effects of b and hro on liquidus phase relations in the ... · pdf fileeffects of b and hro on...

American Mineralogist, Volume 72, pages 1056-1070, 1987

Effects of B and HrO on liquidus phase relations in thehaplogranite system at 1 kbar

Mrcnnr, hcnlvlurCentre de Recherches P6trographiques et G6ochimiques, BP 20,54501 Vandceuvre les Nancy, France

AesrRAcr

Liquidus phase relations have been determined in the system Qz-Ab-Or at I kbar withI and 4.5 wto/o BrO. in the melt under both HrO-saturated conditions and HrO-undersat-urated conditions (4.5 wto/o BrO, in the melt), using a HrO-CO, fluid mixture. Under HrO-saturated conditions, the addition of BrO, leads to a reduction of the liquidus temperaturesfor compositions in the feldspar field; compositions in the quartz field are not significantlyaffected. The minimum liquidus temperature is progressively reduced, and its compositionmoves toward the Ab corner: Qz.rAbrrOrrn,720 "C (B-free, Tuttle and Bowen, 1958),QzruAbrrOrrr, 680 'C (l wto/o BrOr), Qz'AbouOrr,,640 "C (4.5 wto/o BrO3). HrO solubilityin the melt increases with increasing B concentration. Isobaric reduction of the HrO con-tent in the melt by about 50o/o at constant BrO, increases liquidus temperatures for all ofthe compositions studied and shifts the minimum liquidus composition toward the Qz-Or side at constant Qz contents: Qz.,Ab.oOrrr, 750'C (4.5 wto/o BrO3).

The contraction of the liquidus field for Ab suggests that the addition of B partiallydisrupts the aluminosilicate network. B-bearing structural units are formed involving al-kalis (mainly Na) that were previously used for charge-balancing IVAI. A fraction of the Batoms may exist in tetrahedral coordination. The increase in the HrO solubility is attrib-uted to the hydrolysis of the borate units. An isobaric increase of the HrO content of themelt has similar efects on phase relations, indicating that HrO is preferentially associatedwith the Ab-forming components in the melt and that the incorporation of HrO in alu-minosilicate melts may involve expulsion of Al from the network. These results allow anestimation to be made of the individual effects of pressure and of the HrO content of themelt on the phase relations in the Qz-Ab-Or system. The B concentration of naturalmagmas (l wto/o BrO, maximum) is too low to influence significantly the major-elementcomposition of residual melts. However, the minor- and trace-element chemistry of thesemelts may be profoundly affected.

INrnooucrroN cesses. However, they do not give much informationB is concentrated during the late stages of differentia- about the effect of B on aluminosilicate melt structure,

tion of felsic magmas, as shown by the presence of mag- crystal = liquid equilibria, and the composition of resid-matic tourmaline in some leucogranites (Benard et al., ual melts in B-rich natural magmatic systems. Such in-1985) and by the composition of fluid inclusions in some formation is presented in this paper along with experi-pegmatites (London, I 986). It has been demonstrated re- mental data on the effect of B on liquidus phase relationscently that the addition of BrO. has marked effects on in the haplogranite system Qz-Or-Ab at I kbar.solidus temperatures (Chorlton and Martin, 1978; Pi- Most of the new experimental data concern HrO-sat-chavant, 198 1) and on partitioning of elements between urated liquidus phase relations. However, HrO-undersat-melt and vapor in the haplogranite system Qz-Or-Ab (Pi- urated phase relations were also studied, as it was foundchavant, l98l). Solidus temperatures are drastically re- earlier (Pichavant, l98l) that the isobaric solubility ofduced with the addition of BrO, to HrO, by more than HrO increases from B-free to B-bearing melts. Thus it is100 'C for some concentrations (Pichavant, 1981). B is necessary to distinguish between the individual effects ofpartitioned toward the vapor phase relative to melt (Ko : B and HrO on phase relations. This was achieved by de-[BrO3]-.r,/[BrOrJ""oo. : 0.33, Pichavant, 1981). With B termining part of the HrO-undersaturated liquidus phasepresent, aqueous vapor phases coexisting with melts be- relations. Therefore, this study provides experimental in-come enriched in Na relatively to K, and the silicate sol- formation about the isobaric effect of HrO on phase re-ute content of the vapor phase is higher than with HrO lations in the haplogranite system. It should be noted thatalone (Pichavant, l98l). These data are particularly rel- the classical studies ofTuttle and Bowen (1958) and Luthevanttovapor-presentlatemagmaticandpegmatiticpro- et al. (1964) do not allow the effect of HrO on phase

0003-004x/87ll l 12-1056$02.00 l0s6

relations to be separated from that of pressure becausethe HrO content of the melt and the pressure were si-multaneously changed in their experiments. Given thenow-general agreement that most granite magmas areHrO-undersaturated and the current debate about HrOsolution mechanisms in aluminosilicate melts, it is ob-viously of great importance to have experimental infor-mation about the individual effect of HrO on phase re-lations in the haplogranite system.

ExpnnrurNTAL METHoDS

Starting materials

Starting silicate materials were synthetic B-bearing glasses withvarious compositions in the Qz-Or-Ab system. These were madefrom gels prepared using conventional techniques from TEOS(Si), aluminum nitrate, and potassium and sodium carbonates.B was introduced as H.BO.. After drying, the gels were meltedat 1400 'C in welded Pt capsules for 3 to 8 h (two cycles ofmelting with grinding in between). The starting glasses, analyzedfor Si, Al, K, and Na by electron microprobe and for B by a wet-chemical method (discussed below; Table 1), do not deviate sub-stantially from the Qz-Or-Ab plane. There is no significant com-positional difference between various batches ofglasses preparedfrom the same starting gel (Table l). In this paper, liquidus phaserelations have been studied for two average BrO, concentrationsin the melt, i.e., 1 and 4.5 wIo/o (see below). Because of the prob-lem ofthe loss ofB in the vapor phase (see below), each ofthetwo series of starting glasses has been prepared with a slightexcess of BrO, compared to the desired concentration in the melt(glassesnos. l- l1:avg. l . I wto/o; nos. 12-27:avg.4.8 wto/oBrO3;Table l).

For the HrO-saturated experiments [the initial mole fractionof H,O in the fluid phase is 1, i.e., XEo (in.): ll, all chargesconsisted of 900/o of silicate materials (-40 mg of glass groundto about 50 pm) plus 100/o added water (doubly distilled, deion-ized HrO). HrO-undersaturated experiments were carried outwith a HrO-CO. fluid mixture. HrO and AgrCrOo were added ina proportion corresponding to 100/o (HrO + COr) for 900/o ofsilicates (about 40 mg of ground glass). For these experiments,the initial fluid composition was fixed to HrO/(HrO + COr) :

0.6 + 0.02 mole fraction [XEo (in.):0.6]. The charges weregenerally loaded in Au capsules (25 x 2.5 mm; thickness 0.2mm) that were then welded. Pt capsules were also employed.

Apparatus and run procedure

Almost all runs were made in R.en6-41 rapid-quench cold-sealpressure vessels. The longitudinal temperature gradients usuallycharacteristic ofthese vessels were reduced by using special fur-naces with windings condensed toward the colder end of thebomb and by working with Ar as the pressure medium. For eachvessel-furnace pair, the hot-spot zone was determined underpressure by using a sheathed dual chromel-alumel thermocouple(calibrated against melting of NaCl and against a thermocouplecalibrated and certified by the Laboratoire National d'Essais,Paris, France). With the vessel in the optimum position insidethe furnace, maximum temperature gradients do not exceed 3'C along 35 mm at 800 'C. Under run conditions, temperatureswere measured by external unsheathed chromel-alumel ther-mocouples (also calibrated against melting of NaCl and againstthe certified thermocouple) and recorded permanently. Thereadings were corrected against the internal thermocouple (max-imum correction usually less than +5'C with bombs in opti-

r057

mum position). Runs lasted up to 2 months with continuousmonitoring (daily 7 fluctuation less than +2 "C). The overallmaximum error on the recorded temperatures is less than + 10'C. Pressure was measured with a Heise Bourdon tube gauge(error t0.020 kbar). Runs at 860 qC were carried out in an in-ternally heated pressure vessel by Dr. D.A.C. Manning (IJniver-

sity of Manchester).Several capsules tied together were often loaded in the same

bomb to allow direct comparison of results for different com-positions. Heating time was usually between 30 and 60 min, withsimultaneous rise of temperature and pressure from an initiallyapplied pressure of0.8-0.9 kbar. Capsules were quenched underpressure by tilting the bomb out of the furnace (quench time =

5 s). After a check for leaks by weighing, the capsules were opened.The run products were examined by petrographic techniquesusing immersion oils and gypsum plate. Selected experimentalglasses and liquidus feldspars were analyzed by electron micro-probe for the construction of the conjugation lines and three-phase triangles. Care was taken to analyze charges having a min-imum amount of crystals so that the liquid composition re-mained essentially in its constant BrO, plane.

For each starting composition, forward runs were first con-ducted at decreasing temperature intervals of 20 'C up to theappearance of crystalline phases in the run products, as deter-mined by petrographic observation. Reversal experiments con-sisted of two-stage mns. The initial temperature was one knownfrom the forward runs to produce partial crystallization. Then,the temperature was raised isobarically to the final temperatureand kept steady. Temperature intervals between two reversalsare usually 10'C. For a given composition, the liquidus temper-ature is located between the temperature ofa run showing evi-dence of crystai growth (either a forward or a reversal experi-ment) and that ofa reversal run yielding only glass or showingevidence of crystal dissolution.

