, the Mg end-member transition will be out of equilibrium by an isobaric temperature increment T given by: T = G1/Str (2800)/(4 to 5) adopting the average free-energy change and assuming that the transition entropies Str (ortho to clino) are temperature independent and both 4 to 5 J/K.mol (Ghiorso et al. 1995, Table 7; Holland and Powell 1998, Table 5). This back-of-the-envelope calculation suggests that the difference in temperature of the ortho-to-clino transitions between the Fe and Mg end-members is possibly on the order of 500700 C. Given the published record of coexisting cummingtonite and anthophyllite in nature, we have reason therefore to expect that one of the end-member transitions, if not both, may be accessible under the P and T conditions found in the crust. The difference in the Mg and Fe end-member transition temperatures for the analogous, Ca-free pyroxenes (Pbca and C2/c) is just under 600 C at 1 bar (Lindsley 1983; Sack and Ghiorso 1994). The end-member Gibbs free energies of anthophyllite-Pnma and grunerite-C2/m are known from bracketing experiments, and have been incorporated into thermodynamic databases (e.g., Berman 1988; Robie and Hemingway 1995; Gottschalk 1997; Holland and Powell 1998; Chernosky et al. 1998). The solution properties of the cummingtonite-grunerite series have been evaluated from data on temperature-dependent, long-range MgFe order-disorder and experimental equilibria with orthopyroxene, quartz, and H2O (Ghiorso et al. 1995). In this paper, we use new single-crystal X-ray structure refinements of anthophyllite, measurements of its optical indicatrix, and previously published Mssbauer spectra, to compare long-range Mg-Fe order in the orthorhombic series with that of the monoclinic cummingtonite-grunerite series. If the temperature-dependent Mg-Fe order is the same in the two series, it greatly simplifies the problem. Anthophyllite exhibits long-range, non-convergent order of Fe2+ and Mg over the four sites M1, M2, M3, and M4. The strong preference of Fe2+ for the M4 site in anthophyllite has

long been known from X-ray structure refinements and Mssbauer and infrared spectra, as reviewed in Deer et al. (1997). The temperature dependence and kinetics of the orderdisorder process were studied by Seifert and Virgo (1975) and Seifert (1977, 1978). The temperature and composition dependence of order-disorder of Fe2+ and Mg on the M-sites of cummingtonite-grunerite are also known (Hirschmann et al. 1994). We shall show below that systematic differences in the degree of Mg-Fe2+ order-disorder between the two series are not detectable by the methods currently available. This means that the real, small difference in solution properties between the two series, evident from the compositional (XFe) dependence of lnKD, the intercrystalline partition coefficient of Mg and Fe (Evans and Ghiorso 1995), can be accommodated by minor adjustment of one energy parameter in our solution model that is not a function of ordering state.

SINGLE-CRYSTAL REFINEMENTS OF NATURAL AND HEAT-TREATED ANTHOPHYLLITE


X-ray single-crystal refinements were conducted on two samples of anthophyllite (7.3.71.10, with XFe = 0.11, and W82009, with XFe = 0.22), both unheated and after separate hydrothermal heat-treatment at 600, 700, and 800 C followed by rapid quenching. Anthophyllite sample 7.3.71.10 is the same material as studied calorimetrically by Krupka et al. (1985). An X-ray diffraction (XRD) analysis of a third sample (BM 93327, with X Fe = 0.25) was carried out for us by F.C. Hawthorne, before and after heat-treatment at 700 C. The heattreatments were done at 2 kilobars H2O pressure on the C-CH4 buffer, for one month (600 C), two days (700 C), and 3 hours (800 C). Microprobe analyses of the crystals studied by XRD methods are set down in Table 1. The analyses were done on a fully automated JEOL 733 Superprobe at the University of Washington, Seattle, using our library of natural mineral standards, 15 kV accelerating potential, integration times of 10 to 40 seconds, and the correction factors of Armstrong (1988). Formula contents (Table 1) are calculated on the basis of an anion charge of 46 per formula unit (pfu) and all Fe as Fe2+. This standard procedure tends to propagate error particularly in the Si determination to the assignment of all other cations to the C, B, and A groups in the amphibole formula. However, our average calculated ratio VIAl/Al is 0.38, which we suggest is very reasonable given that this ratio is 0.43 in the ideal gedrite end-member (Robinson et al. 1971), a composition believed appropriate for Mg-rich samples (Spear 1980). Of all the formula assignments, the least reliable is that of Na on the B and A sites. The assumption of exclusively Fe2+ in Mg-rich members of the anthophyllite series is supported by Mssbauer absorption spectra, in which Fe3+ is typically barely detectable (Bancroft et al. 1966; Barabanov and Tomilov 1973; Seifert 1978; Stroink et al. 1980; Law 1989; Ferrow and Ripa 1990). The spectrum of only one of four anthophyllites studied by Law (1989), his most Fe-rich sample, with 2.31 atoms pfu Fe, had resolvable Fe3+ (Fe3+/Fe = 0.024). Single-crystal X-ray refinements were done at the University of Washington, Seattle, the Geophysical Laboratory (GL), Washington, D.C., and the University of Manitoba, Winnepeg

642

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE

TABLE 1. Chemical analyses of refined crystals of anthophyllite with formulae


Sample no. Lab. no. SiO2 TiO2 Al2O3 Cr2O3 FeO* MgO MnO NiO CaO Na2O K2O F Cl H2O less O=F,Cl Total unheated 58.9 0.01 0.73 0.15 6.26 30.63 0.12 0.14 0.48 0.08 0.00 0.00 0.00 2.22 0.00 99.72 7.3.71.10 C44 C49 59.4 58.8 0.00 0.01 0.51 1.10 0.03 0.13 6.71 6.57 30.44 30.37 0.15 0.15 0.11 0.11 0.54 0.54 0.06 0.10 0.00 0.00 0.00 0.00 0.00 0.00 2.24 2.23 0.00 0.00 100.19 100.11 C37 59.1 0.01 0.95 0.04 6.74 30.56 0.14 0.12 0.57 0.10 0.00 0.00 0.00 2.24 0.00 100.57 unheated 56.7 0.03 1.09 0.19 12.50 25.55 0.35 0.10 0.65 0.11 0.01 0.03 0.01 2.15 0.02 99.45 W82-009 C57a C25 56.8 57.1 0.02 0.02 1.12 1.13 0.19 0.17 12.62 12.20 25.71 26.02 0.35 0.35 0.10 0.10 0.61 0.71 0.12 0.11 0.00 0.00 0.00 0.00 0.00 0.01 2.17 2.18 0.00 0.00 99.83 100.10 C38 56.8 0.03 1.09 0.18 12.65 25.94 0.35 0.10 0.62 0.11 0.00 0.02 0.00 2.16 0.01 100.04 BM 93327 unheated C93 56.18 56.30 0.07 0.05 2.32 2.25 n.d. n.d. 14.23 14.03 23.96 23.94 0.42 0.40 n.d. n.d. 0.62 0.66 0.16 0.17 0.00 0.00 0.00 0.00 n.d. n.d. 2.16 2.15 0.00 0.00 100.12 99.95

Si 7.923 Al 0.077 T 8.000 VI Al 0.039 Ti 0.001 Cr 0.016 Fe 0.704 Mg 6.142 Mn 0.014 Ni 0.015 Ca 0.069 B Na 0.000 (B + C) 7.000 A Na 0.021 K 0.000 A 0.021 F 0.000 Cl 0.000 OH 1.992 Note: n.d. = not determined. * Total iron. Calculated H2O.
IV

7.962 0.038 8.000 0.043 0.000 0.003 0.752 6.083 0.017 0.012 0.078 0.012 7.000 0.004 0.000 0.004 0.000 0.000 2.003

Atomic proportions based on anion charge of 46 7.892 7.900 7.888 7.879 7.880 0.108 0.100 0.112 0.121 0.120 8.000 8.000 8.000 8.000 8.000 0.065 0.050 0.067 0.062 0.064 0.001 0.001 0.003 0.002 0.002 0.014 0.004 0.021 0.021 0.019 0.737 0.753 1.454 1.463 1.408 6.076 6.090 5.299 5.311 5.353 0.017 0.016 0.041 0.041 0.041 0.012 0.013 0.011 0.011 0.011 0.078 0.082 0.097 0.091 0.105 0.000 0.000 0.007 0.000 0.000 7.000 7.009 7.000 7.002 7.003 0.026 0.026 0.023 0.031 0.029 0.000 0.000 0.002 0.000 0.000 0.026 0.026 0.025 0.031 0.029 0.000 0.000 0.013 0.000 0.000 0.000 0.000 0.002 0.000 0.002 1.996 1.997 1.995 2.002 2.007

