American Mineralogist, Volume 76, pages 470-476, l99l

The effects of fluorine on the vapor-absentmelting of phlogopite * quartz: Implications for deep-crustalprocesses J. W. PrrpnsoN* Department of the Geophysical Sciences,University of Chicago,Chicago, Illinois 60637, U.S.A. T. CH.lcro Department of Geology, University of Alberta, Edmonton, Alberta T6G 2E3, Canada S. M. KuoHNBn Department of Geological Sciences,University of Washington, Seattle,Washington 98195, U.S.A.

Ansrru.cr The vapor-absentpartial melting of biotite-bearing quartzofeldspathicassemblages may play an important role in the petrogenesisof high-temperaturegranitoid rocks. F-bearing biotite has been reported in high-grade terranes; therefore, melting equilibria involving fluorhydroxy micas may be better models for lower-crustal anatexis than those involving the hydroxy end-member. In order to make an initial assessmentof the effect of F on vapor-absent melting, the reaction fluorhydroxy phlogopite + qu"rtz: enstatite + liquid has been investigated experimentally near 8 and 15 kbar, using a synthetic phlogopite of F/(F + OH) : 0.58 (t0.12). Near l5 kbar, melting occursbetween 1020 and 1050"C; at 8 kbar, melting occursbetween 950 and 1000'C. The equilibrium was reversedat l5 kbar by a two-stageexperiment between I100 and 1000 "C, and it was reversedat 8 kbar by a two-stage experiment between 1050 and 950 "C. Reversal criteria included textural evi- dencefor mica regrowth from melt. Results of this study indicate that -60 molo/osubsti- tution of F for OH in phologopite stabilizes the phlogopile + qtrarlz assemblageby as much as 175 "C relative to enstatite and liquid. This significant stabilizing effect indicates that vapor-absentpartial melting of rocks containing high fluorine micas may be restricted to temperaturesin excessof 950 oC,leaving a residual unmelted micaceousassemblage of higher F content. The stabilizing effectofF in this systemis broadly consistentwith current models for the petrogenesisof A-type granites. Quantitative analysesof quenched liquid generatedfrom the vapor-absentfluorhydroxy phlogopite + quartz assemblage,however, indicate melts are lamproitic in composition, not granitic.

Ir.mnonucrroN (Guidotti, 1984),which indicates that the KMASH-F sys- Initial melting of most quartzofeldspathiccrustal rocks tem may be a closer analogueto rocks. It has been dem- occurs at the solidi of constituent mica-bearing assem- onstrated experimentally that the F/(F + OH) ratio of phlogopite blages. Becausethe stability of biotite exceedsthat of increaseswith increasing temperature upon (Munoz muscovite in quartzofeldspathicrocks (Brown and Fyfe, equilibration with a fluid of constant {*. and Lu- presumed 1970), biotite is the mica applicable to the modeling of dington, 1974, 1977). Therefore, it is that F lower-crustal anatexis. To a first approximation, the va- substitution for OH increasesthe decomposition temper- phlogopite pressures. quantitative por-absent melting of biotite * qoartz representsa limit ature of at low No data partial to the beginning of melting in crustal rocks and is con- exist, however, for the vapor-absent melting of phlogopite quartz pressures sequentlyan important constraint on the processofgran- fluorhydroxy + at applicable itoid genesis. to the middle and lower crust. To obtain this informa- The vapor-absentmelting temperatureof the biotite + tion, we have conductedan experimental investigation of phlogopite quartz assemblagehas been determined for an end-mem- the vapor-absent melting of fluorhydroxy + qvartz ber hydroxy phlogopite in the system K,O-MgO-AlrO.- in the system KrO-MgO-AlrO3-SiOr-HrO-F. The principal SiO,-HrO (KMASH) (Peterson and Newton, 1989a); objective of the study was to obtain reversed however, many biotite samplesfrom high-grade terrains experiemntal P-Z determinations of the solidus of the phlogopite quartz: contain substantial fluorine (FJ in place of hydroxyl (OHJ reaction fluorhydroxy * enstatite + liquid. This information provides an initial quantifica- tion of the stabilizing effect of F in biotite, and thereby * Present address: Amoco Production Company, P.O. Box placesfurther constraints on the partial melting processes 3092,Houston, Texas 77253, U.S.A. responsiblefor the genesisof magmas in the lower crust. 0003-004x/9I /0304-0470$02.00 470 PETERSON ET AL.: MELTING OF PHLOGOPITE + QUARTZ 47r