Analytical techniques

Glasses (starting materials and glasses produced in the exper-iments) and feldspars were analyzed using the Camebax electronmicroprobe of the University of Nancy. Analytical conditionswere acceleration voltage, l5 kV; sample current, 6 nA; countingtime on peak, 6 s. The standards involved were albite, K-feld-spar, and AlrOr, and ZAF correction procedures were used. Tominimize the loss of K and Na during glass analysis, the beamwas defocused to a size of 15 s.m (2 pmfor feldspar analyses).The resulting K.O and NarO were all slightly corrected upwardby using calibration curves constructed from glasses analyzed bywet-chemical techniques. The data in Tables 1, 5, and 6 are allaverages of multiple analyses (starting glasses: about five spotseach; experimental glasses: between two and five spots each;feldspars: between one and four crystals analyzed per sample).

The B content of the glasses was analyzed by colorimetry withcarminic acid (Pichavant, 1981) with a maximum uncertaintyof +100/0. In this paper, the BrO, concentrations in glasses aregiven on an anhydrous basis only.

Rosur-rsRun products

Phases present in the run products include glass in allruns, quartz (doubly terminated hexagonal pyramids, sizeless than 5 pm), and alkali feldspars. No other crystallinephase was encountered in the present study. The feldsparsgrow mostly as euhedral tablets (average size, 5-15 pm).

PICHAVANT: B AND H"O IN THE HAPLOGRANITE SYSTEM

l 0 5 8

TaeLe 1. Composition of the starting glasses

PICHAVANT: B AND H,O IN THE HAPLOGRANITE SYSTEM

1 312-t ' lt 0

sior-'Alr03KrONaroB.o"t

Total

QzOrAbCoKSNaS

78.08 78.0911 .32 11 .145.05 6.63.42 2.521 . 1 9 1 . 1 7

0c06 993239.7 39.330.5 39.129.6 21.4o.2 0.00.0 0.10.0 0.1

75.12 74.11't2.83 12.796.46 8.173.51 2.411.0 0.95

9as2- 9-6-frt30.6 29.639.0 49.530.3 20.90.1 0.00.0 0.00.0 0.0

77 .O1 77.4412 .18 11 .944.39 4.764.0 3.761 . 1 9 1 . 1 9

w77 9939

37.1 38.224.2 28.738.1 32.50.6 0.60.0 0.00 0 0 . 0

76.75 77.31 1 -03 10.591 .55 3 .15.65 4.484.78 4.45

99.76 99.92

39.9 41.79.6 18.9

50.4 39.20.1 0.00.0 0.10.0 0.1

79.71 78.9911 . 73 11 .41.64 3.365.68 4.611 . 1 3 1 . 1 7

es3-9 995340.9 40.09.8 20.2

48.7 39.70.6 0.10.0 0.00.0 0.0

75.57 76.38 75.1413.46 13.35 13.131.74 3.03 4.966.97 5.89 4.621 . 1 4 1 . 1 1 . O 7

96-38 es.75- gB52

30.9 29.9't8.2 30.050.5 40.00.4 0.10.0 0.00.0 0.0

79.68 79.1'l9.13 9.053 .1 3 .143.44 3.284.85 5.O2

100.2 99.6

50.1 50.819.2 19.630.5 29.30.1 0.30.0 0.00.0 0.0

29.010.560.30.10.00.0

Note: Qz: quartz, Or : orthoclase, Ab : albite, Co : corundum, KS : KrSiO3, NaS : NarSiO", n.d. : not determined.. Same glass, different batches.'* Average of five microprobe analyses for each glass, uncertainties 0.7ol" (SiO,), 1.8% (AlrO3), 3% (GO), 4.5% (NarO).f Analyzed by wet-chemical techniques (see text).

Skeletal feldspars were also encountered irrespective oftemperature and BrO3 concentration. The silicate meltalways quenched to a fluid inclusion-rich colorless glass.No evidence of liquid immiscibility was found, in agree-ment with the results of Pichavant (198 l). A vapor phaseis present in all runs as indicated by release of a smallamount of aqueous solution on opening [runs with I wto/oB2O3, ,$ro (in.) : ll, by swollen capsules, hissing whenpnnctured [runs with 4.5 wto/o B2O3, XSro (in.) : 0.6] orby the presence of numerous fluid inclusions trapped inglass for all charges. Runs with 4.5 wto/o BrO, and XFro(in.): I appeared "dry" on opening.

In the present study, no attempt was made to analyzethe vapor phase (see Pichavant, l98l). The starting Qz-Or-Ab composition is not modified by incongruent dis-solution (as indicated by the microprobe analysis of someglasses above the liquidus) because the vapor phase ispresent only in very small amounts. In spite of that, B islost in significant quantities to the vapor phase. Boratedaughter crystals appear in fluid inclusions for all exper-iments with 4.5 wto/o BrO, in the melt and XE,o (in.) : I(see also Pichavant, l98l). Mass-balance calculations ofthe distribution of B between melt and vapor phases (us-ing the average BrO, content of each of the two series ofstarting glasses, Ko : 0.33, and the proportion ofphasespresent in the charge) yield 2.9 and 12.8 wto/o B,O, (va-por), I and 4.5 wto/o BrO, (melt) for experiments withXE

" (in.) : l. These vapor-phase concentrations (0.7 and

3.7 molo/o BrOr) suggest that in experiments with -Ypro(in.) : 1, the fugacity of HrO does not significantly differfrom the fugacity of pure HrO under the same T and P(i.e., an o = l). In contrast, for experiments with .Yfiro(in.) : 0.6, the fugacity of HrO in the vapor phase isstrongly reduced (i.e., crro < l).

Analysis of the B content of an experimental glass abovethe liquidus (no. 25, T:740 'C, Table 3) yielded BrO, :4.57 wto/o, in agreement with the concentration in themelt calculated from mass balance (4.5 wto/o). Thus, theexperimental method adopted (small amount of vaporphase present in the charge and starting glasses with Bcontents exceeding the desired values) was able to suc-

cessfully control the B concentration in the experimentalmelts. The presence of CO, increases K. (the absence ofborates in fluid inclusions tells that B partitions more infavor of the melt), and B concentrations in HrO-CO, ex-periments should be slightly higher, although identicalwithin error to those in HrO-saturated experiments.

HrO-saturated liquidus phase relations

Experimental results for -$ro (in.) : I and I and 4.5wto/o BrO, in the melt are given in Tables 2 and 3. Ana-lytical data for feldspars and coexisting glasses are givenin Tables 5 and 6. Liquidus phase relations are repre-sented in Figures 1 and 2. In spite of slight nonternarybehavior (the melts have dissolved HrO and BrO.; alsothe vapor phase contains B and some dissolved silicates),a ternary isobaric representation of the phase relationshas been adopted for each ofthe two BrO, concentrationsinvestigated. Thus the diagrams represent two isobaricsections in 5-dimensional space for two BrO, contents inthe melt (l and 4.5 wto/o) and for their corresponding HrOcontents at saturation (an o = l).

Liquidus temperature brackets are never larger than 20'C except for composition no. 7 (Table 2). In the latter,the liquidus temperature is probably close to 740 "C, asit was not clear whether the crystals were growing or dis-solving in the 710 - 740 "C reversal experiment [com-pare with compositions no. 4 (liquidus temperature 750lC), no. 5 (790 "C) in Table 2 or no. 16 (730 "C) in Table31. For most compositions, in particular all those with4.5 wto/o BrO., the maximum temperature difference be-tween the appearance of crystals in the forward experi-ments and their disappearance in the reversals is 20 "rC.Compositions with I wto/o BrO, in the feldspar field ex-hibit larger discrepancies (up to a maximum of 40 'C

between forward and reversal runs, Table 2). This indi-cates a small problem of supercooling in the forward runs.Because of that problem, the liquidus temperatures forthe I wto/o BrO, compositions have been bracketed byreversals only (Table 2). The conjugation lines and three-phase triangles also point to some problems of super-

PICHAVANT: B AND H,O IN THE HAPLOGRANITE SYSTEM r059

f tate 1.-Continued

16 17 18 18' 18" 19 19. 20 21 22 23 24 25 26 27

75.99 74.56 73.6610 68 10.74 12.824.s5 6.15 1.543.49 2.27 6.854.55 4.98 4.67

00:26 967- 9e5f40.4 40 4 29.828.4 38.8 9.53't 2 20.4 60.60.0 0.4 0.00.0 0.0 0.00.0 0.0 0.1

73.32 73.96 73.3712.65 12.89 12.833.17 3.16 3.245.7 5.87 5.844.7 n.d. n.d.

gCFr s-530 9s28-

73.49 72.58 71.7 71.212.71 12.48 12.33 14.74.75 6.15 7.45 1.64.49 3.31 2.32 7.85n.d. 4.74 5.39 4.84

9E7'4 9926- gg:i-(t t-00:1331.5 31.7 20.438.5 46.9 9.929.6 20.9 69.70.4 0.5 0.00.0 0.0 0.00.0 0.0 0.0

70.75 69.38 70.514.42 14.55 14.093.08 4.75 6.087.O3 s.79 4.455.0s 4.44 4.74

T00-.95 gtg'i 9-s.-6-6

72.5112.514.534.634.89

9e-37

69.32 69.0114.34 14.098.25 9.23.33 2.434.74 4.92

T00i-T 993530.1 29.5 29.2 30.019.5 19.2 19.7 28.450.2 51.1 50.7 41.60.0 0.0 0.0 0.00.1 0.1 0.1 0.00.1 0.2 0.3 0.0

18.9 22.4 19.529.5 37.8 50.951.5 39.6 29.50.0 0.2 0.00.0 0.0 0.10.1 0.0 0.0

30.629.439.8o.20.00.0

20.518.560.40.0o.20.4

20.857.421.70.10.00.0

cooling and incomplete nucleation and gowth in the for-ward experiments: (l) the temperatures indicated by theends of the conjugation lines and three-phase triangles(mainly determined in forward experiments) are generallylower than the actual run temperatures (Figs. I atd 2);(2) the experimental melts in most cases plot close to thestarting compositions, indicating a proportion of crystalsgenerally smaller than the equilibrium proportion (Figs.I and 2). This is true also for the HrO-undersaturatedexperiments (Fig. a). Exceptions are two crystal-richcharges (nos. 17 and22, Fig. 2 and Table 6) that plot faraway from their starting compositions and are possiblyoutside the 4.5 wto/o BrO. plane.