7.861 0.139 8.000 0.039 0.003 0.020 1.464 5.352 0.041 0.011 0.092 0.000 7.022 0.030 0.000 0.030 0.009 0.000 1.994

7.814 0.186 8.000 0.194 0.007 0.000 1.655 4.968 0.049 0.000 0.092 0.035 7.000 0.008 0.000 0.008 0.000 0.000 2.004

7.834 0.166 8.000 0.203 0.005 0.000 1.633 4.966 0.047 0.000 0.098 0.046 6.998 0.000 0.000 0.000 0.000 0.000 2.000

(UM). At the University of Washington, we used a Huber 512 four-circle diffractometer with graphite monochromatized MoK radiation. XRD intensity data from one quadrant of reciprocal space up to 2 = 65 were collected in the -scan mode at a scan speed of 3 per minute. Three standard reflections were checked after every 97 reflections; no systematic or significant variation in the intensities of the standard reflections was observed. All XRD intensity data were corrected for Lorentz and polarization effects, and for absorption by the semiempirical method of North et al. (1968). Only reflections having intensities 2I were considered as observed and included in the refinements, where I is the standard deviation determined from the counting statistics. The laboratories at GL and UM both used a Bruker SMART CCD (charge-coupled device) X-ray diffractometer with graphite monochromatized MoK radiation, but data collection techniques varied slightly: frame widths 0.3 at GL and 0.2 at UM, frame time 30 s at GL and 60 s at UM, a hemisphere of three-dimensional X-ray data collected up to 60 2 at GL and a sphere of data to 60 2 at UM. Unit-cell parameters were refined on the basis of all integrated reflections (>10), using a least-squares technique (GL and UM). The three-dimensional data were reduced and corrected for Lorentz, polarization, and background effects using the Bruker program SAINT (GL and UM). An empirical correction for X-ray absorption was made (GL and UM) using the

program SADABS (G. Sheldrick, unpublished computer program). Equivalent reflections were merged for the final refinements (GL). Preparatory to structure refinement, the following assumptions were made for the assignment of atoms among the crystallographically distinct sites: (1) IVAl = 8.0 Si pfu with all IV Al ordered on T1; (2) VIAl (= Altotal IVAl), Cr, and Ti were confined to M2; (3) Ca and BNa were confined to M4; (4) ANa and K when present were confined to the A-site; and (5) the fractionation of Fe + Mn among the four M cation sites was determined from the refinements. After structure refinement, Mn was proportioned among the M-sites in the same manner as Fe. The results of our ten single-crystal X-ray structure refinements are given in Tables 2 and 3. Site-preferences for Fe2+ are M4 >> M1, M3 > M2; they are smaller at higher temperatures of equilibration. Preferences can be compared with those of similarly equilibrated cummingtonites and grunerites (Hirschmann et al. 1994) with the aid of plots of RTlnKD, the negative ideal ordering energies - Gid, against macroscopic XFe (Fig. 1). Because the energy of exchange of Mg and Fe2+ between the M1 and M3 sites is small, on the order of 1 kJ/Mcation (Fig. 1A), we can combine the occupancies of the M1 and M3 sites for comparison with M2 and M4 according to a 3-site model (Fig. 1B). Finally, by using a 2-site model that

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE TABLE 2. Cell dimensions and refinement statistics of anthophyllite crystals
Sample 7.3.71.10. C44 C49 C37 W82-009 C57 C25 C38

643

T(C) R.T 600 700 800


R.T. 600 700 800

a () 18.5219(9) 18.5008(30) 18.5004(25) 18.4908(36)


18.5705(9) 18.5654(33) 18.5760(10) 18.5668(9)

b () 17.9740(8) 17.9695(22) 17.9646(29) 17.9785(24)


18.0361(8) 18.0271(21) 18.0367(10) 18.0306(8)

c () 5.2725(4) 5.2738(6) 5.2710(7) 5.2743(6)


5.2876(2) 5.2845(6) 5.2847(3) 5.2867(3)

V (3) 1755.3 1753.3 1751.8 1753.4


1771.0 1768.6 1770.6 1769.8

Refls. 1779 1971 1343 2000 2061 1941 1749 1979

R 0.036 0.034 0.033 0.037 0.030 0.036 0.036 0.031

Rw 0.031 0.044 0.043 0.049 0.028 0.045 0.031 0.028

Note GL UW UW UW GL UW GL GL UM UM

BM 93327 R.T 18.5716(9) 18.0307(8) 5.2903(2) 1771.5 1401 0.029 0.023 C93 700 18.5711(6) 18.0256(6) 5.2885(2) 1770.4 1794 0.029 0.024 Note: GL, UW, UM, X-ray data collected at the Geophysical laboratory, the University of Washington, and the University of Manitoba.

TABLE 3. Site occupancies of anthophyllite


7.3.71.10

T (oC) M1 Fe Mg Mn
M2 Fe Mg Al Mn Cr Ti M3 Fe Mg Mn M4 Fe Mg Mn Ca Na

0.017(2) 0.983 0.000 0.005(2) 0.967 0.020 0.000 0.008

C44 600 0.049(3) 0.950 0.001 0.024(3) 0.952 0.022 0.001 0.001

C49 700 0.063(3) 0.936 0.001 0.022(3) 0.938 0.033 0.000 0.007

C37 800 0.065(3) 0.934 0.001 0.038(3) 0.934 0.025 0.001 0.002 0.001 0.046(5) 0.953 0.001 0.258 0.696 0.005 0.041

W82-009 0.049(2) 0.950 0.001 0.014(2) 0.940 0.034 0.000 0.011 0.001 0.034(3) 0.965 0.001 0.672 0.257 0.019 0.049 0.003

C57 600 0.114(3) 0.883 0.003 0.049(4) 0.908 0.031 0.001 0.010 0.001 0.093(5) 0.904 0.003 0.528 0.412 0.015 0.045

C25 700 0.119(3) 0.878 0.003 0.059(2) 0.897 0.032 0.002 0.009 0.001 0.097(4) 0.900 0.003 0.483 0.451 0.014 0.052

C38 800 0.132(2) 0.864 0.004 0.071(2) 0.896 0.020 0.002 0.010 0.001 0.115(3) 0.882 0.003 0.474 0.467 0.013 0.046

BM 93327 C93 700 0.087 0.151(3) 0.910 0.845 0.003 0.004 0.019 0.880 0.095 0.001 0.005 0.058 0.940 0.002 0.704 0.215 0.021 0.045 0.015 0.078(4) 0.815 0.100 0.002 0.005 0.126(4) 0.870 0.004 0.505 0.405 0.015 0.050 0.025

0.022(4) 0.978 0.000 0.328 0.631 0.006 0.035

0.032(5) 0.967 0.001 0.291 0.657 0.007 0.039 0.006

0.048(5) 0.951 0.001 0.266 0.689 0.006 0.039

A Na 0.021 0.004 0.026 0.026 0.023 0.031 0.029 0.030 0.008 0.000 K 0.002 Note: Assignment of Cr to the M3 as well as the M2 site (e.g., Fialips-Gudon et al. 2000) markedly improves the consistency of intrasite Gid values in samples C44, C49, and C37, where Cr is variable.

averages the occupancies of the M1, M2, and M3 sites, we can incorporate into our comparison (Fig. 1C) the Mssbauer data for heat-treated anthophyllite obtained by Seifert (1978). In cummingtonite (but not in grunerite), there is a small calculated temperature dependence of RTlnKD (see Fig. 3 in Ghiorso et al. 1995), reflective of excess entropy related to order-disorder. The precision and accuracy of site occupancies measured by XRD diffraction, in both cummingtonite (Hirschmann et al. 1994) and anthophyllite (this work), are not sufficient to show this dependence over the range 600 to 750 C, although a decrease in RTlnKD for the 2-site model shows clearly in the Mssbauer data of Seifert (1978). More important in the present context is that, allowing for 2 and all other possible uncertainties, Figures 1A, 1B, and 1C show no systematic differences in Fe-Mg order between anthophyllite and cummingtonite-grunerite, at least in the range of XFe of the anthophyllites studied by XRD and Mssbauer spectroscopy (0.11 to 0.25). Except for exchange between the M1 and M3 sites, cummingtonite and anthophyllite in fact seem to share small changes in Gid as the Mg end-members are approached. This similarity may reflect energetic consequences of a compositional change that in cummingtonite eventually induces a change from C2/m to P21/m symmetry.

