American Mineralogist, Volume 72, pages 707-7 il5, 1987 Synthesisand characterization of tremolite in the system HrO-CaO-MgO-SiO, D.lvro M. JnNxrNs Department of Geological Sciences and Environmental Studies,State University of New York-Binghamton, Binghamton, New York 13901, U.S.A. Ansrn,q,cr Attempts at synthesizingtremolite of the composition Ca,Mg5Si8O,,(OH), (: TR) in this and previous studies have repeatedly resulted in good but incomplete yields. This investigation was performed to determine whether the incomplete synthesisof tremolite is the result of sluggish growth rates or is the result of synthetic tremolite not having the ideal composition. The growth rate of tremolite at tlpical synthesisconditions (800'C, 2 kbar, l0 d) does not appearto be a limiting factor, as demonstratedby the partial break- down and essentially complete regrowth (as discernedfrom X-ray diffraction patterns) of both natural and synthetic tremolite with Cal(Mg + Fe) ratios less than 2/5 that werc grown in response to variable silica content in the ambient aqueous fluid. This observation also reveals the pronounced effect that the aqueoussilica content has on the synthesis of tremolite. Tremolite was synthesized from a series of bulk compositions at 5 molo/o incre- ments along part of the join TR-MC (MC : Mg,SirOrr(OH)r) at 850"C and 6 kbar and at 750'C and 13 kbar in order to determine if the CalMg ratio of the starting material affects the synthesisof tremolite. The 6-kbar series(with additional quartz) produced the maxi- mum yield of tremolite from the TReoMCr. bulk composition. Similar results were ob- tained at 13 kbar but were rendered less useful by the spontaneous nucleation of talc in the Mg-rich mixtures. The phaserelations along this join indicate that synthetic tremolite either has the fixed composition of TRnoMC,o or a very narrow compositional rangeabout this value. Attempts at varying the CalMg ratio of synthetic tremolite were unsuccessful. A survey of the purest natural tremolites reported in the literature confirmed that Mg enrichment is common. It is calculated that the presence of l0 molo/o additional Mg on the M4 site of synthetictremolite will increase values of AG! for the TR componentderived from phaseequilibria by up to 2 kJ/mol. INrnonucuoN ing calcic pyroxene plus quartz to form as well? If trem- Many experimental studies involving calcic amphi- olite has a rangeof compositions, can this compositional boles have dealt with tremolite because of its relatively range be correlated with specific geologic conditions of simple chemistry. Somewhat surprisingly, few (if any) pressure and temperature? What effect does variable studies have ever reported that a complete yield of am- tremolite composition have on the thermochemicalprop- phibole could be obtained from a starting mixture of the erties of the tremolite derived from experimental inves- composition Ca,MgrSi,O,r(OH), (or simply TR). Investi- tigations? gators who have documented the synthesisof tremolite A number of experimentshave been performed in this in some detail (Boyd, 1959; Trolt and Gilbert, 1972 studytoaddressthesequestions.First,theresultsoftrem- Wones and Dodge, 1977:' Oba, 1980; Skippen and olite syntheses from a fixed bulk composition over a wide McKinstry, 1985) have all indicated difficulties in ob- rangeofpressuresandtemperaturesareexaminedinterms taining pure yields of tremolite. Usually a mixture of of the phases observed and amount of amphibote ob- phases is obtained consisting of about 90 wto/o tremolite tained. Second,a seriesof experiments is describedthat and l0 fio/o diopsidic pyroxene with or without quartz. was conducted with natural and synthetic tremolite in This rather consistent observation gives rise to several order to investigate the growth rate of tremolite under questions. Have the "right" conditions for synthesizing typical synthesis conditions. Third, the resultsof two iso- pure tremolite simply not been found? Does tremolite thermal, isobaric syntheses of tremolite over a range of growth become prohibitively slow (i.e., kinetically con- bulk compositions are presentedin order to determine trolled) after the synthesisis about 900/o complete, or is the effect of bulk composition on the synthesisof trem- the equilibrium composition of the tremolite formed at olite. Finally, several attempts were made to induce a typical synthesis conditions enriched in Mg thereby caus- shift in the composition of synthetictremolite (i.e.,reverse 0003-004x/87l0708-0707$02.00 707
Transcript
American Mineralogist, Volume 72, pages 707-7 il5, 1987
Synthesis and characterization of tremolite in thesystem HrO-CaO-MgO-SiO,
D.lvro M. JnNxrNsDepartment of Geological Sciences and Environmental Studies, State University of New York-Binghamton,
Binghamton, New York 13901, U.S.A.
Ansrn,q,cr
Attempts at synthesizing tremolite of the composition Ca,Mg5Si8O,,(OH), (: TR) inthis and previous studies have repeatedly resulted in good but incomplete yields. Thisinvestigation was performed to determine whether the incomplete synthesis of tremoliteis the result of sluggish growth rates or is the result of synthetic tremolite not having theideal composition. The growth rate of tremolite at tlpical synthesis conditions (800'C, 2kbar, l0 d) does not appear to be a limiting factor, as demonstrated by the partial break-down and essentially complete regrowth (as discerned from X-ray diffraction patterns) ofboth natural and synthetic tremolite with Cal(Mg + Fe) ratios less than 2/5 that wercgrown in response to variable silica content in the ambient aqueous fluid. This observationalso reveals the pronounced effect that the aqueous silica content has on the synthesis oftremolite. Tremolite was synthesized from a series of bulk compositions at 5 molo/o incre-ments along part of the join TR-MC (MC : Mg,SirOrr(OH)r) at 850"C and 6 kbar and at750'C and 13 kbar in order to determine if the CalMg ratio of the starting material affectsthe synthesis of tremolite. The 6-kbar series (with additional quartz) produced the maxi-mum yield of tremolite from the TReoMCr. bulk composition. Similar results were ob-tained at 13 kbar but were rendered less useful by the spontaneous nucleation of talc inthe Mg-rich mixtures. The phase relations along this join indicate that synthetic tremoliteeither has the fixed composition of TRnoMC,o or a very narrow compositional range aboutthis value. Attempts at varying the CalMg ratio of synthetic tremolite were unsuccessful.A survey of the purest natural tremolites reported in the literature confirmed that Mgenrichment is common. It is calculated that the presence of l0 molo/o additional Mg onthe M4 site of synthetic tremolite will increase values of AG! for the TR component derivedfrom phase equilibria by up to 2 kJ/mol.
