Highly Aluminous Hornblende from Low-Pressure Metacarbonates And

American Mineralogist, Volume 76, pages 1002-1017, 1991

Highly aluminous hornblendefrom low-pressuremetacarbonates and a preliminary thermodynamicmodel for the Al content of calcic amphibole Ar-rnnr Lfcrn, JoHN M. Frnnv Department of Earth and Planetary Sciences,The Johns Hopkins University, Baltimore, Maryland 21218, U.S.A.

Ansrucr Calcic amphiboles in carbonate rocks at the same metamorphic grade from the Waits River Formation, northern Vermont, contain 2.29-19.06wto/o AlrO, (0.38-3.30Al atoms per formula unit, pfu). These Al-rich amphibole samplesare among the most aluminous examples of hornblende ever analyzed. The amphibole-bearing metacarbonatesare interbedded with andalusite-bearingpelitic schists and therefore crystallized at P < 3800 bars. Theseresults demonstrate that factors in addition to pressuremust control Al content in hornblende. We have identified temperature,mineral assemblage,mineral composition, and rock chemistry [especiallyFe/(Fe + Mg)] as other important factors. To explore semiquantitatively the dependenceof the Al content of calcic amphibole on P, T, and coexisting mineral assemblage,a simple thermodynamic model was developed for mineral equilibria involving tremolite-tschermakite ([CarMgrSi'O'r(OH)r]- [CarMg.AloSi6Orr(OH)r])amphibole solutions. The model uses the thermodynamic data base of Berman (1988), with the addition of new values for standard-stateenthalpy and entropy for pure end-member tschermakite derived from experimental and field data on the Al content in tremolite coexisting with diopside, anorthite, and quartz. Calculated phase equilibria lead to three conclusions:(l) At a specifiedP and T, the Al content of calcic amphibole is strongly dependenton the coexisting mineral assemblage.(2) No universal relationship exists between the Al content of amphibole and P. (3) The Al content of amphibole may changedramatically wilh changesin P and 7. Maximum Al contents calculated by the model are - 1.2 Al atoms pfu. These values are far short of the 3.30 Al atoms pfu measuredin some amphibole samplesfrom Vermont. The principal shortcoming of the model is its failure to consider both total Fe in amphibole and the partitioning of Fe and Mg among the Ml, M2, and M3 crystallographicsites.

INrnooucrtoN correlation exists between the Al content of hornblende and pressure(cf. Spear, I 98 l; Pluysnina, 1982;,Cao et al., The chemicalcomplexity of hornblendehas long been 1986;Jenkins, 1988, 1989). ofinterest to petrologistsbecause ofits potential as a rec- It came as a surprise, therefore, that during a regional ord of intensive variables during amphibole crystalliza- study of the progressive metamorphism of marls from tion. Recently, studies have focusedon the Al content of the Waits River Formation, northeastVermont, we found hornblendeas a practical geobarometer.Experimental and some hornblende samples with unusually high Al con- empirical studies, for example, show that a positive cor- tent. The most Al-rich amphibole samplesin this study relation exists between total pressureand the Al content contain over 19 wto/oAlrO, (3.30 Al atoms pfu). These of hornblende in certain assemblagesof minerals and sil- are among the most aluminous hornblende samplesever icatemelt (HammarstromandZnn, 1986;Hollister et al., reported (cf. Leake, 1965a, 1965b, l97l1, Doolan et al., 1987; Johnson and Rutherford, 1989a, 1989b).The Al 1978;Selverstone et al.,1984;Sawaki, 1989). Even though content in hornblende is thus regarded as a useful geo- thesehornblende sampleshave high Al content, they are barometer for appropriate plutonic and volcanic rocks. neither from very aluminous rocks (the host marls have Highly aluminous amphibole is likewise often regarded < 15 wto/oAlrOr) nor from rocks metamorphosedat ele- as an indicator of high-pressuremetamorphic rocks be- vated pressure (the marls are interbedded with andalu- causeit is found coexisting withjadeite, kyanite, or glau- site-bearingschists and crystallized at P < 3800 bars). cophane(Leake, 1965a,1965b; Kostyuk and Sobolev, OursecondmajorobservationisthattheAlcontentof 1969; Raase,1974; Selverstoneet al., 1984).Moreover, hornblende varies greatly (from -0.4 to 3.3 Al atoms some experimental studies of mineral assemblagesrele- pfu), even in marls from Vermont at the same metamor- vant to metamorphic rocks have shown that a positive phic grade, depending on the coexisting mineral assem- 0003-004x/9l/0506- I 002$02.00 I 002 LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE 1003 blage and mineral chemistry. Both observationsindicate that some factor(s) other than pressureand temperature /< a' tr- must exercisea first-order control on the Al content of amphibole.Our study identifiestwo suchcontrols: (l) the assemblageof minerals coexisting with hornblende and (2) the composition of those minerals [particularly Fe/(Fe + Mg) and anorthite content of plagioclasel.We have developeda simple model for the Al content of tremolite- tschermakite solid solutions in the system KrO-CaO- MgO-AIrOr-SiOr-HrO-COr.The model explainsthe de- pendence of the Al content of hornblende both on the coexisting mineral assemblageand the P-7 conditions of crystallization. The model for this simple system, how- Amphibole localities ever, does not predict the high-Al hornblende observed + diopsialezone in some rocks from Vermont. Further researchmust be t tr amphibole zone completed before an adequate quantitative understand- ing of the Al content of calcic amphibole can be attained. ISOGRADS d iops ide amphibole - - - .-biotile Gnor-ocrclr, sET"trNG oF HIGH-AI HoRNBLENDE FROM VERMONT

Hornblende-bearing metamorphosed marls were collected from the Siluro-Devonian Waits River Formation Fig. l. Geologicsketch map of the areastudied. Thick solid : betweenDevonian Gile Moun- in northeasternVermont (Fig. 1),which has beenmapped lines stratigraphicboundary tain (Dgm)and Waits River (Dwr) Formations.Isograds are based by numerousgeologists (Doll, 1951;Dennis, 1956;Hall, on mineralassemblages in metacarbonaterocks and have ha- Konig and Dennis, 1964;Woodland, 1959;Cady, 1960; chureson their high-gradesides. Filled squares: amphibole 1965).The formation consistsof interbeddedlimestone localitieswith geothemometry;empty squares : otheramph! (2- I 5 cm thick), argillaceousmetacarbonates, and minor bolelocalities; filled circles: other geothermometry localities. LM pelitic schists.The Waits River Formation, togetherwith : I-akeMemphremago$ LW : LakeWilloughby; pre-S : pre- the adjacent and more pelitic Gile Mountain Formation, Silurianrocks; CV-GS : ConnecticutValley-Gaspe Synclinor- are part of the Connecticut Valley-Gaspe synclinorium. ium. Geologicmap basedon that of Doll et al. (1961). The structure and metamorphism now observed in the area developed primarily during the Devonian Acadian II{vESTrGATIoN Orogeny (ca. 370-400 Ma). Two major folds, the Brown- Mnrrroos oF ington syncline (Doll, l95l) on the western side of the Thirty-one samplesof metamorphosedimpure carbon- synclinorium and the Strafford-Willoughby arch (Eric and ate rock containing amphiboles were collected from 27 Dennis, 1958) on the eastern side of the synclinorium field localities (Fig. 1). Mineral assemblageswere identi- control the regional attitude ofbeds. Bedsstrike northeast fied by petrographicobservation of thin sections.Mineral except where locally deflectedaround igreous intrusions. compositions in all sampleswere determined by electron The metasedimentaryrocks were intruded by numer- microprobe analysis using the JEOL JXA-8600 Super- ous late- to posttectonic granitic plutons of the New probe at the Department of Earth and PlanetarySciences, Hampshire Plutonic Series(Doll et al., l96l). From the The Johns Hopkins University, and natural mineral stan- distribution of isogradsand metamorphic zones(Fig. l), dards. Data for both silicates and carbonates were re- it is clear that the plutons were local heat sourcesfor the duced with a ZAF correction scheme.Chemical analysis highest gradesof metamorphism attained in the area. With of 23 rock samples for major elements were performed respectto pelitic schists,metamorphic graderanges from by X-ray fluorescence:13 at X-Ray Assay Laboratories, chlorite zone to the sillimanite zone near the plutons. Ltd., Don Mills, Ontario, and ten at the Geochemical Garnet and amphibole porphyroblasts crosscutthe main Laboratories,Department of GeologicalSciences, McGill schistosity,indicating that the peak of medium- and high- University, Montreal, Quebec. grade metamorphism occurredafter the last period of de- AND cHEMrsrRy AND formation. Four metamorphic zones were also mapped MrNnnc.L CoNTENT basedon minerals in the metamorphosedmarls (Fig. l). PETROI,OGIC FEATURES OF ROCKS WITH The biotite isograd separatesthe ankerite zone at lower TTTCTT-AI HORNBLENDE grade from the biotite zone al higher grade. The amphi- Mineral assemblagesobserved in six representative bole isograd separatesthe biotite and amphibole zones, specimensfrom the amphibole zone are listed in Table and the diopside isograd separatesthe amphibole and l. These specimenswere chosen to show the wide range diopside zones.Amphibole describedin this study is from in amphibole compositions (especiallyin AlrO. content) rocks of the amphibole and diopside zones. even in marls metamorphosedat the same gtade. 1004 LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE

nite, sulfides, chlorite, clinozoisite, potassium feldspar, garnet, muscovite, ankerite, and ilmenite. Compositions Al2O3-CaO- of amphibole, biotite, chlorite, calcite, clinozoisite, garnet, and plagioclasefor the six specimenslisted in Table Selected Analyses I are compiled in Table 2. Mineral compositions in Table 2 are typically an averageof five to ten spot analysesof three to five grains. Except where otherwise stated, analyses represent rim compositions. The value of "oxide sum" refers to the conventional sum of metal oxide weight percentageswith all Fe as FeO. No analysesof quartz, FeO + MnO titanite, or sulfideswere obtained. Minerals from the amphibole zone are describedbelow. Amphibole. Calcic amphiboles occur in all 23 specimens studied. They vary widely in composition from AMPHIBOLE tremolite to ferro-tschermakite in the nomenclature of (a) Leake(1978). Amphibole sampleshave 0.38-3.31 Al atoms and 7.81-6.18Si atoms pfu and Fe/(Fe + MC): to calcite 0. l5-0.62. The compositionsof amphibole samplesand coexisting chlorite, calcite, and garnet are plotted on Fig- ure 2a for selectedsamples of metamorphosedmarls from the amphibole zone of the Waits River Formation. The diagram approximatesmineral compositionsby the model All Analyses system Na,O-CaO-MgO-FeO-MnO-AlrO3-SiOr-HrO- a amphibolezone CO,, following the schemeof Robinsonand Jaffe(1969). + diopsidezone 60 (b) The compositions of amphibole samplesalone in all samples analyzedfrom the amphibole and diopside zonesare Fig. 2. (a) Chemographic relationships among minerals in plotted in Figure 2b. five representativesamples from the amphibole zone of Figure Two observations can be made from Figure 2: (l) A 1. Diagram refers to constant P and T and assemblageswith wide range of amphibole compositions crystallized at the plagioclaseand quartz. Note the greater dependenceof AlrO, same metamorphic grade. Amphibole from the diopside content in amphibole on (Fe + Mn)/(Fe + Mg + Mn) than in zone crystallizedat higher temperaturethan that from the chlorite. (b) Sameplot as in note positive a; correlation between amphibole zone but are plotted on the same diagram for (FeO + MnO) and AlrO, in all amphibole samples analyzed comparison. (2) A distinct, positive correlation exists be- from metacarbonatesof the Waits River Formation, northern Vermont. tween the AlrO. content and (Fe + Mn)/(Fe + Mn + Mg) in amphibole. There is also a very weak positive correlation betweenthe Al content and (Fe + Mn)/(Fe + Mn + Mg) of coexisting chlorite and biotite (see Table Mineral assemblages and mineral chemistry 2), but the correlation for chlorite is barely perceptible at Metamorphosed carbonate rocks from the amphibole the scaleof Figure 2. zone in Figure I all contain calcite, qrrartz, plagioclase, Low-Al amphibole samplesgenerally are smaller in size amphibole, and biotite, along with combinations of tita- than Al-rich counterpartsand coexist with muscovite and

TlaLe 1' Mineralabundances" in representativespecimens from the amphibolezone, Waits River Formation.northern Vermont Samplenumber 49-10A 41-31 49-5 48-50 56-11 Amphibole 0.38 0.32 0.82 0.17* 0.70 0.26 Calcite 18.37 14.77 11.09 5.22 7.85 6.70 Ouartz 6.22 12.04 8.20 8.73 14.90 10.86 Biotite 0.34 0.12 0.35 1.56 0.08 0.11 Chlorite 0.07"- 0.09-. 0.01 0.15 0.42* Plagioclase 0.25 0.55 1.O2 3.',t4 0.91 3.04 Garnet 0.18-" 0.14* Clinozoisite 0.03". 0.004* Ankerite tr'* llmenite 0.10 0.09 0.20 Pyrrhotite tr*' 0.26* tr'* 0.37"- 0.88- tr* Titanite 0.002 0.04 0.04 Note.'The abbreviationtr indicates <0.05 vol%. * Abundancesare in moles/literof metamorphicrock, -'Modal calculatedfollowing the method of Ferry (1989). abundanceis determineddirectly by counting 2000 points in thin section. N Froo@@(oolb N F(\l@c)FNrtNo(9|oql(90 Ndt;NN6oi+ o o;o(D(\lo(gFF|r)(o(Oolo o + <5at-i\d6ct+ E Oq-u?Q-9rra?u?qgq ol o J ;;Gtcio6ictcicio;oct- FOFFIOCOO)O)O+[ ddct;dooi o FIOAI (\lrOONO)(ON d--dctF-cict-N F$oy F * F- C) O) (9 (oN F O rO rON F C, rr, ! (9@ oq-{ (pQ @ z N 6 ao o 6 o N ct o 6r (o (o t At q) o o (r) Al t\ c) N O ro qqc.ii (9o eaqqcaeru?cqaqqqqqcq OOOF FFOOO-F(OOOFOOOOFOOF

N FO(orO@(9GrO ts O(?F.olFo.+(D(OrSF-!qO + N+dr6d|o;+ t Od)@Al6l(DC)-N@or

N b6@OO*(')F F- O)OOOn|otOr@Oc)N|oQ o@@@

N ho)F-N(D6O(t (o r(9F-$orOC\lrFS(DQq?O ;6iocjoo@N O -NF--Ftstsc)(oo@(9oO + cqqeqc\\ c a\qu?qqqrga?nccq E o FF(oOoFooOOFOoF cooo(o+(oc)o6@r OOO-OOFF o o qqu?olra?a?ro?q OTFOCO!OO@NCO \t F

N O@OoOotf:f F F@NoOoNooo(l)-Qo + 8N8E38ts3 I R335839=8R8888 o ooci'joci-Ai d ci-(cidct;cidooodd; @@or*

N :trr)F(o(orr)NO - drN(O@\t$NO$(gEr6O + oonoF(\t*F. 6 NCOFOO(?OsfbOC)|o=9 CrcAcqcq q c?au?qqo?rqcr-qqqq o OOOFOOO(9 o OONOOoOOOOOOOF 6IF.NTOOTc|OFO$r () On@FtN*OON 6 OONOF.@Ot(r)N boY $@OO o+OFOOFON*NTOOO

(g N $FNFoNoo * O@$Ot(o(oF-6tb(O+OO FNrOOO@O (t (Do)ONFF(()GOF(D@(DOO o + qrqo?qC\c q olaq-qoqqqqra?gqc o OOOFOOO(', O OOF.OOOOOOOOOOF 6oOooFN@@oI E o$@obNoNN6 oooiodNct+dN o o, !oorv :i \i rr)N cO @ O, O sf @ (9 (o* Al + + t OOI O, N t Q O z FFF(o ('rNO(tO\iSfbNFOF.OFNF(',QQ (! I q ': c9 u? 0c c c a? 6l u? a r q u?a? q a c q I q q o OOoe O(")OOoooNOOOoooOOOOF o o o o 6 E. *61 ol u F'',-:'EnE&i.,* J @ dddCa.a.OE F e.3tE9F s$gF YZZ)>fid>>i= FFsAFEE$gFEfrPSE N(D(oONN$F.*(o n (Dst@O @ oFOFA|r)@-NN66 otf(DF(0F- NOOOONOhNNOO o @ ooooot ro dcido;-cici-Aidci o oooo@o o n

o Ntottrr$oo@oo @ ONFNNO rO o o{tso o s tOOOAION@6$O- o\t\tFO('r@ri F

o o -o :. ro g = o o o(torrfF(9oNo o'E rd)N+Fo N ll Ots('rF

o o E E (! o U) I= U' o oo$o(t@o(DN(9 o = F*rOl-F O (9(9O)rr) N FOONO@ONF@(')O C ro Fo)rFN@Oo o IOFFFION o) ??99499cc!\qa @ € coqqq\qqc? o o)ooooN ooooroooFNroo a s ONFoolN@O :f oooo60 OE @ () 6 a