Mass balance between reactants and products is satis-fied ifaccount is taken ofthe analytical errors on the glasscompositions: feldspars, glasses, and starting composi-tions are roughly colinear; three-phase triangles generallyenclose the starting compositions (see also the HrO-un-dersaturated experiments, Fig. 4). The compositions ofthe experimental glasses do not depart from the Qz-Or-Ab plane in any types of experiments (Table 6).

Structural formulae of the liquidus alkali feldspars (Ta-ble 5) do not reveal any reproducible anomaly that canbe attributed to the substitution of B for Al in the feld-spars (Pichavant et al., 1984, and references therein). Siis usually in slight excess, possibly because of some glasscontamination. However, alkali/Al ratios are close to one.Thus, there appears no detectable Al deficiency in thesefeldspars. However, this does not preclude that B couldbe present at the 100-1000 ppm level.

The addition of B leads to a reduction of liquidus tem-peratures for some compositions. In particular, all com-positions in the feldspar field have reduced liquidus tem-peratures for both BrO. contents. Comparing with B-freecompositions at 1 kbar (Tuttle and Bowen, 1958), reduc-tions are on the order of 30-50 "C for I wt0/0 BrO, and120-150 "C for 4.5 wto/o BrOr. In contrast, compositionsin the quartz field do not have reduced liquidus temper-atures (Fig. 3). The quartz liquidus is not significantlydifferent for B-free and for B-bearing compositions. Thiscontrasts with the increase of quartz liquidus tempera-

tures with F content, as observed in the same system(Manning, 1981, and Fig. 3).

Temperatures and compositions of the minimum liqui-dus-temperature points for I and 4.5 wto/o BrO, andXF," (in.) : 1 are given in Table 7. For both BrO, sec-tions, the quaftz-feldspar field boundaries are located fromthe experiments, and the position of the thermal minimaalong these field boundaries (minimum liquidus compo-sitions) is bracketed from three-phase triangles. Temper-atures of these minimum liquidus compositions are givenby the intersection of the quartz and feldspar liquidi (Fig.3). Compared to the B-free system at the same pressure(Tuttle and Bowen, 1958), the reductions of the mini-mum temperatures are 40 and 80 "C for I and 4.5 wto/oBrOr, respectively. The minimum liquidus temperaturefor 4.5 wto/o BrO. (640 "C, Table 7) lies below the criticaltemperature of the disordered alkali feldspar solvus (about660'C, Smith and Parsons, 1974;I-agache and Weisbrod,1977). Therefore, a change from minimum-type to eu-tectic-type phase relations must occur between the I and4.5 wto/o BrO, sections. Although not experimentally de-termined, the two feldspar-liquid-vapor equilibrium hasbeen represented in Figure 2 as a line originating fromthe minimum liquidus point up to the 660 "C isothermon the liquidus surface. Figures I and2 clearly show thatthe minimum liquidus-temperature points are progres-sively shifted toward the Ab corner with increasing BrOr.

The HrO contents of the experimental glasses have notbeen systematically determined. However, Pichavant(1981) has demonstrated an increase of the solubility ofHrO with increasing BrO, contents for compositionallyidentical melts. In this study, the absence of detectablefree aqueous solution at the opening ofthe charges with4.5 wto/o BrO, and Xfl,o (in.) : I is further indication ofhigher H,O solubilities in melts with 4.5 wto/o BrOr. Es-timations of the HrO contents of melts with I and 4.5wto/o BrO. from the data of Pichavant (198 l) yield, re-spectively, 4.5-5 and 6-6.5 Wo/o HrO (Table 7). A deter-mination of the HrO content of composition no. l8 (800.C), carried out by the weight-loss method, yielded 6. 16wto/o HrO.

r060

TneLe 2. Experimental results for 1 wtTo BrO., XErp (in.) : 1.0


Glass Tempno. fC)

Note: Qz: quartz, Sa : alkali feldspar solid solution, L : liquid, ( ) :trace amount, (diss.) : crystals dissolving, (grow.) : crystals growing.Vapor phase present in all runs.

HrO-undersaturateil liquidus phase relations

Experimental results for {" (in.) : 0.6 and 4.5 wto/oBrO. in the melt are given in Table 4. The analytical datafor feldspars and coexisting glasses are in Tables 5 and 6.Liquidus phase relations are represented in Figure 4. Thephase diagram of Figure 4 is a ternary isobaric sectionfor 4.5 wto/o BrO, in the melt and a given HrO content inthe melt. The latter is fixed by the composition of theHrO-CO, vapor phase but also by the proportion ofHrO + CO, and silicates in the charge (Wyllie and Tuttle,I 960, I 96 l; Boettcher, I 984). In this study, care was takento keep constant the proportion of HrO * CO, in thestarting materials. The HrO content of the melt was de-termined by mass balance from the amount of HrO andCO, in the vapor and melt phases before and after theexperiment. The vapor-phase composition was obtainedfrom the microthermometric study of the three-phase fluidinclusions (liquid HrO, liquid and vapor COr, Pichavantand Ramboz, 1985a, 1985b). The vapor-phase compo-sition changes from H'O : 60 t 2, CO, : 40 + 2 (startingcomposition) to HrO: 35 + 5, CO, : 62 + 5, CHo:3 + I (mol0/0, determined in glass no. 19, Z: 780 "C,Table 4). The presence of CHo indicates partial reductionof the starting H'O-CO, mixture (the f", imposed by thepressure vessel ranges between Ni-NiO and MnO-MnrOo;Montel, 1986). The obtained HrO content of the melt(assuming CO, to be insoluble in the melt) is 2.8 + 0.4wt0/0, in reasonable agreement with HrO contents calcu-lated from the Burnham (1979) model (2.0-2.4 wto/o HrO,written comm., C. Wayne Burnham and H. Nekvasil).

Liquidus temperatures have been determined for fourstarting compositions selected because they are close tothe field boundary determined under HrO-saturated con-ditions (Fig. 2). Temperature brackets ate 20'C exceptfor composition no. 21. The latter has not been properlybracketed (Table 4), and the liquidus temperature wasassumed to be close to 840 'C. Except for that composi-tion, the maximum diference in temperature between theappearance of crystals (forward experiments) and theirdisappearance (reversals) is 20 "C. Determination of theliquidus temperatures is slightly more difficult becausethe HrO-undersaturated subliquidus charges caffy asmaller percentage of crystals than in the HrO-saturatedexperiments.

Reduction of the HrO content of the melt at constantBrO, increases liquidus temperatures for the four com-positions studied. The maximum rise in temperature isfor compositions nos. 18 and 19 (about 120 "C). Figure3 shows that compositions in the quartz field are affectedas much as compositions in the feldspar field. Composi-tion no. 19 has quartz as a liquidus phase, whereas feld-spar is the liquidus phase under HrO-saturated condi-tions. However, there is no significant displacement ofthe quartz-feldspar field boundary compared to HrO-sat-urated conditions. The thermal minimum along the fieldboundary is bracketed by three-phase triangles. Its com-position (Table 7) is shifted away from the HtO-saturatedminimum toward the Qz-Or sideline. The observed shift

Duration(h)

Feldsparcomp.(mot%

Results KAIS|3OB)

780 240760 192

760-770 168'192760-780 168-168760-790 240-288

770 240760 192740 192

750-770 168-16875O-78O 240-288

760 240740 192

74O-750 168-19274O-76O 168*16874O-770 240-288

760 192740 192720 312

740-750 168-192740-760 168-168730-760 168-168

800 68780 192

780-790 96-144780-800 168-168

760 192740 192720 312

740-750 168-192740-760 168-168

740 192720 312700 312

70O-710 168-192710-740 240-288

760 240740 192

740-75O 168-192740-76O 168-168750-770 240-288750'780 168'192

800 68780 192

800-810 96-144800-820 96-72

720 168700 312680 288

680-700 336-288

680-720 336-336

720 168700 312680 288

680-700 336-288

680-720 336-336

LL + Q ZL + Qz (grow.)L + Qz (diss.)L

L + (Oz)L + Q ZL + Qz (grow.)

LL + Q ZL + Oz (grow.)L + Qz (diss.)

LL + (Oz)L + O zL + (Oz) (grow.?)L

LL + (Sa)L + (Sa) (grow.?)L

L + (Sa)L + S AL + (Sa) (grow.?)L

L + (Sa)L + S AL + Sa (grow.)L + (Sa) (grow.?)

L + S AL + Sa (grow )L + Sa (grow.)L + Sa (grow.?)L

L + (Sa)L + S AL + (Sa) (grow.?)L

LL + (Oz)L + Q z + S aL + Q z + S a

(grow.?)L + (Qz) (diss.)

LL + (Oz)L + O z + S aL + Q Z + S A

(grow.?)L + (Qz) (diss.)

JO

3C

7880

1 0

1 1

22

67


TeeLe 3. Experimental results for 4.5 wto/o BrO., X[r,o (in.): 1.0 T^BLE!-Continued

106 I

Feldsparcomp.(mol%

KAtSi306)Glass Temp

no. CC)Duration

(h) ResultsGlass Temp

no fC)Duration

(h)

Feldsparcomp.(mol%

KAtSi3Os)

81LL + Q ZL

1 2 860.840

820-860.