INrnonucuoN ing calcic pyroxene plus quartz to form as well? If trem-Many experimental studies involving calcic amphi- olite has a range of compositions, can this compositional
boles have dealt with tremolite because of its relatively range be correlated with specific geologic conditions ofsimple chemistry. Somewhat surprisingly, few (if any) pressure and temperature? What effect does variablestudies have ever reported that a complete yield of am- tremolite composition have on the thermochemical prop-phibole could be obtained from a starting mixture of the erties of the tremolite derived from experimental inves-composition Ca,MgrSi,O,r(OH), (or simply TR). Investi- tigations?gators who have documented the synthesis of tremolite A number of experiments have been performed in thisin some detail (Boyd, 1959; Trolt and Gilbert, 1972 studytoaddressthesequestions.First,theresultsoftrem-Wones and Dodge, 1977:' Oba, 1980; Skippen and olite syntheses from a fixed bulk composition over a wideMcKinstry, 1985) have all indicated difficulties in ob- rangeofpressuresandtemperaturesareexaminedintermstaining pure yields of tremolite. Usually a mixture of of the phases observed and amount of amphibote ob-phases is obtained consisting of about 90 wto/o tremolite tained. Second, a series of experiments is described thatand l0 fio/o diopsidic pyroxene with or without quartz. was conducted with natural and synthetic tremolite inThis rather consistent observation gives rise to several order to investigate the growth rate of tremolite underquestions. Have the "right" conditions for synthesizing typical synthesis conditions. Third, the results of two iso-pure tremolite simply not been found? Does tremolite thermal, isobaric syntheses of tremolite over a range ofgrowth become prohibitively slow (i.e., kinetically con- bulk compositions are presented in order to determinetrolled) after the synthesis is about 900/o complete, or is the effect of bulk composition on the synthesis of trem-the equilibrium composition of the tremolite formed at olite. Finally, several attempts were made to induce atypical synthesis conditions enriched in Mg thereby caus- shift in the composition of synthetic tremolite (i.e., reverse
0003-004x/87l0708-0707$02.00 707
708
Tnele 1. Natural tremolites used in this studv
JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE
Oxides(wt%) TREM 8. TREM 12'. Cations
medium. The vessels were constructed of Ren6 4l and wereoperated horizontally with a tight-fitting filler rod of Stellite.Temperatures were measured with inconel-sheathed, chromel-alumel thermocouples calibrated against the freezing point ofNaCl (800.5'C). Each thermocouple was located in an externalthermocouple well with the tip positioned near the sample. Acalibration of the sample temperature inside the vessel was madeusing the reaction talc : 3 enstatite + qLraftz + HrO. Talc,enstatite, and quartz were synthesized individually from the ox-ides MgO, SiOr, and H.O and were mixed in the stoichiometricproportions ofthe reaction. At 2.0 + 0.1 kbar, strong growth oftalc was observed at 718 + 5oC, whereas strong growth of en-statite + quartz was observed at74l + 5'C. These results are inexcellent agreement with a temperature of 73l'C for the univar-iant boundary at 2 kbar, which was determined by Chernoskyet al. (1985) using all extant experimental data, and indicate thatthe temperature measured at the thermocouple well is within atleast l0"C of the internal (sample) temperature. Pressures weremeasured intermittently with a bourdon-tube gauge calibratedagainst both a 50 000 and 75 000 psi Heise gauge. Pressure mea-surements are considered accurate to + 100 bars.
Starting materials
All synthetic phases were made from mixtures of reagent-gradeoxides. Calcium oxide was obtained by weighing CaCO. (Baker
and Adamson, lot C224, or Mallinckrodt, JPX), because of itsease in handling, into the mixture and then decarbonating toCaO by heating to 800-900'C in air for 15-30 min. Magnesiumoxide from three sources was used: (l) single-crystal chips ofMgO (Muscle Shoals Electrochemical Corp.), (2) finely powderedMgO (Fisher, lot 706058), or (3) MgO fired from hydrous mag-nesium carbonate (Bakers, lots 91507 and 36416). In each casethe MgO was fired to at least 1000"C for several hours prior toweighing. Several forms of SiO, were used: natural vein quartz
from Lisbon, Maryland (provided by J. R. Goldsmith), ground
fused silica, and cristobalite made by firing silicic acid (Bakers,lot 28298, or Fisher, lot 706058) in air at 1000"C for severaldays. Starting mixtures were prepared in 2-g batches, and eachreagent was weighed to a precision of at least +0.1 wt0/0. With
this precision, mixtures could be prepared to within +0.001 ofthe ideal CalMg ratio of 0.400. No pronounced effect on the rateof growth or the morphology of synthetic amphibole crystals wasobserved with the use of these various forms of MgO or SiOr.