Only rarely is anthophyllite encountered in nature more Ferich than XFe 0.30 without it containing significant amounts of Al. More Fe-rich natural anthophyllite trends compositionally through Al-anthophyllite toward gedrite, although a broad solvus gap below about 600 C separates the two (Robinson et al. 1971; Spear 1980). X-ray and Mssbauer data (Papike and Ross 1970; Seifert 1978) show that the uptake of Al in orthoamphibole is accompanied by increasing Fe2+-Mg disorder. Thus, a direct comparison of order-disorder in the simple system MFSH cannot be taken to more Fe-rich compositions, unless we resort to the use of synthetic anthophyllite. A range of intermediate Fe, Mg-compositions with orthorhombic symmetry have been synthesized in the system FMSH (Cameron 1975; Popp et al. 1976), but their space group symmetry was not determined unambiguously; they are not anthophyllitePnma. This material might be worthy of study in the future.

THE OPTICAL INDICATRIX OF ANTHOPHYLLITE AND CUMMINGTONITE-GRUNERITE


The optical properties of anthophyllite and cummingtonite provide supporting evidence for their indistinguishable intrasite exchange energies. Evans and Medenbach (1997) showed that the shape of the optical indicatrix (2V) of cummingtonite-

644

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE

130

120

110

2Vz

100

90

80

70 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

(Fe+Mn)/(Fe+Mn+Mg)
F IGURE 2. 2V z of anthophyllite-ferroanthophyllite (Table 4) compared to metamorphic cummingtonite-grunerite (Evans and Medenbach 1997). Symbols as in Figure 1. Included are data for two samples of proto-ferroanthophyllite from Sueno et al. (1998).

FIGURE 1. Comparison of RTlnKD vs. macroscopic XFe for site partitioning in heat-treated cummingtonite-grunerite and anthophyllite. A = four-site model, B = three-site model, C = two-site model. Filled circles, cummingtonite-grunerite (Hirschmann et al. 1994); open circles, anthophyllite (this work); triangles, anthophyllite by Mssbauer spectrometry (Seifert 1978). Error bars are 2 for instrumental uncertainty (in most cases smaller than the symbol).

grunerite is a sensitive indicator of Fe-Mg order-disorder in a sample of known composition (XFe). Figure 2 incorporates new measurements made with a double spindle-stage (Table 4) and shows that, in the range XFe = 0.0 to 0.3, the 2Vz of natural anthophyllite is indistinguishable from that of natural cummingtonite. In this compositional range, the state of order for natural samples, as measured by s123, the 2-site order parameter where s123 = XFeM4 average (XFeM1M2M3), changes from 0.0 (end-member cummingtonite) to slightly more than 0.8 at XFe = 0.30 (Evans and Medenbach 1997). The coincidence in the behavior of 2Vz vs. XFe is interpreted as indicative of identical order-disorder in slowly cooled anthophyllite and cummingtonite. The 2Vz of more Fe-rich natural anthophyllite is smaller than that of cummingtonite (Fig. 2), but the differences may reflect the larger amounts of the gedrite component, as well perhaps as different cooling rates. Significantly, anthophyllite shows a minimum 2Vz at XFe 0.3, just like cummingonite, and for the same reason (Evans and Medenbach 1997). We also measured the optical properties of two anthophyllite crystals after equilibration in the laboratory at high temperature, followed by quenching (Table 4). The 2Vz of sample W82009 increased from 83.7 to 93.1 when equilibrated at 600 C, and the 2Vz of sample BM 93327 increased from 80.7 to 92.0

after treatment at 700 C. If the order-disorder behavior of anthophyllite and cummingtonite is assumed to be identical, then we can extract values of the two-site order parameter s123 based on (1) the measured 2Vz and XFe of anthophyllite using the cummingtonite Equation 1 of Evans and Medenbach (1997); and (2) the isothermal fits to the cummingtonite ordering data shown in Figure 2 of Evans and Ghiorso (1995). For sample W82-009 at 600 C these values of s123 are (1) 0.482 and (2) 0.488, whereas the measured value of anthophyllite from Table 3 is 0.477. All three values are in excellent agreement, suggesting that the assumptions made above are reasonable. Corresponding values for sample BM 93327 equilibrated at 700 C are (1) 0.481, (2) 0.486, and a measured value of 0.434. In this case, given XFe and temperature, the 2Vz of anthophyllite is behaving exactly like cummingtonite. The smaller measured s123 of the crystal of BM 93327 is not fully understood; it may reflect the greater amount of gedrite substitution (Al = 0.4 pfu in this sample), or possibly some interlaboratory differences. This sample plots off the general trend for cummingtonite in Figure 1C.

SOURCE OF SOLUTION PROPERTY DIFFERENCES


We are not able to compare the ordering behavior of more Fe-rich anthophyllite with that of compositionally similar members of the cummingtonite-grunerite series because low-Al orthoamphibole with XFe > 0.30 is extremely rare in nature. Such anthophyllite or ferroanthophyllite is largely unstable with respect to cummingtonite-grunerite gedrite (see below) in amphibolite-facies parageneses where it might occur; and, at lower temperatures that would favor anthophyllite rather than cummingtonite, both ferromagnesian amphiboles are unstable with respect to more hydrated and carbonated equivalents, such as minnesotaite, siderite or ankerite + quartz iron oxides. Accordingly, for modeling solution properties, we shall make the crystal-chemically reasonable assumption that the longrange Fe-Mg ordering behavior of more Fe-rich anthophyllite continues to be indistinguishable from its compositionally

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE TABLE 4. Optical properties of anthophyllite


Sample no. Unheated 31 7.31.71.10 96-34 121-73 W82-009 BM 93327 6A9 2H-348
IV

645

Al

VI

Al

Fe 0.025 0.779 0.899 1.156 1.460 1.614 2.282 3.277

Mn 0.062 0.015 0.036 0.029 0.043 0.049 0.110 0.081

Mg

Cations per formula unit Ca Na 0.052 0.077 0.077 0.054 0.090 0.088 0.119 0.048 0.056 0.020 0.017 0.004 0.030 0.024 0.194 0.167

XFe 0.013 0.115 0.136 0.171 0.220 0.246 0.375 0.522

nx

ny 1.6040 1.6248 1.6266 1.6290 1.6359 1.6379 1.6510 1.6670

nz

2Vz 127.0 103.8 99.0 91.3 83.7 80.7 74.0 76.6

0.000 0.049 0.033 0.000 0.154 0.150 0.503 0.571

0.038 0.041 0.024 0.007 0.077 0.153 0.439 0.480

6.730 6.081 5.959 5.731 5.325 5.081 3.986 3.076

1.6120 1.6147 1.6263 1.6292 1.6450 1.6580

1.6336 1.6359 1.6475 1.6507 1.6610 1.6810

Heated W82-009(600) 0.093 0.091 1.457 0.038 5.301 0.097 0.033 0.220 1.6241 1.6359 BM 93327(700) 0.228 0.182 1.657 0.048 4.948 0.135 0.059 0.256 1.6289 1.6401 Notes: XFe = (Fe + Mn)/(Fe + Mn + Mg). Sources of samples. 3-1: Klein (1968), 7.3.71.10: Krupka et al. (1985), 96-34 and 121-73: Fabris and Perseil (1971) (also Seifert 1978, samples AG1 and AG2), W82-009: Hirschmann et al. (1994), BM 93327: British Museum of Natural History, 6A9: Robinson and Jaffe (1969), 2H-348: Guiraud et al. (1996). In parentheses: temperature of heat-treatment.