NOO$OOFONO @ o FONObO O NONN N @FtOOOCtOO o ONNOONNONN+N noNo600(o @ONOr(t OFNNbNOt@€ OOOON(OOOr6*N o, o@@o@Noo olqqg.:ol qqq0qo?gqqo?a oodd;ddo-oidc; * d6i-dAioiaici oooo@o 6 OOOFFOOONO o) o o o

cf o E o (t)

o@(9NtNN(9N@ O cOFONF* Frrr)Ct (o NON6(t(r()NO FFFOT$6Al(o(oNO c) o(Dbo*(ooo o (oFNO@O €NNO,CO(OONNN @OOONNOC)F@OO qqcqq\.qRl ooooN(9 @N(9@OOOOOTS @ :f ctddci-dc,o;A,i rici $ O(r)FoN6lF O oooo@o o ood;;ddciRic; o @ o

E 5 '=

I 6 6 I E> E> E> E (\l - 6 6 6 @ u o.v ar0 J ;L iL o @ E5 .=v '; o.. oc'iX d Olcl. 9. Y:8s9e=f5a5F i 5'8i:4 i; 5,9- ori>>oli o>>ria'ii=

o= !l ii) or9 =E g8E-e FE':9 -lE a@ ag

xo 6[ * F]- 838-; 5t, d?ecaJ EI OOOES 8;' Ft) :tl

o Eq) o ss8-3 92 @FoF-L 6H @ .- o 6 dctdciH 9ci ao 9'; =

.v o o tr S = 6 ,9; '9 .2 s o o ,"9 .9 6.)i 6 o o RR5-B oo o .s s .v 6 o. 333Hs =66d O 5 o il: o E) c o o E 6 a

b qiF o oqrq-+E:8-3 I OOOO(o z*o t =o 9o ai

O,FNNCT,NN OFFOOOO@ (9 fis8-g (o oq9qe?qqqq? o FCTOONOO(O o) 333;d o o E c o

o U) (oc!F-(9(oo@ OFN(O(.)FN:f b38-; (o @oo(9@oob FOOONO(rTF- o, b o :33HR

E '\3 8 E =E (\lI l a ;Y U J o 6 d O)c., i 99F€ P F o>>t

TABLE3. Representativewhole-rock chemical analyses. All samplescontain calcite that is close to pure CaCO, pfu). Samplenumber (0.91-0.99Ca atoms 49-104 41-31 49-5 48-50 56-4C Whole-rock chemistry and the prograde sio, 23.40 34.80 36.70 47.14 48.20 47.80 amphibole-forming reaction Al,o3 2.21 2.45 7.49 14.92 8.35 't2.20 Tio, 0.13 0.17 0.33 0.68 0.37 0.64 Major element chemical analyseswere obtained for all FerO.'* 2.14 2.41 5.53 7.02 8.90 7.39 amphibole-zone specimens from the Waits River For- Mgo 3.78 2.81 4.50 4.09 3.14 2.16 mation. Analyses for the six representative samples in MnO 0.15 0.14 0.58 0.17 0.59 0.29 CaO 37.90 29.72 25.80 14.33 18.70 15.60 Table I are listed in Table 3. The principal observation Naro 0.21 0.21 o.23 0.78 0.42 2.23 from results in Table 3 is that, compared to pelitic rocks, KrO 0.49 0.18 o.57 2.16 0.10 0.15 the Al-rich hornblende is not from Al-rich rocks. All PrO. 0.05 0.07 0.31 0.12 0.25 0.15 LVr I 29.60 27.06 18.20 8.85 11.50 11.60 whole-rock specimens analyzed have <15 wto/oAlrO3. Total 100.06 100.02 100.24 100.26 100.52 100.21 Prograde amphibole-forming reactions were derived us- - Weight percent. Detection limit for maior elements: 0.01%; for P.O.: ing whole-rock chemical and mineral-chemical data in 10 ppm. Tables 2 and 3, following the method outlined by Ferry ** All Fe as FerO3. t Loss on ignition. (1989). The reactions typically are of the form ankerite + quartz * plagioclase* rutile + HrO : amphibole + calcite + chlorite + titanite + COr. potassium feldspar. Amphibole with high AlrO. contents Physical conditions during metamorphism usually occurs as porphyroblasts and coexistswith garnet Metacarbonaterocks from the Waits River Formation and other minerals with high Fe/(Fe + Mg) (i.e., chlorite in northeastern Vermont are interbedded with andalu- and biotite). In some specimens(e.g., 56-4C) amphibole site-bearing metapelites. The study area is close to, but is zoned with higher AlrO, contents at the core than at on the low-pressureside of, the aluminum-silicate triple the rim. In other specimens(e.g., 56-ll), however,am- point isobar as mapped by Thompson and Norton (1968). phibole showsno significant zonation in AlrO, (Table 2). Pressuretherefore was <3.8 kbar (Holdaway, l97l). To Amphibole from specimen56-ll, the sample having a first approximation, the pressureduring metamorphism the highest AlrO, content (19.06 wto/oat the core), was was considered -3500 + 500 bars throughout the study examined with the transmission electron microscope afea. (TEM) at the Johns Hopkins University by E. Smelik ro Peak metamorphic temperaturein the amphibole zone test whether the high Al content was possibly the result was estimated using garnet-biotite, garnet-ilmenite, and of submicroscopic Al-rich inclusions that might not be calcite-dolomite geothermometry. Results for 18 rocks detectedat the scaleof electron microprobe analysis.Re- from nine field localities (see Fig. l) containing appro- sults from TEM observations demonstrate that the am- priate mineral assemblagesare listed in Appendix Table phibole samplesanalyzed. are homogeneous, monominer- l. The preferred estimate of the averagetemperature for alic phasesand contain no submicroscopicinclusions of the amphibole zone, calculated with the three different any kind. geothermometers,is 525 + 30'C. Coexisting minerals. Plagioclaseoccurs in all samples and its composition varies greatly, both within and be- Pnrr,rvrrNlny THERMoDyNAMTC MoDEL FoR tween specimens.Potassium feldspar occursin four spec- Al sunsrrrurroN rN cArcrc AMpHTBoLE imens from the amphibole zone. The range in measured The unusually high Al content of hornblende from potassium feldspar composition is small, K/(K + Na) : metacarbonaterocks of the Waits River Formation and 0.94-0.99. its low crystallization pressuremotivated us to develop a Biotite occurs in 17 specimens with Fe/(Fe + Mg) model for the Al content of calcic amphibole. Becauseof varying between0.1 I and 0.56 and retAl: 0.30-0.52 at- limited thermodynamic data for amphibole components, oms pfu. Chlorite occurs in I I specimenswith Fe/(Fe + our model does not attempt to simulate observed am- Mg) : 0.I l-0.55 and Al : 2.52-2.81atoms pfu. Mus- phibole compositions in the chemically complex marls. ", covite occursin four specimenswith K/(K + Na) : 0.98- Rather, the intention of the model is to explore quanti- 0.99 and 3.17-3.25Si atomspfu. tatively, in a simple analoguesystem, the control of P, T, Garnet coexists with the most aluminous amphiboles and coexisting mineral assemblageon amphibole com- analyzed,(samples 49-1A,56-4C, and 56-l l). The garnets posrtlon. are normally zoned and show near-perfecthexagonal out- line in thin section with radiating inclusion trails. Schis- Model arnphibole tosity is not deflected around the porphytoblasts, indi- The composition of the amphibole samplesare repre- cating that garnet grew during static metamorphism. sented in Table 2 in terms of a single component and Clinozoisite was observedin five specimens;it is close mole fractions of various exchangecomponents, follow- to a Car(Al,Fe)3Si3Orr(OH)solid solution wirh Al/(Al + ing the method of Thompson (1982). Tremolite Fe) : 0.87-0.98. [CarMgrSirOrr(OHL] was chosen as the additive compo- LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE 1009