of the componentswas loaded into an Pt capsulewith an ExprnrvrnlvrAl. PRoCEDUR-ES outside diameter of 8 mm and reacted at 1.5 kbar and Apparatus 750'C for 5 d. Optical examination, as well as the ab- Experimentswere performedat -8 and -15-16 kbar senceof nonmica reflectionson X-ray diffractograms,in- in a solid-medium piston-cylinder apparatus.All exper- dicated a successfulsynthesis. Natural quartz from Lis- iments were conducted at the University of Chicago, ex- bon, Maryland, was used as the SiO, phase in the cept the 8-kbar reversalexperiment (Fl8), which was per- experiments. formed at the University of Alberta. Two types of salt Starting materials of fluorhydroxy phlogopite + quartz (NaCl) pressuremedia were used in the experiments. In (2:3 by weight) were ground in an agate mortar under one type, the sample was surrounded by NaCl cylinders distilled HrO to a fine-grained homogeneous mixture. inside a cylindrical graphite furnace. The furnace assem- Sampleswere dried under a heat lamp at -80 "C for I h, bly wasjacketed by a 1.905cm (o.d.)NaCl outer sleeve. then loaded in 4-7 mg amounts into Pt capsuleswith an In a secondtype, the samplewas placedbetween two soft- outside diameter of 1.5 mm. The encapsulatedsamples glasscylinders (Ca-Na glass,Corning no. 47057), inside werethen dried for a minimum of 1.5h at I l0'C before a graphite furnace that was separatedfrom the NaCl outer sealing of the capsules. sleeveby a l-mm thick soft-glasssheath. The assembly containing soft-glassinternal parts was necessaryfor high- Analytical methods temperature experiments in which the melting tempera- Starting fluorhydroxy phlogopite was characterizedby ture of NaCl was exceeded.Pressure was applied to the powder X-ray diffraction and electron microprobe tech- all-NaCl assembly 3-4 kbar below the desired pressure niques. X-ray diffraction unit-cell refinements,using Ni- at room temperature.A precalibratedamount of thermal filtered CuKa radiation, were obtained by scanning at a expansion allowed final pressuresto be attained within rate of t/q"2-0/min with an internal standard of synthetic 200 bars ofthe desired value upon heating. For the soft- corundum. Refinements were facilitated by the least glassassembly, a cold pressureof 4-5 kbar was initially squares refinement program of Appleman and Evans applied to the assembly.The temperaturewas then raised (1973), for lM (C2/m) mica polytype. The F content of to approximately 520'C, at which point the soft glass the phlogopite was determined by the refinement equa- relaxed enoughto allow the desiredpressure to be applied tion for F-OH substitution in phlogopite of Noda and to the assembly without shattering it. The temperature Ushio (1964). Unit-cell parametersand an F/(F + OH) was then raised to the appropriate conditions. No pres- value for starting mica are shown in Table l. According sure correction was applied to either of the NaCl pressure to the X-ray chancteiaalion, F/(F + OH) : 0.58 (+9.12; media. A third type of assemblywas used for the reversal for the synthetic phlogopite. experiment at 8 kbar. The assembly consisted of talc, Quantitative chemical analysesof starting mica were - pyrex, and alumina. A l0o/ofriction correction was ap- obtained from polished sections on a JEOL 733 Super- plied to this assembly,based on delineation of the reac- probe with Tracor Northern software at the University tion fayalite + quartz: ferrosilite (R.W. Luth, personal of Washington in Seattle.Two different electron micro- communication). Nominal pressureswere measuredwith probe techniques were employed. In the first technique a Heise bourdon tube gauge.Measured pressuresfor all (method l, Table l), major element oxides were analyzed assembliesare considered accurate to within 500 bars, by EDS using 15 keV acceleratingvoltage and l0 nA and precise to within 100 bars. beam current, and resultsare consideredaccurate to with- Temperatures were measured with W-WRe thermo- in 2o/oof the amount present. Quantitative F analyses couples situated directly above the encapsulatedsample. were obtained through WDS methods using a TAP crys- The cold junction of the thermocouple was laboratory tal. A calibration curve of counts vs. concentration was temperature. Temperature was controlled to within a de- constructedusing a natural fluorite (48.7 wto/oF) and syn- greewith an electronic, solid-state controller. No correc- thetic flurophlogopite (9.02 wto/oF) as standards. Con- tion was made for pressure-inducedemf. Temperature centration of F in the starting mica was determined from gradients were measured for the graphite furnaces used the calibration curve by collecting counts over 60 s time in this study. Reported temperaturesare considered ac- intervals. Representativeanalyses are shown in Table I curate to within l0 "C. and indicatean F/(F + OH) :0.52 (a0.03) for the start- Starting material was synthesizedin an interndlly heat- ing mica. In the second technique (method 2, Table l), ed Ar