Melting of a dry peridotite at high pressures and basalt ... · Melting of a dry peridotite at high pressures and basalt magma genesis ErIcst Tera.uesul Institute for Thermal Spring

American Mineralogist, Volume 68, pages 859-879, 1983

Melting of a dry peridotite at high pressures and basalt magma genesis

ErIcst Tera.uesul

Institute for Thermal Spring ResearchOkayama University, Misasa, Tottori-ken, 682-02, Japan

eNo kuo Kusnrno

Geological Institute, Faculty of ScienceTolcyo University, Bunlcyo-ku, Tolcyo, 113, Japan

Abstract

The solidus of a spinel lherzolite (HK66) was determined under dry conditions throughthe pressure range I atm to 30 kbar. It was found that the solidus comprises three curvescorresponding to subsolidus mineral assemblages with cusps at about ll and 26 kbar. Inorder to determine the composition of melt coexisting with peridotite, a thin layer of basaltwas sandwiched between compressed blocks of powdered peridotite minerals and then wasequilibrated with its host at melting temperatures. A time study at 15 kbar and 1300"C (25"Cabove the solidus) showed that the basalt melt embedded in the peridotite was completelyhomogenized with the partial melt in the peridotite matrix within 24 hours. The composi-tion of melt formed along the solidus of the peridotite is a quartz tholeiite at 5 kbar, anolivine tholeiite similar to mid-ocean ridge basalt (ruronn) at 8 and 10 kbar, an alkali-olivinebasalt between 15 and 20 kbar, and an alkali picrite above 25 kbar. The role of K2O in themelting was specifically investigated by adding 20 wt.% potassium feldspar to the startingperidotite. The solidus of the K-rich peridotite is 70o and 150'C lower than that of theoriginal peridotite at 20 and 30 kbar, respectively. Its extension may intersect with thenormal oceanic geotherm at a pressure between 30 and 50 kbar. The melts are leucitephonolitic at the solidus of the K-rich peridotite and change gradually to the composition ofthe alkali picrite at the solidus of the original peridotite. A hypothesis of shallow depthorigin for MoRBs is supported by the fact that the composition of melt formed near thesolidus at 8 kbar (16 wt.% Al2O3, I I wt.% CaO, 9 wt.% MeO) plots in the middle of theIuoRn's cluster in the normative projections. The olivine-liquid Fe/Mg partition coefficient,Kp : (Fe/Mg)or/(Fe/Mg;tio is given by Ko = 0.30 + 0.002 P (kbar) in the pressure range Iatm to 35 kbar. Judging from the Kp values, most previously reported glass compositions inperidotite melting experiments are significantly modified by the overgrowth effect ofresidual solids.

Introduction

It is a well-accepted hypothesis that terrestrial basaltmagmas are formed by partial melting of peridotite in theupper mantle (Bowen, 1928; Ringwood, 1975; Yoder,1976). Both depth and degree of partial melting arebelieved to be important in producing the diversity ofbasalt magma types (Kushiro and Kuno, 1963; Green andRingwood, 1967a; O'Hara, 1965, 1968; Ito and Kennedy,1968). Numerous experiments have been attempted todetermine the compositions of liquids produced by partialmelting of peridotite under various pressure-temperatureconditions. There has been a debate, however, amongexperimental petrologists on the method of determinationof the liquid compositions. Kushiro et al. (1972\, Kushiro

(1973a, b), Mysen and Boettcher (1975) and Mysen andKushiro (1977) reported EIMA (Electron Probe X-rayMicro-Analyzer) analyses of quenched glasses as repre-senting equilibrium liquid compositions. Green (1973) andJaques and Green (1979, 1980), on the other hand, pointedout that the glass compositions are significantly modifiedby overgrowth ofadjacent solids during quenching. Theyestimated equilibrium liquid compositions by mass bal-ance calculations based on compositions and modalamounts of solid residues. However, estimation by thismethod also includes considerable uncertainties becauseof compositional zoning in minerals and uncertainties inthe estimation of their modal amounts.

In this paper we present new experimental results onthe anhydrous melting of a spinel Iherzolite from Salt

0003-004x/83/09 I 0-0859$02. 00

860 TAKAHASHI AND KUSHIRO: PERIDOTITE MELTING

Lake Crater, Hawaii (HK66101703, designated HK66hereafter). The melting behavior of the rock was original-ly studied by Kushiro et al. (1968) both under dry and wetconditions. In order to overcome the quench modificationof liquid compositions, a special experimental techniquewas employed in the present experiments, whereby a thinlayer of basalt powder is sandwiched between com-pressed peridotite minerals and then is equilibrated withits host at melting temperatures (basalt/peridotite sand-wich technique, see Fig. 3). This technique is essentiallythe same as the mineral capsule technique developed byWalker et al. (1979) for one atmosphere experiments andwhich has been used by Stolper (1980) and Takahashi(1980a, c) at high pressures. Melt compositions formednear the solidus of HK66 were determined with the abovetechnique.

It was concluded that most of the previously reporteddirect spl,re glass analyses are in error judging from theFe-Mg partition coefficient between olivine and liquidobtained in our experiments. A general set of isobaricliquid compositional trends for peridotite partial meltingunder dry conditions was obtained in crpw normativeprojections, and the origin of mid-ocean ridge basalts wasdiscussed on the basis of the liquid compositional trend.

Difficulties of peridotite melting experiments

There are at least three major difficulties in conductingmelting experiments of natural peridotite at high pres-sures and determining equilibrium liquid compositions.They are (l) inevitable chemical reactions between theperidotite partial melt and the container, (2) modificationof the partial melt compositions during quenching, (3)

slow reaction rate of residual solids with melt.

Capsule materials

Noble metals such as Pt and Ag-Pd alloy have beenused most frequently as capsule materials in peridotitemelting experiments (e.9., Mysen and Boettcher, 1975;Jaques and Green, 1980). It is well known, however, thatFe and Ni alloy with those noble metals at high tempera-tures, which results in strong compositional zoning insolids (e.9., Jaques and Green, 1980, p. 289). Uponformation of the alloy, excess oxygen is produced andpart of the FeO in the melt is oxidized to FeOl s @.9.,Yoder and Tilley, 1962, p. 391). At 1400"C and 20 kbar,the mole ratio FeO1.5/FeO of a basalt melt embedded inFoes olivine in sealed Pt capsule was found to havechanged from 0.12 (l hr) to 0.50 (48 hr) (Takahashi,1980a). Thus, oxygen fugacities of the charges in thenoble metal capsules change as a function ofrun durationand are more oxidizing than previously expected.

Graphite was found to be the most suitable capsulematerial for the present experiments for the followingreasons. According to Thompson and Kushiro (1972), theoxygen fugacity of the experimental charge in a graphitecapsule is within the stability field of wtistite in the

pressure range between l0 and 20 kbar, and temperaturesbetween 1200" and 1500'C. Precipitation of nickel metalexsolved from olivine was observed by the present au-thors in the melting experiments of peridotite at allpressures and temperatures. However, the amount ofiron alloying with the nickel metal becomes significantonly at pressures less than l0 kbar, hence the Mg/(Mg +Fe) ratios of solids were constant as a function of rundurations at least up to 60 hr.