780760

75O-77O750'780

780760740

740-760740-77074O-780

760740

74O-750740-760

740720

720-73O720-740

720700

700-72O700-730

680660650

650- 660650-670650-680

1681 8

72-168

144288

168 -168336'31 2

1442'16264

24O-288168 -168336-31 2

288288

192-3121 68'384

168312

288-312288-312

240312

240-24O240-240

528528624

336-456336-456336-456

LL + Q ZL + Qz (diss.)L

LL + (Oz)L + Q ZL + O zL + Oz (grow.)L

LL + Q ZL + Qz (diss.?)L + (Oz)

LL + Q ZL + Qz (grow.?)L

LL + Q z + S aL + (Sa) (diss.)L

I

L + (Sa)L + Q z + S aL + Q z + S aL + ( Q z ) + S aL + ( Q z ) + S a

(diss.)

+ S a+ Sa (grow.?)

LL + S AL + Sa (diss.)

LL + S AL + Sa (diss.?)L + Sa (diss.)

LL + S AL + (Sa) (diss.)

LL + S AL + Sa (diss.)L + (Sa) (diss.)

700 456680 456

680-690 480-936680-700 672-504

740 168720 312

720-740 216-312

760 288740 288

740-750 '192-312740-760 168-384

740 288720 456

720-740 240-264

720 240700 312

7O0-71O 288-312700-720 480-21

740 360720 360

720-740 240-28872O-750 24O-240

780 360760 360

760-780 240-288760-790 336'312

800 120780 96

780-790 144-144780'800 144-96

LLLL

1 4

1 5

1 7

1 8

25 LL + S AL + (Sa) (diss.)

L + (Sa?)L + S AL + Sa (diss.)L + (Sa) (diss.)

LL + S AL + Sa (grow.?)L + (Sa) (diss.)

91

19 680660650

650-670

480 L480 L + (Sa)7 4 4 L + Q z + S a 6 6

7 4 4 - 6 9 6 L + ( Q z ? ) + S a(diss.)

is at approximately constant normative-quartz content.The temperature of the minimum liquidus composition,obtained from the intersection ofthe quartz and feldsparliquidus (Fig. 3), increases from 640 "C (HrO-saturated)to 750 "C (HrO-undersaturated, Table 7).

DrscussroN oF THE ExpERTMENTAL DATA

Evaluation of equilibrium

The question of the attainment of equilibrium duringexperimental melting or crystallization of ganitic com-positions deserves detailed discussion (e.g., Johannes,1980). In these liquidus experiments, two critical reactionmechanisms take place: (l) production of an homoge-neous hydrous melt by diffusion of HrO from the vaporphase into the initially "dry" starting glass and (2) nu-cleation and growth of crystalline phases in the hydrousmelt. In order to discuss the attainment of equilibriumin these experiments, it is important to evaluate the ki-netics of each of these two mechanisms.

With an average initial grain size of 50 pm for thestarting ground glass, times of a few tens of minutes aresumcient for HrO to diffuse throughout (D""o : 3.10 8

Note: Qz: quartz, Sa : alkali feldspar solid solution, L : liquid, ( ) :trace amount, (diss.) : crystals dissolving, (grow.) : crystals growing.Vapor phase present in all runs.

' Experiment in internally heated vessel.

cm2.s r, Shaw, 1974). Evidence of an homogeneous dis-tribution of HrO in the melt is furnished (l) by the ho-mogeneous distribution of quartz and feldspar in subli-quidus experiments and (2) by the fluid inclusions inglasses. The population of fluid inclusions is in all casesextremely homogeneous (with respect to phase assem-blage and volumetric proportions), not only for a givenexperiment, but also for a given type of experiment. Thus,dissolution and homogenization of H,O by diffusion inthe melt takes place rapidly. This is also true for the HrO-undersaturated experiments because of the use of a freefluid phase (HrO-COr) to control HrO fugacities. In com-parison, the nucleation of silicate phases requires a muchlonger period of time. For alkali feldspars, nucleation lagtimes of several days have been commonly found in thesystems Or-Ab-H,O and Or-Ab-An-Qz-HrO (Fenn, 1977;Swanson, 1977). For quartz, nucleation lag times rangingfrom several hours to a few days are reported by Swansonand Fenn (1986) in the system Ab-Qz-H,O and for theHarding pegmatite composition. It is therefore concludedthat crystallization did not start before HrO homogeni-zation was achieved by diffusion in the melt.

With the exception of the supercooling and nucleationproblems noted previously, there is overall agreement be-

1062 PICHAVANT: B AND H,O IN THE HAPLOGRANITE SYSTEM

1 kbar1 rM o/o BzOs

xF|,o"n' :1

Fig. l. Liquidus phase relations in the system Qz-Ab-Or at I kbar for I wt0/0 BrO, in the melt under HrO-saturated conditions(see text). Isotherms labeled in .C. O : compositions investigated (Table 1); I : minimum liquidus point of this system; ! :

ternary minimum in the B-free system at 0.98 kbar (Tuttle and Bowen, 1958). Data for the construction of conjugation lines andthree-phase triangles in Tables 5 and 6.

40 50 60 70 80 90

wl o/o

OrAb

tween the forward and the reversal experiments. This isstrong indication that equilibrium is approached. Super-cooling is minor for most runs; complete nucleation andgrowth is not necessary for bracketing liquidus tempera-tures, as only the direction ofreaction is required. On theother hand, superheating problems are not expected inthe reversal runs, given the small and homogeneous sizeof the near-liquidus crystalline phases in the forward ex-periments. Owing to the rapid diffusion of Na and K inthe alkali feldspars, run durations as in these experimentsare long enough to approach equilibrium alkali distribu-tion between melt and alkali feldspars (e.9., Johannes,1985). An approach toward equilibrium is suggested bythe chemical homogeneity of the run products. For a giv-en sample, microprobe data for feldspars and experimen-tal glasses are very constant (see above for the vaporphase). There is no great diference in the composition offeldspars and glasses, whether analyzed in forward or re-versal runs (Table 2). However, complete chemical equi-Iibrium is not attained, and partial equilibrium (exchangeequilibrium, Thompson and Waldbaum, 1968) must pre-vail for compositions showing the largest supercooling.The euhedral shapes ofquartz and feldspars indicates tex-tural equilibrium between melt and crystals. Further-more, the dominant crystal morphologies are consistent

with small values of undercooling (Fenn, 19771' Swansonand Fenn, 1986).

In summary, the experiments closely approached equi-librium with respect to the determination of liquidusphases and temperatures, but complete crystal + liquidchemical equilibrium is not claimed.

Individual effects of B and HrO on phaserelations in the haplogranite system

The experimental data presented above describe theeffect of two components, HrO and BrOr, on phase rela-tions in the haplogranite system Qz-Ab-Or. There is evi-dence of a complex interplay between these two compo-nents, as indicated for example by the increase of the HrOsolubility in the melt accompanying the increase in the Bcontent. Before discussing the structural role of B andHrO in aluminosilicate melts, it is necessary to identifythe individual effects ofeach ofthese components on phaserelations in the haplogranite system. In the following itis assumed that CO, does not dissolve significantly in themelt under the present experimental conditions; conse-quently no significant effect of CO, on melt structure isexpected. This first-order approximation is supported bythe composition of fluid inclusions in H,O-CO, experi-ments and by mass-balance calculations (see above).

PICHAVANT: B AND H,O IN THE HAPLOGRANITE SYSTEM 1063

1 kbar4.5 wt o/o B2Os

Xfil,etrnr = r

Fig.2. Liquidus phase relations in the system Qz-Ab-Or at I kbar for 4.5 wto/o BrO, in the melt under HrO-saturated conditions(see text). Notations, symbols, and sources of data as in Fig. 1. Note the two-feldspar field boundary drawn from the minimumliquidus point up to the 660 'C isotherm on the liquidus surface (see text).

40 50 60 70 80 90

wl o/o

Phase relations have been determined at I kbar in thesystem Qz-Ab-Or for 4.5 wto/o BrO. in the melt, bothunder H2o-saturated (Fig. 2) and HrO-undersaturatedconditions (Fig. a). The compositions of the correspond-ing minimum liquidus points are plotted in Figure 5. Anisobaric increase of the HrO content in the melt leads toa shift of the minimum liquidus composition toward theQz-Ab side, at approximately constant Qz content.Though this shift is related to a significant decrease in theliquidus temperatures (see above), it is concluded that itmainly coffesponds to the effect of isobarically increasingthe HrO content of the aluminosilicate network portionof the melt ("HrO effect"). The shift observed in theB-bearing system is similar in direction to that estimatedin the B-free system at 4 kbar (Steiner et a1., 1975), alsorepresented in Figure 5. A similar shift can be deducedfrom Whitney's (1975) 8-kbar phase diagram for his syn-thetic granite composition and from the 5-kbar fieldboundaries estimated by Huang and Wyllie (1975) in theQz-Ab-Or-HrO system. This "HrO effec!" as determinedexperimentally in the B-bearing system, conflicts withcalculations that predict a shift of the minimum liquiduscompositions toward the Qz apex (away from the feldsparjoin) with isobarically increasing HrO content of the melt(Burnham and Nekvasil, 19861 Nekvasil and Burnham.l 987) .

4 5o/6 B2O3

H2O undersaturated

u/ v "a"o"

4 . 5 %BzOs

/ Ferdspal

*\\ 4./. F

\ . ( M a n n i n g

\. ts8.)