Two natural tremolites were used in this study. One was froma dolomite marble (Mass. 918-l) collected by E-an Zen (USGS)
and kindly donated to J. R. Goldsmith (University of Chicago),from whom a portion ofthe rock was obtained. The other wasfrom a calcite marble in Barrie Township, Ontario, and wasobtained from Wards Scientific. Particular care was taken tohand pick crystals that were as free of carbonate as possible;
however, both samples had to be treated with cold l.2M HCltoremove all carbonate from their X-ray patterns. The tremoliteswere also found to be free of -talc by looking for the distinctive(002) reflection at d :9.35 A in the X-ray pattern' Chemicalanalyses ofthese tremolites are given in Table l.
Procedure
Synthesis experiments were performed by sealing a portion ofthe oxide starting mixture with an additional l5-50 wto/o distilledH.O in a Pt or Au capsule. Some amphibole synthesis experi-ments were performed with an extra 1-10 rto/o SiO, in the start-ing mixture in order to saturate the ambient fluid in SiO, and,thereby, halt the dissolution of this component from the syn-
. Sample: Mass. 918-1, analysis and sample provided by E-an Zen,USGS.
'. Sample: Barrie Township, Ontario, analysis provided by lan M. Steele,University of Chicago
t Total Fe reported as FeO+ P, Cl, and F analyses provided by Richard L. Hervig, Arizona State
University.
the composition) in order to detect what, if any, com-positional dependency tremolite may have on tempera-ture.
ExpnnrurNTAL METHoDS
Apparatus
Three different apparatus were used in this study: piston-cyl-inder presses, internally heated gas vessels, and cold-seal vessels.A11 experiments above 8 kbar were performed in a piston-cyl-inder apparatus at the University of Chicago using %-in. O.D.NaCl pressure media. A detailed discussion of the calibration ofthe sample pressure in this assemblage is given in Jenkins (1981).In brief, sample pressure measurements are believed to be ac-curate to +300 bars, and temperature measurements (with chro-mel-alumel thermocouples) to + 5"C. Individual experiments wereperformed by first pressurizing the NaCl assemblage at roomtemperature to several kilobars below the desired pressure andthen using the (well-calibrated) thermal expansion of NaCl toobtain the desired pressure upon heating. Experiments in thepressure range of 2-8 kbar were performed in internally heatedgas vessels ofidentical design at either the University ofChicagoor SUNY-Binghamton. The pressure medium in these vesselswas Ar, and its pressure was monitored throughout each exper-iment by both a bourdon-tube gauge and manganin cell. Theprecision ofpressure measurement is + 1 5 bars, and the accuracyis about + 50 bars. Temperatures in the internally heated vesselswere monitored with two inconel-sheathed, chromel-alumelthermocouples whose tips were positioned at either end of thesample capsule, as described by Jenkins (1983), allowing directmeasurement of the thermal gradient across the sample. Tem-perature uncertainties associated with experiments performed inthe internally heated gas vessels include both the accuracy (gen-erally +2'C) as well as the thermal gradient across the sample(2-5'C). Experiments at 2 kbar or less were performed in cold-seal vessels at SUNY-Binghamton using water as the pressure
JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE'709
thetic amphibole. Further discussion of this phenomenon is giv-en below.
The integrity ofa sealed capsule was checked at the conclusionof an experiment in two ways: first by examining the capsuleseams for the appearance of water while compressing the capsulewith a pair of pliers, and second by checking the capsule weightbefore and after the experiment.
Analytical methods
All experimental products were examined under the petro-graphic microscope. An oil of refractive index 1.604 (a of trem-olite) proved to be most useful for discerning minor amounts ofenstatite (a : 1.650), diopside (a : 1.664), and quartz G : 1.553)coexisting with tremolite. A1l products were also characterizedwith a Norelco X-ray diffractometer using Ni-filtered Cu radia-tion. Unit-cell refinements were obtained from difractometerpatterns at a scan rate of t/q" 2Olmin using an internal standardof synthetic corundum (a:4.7593 + 0.0003 L; c: 12.991 +0.001 A) and the least-squares progmm of Appleman and Evans(1973). Several experimental charges were examined with an elrn900 scanning-electron microscope (sEM). Modal amounts ofphases were estimated from X-ray diffraction peak-height ratios;reproducibility was hindered, however, by preferred orientationof amphibole grains.
Rnsur,rsTremolite synthesis attempts
A number of attempts were made at the outset of thisstudy to synthesize tremolite from the bulk compositionTR over a wide range of pressures, temperatures, andwith a variety of starting materials. The starting materialsused in these hydrothermal synthesis experiments in-volved the following: oxide mixture, oxide mixture withexcess silica, oxide mixture seeded with 5 wto/o naturaltremolite, gel seeded with natural tremolite, mixture ofsynthetic talc plus Ca(OH), plus quartz, and an oxidemixture encapsulated with a 0.2m CaI, solution insteadof distilled water (H. A. Yin, 1983, pers. comm.). Be-tween 750 and 850'C, clinopyroxene formed with trem-olite, whereas at or below 750"C, talc often nucleatedwith tremolite and clinopyroxene. These results are sum-marized in Figure 1. In short, no synthesis attempt yield-ed 1000/o tremolite when the starting mixture had the bulkcomposition TR or a Ca/Mgrutio of 2/5 or greater.
Tremolite growth rate
The approach adopted here for determining the rate oftremolite growth does not involve a numerical assess-ment of reaction-rate laws but, instead, a simple dem-onstration that the partial dissolution and essentiallycomplete regrowth of tremolite is possible under typicalsynthesis conditions and durations.