It is of considerable importance that the amount ofFe3* in a basalt melt is negligibly small in a graphitecapsule (less than the detection limit of 57Fe Mrissbauerspectroscopy in a number of basaltic glasses; that is lessthan about 5% of total iron: Takahashi, unpublisheddata). Therefore, the Fe2*-Mg partition coefficient be-tween coexisting melt and solids can readily be calculatedfrom the erue analyses.

Quench problem

Results ofline traverse analyses ofolivine and adjacentquenched liquids are shown in Figure 1. The composi-tions of the glass quenched at 15 kbar (Fie. lB) aresignificantly modified at the contact with olivine. Becausethe run product was quenched after a long run duration(61 hr), it is not likely that the zoning represents a growtheffect of the olivine throughout the run time. Alternative-ly, the zoning is considered to have been formed byovergrowth of the olivine during quenching (quenchingrate: 200"C/sec to 400'C/sec). The quench modification ofthe partial melt is more serious as temperature andpressure increase because the melt becomes more picriticat high pressures and temperatures. The extent of thezoning in the quenched liquid also depends on the volumeproportion of the liquid to solids (the smaller the amountof liquid the greater the extent of zoning). The partial meltis virtually unquenchable at pressures above about 30kbar where acicular or dendritic quench crystals areformed instead of quench glass (Fig. 1A). Similar obser-vations have been made by Kushiro (1974\.

Jaques and Green (1979) focussed on this problem inmore detail and concluded that equilibrium compositionof the peridotite partial melt can not be determined bydirect epun analysis of the quenched glass at any pres-sure and temperature conditions. Accordingly, they esti-mated equilibrium liquid compositions by mass balancecalculations. Precision of such calculations, however,may not be sufficient because small error in the estima-tion of modal proportion significantly affects composi-tions of small amounts of melts (Mysen and Boettcher,1975, p. 576-578). In addition, the chemically zonedcrystals in their charges (Jaques and Green, 1980, p.294-298) made it impossible to evaluate standard errors intheir estimation. Therefore, the special experimentaltechnique (basalt/peridotite sandwich technique) was em-ployed by the present authors. Because the glass diskinserted between two peridotite layers (Fig. 3) contains avery small amount of crystals, the quench overgrowth

TAKAHASHI AND KUSHIRO: PERIDOTITE MELTING E6l

A B c

q u e n c h c r y s l o l s I

o l i v i n e s t o s s I o r i v i n e s r o r , I

o l i v i n e

3okb. t51ob, KRg/HK66 15kb, BOOT. U'/HK66 'kb, 12252, KRB/HK66

Fig. l. Line traverse analysis of olivines and adjacent quenched liquids obtained at basalt melt-peridotite boundaries using thebasalVperidotite sandwich technique (c/., Fig. 3). The effect of crystal overgrowth during quenching on compositions of adjacentliquids becomes serious as pressure increases. The glass in Fig, I B shows a variation in MgO from I L0 wt.Vo (left) to 7 .0 wt.Vo(contact). (A, Run #4E;8, Run #40; C, Run #30).

effect of solids on liquid compositions was limited only tothe boundaries with the peridotite (Fig. l). Liquid compo-sitions equilibrated with the peridotite were determinedwith EIMA by analyzing the center of the glass disk.

Slow reaction rates in solids

This problem has been considered very little by previ-ous workers. In Figure 2, an example of relict pyroxenecrystals is shown after a run at 8 kbar and 1200"C for 40hr. The orthopyroxene retains the composition of thestarting material in the core and is surrounded by newlyformed alumina-poor orthopyroxene in the rim. At theboundary between the two pyroxenes, the Ca and Alcontents show an abrupt change, irfdicating that thoseelements were not transferred through the boundary. TheFe and Mg, on the other hand, appear to have exchangedto some extent, judging from the zoning pattern.

Interpretation of the above observation is as follows.Upon melting of a peridotite whose mineral compositionsinherit equilibration at different pressure-temperatureconditions (this is always the case in experimental perido-tite melting), melting is initiated by melting of the meta-stable solids. Chemical reaction between the residues andthe partial melt takes place subsequently, but the rate ofthe reaction is limited by the diffusion process in thesolids, so that the relict compositions remain for a longtime. The run durations of our experiments were intendedto be as long as possible. Achievement of chemical

equilibrium of the run products was examined by timestudies and extensive chemical analyses with epue suchas those by Takahashi (1980b).

Experimental and analytical procedures

Most of the experiments were made with three piston-cylinder apparatus at the Geophysical Laboratory, Car-negie Institution of Washington (Boyd and England'1960). Some of them were made with a piston-cylinder atTokyo University. The piston-out technique was em-ployed. The % inch diameter, talc-pyrex assembly wasused with a graphite heater and sintered alumina as aninner spacer. In order to run more than two chargessimultaneously, the 3h inch diameter assembly with atapered graphite heater and sintered MgO spacer (Ku-shiro, 1976) was used for some runs at pressures less than20 kbar.

Temperatures were measured with ftiPtgtrRhlg ther-mocouples in runs below 1450'C. At higher temperatures,We5-Re5/W75-Re25 thermocouples were used' The twothermocouples were calibrated against each other at a lowoxygen fugacity in a one-atmosphere furnace. The uncer-tainty in their comparison is less than 2"C at 1400-1500'C.No pressure correction was made on the emf of thethermocouple. Run temperatures were controlled to with-in t3'C with a silicon-controlled rectifier. Due to thethermal gradient in the furnace assembly, however, un-certainties in temperature measurements may be greater

862

0.8 6

0 .8 4

TAKAHASHI AND KUSHIRO: PERIDOTITE MELTING

"-.-";;-,"";;i-o-o_o

' -w-Fig. 2. Line traverse analysis ofan orthopyroxene with a relict

crystal in the core (Run #28, KRB/HK66). See Table 4 for fullchemical analysis of the core and the rim.

than l0'C in runs with the Vz inch assembly. In runs withthe % inch tapered furnace assembly, however, tempera-ture variation among the charges run simultaneously isbelieved to be less than lO'C (Kushiro, 1976).

peridoti te

Fig. 3. A schematic iffft*tion showing cross section of theexperimental charge used in order to determine equilibriumliquid compositions (Runs #26-51, 60-64). A thin layer ofpowdered basalt is sandwiched between compressed blocks ofpowdered peridotite minerals.