pressurevessel (design after Holloway, 197l). Pres- all elements were analyzed by WDS, including F with a sure was measured with an Astra bourdon tube gauge, synthetic LDE I crystal. Analytical data were collectedat and temperature was measuredwith two Inconel-sheathed, 15 keV acceleratingvoltage and a 5 nA beam current. chromel-alumel thermocouples. Maximum counting time for each element was 40 s. Ma- jor element counts were calibrated against a natural bio- Starting materials tite standard, and F was calibrated againstan end-mem- The fluorhydroxy phlogopite used in this study was ber fluorphlogopite. The averageF/(F + OH) from this synthesizedhydrothermally from a combination of KF. technique is 0.54 (a0.03). Becauseof the small grain size 2HrO, MgO, ArOr, and SiOr. A homogenized mixture (<5 rrm) of the starting material, only those grains with 472 PETERSON ET AL,: MELTING OF PHLOGOPITE + QUARTZ

Trel-e1. Representativecharacterizations of startingmica (X-ray diffraction and electron microprobe techniques) wt% (method 1) wf/. (method2) Unit-cellparameters sio, 44.50 45.26 46.73 40.64 41.66 42.40 a:5.328 (10.001) Al2o3 12.39 12.65 12.74 11.12 10.76 't1.22 b: 9.194(+0.002) Mgo 30.16 30.58 29.94 27.47 27.24 27.74 c: 10.209(10.001) KrO 12.65 11.26 10.54 10.76 10.55 11.14 d : 99.95"1+9.921 F 4.57 4.58 4.53 5.11 4.65 5.09 v:492.57 A"(*O.tS) counts/min- 2200 2205 2178 averageF/(F + OH): 0.52(+q.m1 averageF/(F + oH) : 0.54(+9.93; F(F + oH): 0.58(+s.12y-

Note.' Method l-Maior elements analyzed with EDS and normalizedto 100%. F analyzedwith WDS, see text. Method 2-All elements analyzed on WDS. See text for details. t Standardizationmethod describedin text. Fluorite(48.7 wt% F) yields 23409 counts/minand full fluorphlogopite(9.02 wt% D yields 4340 counts/ mrn. .t Derivedfrom empiricalcalibration of Noda and Ushio (1964). a basal surfbce normal to the beam could be analyzed. several times larger than residual mica grains. Glass X-ray and electron probe results are in good agreement. (quenched melt) was observed as clusters of low-index In this study, we consider the X-ray determination to be beadletsintimately associatedwith enstatite in near-sol- a more accurate indication of F content becauseof the idus experiments, and larger angular fragments contain- uncertainties associatedwith analyzing grains smaller than ing enstatite and qtnrtz at temperatures well above the 5 pm. solidus. Experimental products were examined optically in re- Glass fragments from a sample quenchedat 14.7 kbar 'C fractive index (RD oil mounts of RI -1.540-1.542. Quartz and I100 (experiment F5, Table 2) were analyzed by was identified by its low positive relief and colorlessap- electron microprobe WDS techniques, using a synthetic pearancein plane light. Enstatite was identified by high LDE I crystal for F. More than ten individual glassfrag- positive relief, weak birefringence, and straight extinc- ments were analyzedwithin the sample. Analytical data tion, if prismatic in habit. Enstatite occurred as abundant were collected at 15 keV acceleratingvoltage and a 5 nA granules (-5 pm diameter) forming a wormlike inter- beam current. Maximum counting time for each element growth with glassand quartz, and also as a small number was 40 s. Major element counts were calibrated againsta of relatively largeprisms (- 100 pm long) at tomperatures natural biotite standard and F against an end-member well above the solidus. Two types of mica were observed, fluorphlogopite (method 2 above).It was determined that both with positive relief. Residual starting mica occurred a surface beam size g.reaterthan I pm diameter would as highly birefringent recrystallizedmats of variable and incorporate adjacent quartz and enstatite in the analysis irregular habit. Mica grown in reversal experiments oc- of glass.Therefore, beam diameter for glassanalyses was curred as euhedral crystals, commonly as hexagonalplates, maintained to < I pm. HrO content was determined by differencefrom a 1000/ototal. Representativeanalyses are given in Table 3. TABLE2. Experimentaldata for syntheticPh + Qz startingas- semblage(F/F + OH: 0.58) ExprnmnNTAL REsuLTs Experi- ment P T Time Experimental data for the reaction fluorhydroxy phlog- qC no. kbar (h) Productassemblage opite + elJartz: enstatite + liquid are listed in Table 2 F11' 8.0* 1030 432 Qz+En+Gl +(Ph) and illustrated in Figure l. At 8.0-8.5 kbar, melting of F12 8.4 1075 231 Qz+En+Gl +(Ph) the starting assemblageoccurs between950 and 1000 "C. F13 8.5 1000 336 Qz+En+Gl +(Ph) In the pressurerange of 15.2-16.0 kbar, melting occurs F18 8.0 1050 51 'C. -8.0 -950 162 Ph+Qz+(En)+(Gt?) between 1020 and 1050 Note that the slope of the Fl 14.8 950 48 Ph+Qz reaction changesfrom undefined in the KMASH system F2 15.5 1000 50 Ph+Qz+(En?)