plagioclase, 7 nent, and the exchangecomponents edenite, -l 29x10 Tschermak are CaAlNa-,Si-,, NaAlSi', and Alr- \ 1,298 and \\ (H,",)= -12,594.67kJ ''' Mg ,Si ,, respectively. Mole fractions of exchangecom- \ - 1.298 7 s38.6J/K ponents were calculated for both the I3CNK and the 6:^ -1.30x10 \^ts?*)= F^ 'at Fe(+2) renormalization schemes(see footnote to Table I 'o'oao.. 2). l@ Considering the calculated mole fractions of exchange oF 7 or -131x10 o\... componentslisted in Table 2, Tschermak'scomponent is N l--l dr*an*qz by far the most important substitution in accounting for Ol l-- forsterite+sPinel the Al content of the analyzedamphiboles. As a first ap- 7 I -1 32x10 proximation, therefore, we assumedall Al in model amphibole is incorporated by Tschermak'ssubstitution, and 600 800 1000 1200 the model adopted for Al-bearing calcic amphibole is a Temperature(K) tremolite-tschermakitesolid solution [CarMgrSirOrr(OH)r- Ca2Mg3l6rAl2lolAlrsi6orr(oH)r.In a later section, we dis- Fig. 3. Experimental data on Al content in tremolite coex- cussthe contribution of the edenite substitution to the Al isting with diopside (di), anorthite (an), and qtartz (qz) (Jenkins, content of amphibole for selectedmineral assemblages. 1988, 1989, personalcommunication) and data for a calcareous hornfels from the Sierra Nevada (Ferry, 1989, sample 5J; plots Activity-composition relations at 693-733 K) were used to derive the standard state enthalpy Crystal structure refinements of amphiboles indicate and entropy for fictive end-member tschermakite (see text for that I6tAl is preferentially partitioned into the M2 am- details).Eo and So values were obtained from a linear regression phibole site (Hawthorne, 198l; Hawthorne and Grundy, of only the data for di + arl + qz. The four data for forsterite for di 1973; Makino and Tomita, 1989).We assumedall t61Al + spinel are tentative but appear consistent with results resides in M2 in the model amphibole. Assuming ideal +an+qz. mixing and local charge balance,the activities, a, of the tremolite (tr) and tschermakite (tk) components are thus among the four minerals at specified P and T, by related to composition @aI^ + 2@."t*,roo.o, + 2c'r;6r', - Gi;.[^ - :0. a',: (X-"-t)' (1) 2G3{irr",.or,r (4) and Each of the partial molar Gibbs energy terms may be expanded: a*: (Xor*r)' (2) where Xrr., and X^'r, are the atom fractions of Mg and G!:,, : 1f o r'zea- ZSs t.:os Al in the M2 site. * dr - r o, Thermodynamic data for tschermakite ['"rt J'"{rr,r", An estimate of the molar volume of tschermakite at 1 fr_ (267 based on unit-cell mea- bar and 298 K .4 cm3/mol), + | V? dP + RI ln a,.,. (5) surements of tremolite-tschermakite solutions by X-ray diffractometry, was provided by D. M. Jenkins (personal communication. 1989). Coefficients for tschermakite in The diopsides in Jenkins' experimentscontained a small Berman's(1988) expression for heatcapacity, ko-kr,were amount of Al, but becausethe actual compositions were calculated from the following algorithm: not available, we assumedunit CaMgSirO6activity. Unit SiO, activity was assumedfor quartz. Activity-composi- - (3) k,,*: k,,n I 2k1.(:u61r5io. 2k."o-u"rrou. tion relations assumed for natural diopside and plagio- Coefrcients for molar volume, V,-Vo, were calculatedin clase were those of Ferry (1989). Using the thermody- an analogous fashion. Enthalpy and entropy of tscher- namic data of Berman (1988) and Equations l-3, the makite at I bar and 298 K were estimated from (l) the only unknown terms in Equation5 are F,ou''t"ond S,o*t':oa. compositions of tremolite-tschermakite solutions equili- Considering Equation 4, eachmeasured amphibole com- brated experimentally with diopside, anorthite, and quartz position coexistingwith diopside, plagioclase,and quartz - (Jenkins,1988, 1989,personal communication), and (2) at known P and T definesa value of F1okr'2e8?"S9*':oa. the composition of amphibole coexisting with diopside, These values for all Jenkins' experimental data and for calcic plagioclase,and quartz in a calc-silicate hornfels the hornfels are plotted in Figure 3 against temperature. (sample5J of Ferry, 1989).The amphibole in the hornfels The horizontal dimension of each rhombus corresponds is unusually Mg rich [Mg/(Fe + MC) : 0.98]. It closely to uncertainty in temperature for each datum. The tem- approximates a tremolite-tschermakite solution, and is perature uncertainty for the hornfels sample (20 "C) rep- believed to have equilibrated with diopside, plagioclase, resentsthe variation in Zrecorded by individual samples and quartz at 2000 bars and 440 + 20 "C. At equilibrium about the averagetemperature estimate for outcrop 5 in l0l0 LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE

tions. Also plotted in Figure 3 are values of II,out':ss- 7"S,0*''tcs,derived from Jenkins' preliminary experimental (g data on the composition of tremolite-tschermakite solu- lt tions, equilibrated with spinel and forsterite which at .Y equilibrium defines the following relationship: o Girtr^+ 2@*L*oo.^- Gf;.!^- [email protected],,o0'o: 0. (6) J tt, The values of fl,out'zra- ZS,out,zsa pre- o derived from these o liminary resultsare in good agreementwith thosederived o- from the assemblageamphibole + diopside * anorthite + quartz. The above values are different from ones derived exclusively from experimental data (Jenkins, personal communication). We included the datum for the natural samplein our analysisbecause without it, derived H2*t.zstand .SPk''2e8values predicted undetectableamounts of Al in amphibole at T - 400-500 "C which is incon- 'r. (b) 4000c sistent with the measured composition of calcic amphi- (g \ 5oooc bole formed during low-temperaturemetamorphism (e.g., lt \r .Y \\ 6oooc Ferry, 1989). We stress,however, two points about our \"\ derived values: (l) They are, with one exception, consis- o \. '. tent with all Jenkins' experimental data (Fig. 3). The f \. .. rhombus at 918-928 K misses the calculated curve by o -300 (2) preliminary o \r only J. They are intended to be o ph+mu+ky+qz'\. t. results subject to modification by later experiment or o. \\ t. mixing models. buffer \ Phase equilibria 0 0.5 1.0 1.5 2.O General relations. For every stoichiometric reaction relationship that can be written among the components i per Altot calcicamphibole formula unit of minerals 7 and the tremolite and tschermakite components of coexisting model amphibole solid solution, there exists a correspondingrelationship among the par- Fig.4. Calculatedtotal Al content,A1,.,, of tremolite-tscher- tial molar Gibbs energies: makiteamphibole as a functionof P coexistingwith (a) clinozoisite(czo) + chlorite(ch) + anorthite(an) + quartz(qz), and (b) phlogopite(ph) + muscovite(mu) f kyanite(ky) + quartz 'C, ,,G!,,,+ v,uG!;!^+ y,,Gl;1^:o (7) at 400"C,500 and600 .rC (see text for details).Nore that the 4 Al contentofamphibole and its P and ?rdependenceare a function of P, Z, and mineralassemblaee. where z, are the stoichiometric coefficients (positive for products; negativefor reactants).At given P, Z, and composition for all minerals except amphibole, Equations l, Ferry (1989). The temperature uncertainty in the exper- 2, 5, and 7 uniquely define the composition of coexisting oC) imental data (5 was taken from Jenkins(1988). The tremolite-tschermakite solution. These equations and vertical dimension correspondsto uncertainty in amphi- Berman's (1988) thermodynamic data constitute our sim- bole composition. For sample 5J, the uncertainty repre- ple model for the Al content of calcic amphibole. sents the observed variation from microprobe analyses. Calculated results. Results for the assemblagesamphi- In Jenkins' experiments, the equilibnum amphibole bole + anorthite + quafiz * clinozoisite + chlorite and composition was bracketed by amphiboles that ap- amphibole + anorthite + quartz + phlogopite + mus- proachedit from initially more Al-rich and more Al-poor covite + kyanite are shown in Figure 4. The curves in compositions. For Jenkins' experimental data, the uncer- Figure 4a are basedon the reaction relationship tainty in amphibole composition correspondsto the dif- ferencebetween the two bracketing compositions at each 14 SiO, * 15 CarMg.Al4Si6Orr(OH), P and T. A straight line was fit to the data by linear + l2 CarAlrSi3Orr(OH) regressionof the values corresponding to the corner of :5 eachrhombus. Its slope representsour estimatefor S0kr,2e8 CarMgrSi8Orr(OH), : 538.6J/K.mol and its interceptour estimatefor F,ou'.rnt * 4 MgrAlrSi3O,o(OH)8 -12534.67 : kJ mol '. These values were derived as- + 44 CaAlrSi,O, (8) suming activity-composition relations (Eqs. I and 2) and therefore should only be applied using these same rela- while the curves in Figure 4b are basedon LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE 1011