\ z ' f

Feldspel

t5F"rd"p*

Fig. 3. Sections of the system Qz-Ab-Or-HrO from the Qzapex through the minimum liquidus compositions to the feld-sparjoin showing the effect ofB and HrO (this study), and ofF(Manning, 1981) on the quartz and feldspar liquidi. HrO-satu-rated conditions except when otherwise stated.

ouartz 55

)'^---2

1064

TeeLe 4. Experimental results for 4.5 wt7" B.O", XFH| (in.) : 0.6

Feldsparcomp.(mol%

Results KAIS|3OB)


Glass Temp. Durationno. fC) (h)

Note: Qz: quartz, Sa : alkali feldspar solid solution, L : liquid, ( ) :

trace amount, (grow.): crystals growing. Vapor phase present in all runs.

As noted previously, an increase in the B content isaccompanied by an increase in the solubility of HrO inthe melt. Consequently, the change in phase relations withthe addition of BrO, under H2o-saturated conditionsshould reflect the combined effects of B and HrO: withprogressively increasing BrOr, the HrO-saturated mini-mum liquidus compositions are shifted toward the Abcorner, with a corresponding reduction of the minimumliquidus temperatures (Fig. 5). As the isobaric solubilityof HrO is known to be much higher in borate melts thanin silicate melts (Franz and Scholze, 1963;Franz, 1966),the observed increase of the HrO content of the melt isattributed to the hydrolysis ofborate groups (see below).Considering that the borate and aluminosilicate groupsform largely separate units in the melt (see below), thehydrolysis of the borate units will not greatly affect thealuminosilicate network portion in the melt. Therefore,it is the progressive addition of B to the system that ismainly responsible for the observed change in phase re-

TABLE 5. Composition of the alkali feldspars

lations under HrO-saturated conditions at I kbar ("B ef-fect"). Comparison of the "HrO effect" and "B effect"(Fig. 5) shows that they are different.

Structural role of B in aluminosilicate melts

This section is a discussion of the structural role of Bin aluminosilicate melts from the phase relations deter-mined above and from spectroscopic information (earlystudies reviewed by Pichavant, 1983a).

There is no evidence from this study for the incorpo-ration of B into the aluminosilicate network. From thecomparison of the Raman spectra of borosilicate and alu-minosilicate glasses of feldspar composition, Konijnen-dijk (1975a) found a strong difference between Al, mainlyincorporated in the network, and B, grouped in metabo-rate units not forming part of the network (see also Pi-chavant, 1983a). Electron microscopy of similar glasses(Konijnendijk, 1975b) suggests an irregular phase-sepa-rated structure, indicating agglomeration to some extentof the metaborate groups (see Pichavant, 1983a, for adescription of the different borate units). These resultsare consistent with the recent NMR spectroscopic and ox-ide-melt solution calorimetric study performed along thejoin NaAlSi.Or-NaBSi.O, (Oestrike et al., 1986) that in-dicates the presence of clustering into regions consistingof a tetrahedral network and regions dominated by borateunits with nIB. The compositions of the liquidus alkalifeldspars analyzed in this study do not reveal the presenceof large amounts of B in the feldspar structure, whichsuggests that B atoms are not significantly incorporatedinto the aluminosilicate network.

The coordination state of B in melts and quenchedglasses obtained in this study is strongly related to thecoordination of Al. Konijnendijk (1975b) has shown thatthe addition of AlrO. to MrO-BrO.-SiO, glasses (M : Naor K) lowers the proportion of IvB present. Because ofthe higher field strength ofAl, the alkali ions previouslyused to charge-balance rvB become bonded to AlOo tetra-hedra and a proportion of the B atoms is forced to thethreefold coordination (see Pichavant, 1983a). In thisstudy, the Qz-Ab-Or compositions are metaluminous [i.e.,AlrOr/(NarO + KrO) is close to ll. Thus, if most of the

LLL + Q z + S aL

L

LL + (Qz)L + O z + ( S a )L + Q z + Q a

1 8

1 0

800 72800 168780 144

780-800 144-72

800 24800 168780 144760 144740 216

780-800 144-72

780 168760 216740 384

760-780 192-168

800 144780 216760 216

800-820 144-48

LL + S a 7 4L + Q z + S a 7 1

20

L + (Sa)L + S AL + S L + Sa (grow.)

6 7 7 I I I 1 0 1 1 1 7 1 8 1 8 1 9 1 9 2 0 2 0720 700 710 740 750 780 680 680 700 650 780 650 740 680 740

2 2 2 2 2 1 1 1 3 2 1 1 1 1 2

Glass no.:Temp ("C):n:

sio,Alro3KrONaro

Total

J I

AIKNaKAlSi306

(mol%)

68.72 66.43 66.3519.01 18.56 1 8.81

1.4 t 9.74 9.5110.71 4.98 5.19s-035 e-g7i gE56

6.03 6.02 6.01.97 1.98 2.00 . 1 6 1 . 1 3 1 . 11.82 0.87 0.91

65.07 64.79 65.1218.32 18.27 18.1213.37 13.42 15.012.47 2.22 1.22

9973 987- 9s76

67 .37 67.24 68.46 67.61 67.8318.24 18.35 18.97 19.2't 18.633.77 11 .28 0.5 0.96 1.008.55 3.72 1 0.92 10.83 10.47

vg T00-.5€ s835- 983-i 9293

66.12 68.7818.98 19 .1911 .65 3.933.87 8.67

T00-62 T00=7

5.98 6.032.02 1.981.34 0.440.68 1.48

66.41 65.5217.58 18.2513.02 12.241.96 3.23

9E3B WZ

6.1 6.021 .9 1.981.53 1.430.35 0.58

56

Structural formulae calculated on the basis of 1 6 oxygens6.0 6.01 6.02 6.07 6.05 6.04 6.0 6.0s1 .99 2.O 1 .98 1 .94 1.95 1.97 2.O1 1 .961.57 1.59 1 .77 0.43 1 .3 0.06 0.1 1 0.110.44 0.40 0.22 1 .49 0.65 1.87 1.86 1.81

7 8 8 0 8 9 2 2 6 7 3 6 6

Note: n : number of microprobe analyses; uncertainties arc 0.51" (SiOr), 1.2% (Al,Oo), 5% (Na,O and KrO).

oo


1 kbar4.5 lvt % BzOg

1065

Xfl,o tinl : o.o

Fig. 4. Liquidus phase relations in the system Qz-Ab-Or at Iconditions. Notations, symbols, and sources ofdata as in Fig. 1.

Al is in fourfold coordination, most of the B has to be inthreefold coordination. On the other hand, if AI is re-moved from the fourfold coordination, an equivalentamount of alkali ions may be used for stabilizing B infourfold coordination. There are a number of factors gov-erning the coordination of Al in aluminosilicate melts(e.g., McMillan and Piriou, 1983). In the absence of directspectroscopic information, it is difficult to evaluate theproportion of IIIB and IvB in the borate units for theseparticular compositions. The analyses of alkali feldspars

Taete 5.-Continued

20 21 21 22 23 24 25 27760 720 760 740 720 700 720 780

kbar for 4.5 vtto/o BrO, in the melt under HrO-undersaturated

indicate that nIB should dominate, as the incorporationofB in feldspar requires the presence oftetrahedrally co-ordinated B atoms (e.g., Pichavant et al., 1984, and ref-erences therein).

The changes in phase relations with the addition of Bunder HrO-saturated conditions (lowering of the liquidustemperatures for compositions in the feldspar field, shiftof the minimum liquidus compositions toward Ab) sug-gest that B may cause partial disruption of the alumino-silicate network. Although the precise nature of the struc-tural units involved is not known, this mechanism maybe schematically written as

B2O3 + NaAlSi'O' :363 + NaBO, + AI3*+ 3SiO,. (l)

BOI- is an isolated unit with B in threefold coordination;NaBO, a unit with B in fourfold coordination formingpart ofa larger borate cluster. Al3* represents an Al thathas been expelled from tetrahedral coordination and hasbecome a network-modifier. Consequently, an equivalentamount of Na is no longer required for charge balanceand may be used for stabilizing a proportion of the Batoms in fourfold coordination, as indicated in the reac-tion. This reaction, basically analogous to that presentedby Burnham and Nekvasil (1986), shows that the maineffect of B is to remove alkali cations from the alumi-nosilicate network. This leads to the partial disruption of

40 50 60 70 80 90

wt 06

66.64 64.26 65.9617.63 18.19 17.8612.06 1s.0 14.942.76 1.09 1.42

9939 s8s4- i00:TE

68.33 67.8 68.2819.2't 19.07 1 8.40.41 1.13 2.89

1 1 .39 10.73 9.499934 e&73- 99T6

65.4 64.3718.51 18.1712.37 15.02.85 0.93

g-E-Tt 9s,4-6

6.01 6.012.O 2.01 .45 1 .790.51 0.17

6 .11 - 91 .410.49

74

6.02.01 .790.2

6.05 6.01 6.01 6.071 .93 1 .99 1.99 1 .931.75 0.05 0.13 0.330.25 1 .94 1.84 1 .63

90 1 7 91

r066 PICHAVANT: B AND H,O IN THE HAPLOGRANITE SYSTEM

TneLe 6. Composition of glasses coexisting with alkali feldspars and alkali feldspars plus quartz

