American Mineralogist, Volume 60, pages 1069-1085, 1975 Amphibolesin Andesiteand Basalt: I. Stability as a Function of P-T-fo, J. C. Alt-nN, Departmentof Geologyand Geography,Bucknell Uniuersity Lewisburg,Pennsyluania 17837 "i,Tl De p a r t m e n t oyc o,,i):,,!, ) ;, uq n ia s ta t e uniu e rs i ty Uniuersitv" Park. Pennsvluania16802 nNn G. M.c,nr-A.No Department of Geography and Geology, Indiana State Uniuersity Terre Haute. Indiana 47809 Abstract The stabilitiesof amphiboleshave been determined in an andesite,three basalts, and an olivinenephelinite in thepresence of HrO vaporat valuesof oxygenfugacity approximating thoseof FerOo-FerOs,Ni-NiO, and FerOr-FeO.The thermalstability decreases with increas- ing activityof silicain the parentrocks. Maximum thermal stability of amphiboleoccurs at 1090"Cat l3 kbar in the Hualalaialkali basalt.Maximum pressure stability occurs at 31.5 kbar at a temperatureof 1030'Cin the olivinenephelinite. Amphibole-bearing assemblages convertto garnet-bearingassemblages at pressuresabove l8 to 30 kbar. The amphiboles straddle the calciferous-subcalciferousboundary, and all of them fall in the pargasite- tschermakite-tschermakitichornblende category. No orthopyroxenewas found in any of the basalticcompositions under any of the conditionsinvestigated, although it did occurin the andesite.This restrictsthe potentialP-T-X conditionsunder which fractionalcrystallization of orthopyroxeneis an important mechanismin the derivationof SiO2-undersaturated magmas. Our resultsare consistentwith thosehypotheses in which someandesitic magmas derive from basalticcompositions through amphibole-liquid equilibria. Fractionation of alkalisby amphibolesmay contribute to theproposed gradient in KrO acrosssubduction zones at island arcs and continentalmargins. The shallowdips and depthsof subductingslabs beneath orogenic-zonevolcanoes is in concertwith this model. Introduction igneousrocks may increasecontinentward from the Knowledge of the P-7 conditions under which oceans, and the fractionation of alkali-bearing hydrous minerals, such as amphiboles,exist is fun- minerals, such as amphibole, at depths may play a damentalto our understandingof melting and related role in this postulated trend. Fractionation of such processesin the crust-mantle system. As is well amphiboles from a basaltic magma may give rise to known, water dissolvesin silicateliquids under pres- andesiticmagmas as proposedby Bowen (1928) and sure and lowers the temperatureof the beginningof others. melting. lf all of the water is bound in amphibolesor To solve these problems, we experimentally es- other hydrous phases,the chemicalpotential of water tablished the stability of amphibolesin an andesite, would be less than if the water were present as a three basalts, and an olivine nephelinite in the nearly pure aqueous fluid, and temperaturesof the presenceof HrO vapor at values of oxygen fugacity beginning of melting would be higher (Burnham, ffi,) approximating those of FerOa-Fe2Os,Ni-NiO, t967\. and FerO.