In experiments for solidus determination (Runs #1-25),a l0 mg charge of powdered peridotite (average grain sizeless than l0 g.m; see Table I for chemical composition)was encased in a graphite capsule (2.5 mm O.D.), whichwas then sealed in a Pt-capsule (3 mm O.D.). In experi-ments for liquid composition determination (Run #26-51),a I mg disk of basalt powder (less than 10 pm in grainsize, see Table I for chemical compositions) was embed-ded in about 10 mg of the peridotite powder which wasthen encased in the above double capsule (Fig. 3). Thebasalt powder melted at the beginning of the experimentsand subsequently mixed with the partial melt formedalong the grain boundaries in the peridotite matrix. Inexperiments along the solidus, the basalt melt stayed inits original position and was quenched as a coherent glassdisk (300-500 pm in thickness) even after more thanseveral days. At temperatures more than 100'C above thesolidus, however, the basalt melt tends to disperse in theperidotite matrix. On the other hand, segregation of the

Peridotite HK66 Basalts

L45ft K2Itt mEtt BOM*cPx' sp'

s i o 2 4 a , o 2 3 a . 7 6

T i O 2 O . 2 2 0 . 0 0

A I z O : 4 . e 8 0 . O 0

F e o " 9 . 9 0 I 4 . 4 I

h o 0 . 1 4 0 . I 7

M g o 3 2 . 3 5 4 5 . 8 9

C a o 2 . 9 ' 7 0 . 0 7

N a 2 O 0 , 6 5 0 . 0 O

" 2 O 0 . 0 7 n . d .

" 2 O 5 O . O 7 n . d .

C r Z o : 0 . 2 5 0 . O 0

N i o n . d . 0 . 3 5

o . l 2

o . 2 L

1 5 . 3 4

0 . 1 3

1 9 . 5 3

0 . 0 4

o . o 0

n . d .

n . d ,

2 . 6 6

o . 4 5

99. '79

5 3 . 5 0 5 0 . 9 3

0 . 1 1 0 . 5 5

8 . 7 1 4 . 3 0

0 . 1 3 0 . 0 9

3 1 . 1 3 7 4 . 6 2

0 . 7 1 1 8 . 7 5

0 . 1 3 I . 9 6

n . d . n . d .

n . d . n . d .

o . t 6 0 . 5 0

n . d . n . d .

9 9 . 8 5 9 9 . 0 5

4 1 . A 4 9 . 5

2 . 4 L . 7

1 7 . 3 1 5 . 1

9 . 3 1 1 . 4

o . 2 0 . 2

8 . 8 8 . 5

8 . 1 1 0 . 6

1 ? ) 1

r . t 0 . 1

o . 1 0 . 2

n . d . n . d .

n . d . n . d .

1 0 0 . 0 1 0 0 . 0

4 9 . 9 4 6 . O 5 7 . 6

2 . 9 2 . O 0 . 3

1 2 . 9 8 . 2 1 5 . 8

r r . 2 1 3 . 4 7 . 2

o . 2 0 . 2 0 . 2

9 . 2 2 7 . O 8 . 5

I 1 . 0 7 . 2 7 . A

2 . 2 7 . 1 1 . 9

0 . 5 0 . 3 0 . 7

n . d . n . d . n . d -

n . d . n . d . n . d .

n . d . n . d . n . d .

Table 1. Compositions of the starting materials

*, EPW anaTqsis ( this studg); l , Ki lshiro et a1.(7968)t +t ' Iakahashi (1978);t , , shi taki and Kutoda (1977) .

t 0 0 , 0

TAKAHASHI AND KUSHIRO: PERIDOTITE MELTING 863

partial melt (mixed with the embedded basalt melt) fromresidual solids was observed in those high temperatureruns.

To drive out moisture from the charge, all the startingmaterials were dried at I100"C and QFM bufer in a one-atmosphere quenching furnace using the method ofPres-nall and Brenner (1974) in order to minimize iron-loss,and then were quenched on a cold metal plate in areduced gas mixture of CO and CO2. The sample assem-bly was stored in an oven at llO'C immediately prior towelding the Pt capsule.

After they were quenched at high pressures, thecharges were sectioned parallel to the length of thefurnace assembly, and phases were identified with areflected light microscope. Run products of subsolidusconditions were examined also with the X-ray difractom-eter. Chemical analyses were made with an automatedelectron microprobe at the Geophysical Laboratory witha specimen current of 0.02 p,A, 2-3 g,m beam diameter,and 20-40 seconds counting time for each element.Additional chemical analyses were made with an electronmicroprobe (JXA-SA) at the Institute for Thermal SpringResearch. For the glass analyses, a large defocussedelectron beam (30-50 pm diameter) was employed inorder to avoid volatilization ofalkalis during the analysis.The analyses of the glasses show no systematic deficien-cy in total (see Table 3), thus it is believed that volatiles(H2O, CO2) were not significantly dissolved in the melt.

Chemical equilibrium in the basalt/peridotitesandwich experiment

Ranges of cationic diffusion coefficients in minerals andmelt relevant to the present experimeirts are summarizedin Figure 4. Characteristic difusion distances (X = VDt,where D represents a cation diffusion coefficient and t is arun duration) are shown as a function of run duration. Thetimes required for solids to achieve chemical equilibriumby diffusion were calculated on the basis of a simplemodel with a spherical boundary condition (t : 0.4 (r2lD),where r denotes radius ofa sphere, Crank, 1956, Fig. 6-l).The typical grain size of residual solids in the quenchedcharges is 10-20 pm, hence it is believed that the Fe-Mgexchange equilibrium between olivine and liquid wasachieved at least in the long runs. Cores of pyroxenecrystals greater than l0 ptm in diameter probably were outof equilibrium even after more than a week if soliddiffusion is the reaction process (cf. Fig.2). Characteris-tic diffusion distances for basaltic melts are more than Imm (Fig. 4), hence, it is likely that homogenization ofbasalt melt with the partial melt of the peridotite wascompleted during the first several hours in the basalt/peridotite sandwich experiments (Fig. 3).

To test the above considerations, a time study wasmade at 15 kbar and 13fi)'C (Runs #38-40). Chargesconsisting of different basaltic rocks sandwiched in theperidotite HK66 were run simultaneously in the large

l y r

I month

lweek

I doy

l h r

l min

Ilog D (cny's)

_ t _6 _g _10 _12 -1 t _16

bosolr metr i i i_"i* _-i,; i ;;_

Fig. 4. Time vs. chemical diffusion relationships for theperidotite melting experiments. Ditrusion coefficients from thefollowing sources: Na, Ca, Sr, Ba in basalt melts, Hofmann(1980) and Watson and Bender (1980); Fe-Mg in olivine, Misener(1974); Ni, Mn, Ca and Mg in olivine, Morioka (1981); Ca-Me inpyroxenes, Takahashi (1980b) and McCallister (1980). Solid linesare difusion distances 1r6il as a function of run duration anddiffusion coefficient. Broken lines indicate approxirnate rangesfor the present experiments.

capacity 3/q inch assembly. Compositions of the basaltmelts initially were greatly diferent (quartz-normative tonepheline-normative), but became very similar after a dayand stayed constant for another 1.5 days (Fig. 5). Nocompositional heterogeneity was observed in the liquiddisk except at the boundary between peridotite and meltwhere quench modification of liquid compositions hadtaken place (Fig. l). Similarities of the final liquid compo-sitions are well illustrated in a crpw normative plot (Fig.6), where compositions of the starting materials areshown for comparison. Also shown in the figure arecompositions of liquids equilibrated with the peridotite atanother pressure and temperature (8 kbar and 1225'C).