1.6482 1.6515

93.1 92.0

equivalent monoclinic analog. Except for M1 vs. M3, intersite (ideal) exchange energies in both the monoclinic and orthorhombic ferromagnesian amphiboles are large in comparison to those related to the accuracy and precision of measured site occupancies, and to those related to possible differences in order-disorder between anthophyllite and cummingtonite-grunerite. The energetic consequence of temperature dependent order-disorder plays a major role in determining the solution properties of cummingtonite-grunerite (see in Fig. 2 Ghiorso et al. 1995), and the same must also be true for the anthophyllite series. The two series are nevertheless not identical in their solution properties. They differ macroscopically because it can be shown that intercrystalline Fe-Mg exchange between them is not independent of their Fe contents. A compilation of data for natural clino-ortho ferromagnesian amphibole pairs was fit to the expression: lnKD = 0.125 + 0.150 XFe + 0.046 XFe wt% Al2O3 (see in Fig. 3 Evans and Ghiorso 1995), where XFe and wt% Al2O3 refer to the orthoamphibole. This expression implies for Al-free pairs a difference of 0.15 in lnKD in going from XFe = 0 to XFe = 1, and zero lnKD at XFe = 0.75. Others have shown that the zero, or extremum, composition is close to XFe = 0.4 (e.g., Elliot-Meadows et al. 1999), but this value applies to cummingtonite coexisting with Al-anthophyllite. Although we cannot rule out small differences in order-disorder behavior between the two series, the macroscopic difference in solution properties is likely to be due, in our opinion, to one or more other crystal-structure factors. The dimorphs differ fundamentally in the manner of stacking parallel to c between adjacent tetrahedral layers (Gibbs 1969; Hawthorne 1983); the sequence in Pnma anthophyllite is (..++ ++ ..) whereas in C2/m and P21/m cummingtonite it is (..++++..). There are also important differences in bond lengths at the M4 site, and kinking of the tetrahedral chains (Boffa Ballaran et al. 2001). These considerations allow us to derive the solution properties of anthophyllite-ferroanthophyllite from those of cummingtonite-grunerite in a straightforward and simple manner. We shall assume that all the solution parameters that are determined by order-disorder in cummingtonite-grunerite, as calibrated by single-crystal X-ray refinements of heat-treated crystals (Table 9 in Ghiorso et al. 1995), also apply to

anthophyllite-ferroanthophyllite. Ghiorso et al. (1995, Eq. 5) expressed the vibrational molar Gibbs free energy of cummingtonite-grunerite solution in terms of a truncated second-order Taylor expansion in composition and ordering variables. Only three coefficients in this expansion are independent of order-disorder. Two of them are determined from standard state properties. The third coefficient, G *r,r, was determined for cummingtonite-grunerite from heterogeneous phase equi librium constraints. G *r,r differs between the two series by an amount that is determined by the difference (0.15) in the lnKD for intercrystalline exchange at the Mg and Fe extremes (Fig. 3). The required adjustment in G *r,r is from 11.2 kJ/mol in cummingtonite-grunerite to 12.2 kJ/mol in anthophyllite ferroanthophyllite. G *r,r is equivalent to W/4 where W is an effective macroscopic regular solution parameter. The absolute values of lnKD (the y-axis of Fig. 3) depend on end-member standard-state properties.

END-MEMBER THERMODYNAMIC PROPERTIES


Our recommended thermodynamic end-member properties are listed in Table 5. For end-member anthophyllite, we use thermodynamic properties given in Ghiorso et al. (1995, Table 7), which are virtually identical to those of Chernosky et al. (1998, Table 6), and consistent with other minerals in the Berman (1988) database. The heat capacity, expansivity, and compressibility of end-member ferro-anthophyllite are assumed to be identical to those of grunerite. The reaction entropy for the ortho clino transition of Fe7Si8O22(OH)2 is taken to be 5 J/Kmol, as was done for Mg7Si8O22(OH)2 (Ghiorso et al. 1995), making the monoclinic structure the high-temperature form. Hirschmann et al. (1994) showed that, corrected Ca- and Mnfree, the cell volume of end-member anthophyllite, if at all, is smaller than that of Mg-cummingtonite by no more than 0.05%. The mean refractive indices (Table 2 and Evans and Medenbach 1997) for the two series between XFe = 0.1 and 0.3 differ by a little more than 0.001, which, from the Gladstone and Dale relation (Jaffe 1988), translates into a density difference of 0.15%. If we split the difference between these two estimates, we arrive at molar volumes that differ by 0.03 J/barmol for anthophyllite and cummingtonite respectively (Table 5). It is not known whether the same volume change holds up to the Fe

646

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE

for specific values of XFe. These data in combination, together with the MgFe solution properties of the two phases derived above, and the remaining end-member properties (Table 5), will constrain the end-member enthalpies.

FIELD CONSTRAINTS
It has long been known that low-Al anthophyllite is a medium-grade metamorphic mineral typical of Mg-rich bulk compositions (metamorphosed ultrabasic rocks), whereas cummingtonite and grunerite are characteristic of intermediate FeMg and Fe-rich compositions (low-Ca amphibolites, low-K pelites, uralitized gabbros, and metamorphosed iron-formation or BIF) equilibrated at metamorphic grades reaching into the granulite facies. Cummingtonite (XFe from 0.29 to 0.5) also occurs as phenocrysts in H2O-rich silicic volcanic rocks and as a possible magmatic mineral in a small percentage of plutonic rocks. With slow cooling, cummingtonite inverts to anthophyllite and liberates a small quantity of Ca-amphibole (Ross et al. 1969; Evans et al. 1974; Carpenter 1982). These observations, and the sign of the partition coefficient lnKD, imply that the transition temperature, at least at first, declines with increasing XFe. The most useful constraints on the T-XFe location of the transition in the FMSH system are supplied by amphibolites and metaperidotites. In Figure 4 we have plotted total Al (apfu) against the XFe for microprobe-analyzed anthophyllite and cummingtonite from five well-documented examples of middle amphibolite-facies metamorphism of low-Ca amphibolites and low K-pelites (Stout 1972; Spear 1980, 1982; Clark 1978; Early and Stout 1991; Schneidermann and Tracy 1991). Cummingtonite and anthophyllite commonly coexist in these samples and in some cases gedrite and/or Ca-amphibole occur as well. Metamorphic temperatures estimated by these authors fall in the range 535 to 575 C, and pressures from 3 to 6 kbar. Possible field boundaries for these conditions have been inserted on Figure 4. Except for the base of the plot and the three-phase assemblage anthophyllite-gedrite-cummingtonite (which is virtually coplanar in composition space), phase boundaries are somewhat dependent on the compositions of coexisting minerals, as

FIGURE 3. Computed correlation of lnKD and XFe of cummingtonitegrunerite for Mg/Fe partition between cummingtonite-grunerite and anthophyllite-ferroanthophyllite. KD = (Fe/Mg)Ath/(Fe/Mg)Cum

end-members, but, for simplicity, we shall assume that it does. These small volume uncertainties affect the pressure dependence of lnKD by a minor amount on the order of 0.001/kbar. Their effect on the temperature of the transition is small but not trivial, however, although a meaningful estimate of the dP/dT of the transition is not possible at the present time. Still to be evaluated are the enthalpies of the end-members ferro-anthophyllite and cummingtonite. The Gibbs free energies of these end-members are extremely close to those of grunerite and anthophyllite, respectively, and, in the absence of any experimental constraints, the only way to evaluate the small differences involved is to analyze data from the field. Two kinds of field data can be used to constrain the free energies of the solutions: (1) field information that serves to bracket the ortho = clino inversion temperature for ideally Al-free (quadrilateral) ferromagnesian amphiboles as a function of XFe; and (2) compositional data that give robust estimates of lnKD for equilibrium pairs of cummingtonite + low-Al anthophyllite

TABLE 5. Internally consistent thermodynamic properties of end-members


H 0 (kJ) 0 f,Tr,Pr S Tr,Pr (J/K) 0 V Tr,Pr (J/bar) k0 k1 102 k2 105 k3 107 v1 106 v2 1012 v3 106 v4 1012 Notes:
anthophyllite Mg7Si8O22(OH)2 12073.132 535.259 26.31 1233.8 71.3398 221.638 233.394 1.1394 28.105 62.894 cummingtonite Mg7Si8O22(OH)2 12067.920 540.259 26.34 ferro-anthophyllite Fe7Si8O22(OH)2 9627.015 720.000 27.81 1347.83 93.5691 202.285 303.919 1.6703 8.68919 28.400 grunerite Fe7Si8O22(OH)2 9623.550 725.000 27.84

o CT,Pr = k 0 +

k1 k k + 2 + 3 T T2 T3

(J/K)

o VT,P 2 2 = 1 + v 1 (P Pr ) + v 2 (P Pr ) + v 3 (T Tr ) + v 4 (T Tr ) (J/bar) o VTr ,Pr

H 0f,Tr,Pr and S 0Tr,Pr of anthophyllite and cummingtonite are given with greater precision than individually known in order to maintain internal consistency.