-FeO, from l0 to 36 kbar, determining the In addition, the KrO contents and KrO/NarO of pressuresat which amphibole-bearingassemblages 1070 ALLEN, BOETTCHER, AND MARLAND transform to garnet-bearing assemblages,i.e., al Tnsle la. Chemical Compositions of Rocks depths(<75-100 km) in the Earth. Used in the Experiments Andesite l)tattz Ollvine AIkaIl Nephe-Linite Experimental Methods tholeiite tholeiite basalt si02 59.10 50.71 49.71 46.01 38.57 Starting Materials Ti02 o.94 r.70 2.5I 2.I0 2.79 Starting mixes were prepared from a Mt. Hood A1^0^ 17.8 L4.48 L2.14 13.89 11.71 andesite,a quartz-normativePicture Gorge tholeiite, cr" 0^ 0. 065 0. 06 a l92l Kilauea olivine tholeiite, a nepheline- Ft2o3 1.78 4,89 3.23 4,29 5,2r FeO 4.83 9.07 8.40 8.96 7.18 normative prehistoric Hualalai alkali basalt, and a Mn0 0.10 0.22 0.r7 0.r9 0.11 Honolulu olivine nephelinite(Table I ). These rocks .ntg0 3.05 4.68 10.31 10.01 13.08 were chosenbecause they encompassa wide range of Ca0 6.85 8.83 10.73 10.36 r2,A4 SiO, and alkali concentrations, and experimental Sr0 0.05 work had previously beenaccomplished on some of Ba0 0. 03 0, 08 theserocks at other conditions (Eggler, 1970,1973; Li,20 < .01 Hill and Boettcher,1970; Holloway, 1970,1971; Hol- Na20 4,27 3.16 r.97 2.59 4.22 loway and Burnham, 1969, 1972;Yoder and Tilley, *zo 1. 08 0.77 0.49 0,75 r.20 Rb20 0.00186 1962;Fudali, 1965;Tuthill, 1968).All sampleswere Hzo* 0.1 I 0.07 0.17 0.59 crushed to -200 mesh under acetone, dried in an lr.o+ 0.00 0.00 0.19 oven at I l0'C, "20- I and stored in sealedvials over KOH P^0- o.22 0.36 o,21 0.33 1.11 in a dessicator. 0.05 0.27 "oz Capsules F 0.04 100.12 99.91 100.00 99.89 99.81 Because the T-fs, in our furnace assemblies is 0=F -0. 02 buffered at values approximating those of the Ni- 99.87 NiO (N-NO) assemblage(Hill and Boettcher, un- Msl (Ms+Fe2+) 0.53 0,48 0.69 0.66 0.75 publisheddata), runs made under N-NO conditions Ms/ (Ms+tFe) O.46 0.38 0.62 0.58 0.65 could be made with larger capsules,for an external r,62 0.54 solid buffer was not required. Thus, for runs under theseconditions, approximately 35 mg of sampleplus 19-26 wt percent HrO were sealedin welded AguoPduo (1975) report 30 and 40 wt percent iron loss from tubing of 2.2 mm I.D. For runs made under andesite-HrOmixtures at 1100'C and 30 kbar in cap- FesO.-FeO (M-W) or FerO.-FerOs (M-H) condi- sulesof AgruPdzuand AgoPdTo,respectively, but find tio-ns, the double-capsule technique (Boettcher, that iron lossis not significantat 1000"Cin runs of l0 Mysen, and Allen, 1973)originated by Eugster(1957) hours. However, our data for runs with the andesite and used previously in piston-cylinderapparatus by in 2.2-mm l.D. AguoPds.capsules under N-NO con- Ganguly and Newton (1968), Lindsley and Munoz ditions at temperaturesbelow 1000'C suggestiron (1969), and Gilbert (1969) was employed. With this lossas high as 40 percent.Run number 50 (Table 2) technique,only about 1.5mg of sampleplus 20-25 wt contained approximately 80 percent glasswith l.0l percent HrO were sealedin AgroPduotubing of 1.5 percenttotal Fe (Table 7),20 percentamphibole with mm I.D. Thesesmaller capsulesplus either the com- 10.57 percent total Fe (Table 6), ( 1.0 percent ponents for the M-W or the M-H buffer were in turn orthopyroxene and a trace of an opaque mineral. sealedin 3-mm O.D. Pt capsules.Possible water loss Thesefigures account for at least58 percentofthe 5.0 during