The fact that liquids initially formed from variousstarting materials converged to very similar compositionsindicates at least two things: (1) homogenization of thepartial melt in peridotite and the embedded basalt meltwas completed within a day; and (2) compositions ofmelts coexisting with the olivine + orthopyroxene +clinopyroxene buffer at given pressure-temperature con-ditions have only limited ranges of variation.

Minor but important diferences in K2O and SiO2 areseen among the liquid compositions. The liquids whichwere initially alkali-olivine basalt (KRB) show higherK2O and SiO2 contents than liquids which were otherbasalt compositions (Table 3, Fig. 5). The difference in K2O

log

/se

8U TAKAHASHI AND KUSHIRO: PERIDOTITE MELTING

w T %o : K R B o ; X 2 1 O : 1 4 5 . : B O N

2

I

M 9 o

F e 0

48 rns 48 tns

Fig. 5. Compositions of melts sandwiched between partially molten peridotite as afunction of time at l5 kbarand 1300"C (Runs

#38-40). Regardless of the starting basalts, the melts are almost identical after one day and remain constant for another 1.5 days.

t 0

l 2

t 0

2424

o K R B , A 1 1 5 , t r K 2 1 , O B O N , v O C E

Fig. 6. Normative compositions of the basalt starting materialslisted in Table I (open symbols); melts equilibrated with theperidotite at 8 kbar and 1225"C (half solid symbols); and those at15 kbar and 1300'C (solid symbols). Regardless of the startingmaterials, melts equilibrated with the peridotite at a givenpressure-temperature condition show similar normativecompositions. After projection scheme described by lrvine andBaragar (1971) based on cation mole percent norms.

is considered to be a signature of the starting material,inasmuch as the peridotite is virtually potassium free.Because there are no peridotite constituent minerals thatincorporate this element in solid-solution, KzO in the meltcan not be buffered but depends only on the volumeproportion of the liquid in the charge. Observed increasein SiOz with K2O accords with the expected shift of theolivine-orthopyroxene liquidus boundary noted by Ku-shiro (1975). Thus, minor differences among the liquidcompositions at a given pressure-temperature conditionreflect the complex variation of chemical equilibria in amulti-component system.

The compositions of solid phases formed during thetime studies were examined with Epue. Cores and rims ofthree or more crystals in the glass disk were analyzed foreach mineral. The minerals in the peridotite matrix wereanalyzed with a grid-mode automated stage motion (3 or 5pm interval) for more than 2fi) points in each sample. Thelatter data included some analyses in which the beamoverlapped more than two minerals and/or liquid. Datacharacterized by abnormal stoichiometry or high K2Ocontent were discarded.

The Me/(Mg + Fe) ratio of olivine shows a bimodaldistribution in short runs because of the higher Me/(Mg +

Fe) ratio of the embedded basalts compared with theperidotite (see Table l). The Mg/(Mg + Fe) of olivine in

Kzo

Noro

-- - -------o

ne '


long runs, however, is limited to a narrow range and thereis no difference between the olivine in the liquid disk andthat in the peridotite matrix (Fig. 7). The Ca content ofpyroxenes, on the other hand, shows a large variationeven in long runs. A large number of relict pyroxenes(both orthopyroxene and clinopyroxene) were found inthe runs particularly at lower pressures (Fig. 2). It shouldbe emphasized that the occurrence of the pyroxene relictsis not limited to the basalt/peridotite sandwich experi-ments but is common in peridotite melting experiments atrelatively low pressure and temperature conditions.

Tsuchiyama and Takahashi (in press) have studiedkinetics of partial melting of a plagioclase feldspar (Anoo)at one atmosphere under dry conditions. According tothem, the melting process is rate-controlled by chemicaldiffusion in the plagioclase and it took several hundredhours even at 14fi)"C before the solid achieved finalchemical equilibrium with the partial melt. Nevertheless,surface chemical equilibrium between the partial melt andthe residue was established and the chemical compositionof the melt, which is identical to the equilibrium liquidcomposition determined by Bowen (1913), was un-changed during the experiment. It is unlikely, therefore,that the liquid compositions are largely out of equilibriumin the peridotite melting experiments even in the chargeswith a great amount of relict crystals.

In summary, it is shown that chemical equilibriumbetween the melt (including the basalt melt and theperidotite partial melt) and olivine is achieved throughoutthe experimental charges at least in the runs more than aday long. Only the rims of pyroxenes are likely to havecompositions in equilibrium with the melt. Core composi-tions of the pyroxenes are nearly isolated from the systemand react slowly.

Solidus

The solidus of the periodotite HK66 was originallystudied by Kushiro et al. (1968) with a tetrahedralanviltype, solid-media high pressure apparatus, but there werelarge uncertainties in the pressure calibration below 30kbar (Akimoto, personal communication, 1980). Further-more, the uncertainty in temperatures may be of the orderof 100'C because of a steep thermal gradient in the smallfurnace assembly used in the previous work (Takahashi eral.,1982). The solidus, therefore, was redetermined in thepresent study with a piston-cylinder apparatus (Runs #l-25, Fig. 8). One-atmosphere experiments (Runs #l-5)were carried out under controlled oxygen fugacity (QFM)using the Pt-loop method of Presnall and Brenner (1974)with crystalline starting material that had been annealedat 15 kbar and 12fi)"C for 2 hr.

The levised solidus is generally at temperatures 70-100"C lower than the original curve. This difference maybe partly due to the large uncertainties in experimentalpressures and temperatures in the previous study, butalso reflects the difference in technique in identifying run

a ih bosoll mcl,, O in peridotita molri,

o l i v i n e

o P r

c P x

100M9/(M9*Fe l

Fig. 7. The Mg-values of olivine and pyroxenes in anexperimental charge quenched after 6l hr at 15 kbar and 1300'C(Run #40, L45/HK66). Crystals in the liquid domain (see Fig. 3)are distinguished by solid squares, those in peridotite matrix byopen squares. The inverted triangles indicate compositions ofstarting materials. Clinopyroxenes rich in Fe are quenchcrystals.

products. Under the reflected light microscope, whichhas been used in the present study instead of the transmit-ted light microscope of the previous study, glassquenched from partial melt is readily identified even insmall amounts, and even quench crystals are disiinct inthat they characteristically show dendritic shapes. Chem-ical analyses and back-scattered X-ray images of the runproducts by eeue were useful in identifying the twopyroxenes and isolating plagioclase from quenched liquid(see Table 2 for mineral assemblages of the run products).