It was found that the hydrothermal treatment of nat-ural tremolite can lead to its incongruent dissolution bythe reactions
Fig. 1. Tremolite synthesis attempts from various startingmaterials of the bulk compositon Ca'Mg'Si'Orr(OH)', with andwithout excess SiO, or Ca2*. Dash-dot curve is the upper stabil-ity oftalc from Chernosky et al. (1985). Solid curve is the upperstability of tremolite from Yin and Greenwood (1983). Dashedcurves are approximated. Abbreviations as in Table 2.
Both reactions indicate that silica is preferentially dis-solved from tremolite, leaving clinopyroxene and eitherorthopyroxene or forsterite as the residual solids. Wheth-er Reaction I or 2 occurs appears to depend on the degreeof silica undersaturation in the aqueous fluid, such thatReaction I occurs in mildly undersaturated fluids, where-as Reaction 2 occurs in strongly undersaturated fluids.This phenomenon of silica dissolution was used to showthat two modally pure natural tremolites and a modallypure synthetic tremolite (CalMg : 1.8/5.2) can be par-tially decomposed by hydrothermal treatment and thencompletely reformed to l00o/o amphibole (as discernedfrom X-ray patterns) as long as the assemblage is satu-rated in silica. Expeimental results are listed in Table 2,and X-ray diffraction patterns for the synthetic tremoliteexperiments are shown in Figure 2. Figure 3a shows ansru image of the surface of natural tremolite sampleTREM 12 after hydrothermal treatment at 800'C and 2kbar for I l5 h with a water/tremolite ratio of 3.6/1.0 (byweight). Notice that the individual crystals have relative-ly smooth and often euhedral terminations, apparently asa result of both dissolution and recrystallizalion of cleav-age fragments. Minor amounts of forsterite and clinopy-roxene (not seen in Fig. 3a) formed as a result of hydro-thermal treatment. Figure 3b shows a relatively largecrystal of tremolite (TREM 12) from experiment no.TREM 12-6 it Table 2. In this experiment, orthopyrox-ene and clinopyroxene almost completely reacted withadded quartz to form tremolite overgrowths on remnanttremolite from a previous hydrothermal treatment. Onecan see the jagged and uneven terminations resulting fromcolumnar growth parallel to the Z axis. In transmittedlight, the tremolite overgrowths are in optical continuitywith the remnant tremolite grains.
It is emphasized that the modal abundances of thephases have been determined primarily from X-ray dif-
E
@Y
NI
SYMBOLSO = T R E M + C P Xo =TREM + CPX + TALC
-1" \ \B l \
' l I 'o l \
f-r." \l . i ' i
r o j o t r o
.r'/ o ,l
TREM UNSTABLE
TALC STABLE
o
TREMUNSTABLE
710 JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE
2 9
? 4 , C u K o
Fig. 2. X-ray diffractometer scans of synthetic tremolite ofthe composition Ca,rMg,rSirorr(OH)r. (a) Tremolite synthe-sized first at 2 kbar, 796C, 285 h, and then at I kbar, 825"C,389 h. (b) The tremolite ofspectrum (a) treated at 2 kbar, 806'C,144 h with a 5/1 ratio of water/tremolite (by weight). Note thepresence of clinopyroxene (Cpx) and onhopyroxene (Opx). (c)The assemblage ofspectrum (b) treated at2kbar,800'C,310 hwith excess quartz. Notice the disappearance of Cpx and Opxand the appearance of quartz (Qtz) in the pattern.
fraction peak heights. Phases present at levels below about5 wto/o are generally not detectable. Careful examinationof the experimental products with the petrographic mi-croscope shows that there are minor amounts (<2 volo/o)of clinopyroxene present in the experiments involvingnatural tremolite (TREM 8-2; TREM 12-6) and virtuallyno clinopyroxene present in the experiment involvingsynthetic rremolite (TREM 5-16). This can be explainedin part by the variable compositions of the three tremo-lites investigated. Both of the natural tremolites have ahigher Cal(Mg + Fe) ratio (1.97/5.01 for TREM 8-2;1.89/5.01 for TREM 5-16) than the synthetic tremolite(1.8/5.2).If there is any tendency toward Mg enrichmentin tremolite at high temperatures, as discussed below,then there should be more clinopyroxene remaining un-reacted in the natural as compared to the synthetic trem-olite experiment, as is observed.
From these experiments, it can be concluded that smallamounts of pyroxenes can be formed from, and readily
Fig. 3. seu images of natural tremolite TREM 12. (a) Surfacetextures resulting from hydrothermal treatment ofcleavage frag-ments at 2 kbar, 800'C, 115 h, water/tremolite ratio of 3.6/1.0(by weight). Notice the relatively smooth, rounded edges of thepartially dissolved and recrystallized cleavage fragments. O) Sur-face textures resulting from hydrothermal treatment of partiallydissolved tremolite fragments plus excess quartz at2kbar,795"Cfor 2l'7 h. Notice the sharp, protruding edges resulting fromtremolite overgrowths parallel to the Z axis of the relic tremolite(approximate boundary is dashed).
react to, tremolite depending on the activity of SiO, inthe ambient fluid. There is no indication that the kineticsat 800-880'C and 2-6 kbar are inhibiting the growh oftremolite, particularly when abundant tremolite nuclei arepresent. Instead, it is the composition of the system thatexerts the dominant control.