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE

647

F IGURE 4. Aluminum per formula unit vs. composition of anthophyllite and cummingtonite in middle amphibolite-facies metamorphism, and an interpretation of phase boundaries. Symbols as in Figure 1. Data sources are given in the text. Four datapoints for low-Al and low-Fe anthophyllite from Telemark (Stout 1972) are from olivine-bearing, plagioclase-free rocks. Four high-Al anthophyllites, three from the Litjorn locality, Telemark, may represent slightly higher temperatures than the other samples.

illustrated, for example, by the reaction: Cum + Plag = Alanthophyllite + Ca-amphibole (Robinson and Jaffe 1969; see also Spear 1982). Extrapolating to zero Al, the data points suggest a narrow gap between anthophyllite and cummingtonite, centered on XFe 0.30. From KD data we know the gap here is about 2 mol% wide. The field of (subsolvus) Al-anthophyllite terminates at these temperatures at XFe 0.40 (Fig. 4), beyond which it reacts to cummingtonite + gedrite (Stout 1972). The many examples of anthophyllite that plot in Figure 4 in the field of cummingtonite + gedrite (and the overlap of anthophyllite and cummingtonite compositions) are interpreted to represent late-stage growth of anthophyllite at lower than maximum metamorphic temperatures, by one (or more) of the following mechanisms; (1) grain-scale exsolution from gedrite; (2) from the reaction gedrite + cummingtonite anthophyllite; or (3) simple inversion from cummingtonite to anthophyllite. Spear (1982) noted in some of his samples that anthophyllite rims gedrite and was the last amphibole to grow. Also, cummingtonite surrounded by a rim of anthophyllite has been observed in many descriptions of amphibolites (Eskola 1914; Tilley 1937; Tella and Eade 1978; Elliott-Meadows et al. 1999). There are, in general, far too many reported examples of two or three coexisting ferromagnesian amphiboles for them all to be considered to represent equilibrium frozen at one set of metamorphic conditions; collectively they represent a very small portion of potential amphibole composition space (Robinson et al. 1982). Because anthophyllite can form from cummingtonite on cooling, Figure 4 should probably be used to extract a maximum XFe for the phase boundary under the average of P-T conditions estimated for the five field areas. Roughly half the samples plotted in Figure 4 were saturated in Ca-amphibole. In these, the anthophyllites and cummingtonites contain 23.5 and 36% Ca-amphibole (actinolite or hornblende) in solid solution, respectively. In various kinds of metaperidotites, low-Ca amphibolites, and metasomatic rocks, numerous analyses in the literature (e.g.,

Rabbitt 1948; Deer et al. 1997) show that low-Al anthophyllite (<0.2 Al apfu) extends from the pure Mg end-member to XFe 0.32, consistent with the data on amphibolites presented in Figure 4. On the other hand, in the last 35 years, Mg-rich cummingtonite in the XFe range from 0.08 to 0.30 has also been reported in metamorphosed ultramafic rocks world-wide (Yakovleva and Kolesnikova 1967; Kisch 1969; Rice et al. 1974: Frost 1975; Pfeifer 1978, 1987; Kamineni et al. 1979; Matthes and Knauer 1981; Evans and Trommsdorff 1983; Andrew 1984; Srikantappa et al. 1985; Matthes 1986; Gole et al. 1987; Kleinschmidt et al. 1987; Dymek et al. 1988; Hirschmann et al. 1994; Droop 1994). In many of these localities, both anthophyllite and cummingtonite are reported. Cummingtonite seems invariably to be accompanied by tremolite. Metamorphic grade is typically high amphibolite-facies. Temperatures for the cummingtonite-bearing metaperidotites have generally been estimated in the range 600-700 C, based in most cases on the location of the anthophyllite + forsterite field in the phase diagram for MSHassuming (not necessarily correctly, see Trommsdorff and Connolly 1996) pure H2O fluid. Among the most Mg-rich cummingtonites known (XFe = 0.080.15) are those occurring in metaperidotite at Cima di Gagnone, Ticino, Switzerland (Rice et al. 1974; Pfeifer 1978, 1987; Evans and Trommsdorff 1983). Here, cummingtonite grew homoaxially on tremolite during cooling and decompression of chloriteenstatite metaperidotite during the main Alpine phase of metamorphism in the Alps. The temperature of this event was estimated from thermobarometry in surrounding rocks to be 600 660 C (Grond et al. 1995), and the locality is just on the high side of the 650 C isotherm of Todd and Engi (1997). A maximum possible temperature for this composition of cummingtonite together with olivine in the system FMSH is 720 C (see Fig. 5 in Evans and Ghiorso 1995). The anthophyllites and cummingtonites of metaperidotite such as at Cima di Gagnone contain on average 3 and 4.3% tremolite in solid solution. The phase loop for Ca-saturated cummingtonite and anthophyllite in the system CMFSH, based on our model for the thermodynamic properties of quadrilateral amphiboles (Ghiorso and Evans, in preparation), is located about 35 to 65 C below that for the system MFSH (Fig. 5). The tremolitesaturated phase loop agrees best with the field constraints from metaperidotites if we adopt an enthalpy of end-member cummingtonite that results in a transition temperature of 800 C (at 5 kbar) for the Mg end-members in the MFSH system. Monoclinic amphibole has never been reported in any of the experiments in the system MSH. Greenwood (1963) synthesized anthophyllite in an obviously metastable assemblage at 830 C and 1 kbar in a 20 hour experiment, and Chernosky and Autio (1979) and Chernosky et al. (1985) treated a similar metastable mixture for a long time at 735 C to give a high yield of anthophyllite for starting material. This material was examined by HRTEM by D.R. Veblen, who reported the presence of 10% triple-chain silicate. About 510% of anthophyllite (optically confirmed by J.V. Chernosky, personal communication, December 20, 1999) was produced at temperatures as high as 761 C in reversal experiments by Chernosky and Autio (1979) on metastable parts of the reactions: Tlc = 3 En + Qtz +

648

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE

FIGURE 5. Calculated TXFe phase loop at 5 kbar for coexisting Al-free orthorhombic and monoclinic Fe, Mg-amphiboles (2 continuous lines), and a phase loop (3 dashed lines) for three coexisting Al-free amphiboles in CMFSH. Arrows denote constraints discussed in the text.

H2O and Tlc + Fo = 5 En + H2O. Although these results are not definitive with respect to the relative stability of cummingtonite and anthophyllite in the system MSH, they are not in conflict with our field-based boundary. One must bear in mind, however, that the apparently orthorhombic Fe, Mg-amphiboles synthesized by Cameron (1975) and Popp et al. (1976) have compositions that place them (from the field evidence) clearly in the T-XFe stability field of the monoclinic form. An occurrence of virtually end-member anthophyllite at the Wight mine, Balmat, New York (sample 31, Klein 1968, with 0.15% F this work), believed to have formed at a maximum of 675 25 C (E.J. Essene, personal communication), provides a low-temperature field bracket on the end-member transition temperature. The constraint on the phase loop supplied by amphibolites (555 C, XFe 0.30) seems to agree better with the MFSH loop than the CMFSH loop, even though hornblende is present in many of the samples. An MSH transition set at 850 rather than 800 C would fit the amphibolite data better, but the metaperidotites less well. An overall flatter dT/dXFe slope for the phase-loop might reasonably fit both the metaperidotites and the amphibolites, but then the standard-state Gibbs free energy of the exchange reaction, as we found by trial and error, would be too small to satisfy the natural KD data, because it would shift the isotherms (Fig. 3) to more positive values of lnKD. Accepting our estimate of the transition temperature for the Ca-free, Mg end-members (800 C, 5 kbar), the remaining unknown, the Gibbs free energy of ferro-anthophyllite, can be obtained from the standard-state Gibbs free energy of the exchange reaction (equivalent to the enthalpy of reaction), which is tightly constrained by the natural KD data, in particular, those from metaperidotites. This provides leverage on the dT/dXFe of the phase loop and the transition temperature at the Fe-end. A value of lnK D = 0.1 for coexisting anthophyllite and

cummingtonite in metaperidotites (XFe = 0.10.2) may be read from the fit expression in Figure 3 of Evans and Ghiorso (1995), Figure 3 here. This value of lnKD is independently supported by a comparison of Fe-Mg partition in a large number of ferromagnesian amphibole + olivine pairs in metaperidotites (Fig. 6). Despite considerable scatter resulting mostly from a lack of perfect exchange equilibrium, Figure 6 shows unmistakably that, for a given olivine composition, cummingtonite is slightly richer in Fe than anthophylliteby an amount corresponding to an average lnKD of 0.1. It should be noted that our enthalpy for end-member cummingtonite is 250 J/mol more negative than adopted before (Table 7 in Ghiorso et al. 1995). Our calculated binary two-phase loop for MFSH (Fig. 5) has a temperature minimum close to XFe = 0.7 and an Fe endmember transition temperature of 450 C, or about 35 C lower in the presence of ferroactinolite. This result is consistent with the common occurrence of cummingtonite-grunerite (ranging in XFe from 0.4 to 0.98) in metamorphosed iron-formation (Klein 1982). The first occurrence of grunerite in the prograde metamorphism of iron-formation is close to the biotite isograd (Klein 1978; Haase 1982), that is, not much higher than 400 C. The only anthophyllite reported in metamorphosed iron-formation occurs in hematite-bearing rocks with high Mg/Fe 2+ ratios (Klein 1972). A flatter dT/dXFe slope would be in poorer agreement with the widespread grunerite in BIF as well as the lnKD constraints. In the Introduction, we guessed at a standard-state exchange reaction free-energy of 0.4 kJ/M-cation. Our recommended dataset (Table 5) has G1 = H1 = 1.75 kJ/mol or 0.25 kJ/M-cation, equivalent to a 350 C difference in transition temperatures from the Mg to the Fe end of the diagram.