It should be noted that no spinel was observed in theexperiments above 15 kbar, although it was reported byKushiro et al. (1968) and would be easily identified inpolished sections because of its high reflectivity andcharacteristic equant shape. There are two possible rea-sons for its absence: (l) with long run durations in thepresent study, all the Al2O3 in the spinel dissolved eitherin the melt or in pyroxenes; (2) because ofthe low oxygenfugacity imposed by the graphite capsule, Ct'* thatstabilizes spinel greatly (e.9., Irvine, 1977) was reducedto Cr2* which then dissolved in olivine and pyroxenes(note higher chromium content of olivine in Table 4, alsocontrasting results with Fe and Pt containers by Jaquesand Green, 1980). The subsolidus mineral assemblage of

o

5 r 0)g

o

9 0


1 8 0 0

1 7 0 0

1 6 0 0

1 5 0 0

1 4 0 0

1 3 0 0

r 2 0 0

1 1 0 0

1 0 0 0

Pressure ( kbor)

Fig. 8. Solidus curve of the peridotite HK66 under dry conditions (circles and solid lines) have cusps at about I I kbar and 26 kbarwhere subsolidus mineral assemblage changes from plagioclase lherzolite to aluminous pyroxene lherzolite, and aluminous pyroxenelherzolite to garnet lherzolite, respectively. Data of previous study (Kushiro et al., 1968) are given for comparison (squares and lightline). Solid symbols denote subsolidus runs and open symbols with quenched liquid.

c(-)

o(-

(*4.,aEq,

f-

5 04 03 02 01 0

HK66 between l0 and 25 kbar corresponds to the alumi-nous pyroxene peridotite of Green and Ringwood0967b).

Additional experiments were made with garnet seeds (3wt.Vo of pyrope-rich garnet) in the peridotite (Runs #20and 23). The amount of garnet increased in the run at 30kbar, whereas it diminished at 25 kbar. Thus, garnet isconsidered to be a stable subsolidus phase above apressure between 25 and 30 kbar, although it was notnucleated in some experiments (see Table 2).

Finally it is noted that the solidus determined in thepresent experiments is not satisfactorily fitted with astraight line. It agrees, however, with the observation byPresnall et al. (1979) and Presnall and Hoover (1982) thatthe solidi for simplified peridotite systems have cusps atpoints where the subsolidus phase assemblage changes.As shown in Figure 8, the present results can be interpret-ed as representing three curves defining cusps at transi-tions from plagioclase lherzolite to aluminous-pyroxenelherzolite at about ll kbar and from aluminous-pyroxenelherzolite to garnet lherzolite at about 26 kbar.

Compositions of melts and solids along the solidus

The compositions of the melts formed at temperaturesjust above the dry solidus of HK66 were determined withthe basalt/peridotite sandwich technique (Table 3). Themelt compositions formed at higher temperatures are alsolisted in the same table. Selected analyses of coexistingsolid phases are given in Table 4.

At 5 kbar, the melt near the solidus, which coexistswith olivine, orthopyroxene and clinopyroxene is aquartz tholeiite. At 8 kbar, the near solidus melt is anolivine tholeiite. At 10 kbar, it is also olivine tholeiite, butis slightly enriched in MgO and Na2O and depleted inSiO2 compared to that at 8 kbar. At 15 and 20 kbar, themelts are nepheline normative alkali-olivine basalt richerin alkalis and poorer in SiO2 than the liquids at 8 and l0kbar. At 25, 30 and 35 kbar, they are picritic meltscharacterized by high MgO contents.

Because the degree of partial melting increases abrupt-Iy in this pressure range (Mysen and Kushiro, 1977),clinopyroxene was absent in residues in all runs above 25

o

I

a

o

I

oo

I t

o

9t

oao ,

Io


Table 2. Run details

867

R u n + P(kbar)

T( . c )

D u ! a - N o . o f S t a r t r n g m a i e r i a l s

t ion charges(hr ) per ido t i te basa l t i c conponents

I 0 . 0 0 1 1 0 s 02 0 . 0 0 1 r r o o3 0 0 0 r l r s o4 0 . 0 0 r r \ 1 55 0 . 0 0 1 1 2 0 0

6 5 r r 5 07 5 1 1 7 58 5 1 2 0 0

9 I 1 1 7 510 8 1200lr I L225

t 2 1 1 1 2 0 01 3 1 1 l 2 l o14 1r 1275

1 5 1 5 1 2 7 51 6 1 5 1 3 0 0

t 7 2 0 1 3 5 0t 8 2 0 1 3 7 5

L 9 2 5 1 3 7 52 0 2 5 1 3 1 52 t 2 5 1 4 0 022 25 1425

l 4 0 0

I 5 0 0

r 2 0 0L225

1 2 0 0t225

t 2 5 0r 2 1 5t27 5

1 2 5 0r215t290

1 2 1 5I 3 0 01 3 0 0r300r 3 5 01 4 0 0

t 3 , ol 7 . o2 3 - 51 4 . 5

3 . 0

1 7 . 52 A

4 6 . 5

2 . O1 , 3

4 1 0

4 02 . O2 . O

1 2 . O6 l . o

2 3 - 5

6 . 02 . 51 . 82 0

1 . 54 - 53 . 0

4 6 - 51 0 . 5

4 0 02 9 . 44 1 . 01 9 . 0

4 - O1 4 . 0

2 6 - O2 5 - O3 0 . 0

1 2 . O4 . 0

2 6 56 1 02 3 . 53 6 . 0

HK56+HK66+

HK65+

HK66+

HK66+

HK65EK66

HK66

HK66

HK66

HK66

HK66

HK66

HK66

HK66

HK66

HK66"

HK66

HK66

HK66 'HK65

HK66

sK66

HK66

HK66

HK66

HK66

HK65

HK66

HK66

HK66

HK66

HK66

HK66

HK66

HK66

HK66

HK66

HK56

HK56

HK56HK66

HK66

o l , o p x , c p x , s p , p I

o 1 , o p x , c p x , s p , p I

o I , opx , cpx , L

o I , opx , cpx , L

o l , o p x , L

o l , o p x , c p x , s p , p l , n

o 1 , o p x , c p x . s p , p l , n

o 1 , o p x , c p x , s p , p l , n . L

o l , opx , cpx , sp , p l , m

o 1 . o p x , c p x , s p , p I , m

o l , o p x , c p x , s p , n , L

o I , opx , cpx , sp , n

o I , o p x , c p x , s P , L

o l . o p x , c p x , I

01 , opx , cpx , sp , n , L?

o l , o p x . c p x , n , L

o 1 , o p x , c p a , m

oI , opx , cpx , m, L

o l , o p x , c p x , s p , n

o I , opx , cpx , sp

o I , o p x , c p x , m , L

o 1 , o p x , c p x , L

o l , o p x , c p x , g t , s P

o l , o p x , c p x , L

3 03 03 0

I O

r ol o . 5

l 51 51 51 5I 5l 5

2 02 0

2 52 5

303 03 03 0

3 5

2 32 42 5

2 62 7

2 A2 93 03 I3 23 3

34

3 5

3 13 a3 940

4 2

4 3

4 54 6

4 74 84 950

5 1

'KB, L45r K B . ! 4 5

r K s , L 4 5.L45, K2IrKre , mE, BON* L 4 5 , K 2 l ,*L45, K2\,L45, \2L

,L45, K2Lr L 4 5 , K 2 l*L45, Kru , OCE

* K B , L 4 5 , K 2 I*Kre , I45 , K21, BON* K m , L 4 5 , K 2 1 , B O Nr K r e , L 4 5 , K 2 1 , B O N