2 KMgrAlSi,O,o(OH),+ l0 SiO, Discussion of calculated results. Calculated results in * 3 CarMg'AloSi6Orr(OH), Figure 5a are in qualitative agreementwith a number of ofAl-bearing cal- : 3 CarMg'Si8Orr(OH), observationson the natural occurrence cic amphibole coexisting with plagioclaseand quartz: (l) + 2KAl.Si3O'o(OH),+ 4Al,SiO5. (9) The least aluminous amphibole samplesare from diopside-bearingrocks (Fig. 2b). (2) Calcic amphibole coex- pfu) is In Figure 4, Al,o, (total Al atoms in amphibole isting with more aluminous mineral such as micas, chlo- 4(Xo,*r) in Equation 2. Calculated results lead to three rite, clinozoisite,and aluminum silicateis more aluminous conclusions: (l) Even at a given P and T in the simple (Table 2; Fig. 2a). (3) Calcic amphiboles mav be relatively model system,the Al content of calcic amphibole is crit- aluminous even when they coexist with calcite and do- ically dependent on the coexisting mineral assemblage. lomite (Table 2). (4) Calcic amphibole coexistswith ky- (2) There is no universal relationship, either quantitative anite (kyanite is metastablebelow P - 4.2 kbars) only at or qualitative, between pressure and the Al content of elevated pressure(fields 3-8 and 13, Fig. 5a); such am- calcic amphibole. With increasing pressure at constant phibole is the most aluminous possible coexisting with temperature,the Al content of amphibole coexistingwith plagioclaseand quartz (cf. Selverstoneet al., 1984). The anorthite, quafiz, clinozoisite, and chlorite increases,but good qualitative agreementbetween natural amphibole it decreasesin amphibole in equilibrium with anorthite, occurrencesand the mineral relationships predicted by quartz, phlogopite, muscovite, and kyanite. (3) The Al the simple model in Figure 5a further emphasizesthat given content of amphibole, even in equilibrium with a both the AI content of calcic amphibole and its pressure mineral buffer assemblage,may vary as a function of ?" dependenceare critically dependent on coexisting min- as well as P. erals. We systematicallyconsidered equilibrium among tremolite-tschermakite amphibole and anorthite, quartz, SHonrcovnNcs oF THE SIMPLE clinozoisite, chlorite, phlogopite, kyanite, diopside, mus- THERMODYNAMIC MODEL covite, potassium feldspar, calcite, and dolomite in the in calcic system K,O-CaO-MgO-AI,O.-SiO2-H,O-CO,. Because The thermodynamic model for Al substitution -0.4- 1.0atoms pfu calcic amphibole from the Waits River Formation co- amphibolepredicts an Al contentof quartz at the exists with plagioclaseand quartz, we restricted our at- in amphibole coexisting with anorthite + Waits River For- tention to assemblageswith anorthite + qvartz. Further, P-Zconditions of metamorphism ofthe Measured equilibria with COr-HrO-bearing fluid were not consid- mation, northern Vermont (3500 bars, 525 "C). aluminous (Ta- ered. Becauseof the interest in the Al content in amphi- amphibole compositions are much more positive correla- bole as a geobarometer,results are presentedin Figure 5a ble 2). Analyses in Table 2 show that a of calcic am- on an isothermal P-AI,", diagram. Unit activity was as- tion exists betweenthe Na and Al contents that, sumed for components, except amphibole, in all miner- phibole. In addition, Figure 2 demonstrates is correlated als. The arrangementof univariant curves in Figure 5a is empirically, high Al content in amphibole consistentwith the method of Schreinemakers.The shapes with high (Fe + Mn)/(Fe + Mg + Mn). Quantitativelv' of the sim- ofthe curves in Figure 5a are robust becausethey depend therefore,it appearsthat a major shortcoming Na and Fe- only on AV and the stoichiometry of the reaction rela- ple model is its failure to incorporate tionship. the shapesare independent of en- Qualitatively Edenite substitution and the Al content of amphibole tropies and enthalpies of reaction. Equilibria in some parts in Table of Figure 5a are metastablerelative to a CO'-HrO fluid. The mole fractions of exchangecomponents is significant Na in natural Al- At 525 "C the assemblageamphibole * anorthite + qtarlz 2 demonstrate that there is largely ac- is stableup to -9000 bars, and amphibole contains -0.2- rich amphibole and that Na in amphibole the 1.2Al atoms pfu dependingon P and coexistingminerals' commodated by the edenite substitution. Potentially Figure 5b shows the compatibility of minerals in the di- edenite substitution could explain why natural amphibole model for variant regions of Figure 5a on a diagram that projects contains more Al than predicted by our simple the edenite through CarMgrSi*Orr(OH)r,CarMgrAloSi6O22(OH)2, a Na-free system. We therefore investigated its contribution to Al for the CaAl2Si2Or,and SiO, (i.e., considering only assemblages content of amphibole and ", in Figure 4. Amphi- that coexist with tremolite-tschermakite amphibole, an- same two assemblagesrepresented and edenite solutions orthite, and quartz). The projection schemein Figure 5b bole with tremolite, tschermakite, rather than pure an- was derived from a transformation of components (see coexists with plagioclasesolutions of such amphi- Thompson, 1982).The alrows pointing outward from the orthite. To calculate the edenite content reaction rela- triangles emphasize that tie lines between kyanite and bole, we used the following stoichiometric edenitecomponents other minerals first passto infinity and then return to the tionship among albite, tremolite, and CaO apex ofthe diagram (cf. tie lines to potassium feld- and quartz: sparon the ThompsonAFM diagram,Thompson, 1957). NaAlSi,O, * CarMgrSi.O,,(OH), clinozoisite and other The tie lines between chlorite or : NaCarMg,Si,AlOrr(OH),+ 4 SiO, (l 0) minerals passdirectly from the CaO apex into the interior of the diagram. and these activity-composition relations: tot2 LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE

tk - tr amphibole + quartz (a) + anorthite T = 525'C

(E6 .ct .!a ftiaN" -.5 "-t fra o o. 3

fluid - lndependent equ il ibria

0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.O

Altot per calcic amphiboleformuta unit

Fig. 5. (a) Calculated total Al content, A1,",,of tremolite- z-i\ .r. /) ^ {o) tschermakite amphibole as a function of P at 525 "C coexisting v/\ @7r with selectedmineral assemblagesin "/\ the system KrO-CaO-MgO- n'/ \ .,/ \ AlrO3-SiOr-HrO-CO, (seetext for details). All assemblagescon- -/f--"-\ tain anorthite + qxartz. Kyanite is metastable belovt p - 4.2 ch do kY do oo ./ kbar. Mineral abbreviations as in Figure 4; ca : calcite; ds : , ,/r' *, dolomite; IC' : potassium feldspar; sill : sillimanite. Circled on 01 oA @n numbers identify divariant regionswith mineral compatibilities ^./ \ */ \ ^/t\ /t\ shown in b. Note that the Al content of amphibole and its p *//\_\ ..//_)__\ dependencecritically depend on the coexisting mineral assem- ".4s\ -4N_\ blage. (b) Compatibility of minerals in the divariant regions - ky ca ky ca of y'*' / /' y'rt a on a diagram that projects through CarMgrSirOrr(OH)r, ;, ; filn raA Ca,Mg.AlnSiuO,,(OH)r, CaAl,SirO,, and SiO, (i.e., considering only assemblagesthat contain tremolite-tschermakite amphi- /t\ bole, anorthite, and quartz). /I\ d -;do ";; a"a:(I - Xu.o)(X*".r), (11) a,"t n au: (Xr.o)(X-u-)2 ^/ \ (r2) il\ a,u: (Xu.o)(Xo,.r)2 (l 3) /\\ "/N\ca oo y'^, where Xu.o is the atom fraction of vacanciesin the amphibole A-site. We used the thermodynamic data baseof Xf KA|O 2 Holland and Powell (1990) for thesecalculations because -"iiir K'ehl e,a I nl Berman's(1988) data basedoes not include edenite.The _/ \ /t\ LryJ/ \ activity coemcientsfor CaAlrSirO, and NaAlSirO, in pla- 1/ \_ gioclase were taken /l\ lo,-" I l___-_JL!d as 2 and I, respectively (Carpenter coz ,F"^" fc-l and Ferry, 1984). For a specified T and aun,Xr^ may be determinedfrom Equations5, 7, and l0-12. Valuesfor LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE r0l3