+(Gl?) F3 16.0 1020 47 Ph+QZ to approximately 140 bars/'C in the F-bearing system. F4 15.7 1020 142 Ph+Qz+(En)+(Gt?) Trace amounts ofglass and enstatite in products quenched F6 15.5 1020 289 Ph+oz+(En)+(Gl?) from 15.5kbar and 1000'C (experimentF2) may be the F8 15.2 1050 68 Qz+En+Gl +(Ph) F5 14.7 1100 218 Qz+En+Gl +(Ph) result of trace amounts of moisture remaining in the sam- Ft 14.9 1100 237 ple before sealing. '14.8 -1000 319 Ph+Qz+(GD The vapor-absentmelting reaction was reversedat -8 Note.'Ph : fluor-hydroxyphlogopite, Qz : quatz, En : enstatite, and - l5 kbar by two-stageexperiments, as illustrated by Gl: glass(quenched melt), ( ): smallto traceamounts, ? : presenceol phaseuncertain, ' : secondstage. brackets in Figure l. For example, the first stageof the 'Temperaturemeasurement within 2OC reversal experiment at 14.9 kbar (experiment F7, Table *'Pressure on this experiment. measurementprecision is +1OObars, accuracy is +SOO 2) consisted of holding a sample of fluorhydroxy phlog- bars. opite + qluartzat 1100 'C for 237 h, a temperature at PETERSON ET AL.: MELTING OF PHLOGOPITE + QUARTZ 473 which a previous experiment (F5) had produced abun- 20 dant enstatite + melt. In the secondstage of the experi- ment, the temperaturewas lowered nearly isobarically to 1000 'C in Ph En the same experiment and held for 3 19 h. Re- L 15 versal of the melting reaction near 8 and 15 kbar was N UL Ph indicated by the absenceof enstatite and textural evi- I En dencefor pronouncedmica regrowth. This mica regrowh LLI tn included -50-pm diameter crystalsas well as finer grained tru l mica textures similar to those reported for vapor-absent a melting in the pure-HrO system (Petersonand Newton, a 1989a).The reader is referred to the sketchesin the last tv F/F+OH F/F*OH referencefor textural details. o_ =0 =058G.12) Representativeanalyses of quenched liquid produced by the melting of fluorhydroxy phlogopite + qvarlz at I 100 and l5 kbar given "C are in Table 3. Melts are 50.8- 750 800 850 900 950 10001050 1 100 1150 62.40/oSiOr, 10.6-13.80/oAl2O3, I1.9-l9.lVo MgO, and 9.4-12.7o/oKrO by weight. The averagemelt lacks nor- TEYPERATURE('C) mative quartz and has a normative olivine to orthopy- Fig. l. P-T diagam illustratingkey experimentaldata from roxene ratio of 6.6:1. The compositions are similar to Table 2. Resultsare comparedto the vapor-absentmelting of those tabulated for lamproites by Bergman (1987). stoichiometrichydroxy-phlogopite + quartz in the F-free KMASH system(Peterson and Newton, 1989a).F/(F + OH) DrscussroN valuesrefer to phlogopite.Open boxes represent experiments Stabilization of phlogopite with productsofabundant enstatite and glass (quenched liquid). Bracketsindicate two-stage reversal in The data of this investigation indicate that -60 molo/o experimentsas described text. Notice that the slope of the reaction changesfrom unde- substitution of F for OH in phlogopite stabilizes the fined in the KMASH systemto approximately140 bars/"Cin phlogopite quartz + assemblagerelative to enstatite + the F-bearingsystem. Ph : phlogopite, : quartz,En : en- -8 'C Qz liquid by 125 "C at kbar and by 175 at - 15 kbar statite,L : liquid. (Fig. l). BecauseF representsan additional component relative to the pure-HrO system,the melting equilibrium is divariant with respect to the KMASH curve. This is manifested by the presenceof recrystallized,presumably Results similar to those of this study were observed more F-rich, phlogopite in experimental products from for the vapor-absent melting of pargasite [NaCar- supersolidusconditions. MgoAl(AlrSiu)O,,(OH,D,l.Holloway and Ford (1975)re- It has been known that F substitution for hydroxyl in- ported that 43 molo/osubstitution of F for OH- increased creasesthe thermal stability of phlogopite (see Munoz, the solidus of pargasite by 70-100 .C at 15 kbar. 1984). Hydroxy phlogopite decomposesat less than 905 Combined with our results, this indicates that F is par- oC at pressureslower than 100 bars (Wones, 1967) as titioned preferentially into phlogopite and pargasiterel- compared to the melting temperature of pure fluorphlo- ative to the vapor-undersaturatedliquids produced upon gopite, which is between -1345 and 1390'C at I bar melting of the respectiveminerals. (Van Valkenburg and Pike, 1952; Shell and Ivey, 1969). Giese(1984) and Munoz (1984)discuss the stabilizing The decomposition temperatures of intermediate fluor- solubility mechanism of F substitution for hydroxyl in hydroxy phlogopitesare unknown, however, as is the na- mica. In trioctahedral phlogopite, the mechanism is based ture ofthe breakdown reactionswith respectto dehydra- on the fact that the H* proton of the (OH-) radical is tion, defluorination, or melting. Likewise, no experimental directed toward the interlayer K* cation by the Mg atoms. data exist on the subsolidus breakdown offluorhydroxy The substitution of F- for OH- is a means by which the phlogopite in the presenceof quartz; therefore, further interlayer cation K*-H* repulsion can be removed, with comparison of fluorhydroxy phlogopite melting to the consequentstabilization of the structure. pure-HrO system (Peterson and Newton, 1989a) is not The data presentedhere show that F substitution in- possible. creases vapor-absent melting temperatures of assem-