KS* K S , L 4 5

rKB, L45

KB

KS

KB

1 3 7 5 2 3 - 51 4 0 0 4 8 . 0

L 4 2 S 1 6 . 5I 4 s O 1 4 . 5

1 5 2 0 1 9 . 01 5 4 0 a , 51 5 5 0 4 . 51 5 5 0 t 2 - 5

1 6 0 0 2 - 5

KPKre

KS

KB

kbar even at temperatures closest to the solidus. Becausepyroxenes in the peridotite matrix frequently show relictcompositions (Fig. 2) and the effect of quench crystalliza-tion on composition of solids is more conspicuous in theperidotite matrix than in the liquid disk (Fig. 7), analysesof coexisting solids within the liquid disk of the chargeswere adopted (Table 4). It is notable that pigeoniticclinopyroxene coexists with orthopyroxene and subcalcicaugite at l5 kbar. According to Lindsley (1980), the lowerstability ofpigeonite in the quadrilateral pyroxene systemis about 1350'C at 15 kbar and Mg/(Me+Fe) : 9.35. tn"present observation suggests that the limit actually ismore than 50'C lower in the presence of other compo-nents (e.9., A12O3).

Efrect of K2O on the solidus and the meltcomPositions

Because K2O is incompatible in all anhydrous perido-

tite minerals, it is likely that K2O plays a role analogousto that of volatiles in the melting of a dry peridotite. Intheir study of the system KzO-MgO-AlzOrSiOz (xues)

at high pressures, Wendlandt and Eggler (1980) showedthat the solidus of a simplified K2O-bearing peridotiteinvolves the following reactions: Sa * En : Fo * L (6-18

kbar): Sa + Fo + En : L (18-19 kbar); Sa * Fo : En *L(19-34 kbar); and Ks * En : Fo + L (above 34 kbar);where Sa, Fo, En, Ks and L denote sanidine, forsterite,enstatite, kalsilite and liquid, respectively.


Table 2. (continued)

R U N } P(kbar )

T{ . c )

d u r a - N o o f S t a r t i n g m t e r i a l s

t ion charqes

in . j ' per ido t i te basa l t i c components

5 25 35 4

5 556

5 859

506 I

6 364

2 0 t 2 5 0 4 . 52 0 1 3 0 0 3 . 52 0 t 3 3 0 2 . 5

3 0 1 3 0 0 2 - o3 0 1 3 5 0 r . o3 0 1 4 0 0 2 . 5

3 7 1 3 0 0 2 - 53 7 1 3 5 0 2 - O

2 0 1 3 0 0 4 . 52 0 1 3 3 0 2 . 5

3 0 1 4 0 0 6 . 03 0 1 4 5 0 2 - O30 L415 4 0

HK66 '

HK66 '

H K 6 6 '

HK66 '

H K 6 6 '

EK66 '

HK56 'HK56 '

HK66"

HK66"

HK66"

HK56"

HK66"

KFSB- I

KFSB_I

KFSB- I

KFSB-2

KFSB-3

o l , o p x , s p , L

o 1 , o p x , L

o l , o p x , c p x , s p , g t ? , s a

o I , o p x , s p , L

o I , o p x , L

o I , o p x , c p x , s p , g t , s a

o I , o p x , 9 t , r

o I , o p x , c p x , s p , L

o I , o p x , L

o I , o p x , c p x , s p , s a , I

o l , opx , cpr , L

0 1 , o p x , c p x , L

1 , fo r runs w i th rc re than one charges , phase assenb laqe o f a represenxax ive charge is g iven ;

HK66+, p reannea ied a t 75 kbar and 12OO"C, EK65ot seeded q i th pgrope- r ich garnex-

HK66 ' , HK66 p lus 20 wt Z o f adu la r ia ; HK66" , EK55 p lDs 2 v t Z o f adu la r ia lKFSB- ] , KRE p l l s 50 wt Z o f adu la r ia ; KFSB-2 , KRB p jus 20 wt Z o f adDjar ia ;KFSB-3 , KRB p lus lO e t * . o f adu la r ia

abbrewia t ions ; o1 , o i i v ine , opx t o r thopgroxene i p jS , p igeon l te ; cpx , Ca- r lch c l jnopgroxene,sp , sp ine l ; p l , p laq ioc lase i g t , qarne t ; n , Fe-N j re taL ; sa , san id jne ; L , 11qu id .

An attempt has been made to examine the anhydrousreactions in a K2O-rich natural peridotite, using theperidotite HK66 plus 20 wt.Vo adularia as a startingmaterial (Runs #52-59). Sanidine was observed in thesubsolidus runs at least up to 37 kbar and the solidus waslowered bv about 100'C relative to that in the rues

system (cf ., Fie. 9 with Fig. 13 of Wendlandt and Eggler,1980). This difference is reasonable in view of the FeOcontent of olivine and pyroxenes, but it may also bepossible that clinopyroxene is involved in the reaction inthe natural peridotite. Compared with the solidus curvedetermined with HK66 (Fig. 8), the solidus of the potassic

Table 3. Compositions of the liquids equilibrated with the peridotite

Run # 26

P (kbar) s

T ( . C ) 1 2 O O

Compo.* KB

2 7 3 0

L225 t225

KB KB

29 29

8 8

1225 t225

L45 K2I

3 0 3 1

8 a

t225 1250

€E L45

3 t 3 3 3 3

a a 8

t250 L215 1275

K 2 l L 4 5 K 2 t

1 0 1 0

1250 L250

L45 K2l

3 5 3 5

I O 1 0

1275 L215

L45 K2I

3 6 3 6

1 0 . 5 1 0 . 5

L290 1290

KS L45

3 6 3 ?