17 18 18 19 19 20 20700 650 780 650 740 680 740

2 3 3 3 3 2 5

7 7 8 8 I 1 0 1 1700 710 740 750 780 680 680

2 2 2 2 3 3 3

Glass no.: 6Temp (€): 720

sio, 71.99At"o. 12.29KrO 3.10NarO 5.32

Total 92.70

7'1.34 70.89 70.9912.34 12.15 '12.14

4 39 4.43 5.974.5 4.41 3.43

92.57 91.88 92.53

71.49 70.3512.27 12.086.15 7 .713.43 2.36

93.34 92.5

71.53 68.4012.45 11.723.25 4.375.5 4.16

92.73 88.65

29.6 31.0205 29.'l49.7 39.70.0 0.20.1 0.00.1 0.0

70.99 67.91 70.6912.58 11.61 12.404.61 5.51 5.954.44 3.25 3.48

@@. 6876 Vfr

72.88 72.88 67.93 69.431161 11 .77 12 .03 12 .123.81 4.38 2.28 2.824.4 ' t 3.81 5.93 5.17

v77 @fr 8&-17 6954Qz 31.5 30.6Or 19.8 28.0Ab 48.6 41.1Co O.2 0.2KS 0.0 0.0NaS 0.0 0.0

30.8 30.528.5 38.140 6 31.40.1 0.00.0 0.00.0 0.0

30.1 29.8 35.238.8 48.7 24.331 .0 21.3 40.40.0 0.0 0.30.0 0.1 0.00.0 0.1 0.0

31.6 29.936.9 38.031.2 31 .80.3 0.30.0 0.00.0 0.0

36.6 28.7 31.927 .9 1 5.0 18.634.7 56.0 48.90.8 0.0 0.60.0 0.1 0.00.0 0.2 0.0

29.729.440.60.30.00.0

Note.-Qz : quartz, Or: orthoclase, Ab: albite, Co: corundum, KS : K,S|O3, NaS : NazSiOg, n: numberof microprobe analyses; uncertaintiesare 0.5/" (SiO, and Al2O3), 2/" (K,O), 3ol" (Na"O).

the network, as a fraction of Al can no longer persist intetrahedral coordination. Reaction I consumes NaAl-SirOr, lowering the activity of the Ab-forming compo-nents in the melt. There should exist an equivalent re-action involving KAlSirOr, but less in favor of the right-hand term than Reaction l This is because the changein phase relations indicates that the Ab-forming compo-nents of the melt are preferentially consumed. In addi-tion, experimental results on the composition of aqueousphases in equilibrium with melts (Pichavant, l98l) oralkali feldspars (Pichavant, 1983b) demonstrate that so-dium borate complexes are largely dominant over potas-sium borate complexes in aqueous solutions and thuspresumably in melts.

Reaction I produces SiO, with a corresponding in-crease in the activity of silica. This may account for thepeculiar effect on quartz liquidus temperatures noted inthis study (unchanged with addition of BrOr), which con-trasts with the important reductions usually observed infeldspar-free systems (e.g., a reduction of the quartz liqui-dus temperature of 215'C with 4.5 wto/o BrO, added inthe system SiOr-BrOr; Rockett and Foster, 1965).

The borate groups as in Reaction I may be isolated orgrouped in larger units, as found in the crystalline alkaliborates (e.g., Konijnendijk, 1975b;- Pichavant, 1983a).They may also be hydrolyzed, with breaking of the B-Obonds (e.g., Franz, 1966) according to a reaction of thetype

BrO4- + H,O: 2HBOr3 . (2)

Reaction 2, written here for a borate group with two IIIB,

may also apply to borate groups with NB. In any case, itIeads to a depolymerization of the borate clusters. Re-action 2 accounts for the observed increase of the solu-bility of HrO in the melt because the isobaric solubilityof water in borate melts is higher than in silicate melts(Franz and Scholze, 19 63; Franz, 19 66).

Implications for the role of H'O inaluminosilicate melts

As noted earlier, the "B effect" differs from the "HrOeffect" at I kbar. It is worth emphasizing that the latterhas been determined by isobarically increasing the HrOcontent at constant B concentrations. Comparison ofthe"B efect" with the effect of increasing P"ro (Tuttle andBowen, 1958; Luth et al., 1964;Lurth,1976; and Fig. 5)can lead to erroneous conclusions (e.g., Manning et al.,1980; Mysen and Virgo, 1985). This is because increasingPrro simultaneously increases pressure (from 0.49 to l0kbar) and the HrO content of the melt (from 3 to l7 wto/o).Therefore, the phase relations obtained (Fig. 5) reflect thecombined effect of both variables rather than the effectof H,O alone. The individual efect of HrO at I kbar canbe deduced from Figure 5. The individual efect ofpres-sure is not well established. From the phase relationsavailable in the dry binary systems Qz-Ab and Qz-Or(Luth, 1969; Boettcher et al., 1984), one can infer a shiftof the ternary minimum toward Ab with increasing pres-sure, in agreement with calculations (Nekvasil and Burn-ham, 1987).

Although the "H,O effect" and the "B efect" are dif-ferent, the change in phase relations in both cases leads

TABLE 7. Compositions and temperatures of minimum liquidus points

Hro-saturated, Bro3-free*Hro-saturated, 1 wtTo BrOs**Hro-saturated, 4.5 wt% BrO3-'H,O-undersaturated, 4.5 wt% B,O3.'

Q z O T A b T P H . O(wt%) (wt%) (wt'/") fC) (kbaO (v,/t'l")

38 29 33 720 0.98136 27 37 680 131 23 46 640 131 35 34 750 1

4.4-4.5-6.5

2.4-i.2

. From Tuttle and Bowen (1958), Luth (1976).". Uncertainties: temperature, + 1 5'C; compositions, +1 %.

to a reduction of the liquidus volume for Ab relative toQz and Or. With the addition of F, Manning (1981) alsofound a similar contraction of the liquidus field for Ab.Therefore, this behavior appears to be a fundamentalproperty of the haplogranite system. It contrasts with thatobserved in less polymerized systems, which generallyshow an enlargement of the liquidus volume of less po-lymerized minerals with increasing HrO contents (e.g.,Kushiro, 1972; Mysen et al., 1980). On the basis of thepresent experiments, it is suggested that there are simi-larities among the structural roles of H2O, F, and B infully polymerized aluminosilicate melts.

It is generally accepted that HrO dissolves in fully po-lymerized melts by hydrolysis with bridging oxygens fromthe network and production of OH groups (e.9., Burn-ham, 1979). In addition to these OH groups, molecularHrO is also present (e.g., Ernsberger,1977; Stolper, 1982).In the Burnham model (Burnham, 1919),the interaction

to6'7

between HrO and the network leads to breaking of T4Tbonds (T : tetrahedrally coordinated cation) and ex-change of H* for Na*, with little change in the Al coor-dination. In contrast, some HrO-solubility (Oxtoby andHamilton, 1978) and Raman spectroscopic studies (My-sen et al., 1980) show that dissolution of HrO is accom-panied by expulsion of Al from the tetrahedral coordi-nation. The similarities noted here between the individualeffect of B (this study), F (Manning et al., 1980; Manning,l98l) and H,O (this study) in the Qz-Ab-Or system at Ikbar further support this hypothesis. In the presence ofF, Manning et al. (1980) and Manning (1981) interpretedthe observed changes in phase relations to indicate theformation of aluminofluoride complexes. In the presenceof B, alkali borate complexes are probably formed (seeabove). Therefore, F and B form structural units that in-volve cations previously forming part of the network. Inboth cases, Al is removed from the aluminosilicate net-work, either directly (formation of aluminofluoride com-plexes) or indirectly (formation of alkali borate com-plexes). This conclusion is supported by somespectroscopic studies (Mysen and Virgo, 1985). The for-mation of these complexes preferentially consumes theAb-forming components in the melt, thereby accountingfor the observed reductions in the liquidus volume forAb. By analogy with F and B, the interaction betweenHrO and the aluminosilicate network is considered toinvolve the formation of similar structural units, proba-bly with Na and Si. The existence of such complexes isrequired by some HrO-solubility studies (Oxtoby andHamilton, 1978). Mysen et al. (1980) presented spectro-scopic evidence for the existence of complexes betweenHrO and alkalis or Si in fully polymerized glasses. In

PICHAVANT: B AND H.O IN THE HAPLOGRANITE SYSTEM

TABLE 6.-Continued

20 21 21 22 23 24 25 27760 720 760 740 720 700 720 780

5 2 5 2 2 2 2 2

71.44 67.9112.17 11 .746.02 7 .143.44 2.27

e3-o-7 69-.06

30.5 30.738.2 47,431.3 21.60 . 0 0 30.0 0.00.0 0.0

70.75 68.4512.OO 13.517.62 2.002.39 6.56

g276 m'52

30.4 25.O47.8 13.121.5 61.30.0 0.60 2 0 00 . 1 0 0

66.74 68.2213.78 13.852.94 4.576.35 5.59

89.81 92.23

66.09 65.0513.46 13.415.82 8.774.45 2.16

Bef2- 69TT

20.5 21.237.8 58.041.4 20.40.0 0.40.1 0.00.1 0.0

20.7 20.719.3 29.759.8 50.20.2 0.00.0 0.20.0 0.2

,ilI V" +ur

Fig. 5. Projections of the minima and eutectics of the system Qz-Ab-Or for different Pr.o (O labeled with pressure, Luth, 1976),for 1 and 4.5 wto/o BrO, in the melt at I kbar under HrO-saturated (l) and undersaturated (il) conditions (this work). The 4-kbardry minimum (O) is taken from Steiner et al. (1975).

' I n2o -sa tu ra teda t

C HzO - undersaturated

o dry

l 068

contrast, Al appears to be unrelated to any hydroxyl group(Oxtoby and Hamilton,1978:- Mysen et al., 1980).

Pnrnor-ocrc APPLTCATToNS

The experimental data presented in this paper haveimportant implications for fractionation processes in fel-sic magmas at low pressures. The petrologic applicationsmay be divided into two groups. The first group is of wideimportance for felsic magmatism: it deals with the roleof water during differentiation in felsic magmas. The sec-ond group of applications more specifically concernsB-rich magmas.