Tmre 3. Representativeanalysis of glassquenched lrom 14.7kbar and 1100 'C in equilibriumwith quartzand enstatite sio, 57.76 52.93 56.37 60.02 59.90 55.24 50.77 56.01 62.37 57.14 55.42 Al,03 12.63 12.03 13.06 12.86 13.17 13.82 12.14 12.49 10.56 13.42 13.01 Mgo 12.52 19.13 13.59 13.66 12.02 11.90 17.50 16.31 12.77 13.37 14.44 KrO 12.64 10.47 12.18 12.06 12.09 13.61 11.00 11.43 9.36 12.32 1r.54 F 3.32 2.74 2.32 0.98 1.88 4.15 3.84 3.00 1.00 2.79 3.86 Total 98.87 97.30 97.52 99.s8 99.06 98.72 95.25 99.24 96.06 99.04 98.27 474 PETERSONET AL.: MELTING OF PHLOGOPITE+ QUARTZ

KAIO Consequently,in the simplified system,the melts change from granitic to lamproitic. It can be seen in Figure 2 that melts become signifi- (wt%) cantly richer in MgO and KAIO, upon addition of F to the KMASH system.F-bearing melts contain an average of l0 wto/omore MgO, 4 wto/omore KrO, and 3 wto/omore AlrO, than the F-freemelts (Petersonand Newton, 1989a). A trajectory from the cluster of analysesof KMASH glass' through the cluster of analysesfrom this study, to the MgO-KAIO, sideline yields the nonvolatile composition- al control on the liquid variation. Accordingly, F en- hancesthe solubility of a fictive mafic-alkali component with a MgO/KAIO, ratio of 55/45 by weight, ot 2/l on a molar basis. The observedchange in quenchedliquid compositions is broadly consistent with the results of studies on other F-bearing systems.In F-bearing, HrO-free systems,spec- troscopic and phase equilibrium data suggest that F preferentially complexes with network-modifying or charge-balancingcations (e.g., K and Mg) in the melt. This complexing results in an expansion of the stability Qz fields of the more polymerized (Si-richer) phases(Foley En et al., 1986; Luth, 1988a, 1988b).This is evidencedin the kalsilite-forsterite-quartz system in which the addi- Fig. 2. Diagram comparingnonvolatile compositions(wto/o) tion of F expands the stability field of enstatite relative of quenchedmelt (glass)from the KMASH andKMASH-F sys- to that offorsterite (Foley et al., 1986) and in the diop- tem. Largeopen circles represent F-bearing melts quenched from side-albite system in which the stability field of albite 14.7kbar and ll00 (-75'C abovesolidus). See Table 3. "C expands relative to that of diopside (Luth, 1988b). In Small solid circlesrepresent F-free melts quenchedfrom 10.0 HrO-F systems,there is an overall depolymerization of kbarand 950 qC (- 100qC above solidus) (Peterson and Newton, of 1989a).Sun symbolrcpresents HrO and F-freebulk composi- the melt structure (causedby HrO), but an expansion tions for both systems.Dotted lines are tie lines for three-phase the stability fields of the polymerized phases(caused by triangle enstatite-melt-quartz.Sa : sanidine.Other abbrevia- F) (Foley et al., 1986) comparedto stability fields of poly- tionsare sameas in Figurel. Seetext for discussion. merized phasesin the pure-HrO system. This is appar- ent in the HrO-saturated haplogranite system, in which the addition ofF expands the liquidus phasefield ofquartz blages in which micas and amphibole are key constitu- relative to those of the feldspars(Wyllie and Tuttle, 1961; ents. The effect of F on the HrO-saturated melting of Manning and Pichivant, 1983). The above findings are mica or amphibole has not been investigated systemati- consistent with the observation in this study that F ex- cally. In the HrO-saturated haplogranite system, Man- pands the stability field of quartzrclative to other phases, ning (1981) demonstrated experimentally that the liqui- resulting in a more mafic-alkalic melt. dus is lowered by 100 "C at I kbar upon addition of 4 wto/oF to the system. Similarly, the vapor-saturated sol- quartz idus is lowered by more than 165 "C. Implications for biotite + assemblagesin granulite terranes Melt compositions It is clear from our results that, under vapor-absent Glass analyses from the F-free KMASH system are conditions, F-rich micas are stable to high temperatures compared on a nonvolatile basis to analysesfrom