The join Ca,Mg.SirOr,(OH)'-Mg'SisOrr(OH)'
Syntheses using a series of bulk compositions along thejoin Ca,Mg,Si8O,,(OH)r-Mg'SirO,,(OH), (TR-MC) wereperformed to determine how variation in the CalMg ratioof the bulk composition afects tremolite formation. Mix-tures were prepared from the oxides in 5 molo/o incre-ments between TRrrMC, and TRr''MCo. Two series of
JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE
TneLr 2. Experimental results on the dissolution and regrowth of natural and synthetic tremolite
t r e m + o p x + c p xt r e m + q t z + [ c p x ]c p x + f o + t r e mt r e m + c p x + o p xt r e m + q t z + [ c p x ]
trem (poorly crystalline)
tremt r e m + c p x + o p xtrem + qtz
825 + 108 0 6 + 58 0 0 + 5
285
2ao
1443 1 0
'Abbreviations: cpx : clinopyroxene; fo : forsterite; opx : orthopyroxene; trem : tremolite; qtz : quartz... The starting material was placed in a crimped (not sealed) Pt capsule to expose the material to the distilled-water pressure medium.
experiments were performed: one at 6 kbar and 850"Cand the other at about 13 kbar and 750'C. An additional2-10 wto/o quartz was added to each of the starting mix-tures in the 6-kbar series in order to saturate the fluid insilica and drive Reactions I and 2 to the left. The resultsof these experiments are listed in Table 3, and the 6-kbarresults are shown in Figure 4. At 6 kbar, mixtures atTR,..MC. and TRnrMC, yielded the assemblage tremo-lite + clinopyroxene + quartz, that at TRe.MCr. yielded
tremolite + qsartz, and those at TR85MC'5, TR8oMC2o,and TRTTMC- yielded tremolite + orthopyroxene +q\arrz. Only the mixture at TReoMCr. yielded amphibolewithout coexisting pyroxene. The experiments at l3 kbar,which were performed without excess SiO2, yielded re-sults similar to those at 6 kbar. The TR,ooMCo andTResMCs mixtures yielded tremolite * clinopyroxenewithout orthopyroxene, and the TRssMCrs and TR8oMC2omixtures yielded tremolite * orthopyroxene without cli-
TneLe 3. Synthesis results from mixtures along the join CarMguSirOrr(OH)r-MgrSi.-o,,(oH), (TR-MC)
Mol%TR in
mixture Run codeP T I
(kbao CC) (h) Products
100
95
90
85
8075
r00
With excess ouartz6.00 851 135 t r e m + c p x + q t z
Notei Abbreviations as in Table 2.- U n i t - c e l l d i m e n s i o n s o t t h e t r e m o l i t e a r e a : 9 . 8 0 6 + 0 0 0 3 : b : 1 8 0 4 6 + 0 . 0 0 6 : c : 5 . 2 7 8 +
0.002 A: d : 104.55' + 0.04' : V: 904.1 + 0.4 A"... Unit-cell dimensions of the tremolite are a : 9.804 + 0.002: b: 18.047 + 0.004: c: 5.280 +
0 001 A ; 6 : 104 .53 ' + O .O2 i V :904 .3 + 0 .3 A3 .t The multiple sets of conditions indicate multiple hydrothermal treatment ot the same material
with intermediate grinding.
7t2 JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE
Fig. 4. Phase relations in a portion (shaded region in inset)of the system HrO-CaO-MgO-SiO. projected from H.O ontothe ternary diagram CaO-MgO-SiOr. Numbers refer to themo10/o of CarMgrSirOrr(OH), in mixtures along the joinCarMgrSirOrr(OH)r-MgrSirOrr(OH)r. Tie lines between Opx, Cpx,and Trem have been inferred. Pyroxene compositional rangesfrom Lindsley and Dixon (1976). Abbreviations given in Ta-ble 2.
nopyroxene. However, the nucleation of talc in two ofthe experiments and the almost certain loss of silica tothe ambient fluid hinders direct comparison with the6-kbar results. From Figure 4 one can see that tremolitesynthesized from a bulk composition of TRr00MC0 willhave an estimated maximum of 93 molo/o of the TR com-ponent.
The synthesis experiments limit the extent of tremolitesolid solution to the compositional range of 90 + 3 molo/oTR. An even narrower range of solid solution, i.e., a sin-gle composition, is suggested by the unit-cell dimensionsof the tremolites produced in the most calcic (TR,ooMCo)and the least calcic (TR?sMCrr) bulk compositions inves-tigated. These unit-cell dimensions are given in the foot-notes to Table 3. Notice that both tremolites have thesame unit cells within the limits of uncertainty. If therewas significant solid solution, one would anticipate the adimension to increase by 0.0035 A"/molo/o TR, or the vol-ume by 0.26 A'/molo/o TR, on the basis of a linear inter-polation between the unit-cell dimensions of tremolite(Borg and Smith, 1969) and the hypothetical phase mag-nesio-cummingtonite extrapolated from the data of Vis-wanathan and Ghose (1965). Using these values, one findsthat the maximum compositional range allowed by theunit-cell dimensions (and their uncertainties) in Table 3is 2 molo/o TR.
Several attempts were made to obtain direct chemicalanalyses on individual synthetic tremolite crystals usingboth an electron microprobe to analyze polished grainmounts and an srv (equipped with a quantitative energy-dispersive unit) to analyze grains dispersed on a graphite
substrate. In general the analyses yielded very low totalsbecause of the small grain sizes and, therefore, substan-dard analyses. However, the CalMg rqtio was consistent-Iy lower than that of ideal tremolite and tended to liewithin the ratge 1.7 5/5.25 to 1.9/ 5.1 for those grains thatgave the highest total weight percentages. The general in-dication is that direct chemical analyses confirm the lowCa/Mg ratio in synthetic tremolite.
CalMg reversal attempts
The evidence presented so far indicates that tremolitesynthesized from a variety of starting mixtures is Mg en-riched. It remains to be shown, however, that this is theequilibrium composition of tremolite. Ideally, one wouldlike to demonstrate that a Ca-enriched and a Mg-en-riched tremolite converge to, or narrowly bracket, a spe-cific composition at a given P and T. Unfortunately, thisis not possible for a substance that displays virtually nosolid solution (Fig. a). An alternative approach is to changethe composition of tremolite by reacting it with otherphases to produce a more calcic or more magnesian trem-olite and thus to demonstrate that this process is revers-ible.