The data and discussion in this paper allow an esti-mation to be made of the individual effects of pressureand of HrO content of the melt on the phase relations inthe Qz-Ab-Or system. Consequently, the efect of thesetwo independent parameters during fractional crystalli-zation of granitic magmas may be evaluated. A detaileddiscussion of case studies is unwarranted here. However,there are several possible natural situations where one ofthese parameters may be changed independently of theother. For example, HrO enrichment in the melt underroughly isobaric conditions may be a common situationin magma chambers undergoing fractional crystallizationof anhydrous phases. In contrast, a magma adiabaticallyascending from its source region may undergo a pressuredrop without any significant change of the HrO content.It is worth emphasizing, however, that the "HrO efect"determined here at I kbar may not be necessarily iden-tical at higher pressures. There are also few data concern-ing the effect of pressure. More work is needed, involvingboth experiments and calculations, in order to distinguishbetween the individual effects of pressure and HrO con-tent of the melt during differentiation of granitic magmas.

With respect to B-bearing magmas, applications con-cerning their low melting temperatures have been pre-sented elsewhere (Pichavant, 1981) and are clearlystrengthened by the liquidus data. Here, the focus is onthe effect of B on the composition of residual magmasand on some implications of melt structure.

B is present in elevated concentration in tourmalineleucogranites, aplites, and pegmatites (e.g., Pichavant andManning, 1984). In these rocks, the average B content(mainly tied up in tourmaline) is around 0.2-0.3 wto/oBrO, @ichavant and Manning, 1984). B may also be foundin elevated amounts in some race-element, Li- and F-en-riched pegmatites (e.g., Joliffet al., 1986; London, 1986).The amount of tourmaline present in the rock may notbe a valid indication of the amount of B in the melt.Tourmaline crystallization is controlled by many otherfactors than the B content of the melt, including p, T, fo,and several compositional factors (Benard et al., 1985).B may also be lost to the vapor phase because of its melVvapor partition coefficient (see above). More direct infor-mation regarding the B content of natural felsic meltsmay be obtained from the composition of some obsidianglasses. The Macusani glasses from southeastern Peru areperaluminous obsidians of granitic composition; they


contain up to 0.62 wto/o BrO, (Pichavant et al., 1987, andreferences therein) and may be useful to establish a pos-sible upper limit of BrO3 concentrations in natural felsicmagmas at around I wto/0.

The CIPW normative composition of some selectedB-rich felsic rocks is plotted in Figure 6, together withminimum liquidus compositions in the HrO-saturated Qz-Ab-Or system, with and without added B. The rocks plot-ted are characterized by magmatic tourmaline crystalli-zation. Therefore, the B enrichment is related to mag-matic processes, as opposed to hydrothermaltourmalinization (e.g., Charoy, 1982). The normativecompositions generally cluster near the B-free, low P"rominima of the Qz-Ab-Or system (see also Pichavant andManning, 1984). This is consistent with crystallization atlow pressures, under roughly HrO-saturated conditionsfor most of the rocks plotted (one Manaslu leucograniteplots close to the 3-kbar Prro minimum). There is nosystematic shift of the normative compositions towardAb, in contrast to that observed for the F-rich magmaticrocks (Kovalenko and Kovalenko, 1976; Manning, 1982Christiansen etal., 1984; Pichavant and Manning, 1984).This is consistent with the phase relations determined inthis paper: the minimum liquidus composition with Iwto/o B,O, added is not drastically shifted from the l-kbarPrro minimum (Fig. l). It is therefore concluded that theB concentration of natural magmas is too low to haveany large effect on the major-element composition of re-sidual melts.

However, B enrichment in natural magmas may be ac-companied by enrichment in F and Li, as in some rare-element pegmatites (e.g., London, 1987). The Macusaniglasses (compositionally equivalent to some rare-elementpegmatite magmas, London, 1987; Pichavant et al., 1987)are remarkable examples of this behavior: besides 0.62wto/o BrOr, they contain up to 1.3 wto/o F and 0.74 wto/oLi,O (London, 1987; Pichavant et al., 1987). The nor-mative composition of a Macusani glass (Fig. 6) reflectsthe combined efects of B, F, and Li on phase relationsand composition of residual liquids (Pichavant et al.,1987). Therefore, the effects ofcomponents such as F andLi may combine with the effect of B to affect significantlycrystal + liquid equilibria and the composition of resid-ual felsic liquids.

As detailed above, B causes a major reorganization ofmelt structure: the aluminosilicate network is partiallydisrupted with the formation of structural units that in-volve cations derived from the network; network-modi-fying cations are created in this process. The overall effectis to promote depolymerization of the melt with impor-tant consequences for its physical and chemical proper-ties. Liquidus (and solidus) temperatures are reduced.Viscosities are expected to be decreased and diffusioncoefficients to be increased. Crystal/liquid partition coef-ficients may also decrease, and the solubilities of highfield-strength cations (e.g., Sn, W, Ta, U) may increasebecause depolymerization creates structurally favorablesites in the melt. The solubility of some "refractory phas-

es" may also increase as a result of the formation of ad-ditional structural units. Therefore, the minor- and trace-element chemistry of the liquid may be profoundly mod-ified, in contrast with modest changes for the major ele-ments. A better knowledge of the structure of residualfelsic magmas is clearly necessary before attempting tomodel their chemical and physical properties.

AcrNowr,BocMENTS

David Manning is gratefully acknowledged for the experimental runs,for many discussions, and for review of the final manuscripl. Many stim-ulating discussions and a very careful review of the final manuscript byC. Wayne Burnham led to improvement of this paper. However, theinterpretations and conclusions are the author's responsibility Discus-sions and reviews of early drafts by Bemard Charoy and Mike Hendersonwere helpful. Final presentation was improved by the comments of M.Barton, D. Burt, and F. Spera. David London is acknowledged for sendingpreprints and reprints Hanna Nekvasil provided the calculations ofHrOcontents. This sludy was supported by CNRS, A.T P G6ochimie et M6-tallog6nie 1981. CRPG Contribution No. 714.

RnrrnpNcns crrrnBenard, F, Moutou, P., and Pichavant, M. (1985) Phase relations of

tourmaline leucogranites and the significance of tourmaline in silicicmagmas. Joumal of Geologt, 93, 27 l-29 |

Boettcher, AL (1984) The system SrO.-H,O-CO,: Melting, solubilitymechanisms of carbon, and liquid structure to high pressures. Ameri-can Mineralogist, 69, 823-833.

Boettcher, A, Guo, Q., Bohlen, S., and Hanson, B. (1984) Melting infeldspar-bearing systems to high pressures and the structures of alu-minosilicate liquids. Geology, 12, 202-204.

Burnham, C.W. (1979) The importance of volatile constituents. In H.S.Yoder, Jr., Ed., The evolution ofthe igneous rocks, p 439-482. Prince-ton Universitv Press. Princeton. New Jersev.

1069

71-------------O,

Bumham, C.W., and Nekvasil, H. (1986) Equilibrium properties of gran-

ite pegmatite magmas American Mineralogist, 71,239-263.Charoy, B (1982) Tourmalinization in Cornwall, England. In A.M Ev-

ans, Ed., Metallization associated with acid magmatism, p. 63-70. Wi-ley, New York

Chorlton, L.8., and Martin, R.F. (1978) The effect of boron on the granite

solidus. Canadian Mineralogist, 16, 239-244.Christiansen, E.H., Bikun, J.V., Sheridan, M.F., and Burt, D M. (1984)

Geochemical evolution oftopaz rhyolites from the Thomas Range andSpor Mountain, Utah. American Mineralogrst, 69,223-236.

Dietrich, V., and Gansser, A. (1981) The leucogranites ofthe BhutanHimalaya Schweizerische Mineralogische und Petrographische Mittei-lungen,6l , 177-2O2.

Ernsberger, F.M. (1977) Molecular water in glass. American Ceramic So-ciety Journal, 60, 9l-92.

Fenn, P.M. (1977) The nucleation and growth of alkali feldspars fromhydrous melts. Canadian Mineralogist, 15, 135-161.

Franz, H. (1966) Solubility of water vapour in alkali borate melts. Amer-ican Ceramic Society Joumal, 49, 473-477.

Franz, H., and Scholze, H (1963) Die Ltislichkeit von H'O-Dampf inglass schmelzen verschiedener Basizitat. Glastechnische Berichte, 36,347-355

Gouanvic, Y., and Gagny, C (1985) Reflexion sur I'utilisation des exp6r-imentations pour la compr6hension de la gendse des aplopegmatitesIit6es (cas de Santa Comba)-R6ponse i la discussion de M. Pichavant(1984) Bulletin de la Soci6t6 G6ologique de France, 8,273-276.

Huang, W.L., and Wyllie, P.J (1975) Melting reactions in the systemNaAlSi.Os-KAlSi3O8-SiOr to 35 kilobars, dry and with excess water.

Joumal of Geology, 83, 7 37 -7 48Johannes, W (1980) Metasrable melting in the granite system Qz-Or-Ab-

An-H,O. Contributions to Mineralogy and Petrology, 72,73-80.-(1985) The significance of experimental studies for the formation

of migmatites. In J.R Ashworth, Ed., Migmatites, p. 36-85. Blackie,Glasgow, Scotland.

Jollifl B.L., Papike, J J., and Shearer, C.K. (1986) Tourmahne as a re-corder of pegmatite evolution: Bob Ingersoll pegmatite, Black Hills,South Dakota. American Mineralogist, 71, 412-500.