this in quartzofeldspathic rocks. Quartzofeldspathic rocks study in Figure 2. Although compositions of the F-bear- containing biotite with l-5 wt9o F have been reported ing glassesare more variable than the F-free glasses,Fig- from many granulite terranes including southern India ure 2 demonstratesclearly that quenched melt from this (Leenanandam, 1969; Janardhan et al., 1982; Chacko et study contains significantly lessSiO, than KMASH melts. al., 1987), the Grenville Province of Canada (Lonker, The presence of l-4 wto/oF in the melt increasesthe 1979), and Namaqualand, South Africa (Waten, 1988). proportions of quartz to other phases,relative to the F-free The higher F contents tend to be associatedwith the system.This is apparent if the two three-phasetriangles, metasedimentary lithologies, although orthogneissesare enstatite-quartz-melt, are examined with reference to the also included in the referencesmentioned. The significant nearly identical nonvolatile bulk composition for both stabilizing effect of F indicates that vapor-absent partial systems.The liquid composition shifts from the SiOr-rich melting during granulite grade metamorphism may be 'C, side of the enstatite-sanidinejoin to the SiOr-poor side. restricted to temperatures exceeding950 if the rock PETERSON ET AL.: MELTING OF PHLOGOPITE + QUARTZ 475

contains F-rich micas. Furthermore, the residual micas in the HrO and H,O-F systems(Fig. 2), a genetic con- will be progressivelyF rich. nection betweengranitic melts and lamproitic melts may The vapor-absentsolidus temperaturesof fluorhydroxy involve heterogeneitiesin the F contents of the lower mica-bearing natural rocks may be different than those crust. These heterogeneitiescould arise becauseof pri- given by our experimental work. First and most impor- mary lithologic variations in the lower crust or because tantly, a relatively sodic plagioclasebecomes an addi- of the spatially variable influx of halogen-rich mantle- tional reactant phasein the melting reaction. By analogy derived fluids (e.g., Bailey, 1978; Harris and Marriner, with the pure HrO system,this may lower the solidus by 1980). Whatever the causeof the heterogeneities,under 'C 20-50 (e.g.,Thompson and Tracy, 1979;Waters, 1988). vapor-absent conditions, F-rich pockets would remain Secondly, the significant Fe content of natural biotite is melt-free or melt-poor during the lower-temperature likely to mitigate somewhat the stabilizing effect of F be- melting events that produced granitic melts in adjacent causeof Fe-F avoidance in micas (Rosenbergand Foit, F-poor rocks. A subsequenthigher-temperature melting 1977;'Yalleyet al., 1982).On the other hand, many bi- event could then generatemelts of lamproitic composi- otite samplesfrom granulites have high Ti contents,and tion from these F-rich regions. The potential genetic re- it has been demonstraredthat Ti stabilizesbiotite to high- lationship betweengranitic and lamproitic-lamprophyric er temperatures(Forbes and Flower, 1974; Tronnes et al., rocks (seeRock, 1987) warrants further investigation in 1985). The cumulative effect of additional componentsis light ofthe results ofthis study. difficult to ascertain. We have consideredthe melting of fluorhydroxy phlog- opite + qrrartz, and the resultant magmas, only in the Implications for the origin of A-type granites context of vapor-absent melting; alternate scenariosare Our experimental data also bear on models that have possible. Severalworkers have suggestedthat a COr-rich been proposed for the origin of A-type granites (Loiselle fluid may play an important role in the evolution of an- and Wones, 1979;Collins et al., 1982).A-type, or anoro- orogenicgranites, granulites, and associatedrocks (Taylor genic granites, were originally thought to be restricted to et al., 1980;Wendlandt, l98l; Santoshand Drury, 1988). rift-related tectonic settings (Bailey, 1978; Loiselle and Data from recent experiments indicate that the addition Wones, 1979).However, recent studiesindicate that they of a l:l molar, COr-HrO fluid significantly lowers the may be found in a variety of tectonic settings,including solidus ofphlogopite + qvartz assemblagesrelative to the transcurrent faulting zones,continent-continent collision vapor-absent and pure H2o-saturated solidi (Peterson and zones or even subduction zones (Whalen et al., 1987: Newton, 1989b), with attendant melts rangrngfrom gra- Sylvester, 1989). Geochemically, A-type granites are nitic to lamprophyric in composition. The effect of this characterizedby high contents of Si, Na I K, Ga, Zr, carbonic fluid on melting temperaturesin the F-bearing Nb, and F, and low contents of Ca, Mg, Ba, and Sr rel- system remains