CONCLUDING REMARKS
We believe that our phase diagram (Fig. 5) successfully accounts for the occurrence in various rock types of low-Al anthophyllite, cummingtonite, and grunerite. In addition, the diagram opens up the possibility of a stability field at low temperature for pure ferro-anthophyllite. End-member Fe7Si8O22(OH)2, usually somewhat manganoan, forms in nature by cooling and hydration of fayalite, and has been described

F IGURE 6. Mg/Fe partition between natural cummingtonite, anthophyllite, and olivine as a function of Fe/(Fe + Mg) in olivine. Symbols as in Figure 1. KD = (Fe/Mg)0l/(Fe/Mg)Amph.

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE

649

as grunerite (e.g., Bowen and Schairer 1935; Bonnichsen 1969; Floran and Papike 1978; Vaniman et al. 1980; Janeczek 1989). Recently, it has been shown that proto-ferroanthophyllite-Pnmn (Sueno et al. 1998) and ferro-anthophyllite-Pnma (Bozhilov and Evans 1999) are alternative possibilities. These observations are not fully understood at present (and MnO may play a role in the occurrence of protoamphibole), but they certainly hint at the possibility that an orthorhombic end-member is more stable than grunerite at low temperature, but not so low a temperature as to preclude growth of any Fe-amphibole. This lower limit is about 340 C, because below this temperature we can expect reaction of end-member Fe 7Si 8O 22(OH) 2, whether grunerite or ferroanthophyllite, to minnesotaite in the presence of quartz and H2O (Miyano 1978; Rasmussen et al. 1998). Thus, Fe end-member, Al-free orthorhombic amphibole may have a stability field above 340 C and below 450 C. Probably the best place to look for ferro-anthophyllite is in high Fe/Mg hydrothermal ore deposits; in low-grade metamorphosed BIF, the required temperature may be too low and, furthermore, grunerite will be favored in the presence of Ca. Our reference-state Gibbs free-energies of formation for monoclinic vs. orthorhombic FeMg-amphibole, not surprisingly, are quite similar at 298 K and 1 bar, with GAthCum = 3.875 kJ/mol and GFathGru = 2.125 kJ/mol. When free energies are extracted from different sources along independent paths, the differences, and the relative uncertainties, tend to be considerably larger in the case of polymorphs. For example, the corresponding reference-state transition free energies from Holland and Powell (1998, Table 5) are 10.61 and 12.63 kJ/mol respectively, and there is no stability field for either of the monoclinic end-members. None of the published papers on the subsolidus phase diagram for MSH have considered the possibility of a stability field for cummingtonite. Our data suggest that, close to its breakdown to enstatite, quartz, and H2O (790 C at 5 kbar), the stable amphibole may indeed be anthophyllite, but the uncertainties are such that it could be cummingtonite. Additional study of experimental run-products in the 800 C region in this system, by HRTEM for example, would certainly be helpful. In the presence of tremolite, the field of anthophyllite is sufficiently reduced (Fig. 5) that it is unlikely to reach that of enstatite, quartz, and H2O. In an attempt to check of our proposed phase loop, D. M. Jenkins (personal communication, October 1998) hydrothermally treated anthophyllite sample BM 93327 (XFe = 0.25, 12.5 wt% Al2O3) at 5 kbar. He found no indication of conversion to cummingtonite at 700 C for 20 days, at 800 C for 12 days, and at 850 C for 5 days (at which temperature partial breakdown to orthopyroxene took place). The free-energy drive for the inversion reaction is so small that direct experimental bracketing of the inversion curve will clearly be extremely difficult. On the other hand, crystallization experiments at high oxygen fugacity on hydrous, high Mg/Fe rhyolite and dacite melts might be able to overcome the kinetic problem, in that ferromagnesian amphiboles grow readily from such melts (e.g., Nicholls et al. 1992; Rutherford and Devine 1996; Scaillet and Evans 1999). Figure 4 shows the stabilizing effect of Al on anthophyllite, but otherwise the diagram should only be regarded as a tenta-

tive illustration of the mutual relations of cummingtonite, anthophyllite, and gedrite in the system NFMASH. For example, it would be interesting to know if field evidence can be brought to bear on the T-XFe behavior of the reaction terminal to anthophyllite: Al-Ath = Ged + Cum. The analogous inversion in Fe, Mg-pyroxene from orthopyroxene to clinopyroxene shows the same behavior as the ferromagnesian amphiboles in the T-XFe section, with a minimum close to the Fe-end (Fig. 3 in Davidson et al. 1988, Fig. 7 in Sack and Ghiorso 1994). The 350 C difference in the transition temperature from the Mg to Fe end for Fe, Mg-amphibole may be normalized to the average temperature (in K): 350/898 = 0.390. This normalization yields a number that is essentially identical to the corresponding value for pyroxene: 590/1478 = 0.399 (Fig. 7 in Sack and Ghiorso 1994). The correspondence may be understood by multiplying both numerator and denominator of this ratio by the appropriate averaged heat capacity of the phase, i.e. (T) (CP,ave) / (Tave ) (CP,ave), which may be written (H)/(Have). This demonstrates that the T normalized to the average temperature of the transition is a proxy for the enthalpy difference between the Mg and Fe end-members of the series, rendered as an intensive quantity through division by the average enthalpy. The fact that structures and transition mechanisms are similar between amphibole and pyroxene suggests that normalized enthalpy differences between Mg and Fe end-members of both series should compare favorably. In fact, this calculation could have been used to predict the transition temperatures in the amphibole series from knowledge of those in the pyroxenes, and it could be argued that the favorable correspondence is an independent test of the observations that led us to select the transition temperatures adopted for the Fe, Mg-amphiboles.

ACKNOWLEDGMENTS
We thank J. Fabris, M. Guiraud, C. Klein, the Natural History Museum London, J.S. Schneidermann, and R.J. Tracy for the loan of samples, M. Schindler and F.C. Hawthorne for the refinements of sample BM 93327, and D.M. Jenkins for high PT work on this sample. Critical reviews by G. Droop, J.C . Schumacher, and V. Trommsdorff are greatly appreciated. This study was supported by the National Science Foundation, grant EAR 97-06326.

REFERENCES CITED
Andrew, A.S. (1984) P-T-X(CO2) conditions in mafic and calc-silicate hornfelses from Oberon, New South Wales, Australia. Journal of Metamorphic Geology, 2, 143164. Armstrong, J.T. (1988) Bence-Albee after 20 years: review of the accuracy of factor correction procedures for oxide and silicate minerals. Microbeam Analysis, Applications in Geology, 469476. Bancroft, G.M., Maddock, A.G., Burns, R.G., and Strens, R.G.J. (1966) Cation distribution in anthophyllite from Mssbauer and infra-red spectroscopy. Nature, 212, 913915. Barabanov, A.V. and Tomilov, S.B. (1973) Mssbauer study of the isomorphous series anthophyllite-gedrite and cummingtonite-grunerite. Geochemistry International, 12401267. Berman, R.G. (1988) Internally-consistent thermodynamic data for stoichiometric minerals in the system Na2O-K2O-CaO-MgO-FeO-Al2O3-SiO2-TiO2-H2O-CO2. Journal of Petrology, 29, 445522. Boffa Ballaran, T., Carpenter, M.A., and Domeneghetti, M.C. (2001) Phase transitions and thermodynamic mixing behaviour of the cummingtonite-grunerite solid solution. Physics and Chemistry of Minerals, in press. Bonnichsen, B. (1969) Metamorphic pyroxenes and amphiboles in the Biwabik Iron Formation, Dunka River area, Minnesota. Mineralogical Society of America Special Paper 2, 217239. Bowen, N.L. and Schairer, J.F. (1935) Grunerite from Rockport, Massachusetts, and a series of synthetic fluor-amphiboles. American Mineralogist, 20, 543 551.