Ir was noted by Robinson et al. (1982, Fig. 5l) thattremolite in ultramafic rocks shows a tendency towardCa enrichment at lower metamorphic grades. According-ly, several experiments were performed at 490"C with atremolite that was synthesized at 800"C in order to inducean increase in its CalMg ratio by reaction with clinopy-roxene and quartz (Fig. a). A mixture of synthetic trem-olite, diopside, and quartz was hydrothermally treated at490"C and 2.5 kbar for 2166 h with grinding after 531 h.No perceptible reaction occurred. An additional treat-ment of the same material with a 0.2m CaI, solution for679 h also yielded no reaction. Talc was not observed.Additional experiments at pressures above 8 kbar mightprove successful in view of the enhanced reaction ratesat higher pressures (e.g., Goldsmith and Newton, 1974),but such experiments have not yet been performed be-cause of the extreme durations involved.
In summary, experimental proof of the Mg enrichmentof synthetic tremolite or of any variation in its CalMgratio with temperature has yet to be obtained. The onlyargument that can be made at present for the stable ex-istence of Mg enrichment in synthetic tremolite is thatmany different starting materials (i.e., oxide mixture, gel,crystalline phases) produce the same assemblage of trem-olite plus clinopyroxene at temperatures of 750-850"Cand that the process does not appear to be path depen-dent.
DrscussroN
Applications to natural trernolite
The presence of Mg enrichment, or at least Ca deple-tion relative to octahedral cations, in natural tremoliteshas been noted by Wones and Dodge (1977) and Rob-inson et al. (1982) based on surveys ofnumerous calcic
T r e m * C p x * Q t z
T r e m t O p x * Q l z
MorSiaOa, (OH),
o 3 4 6 o 3 6 9 : ! 0 o r 5 I ' E A L T R E M
t l lt t l{ { t
JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE 7 t3
ozl
UEL
C o l I M E
Fig. 5. Histogram ofthe ratio ofca to the sum ofall cationsassigned to octahedral sites (CalZ"IM) of 25 natural tremolitesselected according to the criteria discussed in the text. Aruowsindicate the values for synthetic tremolite (0.346), mean for thenatural tremolites (0.369), and ideal tremolite (0.40).
amphibole analyses. A survey of natural tremolite com-positions was also performed in this study in order todetermine the range of cation variations within the "pur-est" tremolites reported in the literature. Out of a totalof 123 tremolite analyses surveyed from the literature(Deer et al., 1963:, Leake, 1968; Ross et al., 1969:Trommsdorffand Evans, 19721-FrosI, 1975; Misch andRice, 1975; Slaughter et al., 1975; Vance and Dungan,1977;Yalley and Essene, 1980; Sanford,1982), only 25were found to obey the following criteria: AlrO3 < 1.0wt0/0, FeO < 2.0wto/o (NarO + KrO; = 0.5 wt0/0, FerO. <1.0 wto/o (if given), and the sum of Ca plus the cationsMg * Fe2* + Ni + Mn + Ti + vIFe3+ + vrcr + vIAl(:
)vIM) being 7.0 + 0.1 on the basis of 23 oxygens. Theassignment of cations to the octahedral sites was made inaccord with the assignment scheme of Leake (1978) withthe exception that the total number ofoctahedral cationscould exceed 5.0 and that Na. K. and Ca could not beassigned to octahedral sites. These selection criteria wereused in order to identify tremolites that are as free aspossible of non-tremolite components and that have fulloccupancy of the Ml to M4 sites. A histogram of the Cal)v'M ratios for these 25 analyses is shown in Figure 5, aswell as the ratios for the ideal tremolite composition(2.0/5.0 : 0.4) and the synthetic tremolite (1.8/5.2: 0.346).Although the database is admittedly small, Figure 5 showsthat none of the natural tremolites considered here hasthe ideal ratio of 2/5 and that most have values lyingmidway between ideal and synthetic tremolite. In otherwords, even the purest tremolites in nature frequentlypossess an excess of cations assigned to the octahedralsites, indicating the presence of (mostly) Mg and Fe2* onM4 sites.
There are several lines ofevidence to suggest that theCa/2vtM ratio of tremolite varies inversely with temper-ature. First, synthetic tremolite has a smaller Cal)"IMratio than that of the most natural tremolites (Fig. 5). Itis reasonable to assume that tremolite synthesized in therange 750-850"C has formed at temperatures much higher
MorSiro22(OH)2 Mol % CotMeUSiaOt2(OH)t
Fig. 6. Approximate phase diagram for the pseudobinaryjoinMg,SirO,,(OH)r-CarMgrSirO,dOH)' in the system HrO-CaO-MgO-SiO, at Pr,o:2 kbar. HrO is present in all fields. Shadedregion denotes range of tremolite solid solution. Dashed linesare approximate boundaries. Upper thermal stability of pure an-thophyllite (Anth) is from Chernosky et al. (1985). Upper ther-mal stability of tremolite is from Fig. l. Other abbreviations asin Table 2.