to be evaluated. ative to S- and I-type granites (Collins et al., 1982; Wha- len et al., 1987). Petrographic and experimental studies AcxNowr,BocMENTS indicate that these granites were emplaced in an almost Thanks are in order to Julian R. Goldsmith and Robert C. Newton for completely molten stateat temperaturespossibly exceed- use ofthe high-pressure laboratory and facilities at the Univenity ofChi- ing 900 'C (Collins et al., 1982; Clemenset al., 1986), cago. A very thorough review by Steve Foley improved the manuscript considerably. Discussions with Bob Luth were helpful. Work at the Uni- significantly higher than temperatures of many other versity of Alberta was supported by a CRF-NSERC grant (T.C.). granitoid magmas. To account for their high-temperature character and RnrunnNcBs crrED also for the fact that they commonly postdate the intru- sion ofother spatially associatedgranitoids, one currently Appleman, D.8., and Evans, H.T. (1973) Job 9214: Indexing and least- squares refinement of powder popular model maintains granites diftaction data. United States Depart- that A-type were de- ment of Commerc€, National Technological Information Service, Pub- rived by the high-temperature,fluid-absent partial melt- lication 216188. ing of previously melt-depleted source regions that con- Bailey, D.K. (1978) Continental rifting and mantle degassing.In E.R. tained halogen-enrichedmicas and amphibole (Collins et Neumann and I.B. Ramberg, Eds., Petrology and geochemistry of con- tinental rifts, p. l-13. al., 1982; Clemens et al., 1986). Our experimental data Reidel, Holland. Bergman,S.C. (1987) Iamproites and other potassium-rich rocks: A re- are broadly consistent with this model, in that the sub- view of their occurrcnce, mineralogy and geochemistry. In J.G. Fitton stitution of F for OH- in phlogopite does indeed stabi- and B.G.J. Elpton, Eds., Alkaline igneousrocks, p. 103-190. Geolog- lize the mica * quartz assemblage,thereby forcing the ical Society Special Publication 30. melting of any residual F-rich biotite to occur at relatively Brown, G.C., and Fyfe, W.S. (1970) The production of granitic rnelts high temperatures. gross during ultrametamorphism. Contributions to Mineralogy and Petrol- The inconsistencyis, however, ogy, 28, 310-318. the somewhat surprising result that the melts generated Chacko, T., Ravindra Kumar, G.R., and Newton, R.C. (1987) Metamor- are lamproitic in composition, not granitic as may have phic P-T conditions in the Korla (South India) Khondalite Belt: A been predicted. It is probable that less mafic, more gran- granulite-faciessupracrustal terrain. Journal of Geology, 95, 343-358. iteJike melts are produced by the melting of micas with Clemens, J.D., Holloway, J.R., and White, A.J.R. (1986) Origin of an A-type granite: Experimental constraints. American Mineralogist, 71, lower F contents. 3r7-324. Considering the vapor-undersaturatedmelts generated Collins, W.J., Beams, S.D., White, A.J.R., and Chappell, B.W. (1982) 476 PETERSON ET AL.: MELTING OF PHLOGOPITE + QUARTZ

Nature and origin ofA+ype granites with particular references10 south- Noda, T., and Ushio, M. (1964) Hydrothermal synthesisof fluorine-hy- eastern Australia. Contribulions to Mineralogy and Petrology, 80, 189- droxyl-phlogopite; part two: Relationship between the fluorine content, 200. lattice constants, and the conditions of synthesis of fluorine-hydroxyl- Foley, S.F., Taylor, W.R., and Green, D.H. (1986) The effect of fluorine phlogopite. GeochemistryInternational, l, 96-104. on phaserelationships in the systemKAlSiOo-MgrSiOn-SiO, at 28 kbar Peterson,J.W., and Newton, R.C. (1989a) Reversedexperiments on bi- and the solution mechanism offluorine in silicate melts. Contributions otite-quartz-feldspar melting in the system KMASH: Implications for to Mineralogy and Petrology, 93,46-55. crustal anatexis.Journal of Geology, 97, 465-485. Forbes, W.C., and Flower, M.F.J. (1974) Phaserelations of titan-phlog- -(1989b) COr-enhancedmelting of biotite-bearing rocks at deep- opite, $MgTiAlrSiuO.(OH)o: A refractory phasein rhe upper man- crustal pressure-tomporatureconditions. Nature, 340, 378-380. tle? Earth and PlanetaryScience Letters, 22,60-66. Rock, N.M.S. (1987) The nature and origin of lamprophyres: An over- Giese,R.F., Jr. (1984) Eloctrostaticenergy models of micas. In Mineral- view. In J.G. Fitton and B.G.J. Elpton, Eds., Alkaline ig.eous rocks, ogical Societyof America Reviews in Mineralogy, 13, 105-141. p. 19l-226. Geological Society SpecialPublication 30. Guidotti, C.V. (1984) Micas in metamorphic rocks. In Mineralogical So- Rosenberg,P.E., and Foit, F.E., Jt. (1977) Fe'+-F avoidance in silicates. ciety of America Reviews in Mineralogy, 13,357-467. Geochimica et CosmochirnicaActa, 41, 345-346. Harris, N.B.W., and Marriner, C.F. (1980) Geochemistry and petrogen- Santosh, M., and Drury, S.A. (1988) Alkdi granites with Pan-African esis of a peralkaline granite complex from the Midian Mountains, Saudi affinities from Kerala, S. India. Joumal of Geolory,96,616-626. Arabia. Lithos. 