650

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE


equilibria. Economic Geology, 77, 6081. Hawthorne, F.C. (1983) The crystal chemistry of the amphiboles. The Canadian Mineralogist, 21, 173480. Hirschmann, M., Evans, B.W., and Yang, H. (1994) Composition and temperature dependence of Fe-Mg ordering in cummingtonite-grunerite as determined by X-ray diffraction. American Mineralogist, 79, 862877. Holland T.J.B. and Powell, R. (1998) An internally consistent thermodynamic data set for phases of petrological interest. Journal of Metamorphic Geology, 16, 309343. Jaffe, H.W. (1988) Introduction to crystal chemistry. Cambridge University Press, New York, 161 p. Janeczek, J. (1989) Manganoan fayalite and products of its alteration from the Strzegom pegmatites, Poland. Mineralogical Magazine, 53, 315325. Kamineni, D.C., Jackson, G.D., and Bonardi, M. (1979) Coexisting magnesian and calcic amphiboles in meta-ultramafites from Baffin Island (Arctic Canada). Neues Jahrbuch fr Mineralogie, Monatsheft 12, 542555. Kisch, H.J. (1969) Magnesiocummingtonite-P2 1/m: A Ca- and Mn-poor clinoamphibole from New South Wales. Contributions to Mineralogy and Petrology, 21, 319331. Klein, C. (1968) Coexisting amphiboles. Journal of Petrology, 9, 281330. (1978) Regional metamorphism of Proterozoic iron-formation, Labrador Trough, Canada. American Mineralogist, 63, 898912. (1982) Amphiboles in iron-formations. In D.R. Veblen and P.H. Ribbe, Eds., Amphiboles: petrology and experimental phase relations, vol. 9B, p. 8898. Reviews in Mineralogy Mineralogical Society of America, Washington, D.C. Kleinschmidt, G., Schubert, W., Olesch, M., and Rettmann, E.S. (1987) Ultramafic rocks of the Lanterman Range in North Victoria Land, Antarctica. Petrology, geochemistry, and geodynamic implications. Geologisches Jahrbuch, Hannover, B 66, 231273. Krupka, K.M., Hemingway, B.S., Robie, R.A., and Kerrick, D.M. (1985) High temperature heat capacities and derived thermodynamic properties of anthophyllite, diopside, enstatite, bronzite, and wollastonite. The American Mineralogist, 70, 249260. Law, A.D. (1989) Studies of the orthoamphiboles. IV. Mssbauer spectra of anthophyllites and gedrites. Mineralogical Magazine, 53, 181191. Lindsley, D.H. (1983) Pyroxene thermometry. American Mineralogist, 68, 477 493. Matthes, S. (1986) Die prograde Kontaktmetamorphose von Serpentiniten des Oberpflzer Waldes. Geologica Bavarica, 89, 720. Matthes, S. and Knauer, E. (1981) The phase petrology of the contact metamorphic serpentinites near Erbendorf, Oberpfalz, Bavaria. Neues Jahrbuch fr Mineralogie Abhandlungen, 141, 5989. Miyano, T. (1978) Phase relations in the system Fe-Mg-Si-O-H and environments during low-grade metamorphism of some Precambrian iron formations. Journal of the Geological Society of Japan, 84, 679690. Nicholls, I.A., Oba, T., and Conrad, W.K. (1992) The nature of primary rhyolitic magmas involved in crustal evolution: evidence from an experimental study of cummingtonite-bearing rhyolites. Geochimica et Cosmochimica Acta, 56, 955 962. North, A.C.T, Phillips, D.C., and Matthews, F.S. (1968) A semi-empirical method of absorption correction. Acta Crystallographica, A24, 351359. Papike, J.J. and Ross, M. (1970) Gedrites: crystal structures and intracrystalline cation distributions. American Mineralogist, 55, 19451972. Pfeifer, H.R. (1978) Hydrothermal Alpine metamorphism in metaperidotite rocks of the Cima Lunga zone, Valle Verzasca, Switzerland. Schweizerische Mineralogische und Petrographische Mitteilung, 58, 400405. (1987) A model for fluids in metamorphosed ultramafic rocks: IV. Metasomatic veins in metaharzburgites of Cima di Gagnone, Valle Verzasca, Switzerland. In H.C. Helgeson, Ed., Chemical Transport in Metasomatic Processes, NATO Advanced Study Institute Series, Series C, p. 591632. D Reidel Publishing Company, Dordrecht. Popp, R.K., Gilbert, M.C., and Craig, J.R. (1976) Synthesis and X-ray properties of Fe-Mg orthoamphiboles. American Mineralogist, 61, 12671279. Rabbitt, J.C. (1948) A new study of the anthophyllite series. American Mineralogist, 33, 263323. Rasmussen, M.G., Evans, B.W., and Kuehner, S.M. (1998) Low-temperature fayalite, greenalite, and minnesotaite from the Overlook gold deposit, Washington: Phase relations in the system FeO-SiO2-H2O. Canadian Mineralogist, 36, 147162. Rice, J.M., Evans, B.W., and Trommsdorff, V. (1974) Widespread occurrence of magnesiocummingtonite in ultramafic schists, Cima di Gagnone, Ticino, Switzerland. Contributions to Mineralogy and Petrology, 43, 245251. Robie, R.A. and Hemingway, B.S. (1995) Thermodynamic properties of minerals and related substances at 298.15 K and 1 bar (105 pascals) pressure and at higher temperatures. U.S. Geological Survey Bulletin 2131, 461p. Robinson, P. and Jaffe, H.W. (1969) Chemographic exploration of amphibole assemblages from central Massachusetts and southwestern New Hampshire. Mineralogical Society of America, Special Paper 2, 251300. Robinson, P., Ross, M., and Jaffe, H. (1971) Composition of the anthophyllite-gedrite series, comparisons of gedrite and hornblende, and the anthophyllite-gedrite solvus. American Mineralogist, 56, 10051041.