30 40 50Wl o/o

Fig. 6. CIPW normative composition of selected B-rich granitic rocks, compared with HrO-saturated minima and eutectics inthe Qz-Ab-Or system (symbols as in Fig. 5). Sources of data: Tourmaline leucogranites-Dietrich and Gansser (1981), k Fort(1981), Pichavant and Manning (1984), Benard et al. (1985), Gouanvic and Gagny (1985). Tourmaline pegmatites-Manning(unpub.), Pichavant and Manning (1984). Tourmaline aplite-Nemec (1978). Macusani glass-Pichavant et al. (1987).

1070

Konijnendijk, W.L. (1975a) Structural differences between borosilicateand aluminosilicate glasses studied by Raman scattering. Glastech-nische Berichte, 48, 216-218.

- ( I 975b) The structure ofborosilicate glasses. Philips Research Re-ports Supplement l, Eindhoven.

Kovalenko, V.I., and Kovalenko, N I. (1976) Ongonites (topaz-bearingquartz keratophyres): Subvolcanic analogues of rare-metal Li-F gran-ites. Joint Soviet-Mongolian Scientific Research Geological Expedition,Nauka, Moscow

Kushiro, l. (1972) Effect of water on the composition of magmas formedat high pressures. Joumal ofPetrology, 13,3ll-334

Lagache, M., and Weisbrod, A, (1977) The system two alkali feldspars-KCI-NaCI-HrO at moderate to high temperatures and low pressuresContributions to Mineralogy and Petrology, 62,77-102.

Le Fort, P. (1981) Manaslu leucogranite: A collision signature ofthe Hi-malaya. A model for its genesis and emplacement. Journal of Geo-physical Research, 86, I 0545-1 0568.

London, D. ( I 986) Magmatic-hydrothermal transition in the Tanco rare-element pegmatite: Evidence from fluid inclusions and phase-equilib-rium experiments American Mineralogist, 7 l, 37 6-39 5.

-(1987) Internal evolution of rare-element pegmatites: Effects of F,B, Li and P. Geochimica et Cosmochimica Acta, 51, 403-420.

Lu1h, W.C. (1969) The system NaAlSi3Or-SiO, and KAlSi,Or-SiO, to 20kb and the relationships between H,O content, PH2o and P,"o, in graniticmagmas. American Journal of Science, 267 -A, 325-341 .

-(1976) Granitic rocks In D.K. Bailey and R. MacDonald, Eds.,The evolution of the crystalline rocks, p. 335-417. Academic Press,London.

Luth, W.C., Jahns, R H., and Tutlle, O.F. (1964) The granite system atpressures of 4 to l0 kilobars. Journal of Geophysical Research, 69,7 59-77 3

Manning, D.A C ( 1 98 I ) The effect of fluorine on liquidus phase relation-ships in the system Qz-Ab-Or with excess water at I kbar. Contribu-tions to Mineralogy and Petrology , 7 6, 206-215.

- (l 982) An experimental study ofthe effects of fluorine on the crys-tallization of granitic melts In A M. Eyans, Ed., Metallization associ-ated with acid magmatism, p. l9l-203 Wiley, New York.

Manning, D.A.C, Hamilton, D.L., Henderson, C.M.B., and Dempsey,M.J. (1980) The probable occurrence of interstitial Al in hydrous,F-bearing and F-free aluminosilicate melts Contributions to Mineral-ogy and Petrology, 7 5, 25'l-262

McMillan, P, and Piriou, B (1983) Raman spectroscopic studies of sili-cate and related glass structure: A review. Bulletin de Min6ralogie, 106,57-7 5

Montel, J.-M. (1986) Experimental determination of the solubility of Ce-monazite in SiO,-AlrOr-KrO-NarO melts at 800.C, 2 kbar, under HrO-saturated conditions Geology, 14, 659-662

Mysen, B.O, and Virgo, D. (1985) Structure and properties offluorine-bearing aluminosilicate melts: The system NarO-AlrOaSiO,-F at I atmContributions to Mineralogy and Petrology, 91, 205-220.

Mysen, B O., Virgo, D., Harison, W J , and Scarfe, C.M. (1980) Solubil-ity mechanisms of H,O in silicate melts at high pressures and temper-atures: A Raman spectroscopic study. American Mineralogist, 65, 900-914 .

Nekvasil, H., and Burnham, C.W. (1987) The calculated individual effectsofpressure and water content on phase equilibria in the granite systemIn B.O. Mysen, Ed , Magmatic processes: Physiochemical pnnciples,Geochemical Society Special Publication 1, 433-445.

Nemec, D. (1978) Genesis of aplite in the Ricany massif, central Bohemia.Neues Jahrbuch fiir Mineralogie Abhandlungen, 132, 322-339.

Oestrike, R., Geisinger, K., Navrotsky, A., Turner, G.L., and Kirkpatrick,R.J. (1986) Stmcture and thermochemistry of glasses along the joinNaAlSi,Or-NaBSi,O,: The effect of boron. Geological Society of Amer-ica Abstracts with Programs, 18, 709.

Oxtoby, S., and Hamilton, D.L. (1978) The discrete association of warerwith NarO and SiO, in NaAl silicate melts Contributions to Mineral-ogy and Petrology, 66, 185-188


Pichavant, M. (1981) An experimental study ofthe effect ofboron on awater-saturated haplogranite at I kbar pressure: Geological applica-tions Contributions to Mineralogy and Petrology, 76,430-439.

-(1983a) Melt-fluid interaction deduced from studies of silicate-BrOr-HrO systems at I kbar Bulletin de Min6ralogie, 106,201-211.

-(1983b) (Na,K) exchange between alkali feldspars and aqueoussolutions containing borate and fluoride anions; experimental resultsat P : I kbar (abs.). 3rd NATO Advanced Study Institute on Feldspars,Feldspathoids and their Parageneses.

Pichavant, M., and Manning, D.A.C. (1984) Petrogenesis of tourmalinegranites and topaz granites: the contribution of experimental data.Physics ofthe Earth and Planetary Interiors, 35, 31-50

Pichavant, M., and Ramboz, C. (1985a) Liquidus phase relationships inthe system Qz-Ab-Or-BrOyHrO at I kbar under Hro-undersaturatedconditions and the effect of H,O on phase relations in the haploganitesystem (abs.). Terra Cognita, 5, 230

-(i985b) Premidre d6termination exp6rimentale des relations dephases dans le systbme haplogranitique en conditions de sous-satura-tion en HrO. Comptes Rendus de I'Acad6mie des Sciences, 301, 607-6 1 0 .

Pichavant, M, Schnapper, D, and Brown, W.L. (1984) Al=B substitu-tion in alkali feldspars; preliminary hydrothermal data in the systemNaAlSi,O.-NaBSi 0, Bulletin de Min6ralogie, lO7, 529-537.

Pichavant, M., Valencia Herrera, J., Boulmier, S., Briqueu, L., Joron, J.L.,Juteau, M., Marin, L., Michard, A., Sheppard, S M.F., Treuil, M., andVernet, M (1987) The Macusani glasses, SE Peru: Evidence ofchem-ical fractionation in peraluminous magmas. In B.O. Mysen, Ed., Mag-matic processes: Physicochemical principles Geochemical Society Spe-cia l Publ icat ion l , 359-373.

Rockett, T.J , and Foster, W.R. (1965) Phase relations in the system SiOlB,O.. American Ceramic Society Journal, 48, 75-80

Shaw, H.R (1974) Diffusion of HrO in granitic liquid. I-Experimentaldata. II-Mass lransfer in magma chambers. In A W. Hofmann, B.J.Giletti, H.S. Yoder, Jr., and R.A. Yund, Eds., Geochemical transportand kinetics. Camegie Institution of Washington Publication 634, pI 39- I 70.

Smith, P, and Parsons, I. (1974) The alkali leldspar solvus at 1 kbarwater vapour pressure. Mineralogical Magazine, 39,7 47-7 67 .

Steiner, J.C., Jahns, R.H., and Luth, W.C. (,l975) Crystallization of alkalifeldspar and quartz in tbe haplogranite system NaAlSirO8-KAlSijO8-SiOr-HrO at 4 kb Geological Society of America Bulletin, 86, 83-98.

Stolper, E. (1982) Water in silicate glasses: An infrared spectroscopic study.Contributions to Mineralogy and Petrology, 81, l-17.

Swanson, S.E (1977) Relation of nucleation and crystal-growth rate tothe development of granitic textures. Amencan Mineralogist, 62,966-978.

Swanson, S E., and Fenn, P M (1986) Quartz crystallization in igneousrocks. American Mineralogist, 7 1, 331-342.

Thompson, J B , Jr , and Waldbaum, D R (1968) Mixing properties ofsanidine crystalline solutions. I-Calculations based on ion exchangedata. American Mineralogist, 53, 1965-1999.

Tuttle, O.F, and Bowen, N.L. (1958) Origin of granite in the light ofexperimental studies in the system NaAlSirOr-KAlSi3O8-SiOr-HrOGeological Society of America Memoir 74

Whitney, J A. (1975) The effects of pressure, temperature and X(HrO) onphase assemblage in four synthetic rock compositions Joumal of Ge-ology, 83, l -31

Wyllie, P.J., and Tuttle, O.F. (1960) Experimental investigation of silicatesystems containing two volatile components: I-Geometrical consid-erations American Joumal of Science. 258, 497 -5 11.

-(1961) Experimental investigation of silicate systems containingtwo volatile components: II-The effects of NH, and HF, in additionto H2O, on the melting temperatures of albite and granite AmericanJournal of Science, 259, 128-143.

M,q.Nuscprm RECEIVED MatcH 19, 1987MeNr;scnrpr AccEprED Jum 26. 1987

effects of b and hro on liquidus phase relations in the ... · pdf fileeffects of b and hro on...

Documents