1 0 . 5 1 5

1290 1275

OCE KB

1 5 t 5

1 3 0 0 1 3 0 0

KB L45

s i o 2 5 1 . 9 4r i o 2 1 . 9 IA l 2 o 3 1 5 . 6 7F e O " 7 . 5 0h o 0 . t lM g O 7 . 4 2

c a o 9 . 0 9N a 2 o 2 . 9 OK 2 O 1 . 1 7P 2 o 5 o . 3 9c r 2 o 3 0 . 0 5

F t a l 1 0 0 . 1 5

5 4 . 4 0 5 0 . 9 r1 . 9 9 1 . 4 1

1 6 . 2 0 1 6 . 1 66 . 6 4 7 9 8o . 0 3 0 - 2 17 . 9 I 8 . 0 0

8 . ? 8 I 0 . 2 92 . 0 3 2 . 8 t0 . 9 5 0 . 7 2o - 4 2 0 . 2 2o . 0 9 0 . 0 7

9 9 - 4 4 9 4 . 7 4

4 9 . 7 s 5 0 . 0 2I . 3 3 r . 5 4

1 5 . 9 7 1 5 . 9 58 . 8 7 8 , 9 3o . 1 8 0 . 1 29 . 0 5 8 . 9 1

1 0 . 6 1 f 0 8 02 , 4 9 2 . 4 I0 . 1 6 0 - 2 60 . 0 4 0 . 1 60 . 0 8 0 . 0 9

9 8 . 5 3 9 9 , 1 9

5 0 . 3 a 5 I . 7 4L - 5 2 1 . 0 ?

1 6 . O 2 L 5 - 2 38 9 5 9 . 4 10 . 1 5 0 . 1 9a - 2 5 9 , 8 I

I 1 . 0 8 I 0 . 3 82 . 5 L 2 - 3 40 . 3 1 0 . 0 80 . 1 8 n . d .0 0 7 0 . 0 8

9 9 . 4 2 1 0 0 . 3 1

5 0 4 3 5 0 . 9 5 5 0 . 4 81 . 5 8 t . 1 6 1 . 7 0

1 4 . 3 6 1 4 . 3 0 1 3 8 99 . 6 3 9 . 7 9 1 0 . 0 90 . 1 8 0 . 1 8 0 . 1 49 . 6 3 1 O . 6 2 1 0 . 2 8

r o . a o 9 3 6 9 - 5 12 . I I 2 - 3 6 2 . 2 2o . 2 2 0 . 0 9 0 . 2 5n . d . n . d . n . d .0 . 0 9 0 . 1 0 0 . 1 1

9 9 . 0 3 9 8 . 9 1 9 8 . 7 3

4A-65 49-11r . 3 9 r . 4 5

t 6 . 5 9 1 7 . 0 39 . 8 8 9 4 30 . 2 r o 1 9a - 7 0 8 . 4 9

I O . O 7 1 0 5 43 . 1 9 2 . 9 40 . 1 3 0 . 2 60 . 0 8 0 . 1 70 . 0 8 0 0 9

9 8 . 9 7 1 0 0 . 3 4

4 9 5 2 4 9 . 2 51 3 0 1 6 0

1 5 . 0 2 1 4 . 5 49 . 5 5 9 - 4 20 . 1 9 0 . I 89 . 9 5 1 0 . 2 6

1 0 . 4 0 1 0 8 42 - 7 4 2 . 4 40 . r 0 0 - 2 60 . 0 1 0 . 1 60 . 0 7 0 . 0 7

94.49 99.42

5 0 , 2 7 4 9 . 1 2r . 4 5 I . 2 9

1 5 . 3 7 1 4 . 6 98 . 9 5 9 . 9 Ao , 1 3 0 . 1 8

1 0 . 1 8 1 0 . 9 3

9 6 1 1 0 . 3 62 . 5 2 2 . 3 9o . 1 9 0 . L 2n - d . n . d -o . 0 5 0 . 0 6

9 9 - 3 2 9 9 - 1 2

4 9 . r 2 4 9 - 7 41 . 5 4 I 5 9

1 5 . 1 5 1 7 . 4 8l 0 . o o 8 . 5 9

0 . 1 4 0 . 1 19 - 6 4 L 4 5

t ] 3 4 1 . 6 32 . 2 t 4 . 0 0o . 2 9 1 . 5 5n . d . n . d .0 . 0 6 0 . 0 4

9 9 . 5 0 9 8 . 5 8

4 A - 5 2 4 A . 4 3r - 6 6 I . 1 3

1 5 . 3 0 1 5 . 8 99 . 3 4 1 0 . 3 4o . t 5 0 . 2 I9 , 9 5 1 0 . 8 6

9 . 3 0 1 0 . 7 12 . 9 4 2 . 4 91 . 0 3 0 . 1 3n . d . n . d ,o . 0 9 0 . 0 8

9 A - 3 2 r O O . 2 7

CIPW rcmg I . 7 6 7 . 3 8o r 6 . 9 0 5 . 6 5& 2 4 - 5 0 t 1 . 2 7

2 6 . 2 4 3 2 . 4 7

1cd i 1 3 . 2 1 6 . 9 3h y 2 2 . 7 9 2 5 - 4 Oo 1i l 3 . 6 2 3 . A Oa p 0 . 9 0 0 . 9 8

0 . 0 7 0 1 3

l . a 4 0 - 4 12 I - 3 6 t 9 . 7 33 1 . 7 1 3 0 7 1

t . 3 I O - 5 4 L 5 01 8 . 0 3 2 0 . I 9 1 9 . 0 32 9 - 3 5 2 4 . 4 1 2 7 5 5

2 0 . 1 9 1 5 . 0 5 1 6 . 7 32 2 . 9 I 2 6 - 6 6 2 6 . 7 5

5 . 0 6 6 - 7 3 5 . 6 23 0 3 2 - 2 3 3 . 2 7

0 . 1 3 0 . 1 5 0 . 1 6

0 . 7 8 I 5 32 1 2 7 2 5 - 1 33 0 . 8 8 3 2 . 2 1

1 5 . 6 2 1 5 . 3 80 - 4 6 5 , 6 5

2 2 . 0 2 1 6 - 4 52 - 6 7 2 7 4o . 1 9 0 . 3 9o , L 2 0 , 1 3

0 . 6 0 r . s 52 3 . 1 9 2 0 - 1 12 8 - 5 3 2 8 . I 2

t 9 . 2 2 2 0 . 3 08 . 3 6 1 0 . 1 7

1 6 . 8 e 1 5 , 5 ?2 . 5 0 3 . 0 6o . o 2 0 . 3 70 . r 0 0 . r 0

4 . 3 I2 4 . O 12 9 - 7 2

1 6 - 1 21 3 . 9 0

1 . 9 62 . 7 7o - 5 20 , 1 0

0 . 9 6 I . 5 52 l . 3 8 2 0 . 5 63 2 - 4 0 3 2 . 2 0

1 6 . 9 2 1 6 - 9 3 1 8 . 3 0 1 6 . 8 11 5 . 8 2 1 7 . 5 1 1 6 . 3 8 2 5 . 4 0

9 . 7 4 7 . 8 1 6 . 9 8 4 . 7 32 . 5 6 2 - 9 5 2 - 9 0 2 . O 30 . 0 9 0 . 3 7 O . 4 2o . I 2 0 . 1 3 0 . I 0 0 . I 2

4 . 7 0 0 1 I I - 7 2 9 . 2 9 6 . 1 9 0 - 7 72 L 4 7 2 0 . 2 4 L A - 1 9 2 4 . 6 1 2 3 . 0 4 2 I . 0 22 A 4 9 2 9 - O a 3 0 , 7 4 2 5 - 5 1 2 5 . 7 6 3 I . ' 1 I