Bozhilov, K.N. and Evans, B.W. (1999) A HRTEM study of grunerite and ferroanthophyllite from Rockport, Massachusetts. Transactions of the American Geophysical Union, 80, no. 46, F1107 (abstract). Cameron, K.L. (1975) An experimental study of actinolite-cummingtonite phase relations with notes on the synthesis of Fe-rich anthophyllite. American Mineralogist, 60, 375390. Carpenter, M.A. (1982) Amphibole microstructures: some analogies with phase transformations in pyroxenes. Mineralogical Magazine, 46, 395397. Chernosky, J.V. and Autio, L.K. (1979) The stability of anthophyllite in the presence of quartz. American mineralogist, 64, 294303. Chernosky, J.V., Day, H.W., and Caruso, L.J. (1985) Equilibria in the system MgOSiO2-H2O: experimental determination of the stability of Mg-anthophyllite. American Mineralogist, 70, 223236. Chernosky, J.V., Berman, R.B., and Jenkins, D.M. (1998) The stability of tremolite: New experimental data and a thermodynamic assessment. American Mineralogist, 83, 726738. Clark, M.D. (1978) Amphibolitic rocks from the Precambrian of Grand Canyon: mineral chemistry and phase petrology. Mineralogical Magazine, 42, 199207. Davidson, P.M., Lindsley, D.H., and Carlson, W.D. (1988) Thermochemistry of pyroxenes on the join Mg2Si2O6-CaMgSi2O6: A revision of the model for pressures up to 30 kbar. American Mineralogist, 73, 12641266. Deer, W.A., Howie, R.A, and Zussman, J. (1997) Rock-forming minerals, Volume 2B, Double-chain silicates. The Geological Society, London, 764p. Droop, G.T.R. (1994) Triple-chain pyriboles in Lewisian ultramafic rocks. Mineralogical Magazine, 390, 120. Dymek, R.F., Brothers, S.C., and Schiffries, C.M. (1988) Petrogenesis of ultramafic metamorphic rocks from the 3800 Ma Isua supracrustal belt, West Greenland. Journal of Petrology, 29, 13531397. Early, D. III and Stout, J.H. (1991) Cordierite-cummingtonite facies rocks from the Gold Brick District, Colorado. Journal of Petrology, 32, 11691202. Elliott-Meadows, S.R., Froese, E., and Appleyard, E.C. (1999) Cordieriteanthophyllite-cummingtonite rocks from the Lar deposit, Laurie Lake, Manitoba. Canadian Mineralogist, 37, 375380. Eskola, P. (1914) On the petrology of the Orijrvi region in southwestern Finland. Bulletin de la Commission Gologique de Finlande 40, 277. Evans, B.W. and Ghiorso, M.S. (1995) Thermodynamics and petrology of cummingtonite. American Mineralogist, 80, 649-663. Evans, B.W. and Medenbach, O. (1997) The optical properties of cummingtonite and their dependence on Fe-Mg order-disorder. European Journal of Mineralogy, 9, 9931003. Evans, B.W. and Trommsdorff, V. (1983) Fluorine hydroxyl titanian clinohumite in Alpine recrystallized garnet peridotite: compositional controls and petrologic significance. American Journal of Science, 283-A, 355369. Evans, B.W., Ghose, S., Rice, J.M., and Trommsdorff, V. (1974) Cummingtoniteanthophyllite phase transformation in metamorphosed ultramafic rocks, Ticino, Switzerland. Transactions of the American Geophysical Union, 55, 469. Fabris, J. and Persil, E.A. (1971) Nouvelles observations sur les amphiboles orthorhombiques. Bulletin Societ Franaise Mineralogie et Cristallographie, 94, 385395. Ferrow, E. A. and Ripa, M. (1990) Al-poor and Al-rich orthoamphiboles: a Mssbauer spectroscopy and TEM study. Mineralogical Magazine, 54, 547552. Fialips-Gudon, C.-I., Robert, J.-L., and Delbove, F. (2000) Experimental study of Cr incorporation in pargasite. American Mineralogist, 85, 687693. Floran, R.J. and Papike, J.J. (1978) Mineralogy and petrology of the Gunflint iron formation, Minnesota-Ontario: Correlation of compositional and assemblage variations at low to moderate grade. Journal of Petrology, 19, 215288. Frost, B.R. (1975) Contact metamorphism of serpentinite, chloritic blackwall, and rodingite at Paddy-Go-Easy Pass, Central Cascades, Washington. Journal of Petrology, 16, 272313. Ghiorso, M.S., Evans, B.W., Hirschmann, M., and Yang, H. (1995) Thermodynamics of the amphiboles: Fe-Mg cummingtonite solid solutions. American Mineralogist, 80, 502519. Gibbs, G.V. (1969) Crystal structure of protoamphibole. Mineralogical Society of America Special Paper 2, 101109. Gole, M.J., Barnes, S.J., and Hill, R.E.T. (1987) The role of fluids in the metamorphism of komatiites, Agnew nickel deposit, Western Australia. Contributions to Mineralogy and Petrology, 96, 151162. Gottschalk, M. (1997) Internally consistent thermodynamic data for rock-forming minerals. European Journal of Mineralogy, 9, 175223. Greenwood, H.J. (1963) The synthesis and stability of anthophyllite. Journal of Petrology, 4, 317351. Grond, R., Wahl, F., and Pfiffner, M. (1995) Mehrphasige alpine Deformation und Metamorphose in der nrdlichen Cima-Lunga Einheit, Zentralalpen (Schweiz). Schweizerische Mineralogische und Petrographische Mitteilungen, 75, 371 386. Guiraud, M., Powell, R., and Cottin, J.-Y. (1996) Hydration of orthopyroxene-cordierite-bearing assemblages at Laouni, Central Hoggar, Algeria. Journal of Metamorphic Geology, 14, 467476. Haase, C.S. (1982) Metamorphic petrology of the Negaunee Iron Formation, Marquette District, Michigan: mineralogy, metamorphic reactions, and phase

EVANS ET AL.: THERMODYNAMICS OF ANTHOPHYLLITE-FERROANTHOPHYLLITE


Robinson, P., Spear, F.S., Schumacher, J.C., Laird, J., Klein, C., Evans, B.W., and Doolan, B.L. (1982) Phase relations of metamorphic amphiboles: natural occurrence and theory. In D.R. Veblen and P.H. Ribbe, Eds., Amphiboles: petrology and experimental phase relations, vol. 9B, p. 1227. Reviews in Mineralogy Mineralogical Society of America, Washington, D.C. Ross, M., Papike, J.J., and Shaw, K.W. (1969) Exsolution textures in amphiboles as indicators of subsolidus thermal histories. Mineralogical Society of America, Special Paper, 2, 275299. Rutherford, M.J. and Devine, J.D. (1996) Pre-eruption pressure-temperature conditions and volatiles in the 1991 eruption of Mount Pinatubo magma. In: Newhall, C.G. and Punongbayan, R.S., eds. Fire and Mud. Eruptions and Lahars of Mount Pinatubo, Philippines. Seattle: University of Washington Press, pp. 751766. Sack, R.O. and Ghiorso, M.S. (1994) Thermodynamics of multicomponent pyroxenes: II. Applications to phase relations in the quadrilateral. Contributions to Mineralogy and Petrology, 116, 287300. Scaillet, B. and Evans, B.W. (1999) The 15 June 1991 eruption of Mount Pinatubo. I. Phase equilibria and pre-eruption P-T-fO2 fH2O conditions of the dacite magma. Journal of Petrology, 40, 381411. Schneidermann, J.S. and Tracy, R.J. (1991) Petrology of orthoamphibole-cordierite gneisses from the Orijrvi area, southwest Finland. American Mineralogist, 76, 942955. Seifert, F.A. (1977) Reconstruction of rock cooling paths from kinetic data on the Fe2+-Mg exchange reaction in anthophyllite. Philosophical Transactions of the Royal Society London A, 286, 303311. (1978) Equilibrium Mg-Fe2+ cation distribution in anthophyllite. American Journal of Science, 278, 13231333. Seifert, F.A. and Virgo, D. (1975) Kinetics of the Fe2+-Mg order-disorder reaction in anthophyllites: quantitative cooling rates. Science, 188, 11071109. Spear, F.S. (1980) The gedrite-anthophyllite solvus and the composition limits of orthoamphibole from the Post Pond Volcanics, Vermont. American Mineralogist, 65, 11031118. (1982) Phase equilibria of amphibolites from the Post Pond Volcanics, Mt. Cube quadrangle, Vermont. Journal of Petrology, 23, 383426.

651

Srikantappa, C., Raith, M., and Ackermand, D. (1985) High-grade regional metamorphism of ultramafic and mafic rocks from the Archaean Sargur Terrane, Karnataka, South India. Precambrian Research, 30, 189219. Stout, J.H. (1972) Phase petrology and mineral chemistry of coexisting amphiboles from Telemark, Norway. Journal of Petrology, 13, 99146. Stroink, G., Blaauw, C., White, C.G., and Leiper, W. (1980) Mssbauer characteristics of UICE standard reference asbestos samples. Canadian Mineralogist, 18, 285290. Sueno, S., Matsuura, S., Gibbs, G.V., and Boisen, M.B.Jr. (1998) A crystal chemical study of protoanthophyllite: orthoamphiboles with the protoamphibole structure. Physics and Chemistry of Minerals, 25, 366377. Tella, S. and Eade, K.E. (1978) Coexisting cordierite-gedrite-cummingtonite from Edehon Lake map area, Churchill structural province, District of Keewatin. Geological Survey of Canada, Paper 78-1C, 712. Tilley, C.E. (1937) Anthophyllite-cordierite granulites of the Lizard. Geological Magazine, 74, 300309. Todd, C.S. and Engi, M. (1997) Metamorphic field gradients in the Central Alps. Journal of Metamorphic Geology, 15, 513530. Trommsdorff, V. and Connolly, J.A. (1996) The ultramafic contact aureole about the Bregaglia (Bergell) tonalite: isograds and a thermal model. Schweizerische Mineralogische und Petrographische Mitteilungen, 76, 537547. Vaniman, D.T., Papike, J.J., and Labotka, T. (1980) Contact metamorphic effects of the Stillwater Complex, Montana: the concordant iron formation. American Mineralogist, 65, 10871102. Yakovleva, A.K. and Kolesnikova, V.V. (1967) Characteristics of high-magnesium cummingtonite from ultrabasic rocks. Mineralogical Abstracts, 1969, Vol. 20, No. 4, p. 314, abstract 69-3245.

MANUSCRIPT RECEIVED APRIL 3, 2000 MANUSCRIPT ACCEPTED JANUARY 31, 2001 MANUSCRIPT HANDLED BY JOHN C. SCHUMACHER

Vous aimerez peut-être aussi