than those of natural tremolites from greenschist or low-er-amphibolite facies mafic, ultramafic, or siliceous car-bonate rocks. Second, the experimental study by Cam-eron (1975) on cummingtonite-actinolite phase relationsin the system HrO-CaO-MgO-FeO-SiO, at 2 kbar and500-800"C suggested that the Ca/(Mg + Fe) ratio of ac-tinolite decreased with increasing temperature. Third,Robinson et al. (1982) observed a decrease in the Cal)"IM ratio of calcic amphiboles with increasing meta-morphic grade in progressively metamorphosed ultra-mafic rocks. Combining these observations with the in-formation on tremolite compositions obtained in thisstudy, one can approximate the phase relations along thepseudobinary join MC-TR, as shown in Figure 6. Thecritical feature to be noted in this diagram is the verynarrow range of solid solution indicated for tremolite(shaded region). Unlike the analogous single-chain sili-cate join MgrSi,Ou-CaMgSirOu (enstatite-diopside), thereis no indication ofa range ofextensive solid solution foramphiboles along this join at high temperatures. This israther surprising in view of the general similarities be-
C)o
Fi
P6a9=2 kbor , H2O EXCESS
C p x * 0 p x * Q t z
T r e m + O p x + Q l z
An th +0px +Q lz
Trem +An lh
)-''IIIII
III-lIII
7 t 4 JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE
tween the pyroxenes and amphiboles as demonstrated,for example, by their similar structural transformations(Carpenter, 1982). Nevertheless, the experimental evi-dence does not support the presence ofan extensive rangeof solid solution near the maximum temperatures of am-phibole srability (750-8 50"C).
Thermodynamic considerations
Mg enrichment in synthetic tremolite will affect thethermodynamic properties of the TR component derivedfrom phase equilibria. At a given temperature and pres-sure, the change in the chemical potential of the TR com-ponent (pr") resulting from solid solution in amphibole(amph) is given by the relation
pitph - p$* : RI ln aiirh : Aprn, (3)
where pgn is the chemical potential of the pure phase ofcomposition TR and cttph is the activity of TR in theamphibole solid solution. At present there is little exper-imental information on the activity-composition rela-tions of amphiboles. However, one can examine severalpossible activity expressions and from them ascertain arange of vaiation in derived thermochemical properties.
The simplest assumption for activity is that of idealmixing of unlike cations on each crystallographic site. Inthe case of tremolite, the ideal activity expression is
where -{ is the mole fraction of species I on site /<, !stands for vacancy, and site nomenclature is that of Haw-thorne ( I 98 1). The reader is directed to Wood and Fraser(1976, Chap.3) or Nordstrom and Munoz (1985, App.C) for the theory behind the derivation of ideal activityexpressions. For the system investigated in this study, allvalues of )(} in Equation 4 are unity except X}"a, which is0.9 because of the l0o/o additional Mg in synthetic trem-olite. Thus, Equation 4 simplifies to
a+Reh : (xg"), : (0.9), : 0.81. (5)
At 800"C, for example, this activity yields a Ap,* of - 1.88kJlmol.
An alternative activity expression was proposed byPowell (1975), who treated natural amphiboles as regularsolutions (e.g., Wood and Fraser, 1976) of eight compo-nents. Using a suite of coexisting cummingtonite-horn-blende pairs from New Zealand rhyolites, Powell (197 5)was able to derive a set of seven mixing parameters byknowing the compositions of the amphiboles and theirtemperature of equilibration as determined by Fe-Ti ox-ide geothermometry. Application of Equations 6 and 8of Powell (1975) to the synthetic tremolite of this studyyields the following expressions for the mole fraction(X+Pn), activity coefficient (,y?t'n), and activity of the TRcomponent:
At 800"C, Equations 6a-6c yield 411*n : 0.857, which inturn, yields & Ap". of - 1.38 kJ/mol.
It is apparent from the above discussion that reason-able choices for aii*h will increase the derived Gibbs freeenergy of formation, AGp, of stoichiometric tremolite byup to 2 kJ/mol. At present, this adjustment is muchsmaller than the range in the reported values of AGP (298K, I bar) of tremolite, which vary from - ll 592.55 kJ/mol (Helgeson et a l . , 1978) to - \ t627.91 + l7 kJ/mol(Robie et al., 1978). However, a 2-kJ increase in theAGP of tremolite translates to a decrease of 20-30"C inthe calculated position ofthe reaction
Financial support for this project was provided by National ScienceFoundation (NSD Grant EAR-8507752 to the author and by NSF GrantEAR-8305904 to Julian R. Goldsmith at the University of Chicago.
Special thanks are exended to Jon Palmatier for fabrication ofthe in-ternally heated gas vessels at SUNY-Binghamton and to William Black-burn for assistance with the serrr analyses. Chemical analyses ofthe BarrieTownship, Ontario, tremolite were kindly provided by Ian Steele (Uni-versity of Chicago) and fuchard Hervig (Arizona State Universi$. Thanksalso to Katherine Holley, Olga Kurty, and David Tuttle for manuscriptpreparation The manuscript was greatly improved by the reviews ofJo-seph V Chemosky, Jr, Frank S. Spear, and Warren M. Thomas.
RnrnnnNcBsAppleman, D.E, and Evans, HT, Jr (1973) Iob 9214: Indexing and
least-squares refinement of powder diflraction data. U.S. NationalTechnical Information Service, PB-216- 188.
Borg, I.Y., and Smith, D K (l 969) Calculated X-ray powder patterns forsilicate minerals. Geological Society of America Memoir 122.
Boyd, F.R. (1959) Hydrothermal investigations of amphiboles In P HAbelson, Ed., Researches in geochemistry, vol. I, p. 377-396 wiley,New York
Cameron, K L. ( I 975) An experimental study of actinolite-cummington-ite phase relations with notes on the synthesis ofFe-rich anthophylliteAmerican Mineralogist, 60, 375-390.
Carpenter, M.A (1982) Amphibole microstructures: Some analogies withphase iransformations in pyroxenes. Mineralogical Magazine, 46, 39 5'397 .