Xu.o are then included in the tremolite and tschermakite 10 activity terms (Eqs. l2 and I 3); valuesfor Xrry2 ?rd Xervz (g are finally calculatedusing Equations 5, 7, 12, and 13. .ct A1,",for tremolite-tschermakite-edeniteamphibole is (l .Y - X".) + 4(l - X*"*,). aan = 0'1 o Calculated results (Fig. 6) show that the Al content of d"n =0.4 f tremolite-tschermakite-edeniteamphibole solutions can o ?"n = 1'o significantly differ from that of tremolite-tschermakite o o amphibole solutions dependingon the extent of the eden- czo+ch+pl+qzbuffer ite substitution as controlled by the composition o. of co- T = 500oC existing plagioclase(by Reaction l0). For equilibria that do not involve the anorthite component, such as phlogopite-muscovite-kyanite-quartz-plagioclase-amphibole (Reactions9 and l0), increasingedenite substitution (i.e., increasing a"b)results in an increasein the Al content of amphibole (Fig. 6b). In general,however, the increaseis small (<0.5 Al pfu) and inadequate to explain the very 10 high Al content of the amphibole analyzed.For equilibria c li (l 98 ph+mu+ky+ rl that involve the anorthite component, such as clinozois- EI ite-chlorite-plagioclase-quartz-amphibole(Reactions .:( 8 v6 qz+plbuffer \\ and l0), however, r\ increasingedenite substitution actually o 'i. resultsin a reduction in the Al content of amphibole (Fig. T = 500oC 54 \\ 6a). The reduction in Al", occurs becausethe effect ofan o =0'1 '.t. increase in edenite content caused by increasing a"o is o "an \ offset by a decreasein tschermakite content caused by E2 ?'^n = 0'4 decreasinga"n. Consideration of Na - substitution in am- o. a an= 1'o \\ phibole by the edenite substitution, therefore, fails to ex- 0 plain the discrepancybetween the Al content ofanalyzed amphiboles and that predicted by the model. 0 0.5 1.0 1.5 2.0 Calculated results in Figure 6a neverthelessprovide a simple thermodynamic model for the difference in Al Al. per calcicamphibole formula unit tol content of calcic amphibole between mafic greenschists and amphibolites. Mafic gr€enschistsand amphibolites Fig. 6. Calculatedtotal Al content,A1.,, of tremolite-tscher- typically contain the samemineral assemblage(clinozois- makite-edeniteamphibole coexisting with the samemineral as- ite + chlorite + plagioclase + quartz * calcic amphi- semblagesas in Figure4 as a functionof 4", for a^": 0.1, 0.4, (see bole). Greenschists,however, contain Na-rich plagioclase and 1.0 text for details). and Al-poor actinolite, while amphibolites contain calcic plagioclase and Al-rich hornblende (Laird, 1982). The qnartz + chlorite + calcite + ankerite. In both casesFe/ correlation betweenNa-rich plagioclaseand Al-poor am- (Fe + Mg) in chlorite was used to monitor Fe-Mg sub- phibole and between calcic plagioclaseand Al-rich am- stitution in amphibole. Table 4 contains the components phibole is the result ofthe dependenceofthe edenite and used to model the compositions of the minerals and the tschermakite contents of amphiboles with tremolite, linearly independent stoichiometric reaction relation- tschermakite,and edenite solutions, coexistingwith chlo- ships that can be written among components in each as- rite, clinozoisite, quartz, and plagioclase,on the plagio- semblage.The Gibbs method was used to derive expres- clasecomposition. sions for (dpoh^/dXF""h)at conditions of constantpressure, temperature, and composition of coexisting plagioclase Fe-Mg substitution and the Al content of amphibole and clinozoisite by solving a set ofdifferential equations We verified that Fe-Mg substitution has an effect on that includes a Gibbs-Duhem equation for each phase the Al content of calcic amphibole by deriving expres- and the differentials of the conditions of heterogeneous sions for the relationship betweenthe chemical potential equilibrium implied by the reaction relationships in Ta- of an aluminous component of calcic amphibole (po,"-) ble 4 (see Spear et al., 1982 for further details). Ideal and the Fe/(Fe + Mg) of coexistingminerals i (X..,) using mixing of the Fe and Mg componentsin the chlorite solid the Gibbs method (Spearet a1.,1982). The composition solution was assumed.Results are presentedin Table 4. of coexisting Fe-Mg minerals indirectly monitors Fe-Mg Using measuredmineral compositions (Table 2) and an substitution in amphibole. Two assemblagesrelevant to inferred temperature of metamorphism of 525 oC,values metamorphosed marls from the Waits River Formation of (lpo,^ /0Xr".n)r,r.xon.xuowere computed for three sam- were considered, amphibole + plagioclase * quartz + ples of impure carbonaterock from the Waits River For- chlorite + clinozoisite and amphibole * plagioclase * mation (Table 4). All calculated values are positive. Al- t0l4 LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE

TABLE4. Predictionof Al-Fe correlation in amphibole by the Gibbs method

Components used to model compositions of minerals Method for computing Xi Symbol Formula from ineral analyses

amphibole(amp) Feam Ca"FesSisoo(OHL X.*: Fe(Fe + Mg + I6|AD Mgam CarM95SioOr,(OH), X"n".: Mg/(Fe+ Mg + i6rAD Alam CarAl,oSisolOH), X"-: tot1171P"+ Mg + I6rAD chlorite (ch) Fech Fe5Al,Si301o(OH)o xtu": Fe/(Fe + Mg) Mgch Mg5AlrSisolo(OH)6 XM$h: Mg/(Fe+ Mg) dolomite (do) Fedo CaFe(COs), X.*: Fe(Fe + Mg) Mgdo CaMg(COJ, XMsd": Mg/(Fe + Mg) clinozoisite(czo) Feczo Ca2FeoSi301,(OH) Xttgo: Fe*(FeP* + AD Alczo Ca,AlsSisO,lOH) X^8": AY(FeF.+ AD plagioclase(pl) an CaAl,SirOB X..: Cal(Ca + Na) ab NaAISLOs X"": Na/(Ca+ Na) calcite (ca) ca Ca(COJ quartz (qa) qz sio, Reaction relationships 1.Exchange reactions 1 Feam + 1 Mgch : 1 Fech + 1 Mgam 5 Fedo + 1 Mgam : 5 Mgdo + 1 Feam 2. Net-transfer reactions 2 Mgch + 22 an:3 Alam + 2 Mgam + 6 Alczo + 7 qz 5 Mgch+ 15Mgdo+ 55an: 12Alam+ SMgam+ 30ca + 25qz System of equations in matrix form for the equilibrium

0000 0pr*l0Xr-n 0 X"no" 0 0 0 6pu*nl0Xr*n 0 0 X*"- X6"^ X^* 6pa"l0Xw 0 0-1 10 0p"sl0Xra, 0 51-1 0 dpn" 0 't2 l|X,*n -15 0 I 0p""^ldXr*n 0 0000 0p^" ldX,* -Gxx

o o 0l X..". X"n- X^* -1 | 1 0 | 0231 0 0 0l 1".,l where G- : (&GJaX?*)p,r: 5RI/X.*nX"*n Vafues ot apn" lilG"^ tor three diflerent amphibole-zone specimens trom the Waits River Formation, northeastern Velmont in ioules (lpo'*J04*o1,,,,,- (0ponJiX*)p,la'a amp-pl-qz- amp-pl-qz- Sample number X,* Xruo Xrn"- ca-ch-do ch - czo 57-13A 0.27 0.19 0.26 0.61 7334 56-4C 0.48 0.28 0.37 0.39 9185 768 56-11 0.55 0.41 0.29 4468 though it is not possible to simply relate pAr.- those in marls from the Waits River Formation, probably quantitatively to the Al content of calcic amphibole in are characterizedby Klat-uz << 0.3. atoms pfu, the Gibbs method indeed predicts that sub- The effect of Fe-Mg ordering among the amphibole M stitution of Fe for Mg in amphibole resultsin elevatedAl sites on the Al content of calcic amphibole can be seen contents.The results in Table 4 were obtained by consid- by considering the specific caseof coexisting amphibole eration of mineral equilibrium in individual samples. The ICa,(Fe,Mg).(Al,Fe,Mg),(Al,Si)8Orr(OH),], chlorite results demonstrate that the Fe-Al correlation in amphi- [(Fe,Mg),Al,Si3O,o(OH)8],ankerite [Ca(Fe,Mg)(CO,),], bole illustrated in Figure 2 should have been anticipated. calcite [Ca(CO.)], anorthite [CaAl,SirOr], and qaartz The positive correlation betweenthe Fe and Al content [SiO,]. The following reaction relationship can be written of calcic amphibole is probably, in part, the result of Fe- among the Fe-free mineral components: Mg ordering among the M I , M2, and M3 crystallographic sites in amphibole. Crystal structure refinements of am- 6 CarMgrAloSi6Orr(OH),+ 6 CaCO, + 5 SiO, phibole indicate that Fe is partitioned into the Ml and : 2 Ca,Mg,Si8O,,(OH), + 3 CaMg(COr), M3 sitespreferentially over the M2 site (e.g.,Hawthorne, + ll CaAl,Si,Os+ MgrAl,SirO,o(OH)r. (14) l98l; Makino and Tomita, 1989).For example,Makino and Tomita (1989) estimated that even in calcic amphibole that crystallized at high temperature in granulites At equilibrium (assumingideal mixing, local chargebal- and volcanic rocks, Kl{r-M2: (Mg/Fe)r,/(Mg/Fe)r, : 9.3. ance, and identical site occupancies of Ml and M3), Amphiboles that crystallized at lower temperatures,like Equation l4 implies an equilibrium constant, LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE l0l5