preparation was determined by weighing the percent total Fe in the starting material (Table l). capsulesbefore and after welding. Potential leaks in Our runs with the andesiteunder M-W and M-H these capsuleswere detected by heating the capsules conditions were made with smallercapsules (1.5-mm at I l0'C for 12-24 hours and reweighingthenr. Op- LD.) and afford a bettercomparison of our data with tical and X-ray diffraction techniques were used to that of Stern and Wyllie. For example,run number determine whether the M-W or M-H assemblages 68 (Table 2) contains at least 85 percent glasswith lasted the time of the experiments. 1.32 percent total Fe (Table 7), at least 5 percent The sampleswere sealedin AguoPduorather than Pt amphibole with 6.4 percent total Fe (Table 6), and in order to preventloss of iron from the sampleto the ( l0 percent opaque mineral. These figures indicate capsule(Merrill and Wyllie, 1973).Stern and Wyllie no absorption of iron by the capsule.This is ex- AMPHIBOLES IN ANDESITE AND BASALT l07l TneLr lb. Normative Compositions(C.l.P.W.) of Tngr-E 2. Run Data Rocks Used in the Experiments @wtzDura- Phase NepheliniEe kbars 'c /l Mterier It20 tion e6sdblage nr 9. (OP) 10.14 3 .82 LO 920 140 andeslte N-NO 25.0 7.0 L,V,Am, qM' ' ?opx 757 andesite N-NO 25.0 7.O r,V, An, qM or Ehoclase 6.38 2.9D 4.43 IO 940 l0 960 r9Z andeslte N-No 20.2 6.5 1,?V,?qM albite 36.13 26-14 L6 67 21.54 900 50 andesite N-No 20.0 12.0 L, ?V,h,opx,OP 920 sz andeslte N-No 20.0 12.5 L,v,An, Opx,0p,qM 26 22 23 05 24,4t 24.06 947 940 54 atrd es iEe N-No 19.9 12.5 L, ?v,An,opx,qM 0 .20 19.34 13 960 56 andeslte N-No 20.0 13.3 L,?V qM,Ru 18 920 102 andesi!e N-No 20.0 1r.3 L, ?V,An,Opx,Ga'OP r leuc ite 5.56 18 940 110 andeslte N-NO 22.6 12.5 L,V,An,Ga,Op,qM 15.18 21.67 20.28 26.1-6 18 960 104 andesite N-No 25.0 12.0 L, V , q!,1 I9 850 130 andesite N-No 25.0 22 0 1-,V,An,Ga,opx 'Ru (2.65) (7.69) (11.28) (r0.52) (13.80) (Op),qM L9 920 \26 andesite N-No 25. 0 23.0 L,v,Ga, 22.3 L, ?V,Ga Op,qM (1.43 (3. (7.1r) (7.00) (r0 56) 20 920 l!4 andesite N-No 24.9 , ) ee) (Op),qM 20 960 425 andesite N-No 20.7 26.5 L,v,Ge, (1.8r) (r.r3) (3.s0) (2.68) (2.75) 20 975 230 and es i!e N-No 33.3 3.5 L,v,Ga,(Op),qM 20 980 423 and es 1te N-No 20,6 22.5 I,v, (0p) 'qM 11.07 (0P),qM 23 1000 429 andeslte N-No 21.0 22.5 L,V,Ga' (op),qM (6.17 ) Q .61) (r7.80) 23 to20 424 andesice N-NO 22.4 22.5 L,v, (oP) 10 940 202 andesl! e M-W 2t.7 4.5 L,V,An, fs (4 90) (6.7 2) (6.19) 10 960 209 andesite M-W 20.2 4.5 L,V,Ao,Op,qM olivine 0.16 18.00 22.6a l0 970 233 and es l!e M-W 24.8 4.5 L,v,qM L3 900 94 andesite M-W 20.0 4.5 L,?V,AD, opx,OP,SP,Ru to (0.12) (r2.56) (1s.43) (0p) I3 940 88 andeslte M-W 20.0 4.5 L,?V,Am, qM fa (o.04) (s.44) (2.91) 13 950 93 andeslte M-W 20,O 4.5 L, V, M-H 25.0 3,0 L,V,An,Op,qM (4.34) 10 940 227 andesi!e 10 960 222 andesite M-H 25 0 3.0 t ,v,op 20.0 5.2 L,?V,Am,P1'OP magnetite 2.58 7.09 468 6.22 7.55 880 12 and es lte M-H 13 920 68 andesite M-lt 20.0 3,2 r, ?v,h,0P iLnenite 1.7 9 3.23 4 -71 3 .99 5.30 13 940 70 and esite M-H 20,0 4.o L,V,An,Op,qM L,V,0p,qM -77 2.51 I3 960 223 andesite M-rl 25.0 4.3 10 920 r79 qz thol N-No 21.0 15.5 L,V,h,Cpx,0P'qM qM 10 940 r94 qz thof N-No 20.0 5.0 L,V,An, Cpx' OP' itlt.