Chernosky, J.V., Jr., Day, H W, and Caruso, L.J (1985) Equilibria in thesystem MgO-SiOlHrO: Experimental determination of the stability ofMg-anthophyllite. American Mineralogist, 7 0, 223-236.
Frost, B R ( I 975) Contact metamorphism of serpentinite, chloritic black-wall, and rodingite at Paddy-Go-Easy Pass, central Cascades, Wash-ington Journal of Petrology, 16, 272-3 13.
Goldsmith, J R., and Newton, R C (1974) An experimental determina-tion of the alkali feldspar solvus. In W S. MacKenzie and J. Zussman,Eds , The feldspars, p. 337-359 Manchester University Press, Man-chester
Hawthorne, F.C. (1981) Crystal chemistry of the amphiboles. Mineral-ogical Society of America Reviews in Mineralogy, 9A., l-102.
Helgeson, H.C., Delany, J.M , Nesbitt, H.W., and Bird, D.K. (1978) Sum-mary and critique of the thermodynamic properties of rock-formingminerals American Journal of Science, 27 8-4.
modelled in the system HrO-CaO-MgO-SiOr. Contributions to Min-eralogy and Petrology, 77,166-176
-(1983) Stability and composition relations of calcic amphiboles inultramafic rocks. Contributions to Mineralogy and petrology, 83, 375-384
Leake, B.E. (1968) A catalog of analyzed calciferous and subcalciferousamphiboles together with their nomenclature and associated minerals.Geological Society ofAmenca Special Paper. 98.
- ( I 978) Nomenclature of amphiboles. American Mineralogist, 63,I 023- l 05 2.
Lindsley, D.H., and Dixon, S.A. (1976) Diopside-ensratite equilibria at850" to 1400'C, 5 to 35 kbar. American Journal ofScience,276,1285-I 301
Misch, P , and Rice, J M. (1975) Miscibility of tremolite and hornblendein progressive Skagit metamorphic suite, north Cascades, Washington.Journal of Petrology, 16, l-21.
Nordstrom, D.K, and Munoz, J L (1985) Geochemical thermodynamics.Benjamin/Cummings, Menlo Park, California
Oba, T (1980) Phase relations in the tremolite-pargasitejoin. Contribu-tions to Mineralogy and Petrology , 7 l, 247-256
Powell, R (1975) Thermodynamics of coexisting cummingtonite-horn-blende pairs Contributions to Mineralogy and petrology, 51,29-3'7.
Robie, R.A., Hemingway, B S., and Fisher, J R (1978) Thermodynamicproperties of minerals and related substances at 298 15 K and I bar(105 pascals) pressure and at higher temperatures. U.S. Geological Sur-vey Bulletin 1452.
Robinson, P, Spear, FS, Schumacher, J.C., Laird, J., Klein, C., Evans,B.W., and Doolan, B.L. (1982) Phase relarions of metamorphic am-phiboles: Natural occurrence and theory Mineralogical Society ofAmerica Reviews in Mineralogy, 98, l-228.
Ross, M, Papike, JJ., and Shaw, K.W. (1969) Exsolution rexrures rnamphiboles as indicators of subsolidus thermal histories. MineralogicalSociely ofAmerica Special Paper 2,275-299
Sanford, R F. (1982) Growth of ulrramafic reaction zones in sreenschist
I I J
to amphibolite facies metamorphism American Journal of Scrence,282,543-616.
Skippen, G, and McKinstry, B W (l 985) Synthetic and natural tremolitein equilibnum with forsterite, enstatite, diopside, and fluid. Contribu-tions to Mineralogy and Petrology , 89, 256-262
Slaughter, J., Kenick, D.M., and Wall, V.J. (1975) Experimental andthermodynamic study of equilibria in the system CaO-MgO-SiOr-H.O-CO, American Journal ofScience,275, 143-165
Troll, G., and Gilbert, M.C. (1972) Fluorine-hydroxyl substitution intremolite American Mineralogist, 57, 1386-1403
Trommsdorfi V , and Evans, B.W. (1972) Progressive metamorphism ofantigorite schist in the Bergell tonalite aureole (Italy). American Joumalof Science, 272, 423-437 .
Valley, J W , and Essene, E J ( I 980) Calc-silicate reactions in Adirondackmarbles: The role of fluids and solid solutions. Geological Society ofAmerican Bulletin, Part II, 91, 720-815
Vance, J A., and Dungan, M.A (1977) Formation of peridotites by de-serpentinization in the Darnngton and Sultan areas, Cascade Moun-tains, Washington. Geological Society of America Bulletin, 88, 1497-I 508
Viswanathan, K., and Ghose, S. (1965) The effect ofMg'?*-Fe'?* substitu-tion on the cell dimensions of cummingtonites. American Mineralogist,5 0 , l 1 0 6 - 1 1 1 2
Wones, D R , and Dodge, F C W (1977) The stability of phlogopite inthe presence of quartz and diopside. In D G. Fraser, Ed., Thermody-namics in geology, p. 229-248 D. Reidel, Dordrecht, Holland.
Wood, B.J , and Fraser, D G. (1976) Elementary thermodynamics forgeologists. Oxford Press, Oxford.
Yin, H.A., and Greenwood, H.J. (1983) Displacement of equilibria ofOH-tremolite and F-tremolite solid solution. I. Determination of theequilibrium P-Z curve of OH-tremolite (abs ). EOS, 64, 347
MlNuscrrpr REcETvED Decruesr l. 1986Mnxuscnrrr AccEPTED Apnrl 3. 1987
JENKINS: SYNTHESIS AND CHARACTERIZATION OF TREMOLITE