E4 tents in amphibole becauseif Fe and Mg mix randomly 3 T = 525oC over the Ml, M2, and M3 sites(K- : 1), Fe-Mg substi- s tution has a negligible effect on A1 The simple analysis 5 P = 5000 bars .,. E shows that substitution of Fe for Mg in calcic amphibole b3 ca+do+an+qz+chbuffer only results in a dramatic increasein its Al content when o K (FelMg)M2 Fe shows a strong preferencefor the Ml and M3 sites 6 M= over the M2 sites (i.e., K- < 0.3). The effect may be : (Fe/Mg)M = r-Ms greater if Fe-Mg mixing is nonideal. cL2 E Consideration of edenite substitution (Fig. 6), Fe-Mg o substitution, and Fe-Mg ordering in amphibole (Fig. 7) I 9 fails to reproduce the Al contents of natural calcic am- 3t phibole (Table 2). These results demonstrate that before o the Al content of amphibole can generally be used as a CL geothermobarometer,much remains to be learned about o the thermodynamic properties of amphibole solid solu- a-0 tions. Our study suggeststhat, as a beginning, the correlation Fe and Al contents in calcic amphibole 0.0 o.2 0.4 0.6 0.8 between must be calibrated either empirically or experimentally in the laboratory. Experiments should include not only Fe / (Fe+Mg)in amphibole studies of CarFerSirOrr(OH)r-CarFerAloSiuOr'(OH),solid Fig.7. Calculateddependence of the Al content of tremolite- solutions but also of the partitioning of Fe and Mg among tschermakite amphibole, A1.,, coexisting v/ith calcite + dolo- amphibole M sites. mite + anorthite + chlorite + quartz aI 525 "C,5000 bars on + Fe/Fe Mg in amphibole. See text for details. Note that the CoNcr.usroNs dependenceof A1,",on FelFe + Mg dependscritically on K-, a measureof the partitioning of Fe and Mg betweenthe Ml, M2, Microprobe analyses reveal that calcic amphibole, and M3 sites in amphibole. crystallized at the same metamorphic grade in carbonate rocks from northern Vermont, has a great range in Al content, 0.38-3.30 atoms pfu. The metacarbonatesare (XR rr)(Xitr"rXXtrr.-*)(Xirr.n\ K'\q - r."r = (15) interbeddedwith andalusite-bearingschists. These results (XL'r*,-"r)(Xl?"r) lead to three conclusions: (l) High-Al amphibole is not where the X, terms refer to the atom fraction of i in the restricted to high-pressuremetamorphic environments, appropriate site. Using the thermodynamic data for becausethe amphiboles from Vermont are some of the tschermakite in this study and for the other Fe-free min- most aluminous ever reported and yet crystallized at P eral components in Berman (1988) at 525 "C,5 kbar, < 3800 bars. (2) High-Al amphiboles is not restricted to highly rocks, the amphiboles from (K;"o aluminous because Ko: Vermont crystallized in carbonate rocks with < 15 wtolo (X A1rO3.(3) The Al content of calcic amphibole is not just : y*.",)(X?Jn\,^")!X ?lL."n) : 62.387 . (16) (Xi'r..'-'')(Xi'..r) a function of pressure,but also of temperature, mineral assemblage,composition of coexisting minerals [princi- Partitioning of Fe and Mg between the Ml -M3 and M2 pally Fel(Fe + Mg) and plagioclase compositionl, and sites is quantitatively describedby partitioning of Fe and Mg among the Ml, M2, and M3 amphibole crystallographicsites. (r7) AcxNowr.nIGMENTS

The bulk Fe-Mg partition coefficients for chlorite and Acknowledgrnentis madeto the National ScienceFoundation (grant of the PetroleumResearch amphibole and for ankerite and amphibole were esti- EAR.-8903493to J.M.F.)and to the donors maiedfrom the composition orcoeiisting -rn"'ur' in #:#:.HT::,#l|jf1trffff,TilTjffiLT*?ffTl'."* sample 56'4C: by NSF grantfan,eoooAO+. Field work by A.L. wasmade possible in K3-.r,- (Fe/Mg)"*/(Fe/Mg)"n : r.06 (r8) iTl'.ilT#1,*'#I:lT"Hffi?"#fftll,l#i?T;llffil Km{nk: (Fe/Mg).-/(Fe/Ms)*u : 2.42. (re) [Tl#rYiliJJlr[:Jf,1"",:::nl:ffl':,jil,J;;:::*::;::il: studycould not havebeen possible without David M. Jenkinswho kindly For a specified value of K- and Fe/(Fe + Mg) in amphi- ,uppti"aus with his experimentaldata prior to publication.Jenkins'ex- bole, the Al content of calcic amphibole can be computed perimentalprogram was supportedby NSF granr EAR-8E03437' We us to manv of his from Equations l, 2, and 16-19. Results are illustrated warmlv thank J.B. Thornpson,Jr., for introducing inFigurl 7 ror represenrative resionar rn"tu-o'pr'l" r-i :il|ililXlfiff:iflnjffi"j;: ffi:ffiT:|'jf".3Jffff*'fl1"{ conditions and for four values of K-. It is not simply the u.Jvrir, p.oi."ted phasediagrams, and rhe cibbs method.The manu- substitution of Fe for Mg that results in elevated Al con- scriptwas improved by helpfulreviews from M. Choand C.R. Nabelek. t0l6 LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE

RnrnnnNcps crrED Johnson, M.C., and Rutherford, M.J. (1989a) Experimental calibration of the aluminum-in-hornblende geobarometerwith application to Long Valley caldera (California) volcanic rocks. Geology, 17,837-841. Anovitz, L.M., and Essene,E.J. (1987) Phase equilibria in the system - (l 989b) Experimentallydetermined conditions in the Fish Canyon CaCO.-MgCO,-FeCO,.Journal of Petrology, 28, 389-414. Tufl Colorado, magma chamber. Joumal of Petrology,30, 7l 1-738. Berman, R.G. (1988) Internally-consistentthermodynamic data for min- Konig, R.H., and Dennis, J.G. (1964) The geologyof the Hardwick area, eralsin the systemNa,O-KrO-CaO-MgO-FeO-Fe,O,-AlrOr-SiOr-TiOr- Vermont. Vermont Geological Survey Bulletin, 24,57 p. HrO-CO?.Joumal of Petrology, 29, 445-522. Kostyuk, E.A., and Sobolev, V.S. (1969) Paragenetictlpes ofcalciferous Cady, W.M. (1960) Stratigaphic and geotectonicrelationships in north- amphiboles of metamorphic rocks. Lithos, 2, 67-81. ern Vermont and southernQuebec. Geological Society ofAmerica Bul- Iaird, J. (1982) 4rp616oles in metarnorphosedbasaltic rocks: Green- letin. 71. 531-576. schist facies to amphibolite facies. In Mineralogical Society of America Cao, R., Ross, C., and Emst, W.G. (1986) Experimental studiesto l0 kb Rewiewsin Mineralogy, 98, 1l3- I 37. of the bulk composition tremoliteso- tschermakitero+ exc€ssHrO. Leake, B.E. (1965a)The relationship betweentetrahedral aluminum and Contributions to Mineralogy and Petrology, 93, 160-167. the maximum possibleoctahedral aluminum in natural caciferousand Carpenter, M.A., and Ferry, J.M. (1984) Constraints on the thermody- subcaliferousamphiboles. American Mineralogist, 50, 843-85 l. namic mixing propertiesof plagioclasefeldspars. Contributions to Min- -(1965b) The relationship betweencomposition of calciferousam- eralogyand Petrology,87, 138-148. phibole and gradeof metamorphism. In W.S. Pitcher and G.W. Flinn, Dennis, J.G. (1956) The geologyofthe Lyndonville area, Vermont. Ver- Eds., Controls of metamorphism, p. 299-318. Wiley, New York. mont Geological Survey Bulletin, 8, 98 p. -(1971) On aluminous and edenitic hornblendes. Mineralogical Doll, C.G. (1951) Geology of the Memphremagog quadrangle and the Magazine, 38, 389-407. southeastern portion of the Irasburg quandrangle, Vermont. Vermont -(1978) Nomenclature of amphiboles. American Mineralogist, 63, Geological Survey Bulletin, 3, I 13 p. 1023-1052. Doll, C.G., Cady,W.M., Thompson,J.8., Jr., and Billings,M.p. (1961) Makino, K., and Tomita, K. (1989) Cation distribution in the octahedral Centennial geologic map of Vermont. Vermont Geological Survey, sites of hornblendes.Arnerican Mineralogist, 74, 1097-l 105. Burlington, Vermont. Pluysnina,L.P. (l 982) Geothermometryand geobarometryofplagioclase- Doolan, 8.L., Z.en,E., and Bence, A.E. (1978) Highly aluminous hom- hornblende bearingassemblages. Contributions to Mineralogy and Pe- blendes:Compositions and occurrencesfrom southwesternMassachu- trology,80, 140-146. setts.American Mineralogist, 63, 1088-1099. Pownceby,M.I., Wall, V.J., and O'Neill, H.SI.C. (1987) Fe-Mn partition- Eric, J.H., and Dennis, J.G. (t958) Geology of the Concord-Waterford ing between garnet and ilmenite: Experimental calibration and appli- area, Vermont. Vermont GeologicalSurvey Bulletin, I l, 66 p. cations. Contributions to Mineralogy and Petrology,97 , 116-126. Ferry, J.l,t. (1989) Contact metamorphism of roof pendantsat Hope Val- Raase,P. (1974) Al and Ti contentsofhornblende, indicators ofpressure ley, Alpine County, California, USA: A record of the hydrothermal and temperatureof regional metamorphism. Contributions to Miner- system of the Sierra Nevada batholith. Contributions to Mineralogy alogy and Petrology, 45,231-236. and Petrology, 101, 402-417. Robinson, P., and Jaffe, H.W. (1969) Chemographicexploration of am- Ferry, J.M., and Spear,F.S. (1978) Experimental calibration of the par- phibole assemblagesfrom central Massachusettsand southwestem New titioning of Fe and Mg between biotite and garnet. Contributions to Hampshire. Mineralogical Society of America Special Paper 2, 251- Mineralogy and Petrology,66, ll3-117. 274. Hall, L.M. (1959) The geologyofthe St. Johnsburyquadrangle, Vermont Sawaki,T. (1989) Sadanagaiteand subsilicic ferroan pargasitefrom ther- and New Hampshire. Vermont Geological Survey Bulletin, 13, 105 p. mally metarnorphosed rocks in the Nogo-Hakusan area, central Japan. Hammarstrom, J.M., and Z,en,E. (1986) Aluminum in hornblende: An Mineralogical Mag;azine,53, 99- 106. empirical igneousgeobarometer. American Mineralogist, 7 l, I29j-1j13. Selverstone,J., Spear, F.S., Franz, G., and Morteani, G. (1984) High- Hawthorne, F.C. (1981) Crystal chemistry of the amphiboles. In Miner- pressuremetamorphism in the SW Tauem Window, Austria: P-T paths alogical Society of America Reviews in Mineralogy, 9A,, l-102. from hornblende-kyanite-stauroliteschists. Journal of Petrology, 25, Hawthorne, F.C., and Grundy, H.D. (1973) The crystal chemistry of the 50l-53 l. amphiboles.I. Refinementof the crystal structureof ferrotschermakite. Spear, F.S. (1981) An experimental study ofhornblende stability and Mineralogical Magazine, 39, 36-48. compositional variability in arnphibolite. American Journal of Science, Hodges, K.V., and Spear, F.S. (1982) Geothermometry, geobarometry 28t,697-734. and the AlrSiO5 triple point at Mt. Moosilauke, New Hampshire. Spear,F.S., Ferry, J.M., and Rumble, D., m. (1982) Anal).tical formu- AmericanMineralogist, 67, I I l8-1134. lation of phaseequilibria: The Gibbs'method. In Mineralogical Society Holdaway, M.J. (1971) Stability of andalusiteand the alurninum silicate of America Reviews in Mineralogy, 10, 105-206. phasediagram. American Journal of Science,27 | , 97- l3l . Thompson, J.B., Jr. (1957) The graphicalanalysis ofmineral assemblages Holland, T.J.B., and Powell, R. (1990) An enlargedand updated inter- in pelitic schists.American Mineralogist, 42,842-858. nally consistent thermodynamic dataset with uncertainties and corre- -(1982) Compositional space: An algebraic and geometric ap- lations:The systemKrO-NarO-CaO-MgO-MnO-FeO-Fe,Or-AlrO,-TiO,- proach. In Mineralogical Society of America Reviews in Mineralogy, SiOr-C-Hr-O,. Journal of Metamorphic Geology, 8, E9-124. 10,l-51. Hollister, L.S., Grissom, G.C., Peters, E.K., Stowell, H.H., and Sisson, Thompson, J.B., Jr., and Nonon, S.A. (1968) Paleozoicregional meta- V.B. (1987) Confirmation of the empirical correlation of Al in hom- morphism in New England and adjacent areas.In E-anZcn, W.S. White, blende with pressureof solidification of calk-alkaline plutons. Ameri- J.B. Hardley, and J.B. Thompson, Jr., Eds., Studies of Appalachian can Mineralogist, 72, 231-239. geology-Northern and maritime, p. 203-218. IntersciencePublishers, Jenkins, D.M. (1988) Experimenral study of the join tremolite-tscher- New York. makite: A reinvestigation. Contributions to Mineralogy and Petrology, Woodland, B.G. (1965) The geologyofthe Burke quadrangle,Vermont. 99,392-400. Vermont GeologicalSurvey Bulletin, 28, 15l p. -(1989) Experimental reversal of the Al content of tremolitic amphiboles coexisting with diopside, anorthite, and quartz. Geological Ma.NUSCnrr'rREcEIvED JeNuanv 5, 1990 Society of America Abstract with Programs,2l , Al 57. MeNuscnrpr AcCEFTEDMencH l, l99l LEGER AND FERRY: AI-RICH CALCIC AMPHIBOLE 1017

AppeNotxTlele 1. Biotite-garnet, calcite-dolomite, and ilmenite-garnetgeothermometry for 18 samples from the Waits River Formation, northern Vermont

Mine?al composition data Calculated temperatules Sample Xal,ga Xpy,ga Xsp, ga Xgr, ga XFe,bi XMg, bi XFe,il XMn, il XFe,ca XMg, ca G-B- G-B'. Glt C-D+ 48-1 0.758 0.082 0.113 o.o47 0.519 0.481 451 469 48-1A 0.712 0.121 0.031 0.137 0.459 0.541 0.994 0.006 508 s60 551 48-18 0.764 0.115 0.023 0.099 0.490 0.510 508 546 48-2A 0.680 0.102 0.082 0.135 0.502 0.498 0.968 0.032 522 573 48-34 0.671 0.082 0.044 0.204 0.477 0.523 0.992 0.008 440 513 517 49-14 0.580 0.080 0.100 0.240 0.500 0.500 495 583 49-18 0.534 0.061 0.139 0.266 0.517 0.483 459 558 56-4A 0.541 0.095 0.126 0.238 0.408 0.592 0.972 0.028 0.031 0.042 460 546 521 525 56-48 0.560 0.090 0.110 0.240 0.430 0.570 0.978 0.022 0.034 0.o42 461 551 500 528 56-4C 0.570 0.090 0.110 0.240 0.518 o.482 0.034 0.037 560 647 509 56-8A 0.609 0.081 0.102 0.208 0.470 0.530 0.974 0.026 454 529 598 56-11 0.620 0.070 0.090 0.220 0.555 0.445 0.984 0.016 501 582 492 57-40 0.530 0.068 0.086 0.316 0.984 0.016 466 57-4E 0.473 0.039 0.034 0.454 0 988 0.012 0.034 0.041 524 57-4F 0.010 0.040 496 57-4H 0.017 0.041 508 57-41 0.005 0.050 531 57-13A 0.020 0.045 azo Average ('C) 478$ s46$ 521 518 Std. ctev.('C) 29$ 33$ 43 12 /Vote.'TheX,n" : mole fraction component iin garnet (al : almandine;py : pyrope; sp : spessartine;gr: grossular).X,,o,: i(Fe + Mg) in biotite; X,,t: i(Fe + Mg) in ilmenite;X* : l(Ca + Fe + Mg + Mn) in calcite. - Garnet-biotitethermometry, calibrationof Ferry and Spear (1978). -- Garnet-biotitethermometry, calibrationof Hodges and Spear (1982). f Garnet-ilmenitethermometry, calibrationof Powncebyet al. (1987). f Calcite-dolomitethermometry, calibration of Anovitz and Essene(1987, Equation26). $ Average temperature and standard deviation (1 o) values exclude results for sample 56-4C.