13. 325-327. Shell, H.R., and Ivey, K.H. (1969) Fluorine micas. United StatesBureau Holloway, J.R. (1971) Internally heatedpressue vessels.In G.C. Ulmer, of Mines Bulletin, 647, 291 p. Ed., Research techniques for high pressure and high temperature, p. Sylvester,P.J. (1989) Post-collisionalalkaline granites.Journal ofGeol- 217-258. Springer,New York. ogy, 97, 261-280. Holloway, J.R., and Ford, C.E. (1975) Fluid-absentmelting of the fluoro- Taylor, R.P., Strong, D.F., and Kean, B.F. (1980) The Topsails Igneous hydroxy amphibole pargasite to 35 kilobars. Earth and Planetary Sci- Complex: Silurian-Devonian peralkaline magmatism in westem New enco Isters. 25.44-48. Foundland. CanadianJournal ofEarth Sciences,17, 425-439. Janardhan, 4.S., Newton, R.C., and Hansen, E.C. (1982) The transfor- Thompson, A.B., and Tracy, R.J. (1979) Model systemsfor anatexis of mation of amphibolite facies greiss to charnockite in southem Kar- pelitic rocks. II. Facies series melting and reactions in the system CaO- nataka and northern Tamil Nadu, India. Contributions to Mineralogy KAIO,-NaAlOz-AlrOj-SiOr-HrO. Contributions to Mineralogy and Pe- and Petrology, 79, 130-149. trology,70, 429-438. I-eenanandam,C. (1969) HrO*, F and Cl in the chamockite biotites from Tronnes, R.G., Edgar,A.D., and Arima, M. (1985) A high pressure-high Kondapalli, India. Neues Jahrbuch fiiLr Mineralogie Monatshefte, 10, temperatue study of TiO, solubility in Mg-rich phlogopite: Implica- 46t-468. tions to phlogopite chemistry. Geochimica et Cosmochimica Acta, 49, Loiselle, M.C., and Wones, D.R. (1979) Characteristicsand origin of an- 2323-2329. orogenic granites. Geological Society ofArnerica, Abstracts with Pro- Valley, J.W., Petersen,E.V., Essene,E.J., and Bowman, J.R. (1982)Fluor- grams,I l, 468. phlogopite and fluortremolite in Adirondack marbles and calculated Lonker, S.W. (1979) Conditions of metamorphism in high grade pelites C-O-H-F fluid compositions.American Mineralogist, 67, 545-557. from the Frontenac Axis, Ontario, Canada, 361 p. Ph.D. thesis, Har- Van Valkenburg, A., and Pike, R.G. (1952) Synthesisof mica. Joumal of vard University, Cambridge, Massachusetts. Researchofthe National Bureau ofStandards, 48, 5, 360-369. Luth, R.W. (1988a) Raman spectroscopicstudy of the solubility mecha- Waters, D.J. (1988) Partial melting and the formation of granulite facies nisms of F in glassesin the system CaO-CaFr-SiOr.American Miner- assemblagesin Namaqualand, South Africa. Journal of Metamorphic alogist, 73, 297-305. Geology, 6, 387-404. -(1988b) Effectsof F on phaseequilibrium and liquid structure in Wendlandt, R.F. (1981) Influence of CO, on melting of model granulite the system NaAlSiO4-CaMgSirO6-SiOr.American Mineralogis! 73, 30w< facies assemblages:A model for the genesisof chamockites. American 312. Mineralogist, 66, | 164-l l7 4. Manning, D.A.C. (1981) The effectof fluorine on liquidus phaserelation- Whalen, J.B., Currie, ICL., and Chappell, B.W. (1987) A-type granites: ships in the system Qz-Ab-Or with excess water at I kbar. Contribu- Geochemical characteristics, discrimination and petrogenesis. Contri- tions to Mineralogy and Petrology,76, 206-215. butions to Mineralogy and Petrology, 95, 407-419. Manning, D.A.C., and Pichivant, M. (1983) The role of flourine and bo- Wones, D.R. (1967) A low pressureinvestigation of the stability of phlog- ron in the generation of granitic melts. In M.P. Atherton and C.D. opite. Geochimica et CosmochimicaActa, 31,2248-2253. Gribble, Eds., Migmatites, melting, and metamorphism, p. 94-109. Wyllie, P.J.,and Tuttle, O.F. (1961)Experimental investigation of silicate Shiva, Nantwich. systems containing two volatile components. Part II. The effects of NH, Munoz, J.L. (1984) F-OH and CI-OH exchangein micas with applications and HF, in addition to HrO on the melting temperatures of albite and to hydrothermal ore deposits. In Mineralogical Society of America Re- granite. American Joumal of Science,259, 128-143. views in Mineralogy, 13,469-491. Munoz, J.L., and Ludington, S.D. (1974) Fluorine-hyfuoxyl exchangein biotite. Arnerican Journal of Science, 274, 396-41 3. - (1977) Fluorine-hydroxyl exchange in synthetic muscovite and its applications to muscovite-biotite assemblages.American Mineralogist, Mar.ruscnrrr RECETvEDMencn 30, 1990 62.304-tO8. Mrxuscnrp,r AccEprEDDscErdsen 26, 1990