5 - 2 3 I . 4 1

1 5 . 8 5 t 8 , 2 8 2 I . 0 6 1 0 . 4 9 1 1 . 2 2 t 7 . 3 6I 2 . O I 1 3 . 5 7 1 0 . 3 9 2 - 7 31 4 . 6 3 1 5 5 4 1 4 . 2 6 2 I . 6 6 2 3 . O 4 2 4 . ! 1

2 . 7 7 2 - 4 6 2 - 9 4 3 . 0 6 3 - 2 I 2 . I 4

0 . o 7 0 . 0 9 0 - 0 9 0 . 0 6 0 . I 3 0 . t 2

F o r r 8 5 . 2 8 6 . 8

K D * * * . 3 0 5 - 3 2 3

8 5 . 5 8 5 . 2 A 5 . 4

. 3 0 4 . 3 1 6 . 3 0 6

a 5 - 2 8 5 . 8 A 5 2 8 5 . 9

. 2 4 6 . 3 0 8 . 3 0 9 , 3 1 7

8 5 . 6 A 3 2 8 4 . 1 8 5 . 2

. 3 0 6 . 3 0 3 . 3 1 8 . 3 2 4

8 5 . 2 8 6 , 1 8 5 . 9 8 5 . 8 A 4 4 8 6 . 6 a 5 . 8

, 3 2 3 . 3 2 6 . 3 1 9 - 2 A 4 . 3 2 6 - 2 9 4 . 3 1 1

' , add i t iona l basaf t i c cokpnenxs , ++, to rs te r i te rc je Frcent o t coex is t jng o f i v jne*** . o l i v ine / l iqu id i ron-nagnes iun Fr t iX ion c@f f ic ien t


Table 3. (continued)

E69

Run # 40 39 41 42

P (kbar) 15 15 15 15

T ( " C ) I 3 O 0 1 3 O O 1 3 5 0 1 4 0 0

Conpo * K2I BON KB Kre

4 2 4 3 4 4

1 5 2 0 2 0

1 4 0 0 1 3 7 5 1 4 0 0

L45 KS KM

4 6 4 1 4 A 4 9

2 5 3 0 3 0 3 0

I4s0 1520 1540 1550

KS KB KB KS

4 5

2 5

1425

KS

50

3 0

t 5 5 0

5 1

3 5

t600

Krc

6 0 6 1

2 0 2 0

1 3 0 0 1 3 3 0

KFSB KFSB

63 64

3 0 3 0

1 4 5 0 1 4 7 5

rc58 trSB

6 2

3 0

f400

KFSB

s i o 2 4 7 . 5 1 5 1 3 0 4 9 . 5 2 5 0 . 1 5 4 9 - 8 9 4 7 . 8 o 4 9 . f 2 4 A 7 7 4 9 . 3 a 4 6 . 5 o 4 1 . A 4 4 7 . 3 2 4 7 . 5 O 4 A . 6 3 5 4 . 7 5 5 3 . 3 9 s f . o 0 4 9 . 4 L 4 7 . 3 7r i o 2 1 . 3 6 0 . 4 6 1 . 0 6 O . 1 2 O . 8 1 1 5 5 1 . 1 4 I 6 6 I t 8 1 . 7 9 I . I o l . 1 6 I . 0 8 l . O o o . ? 8 o . 1 4 0 . 8 5 \ . 1 9 2 . O 3A l 2 o 3 1 6 . o O 1 6 . 5 5 1 3 . 3 9 1 2 6 5 1 2 . 9 e t 5 . 6 0 1 3 , 5 1 1 5 . 3 2 1 3 . 7 8 1 2 9 l f I . 8 5 1 1 . 3 4 1 1 , 8 8 l o . 3 0 1 9 . 3 3 1 ? . 3 6 1 5 . 3 9 1 5 . 5 7 1 4 . 6 9F e O o 1 0 . 3 2 9 . r 8 8 . 6 1 1 r . 0 8 1 0 . 6 4 9 . 9 1 r 0 3 6 S . 7 0 1 0 . 4 1 t r . O 3 r O t 2 1 0 . 6 4 9 . 2 r r t . 2 0 4 . 3 6 5 9 4 7 - 4 A A . 4 4 9 . 5 6h o 0 . 1 2 0 . 1 8 0 . I I 0 1 9 0 . 2 5 0 . 1 9 O . I t o 1 4 0 . 1 4 0 1 6 0 _ 2 1 0 . 1 8 0 . 2 t 0 . t 2 0 . 0 8 0 . 1 0 0 . I l o . I 2 0 . I 4M g O 1 0 , 3 4 9 . 7 6 1 2 - 3 9 1 4 . 8 6 1 4 . 3 0 I \ . 2 O L 4 . 2 5 1 2 . 1 8 1 4 . 3 9 1 6 , 5 6 1 8 . 4 7 1 8 . 9 1 1 8 , ? t 2 t . 4 1 3 . a O 6 . 1 0 8 . 9 5 A . 1 9 9 . 8 9

c a o 1 0 . 7 9 9 . 6 8 1 0 . 7 0 8 . 1 5 4 . 8 6 9 . 1 7 a . O 2 8 . 8 0 8 . 3 4 7 . 2 5 a . o a 6 . 1 9 8 . 8 4 6 . 2 3 3 . 3 4 5 . 0 7 4 . 4 6 6 . O 2 6 . t BN a 2 O 2 . 3 O 3 . 2 0 2 - 2 O I . 4 2 l . 9 I 3 . 3 3 2 . 3 4 3 . 0 9 2 . 2 O 2 , 1 2 I . 8 4 1 . 8 8 2 1 r l - 6 4 2 . 5 7 2 . 2 1 2 . L 6 2 , 9 9 3 . 5 1K 2 o A - 2 5 0 . 8 7 0 , 5 I O . 2 3 0 . 0 7 f . 0 0 0 . 6 3 L . 2 6 O . 1 j I 2 3 0 . 5 5 0 . 6 8 0 . O 9 0 . 6 4 1 0 . 1 f ' t . 7 3 8 . 4 5 4 . 9 3 3 . 9 3P 2 o 5 n . d . n . d . n . d . n . d . n . d . 0 . 3 1 0 - 1 6 0 . 3 0 0 . 2 1 n - d 0 . 0 7 0 . 1 5 n . d . 0 . 0 9 n . d . n . d . n d . 0 . 3 9 0 - 5 3c r 2 o 3 0 . 0 4 0 . 0 7 o . f o 0 . I 9 0 . 1 8 0 . 0 4 O . t 5 0 . 1 0 0 . 1 8 0 . 0 8 0 . 1 6 0 1 9 O . t l O . 2 3 0 . 0 5 O . 0 5 0 . 0 9 n . d . 0 . 0 7

T o t a r 9 9 . 0 1 1 0 1 2 5 9 8 . 5 9 9 9 6 4 9 9 , A 9 1 0 0 6 4 1 0 0 . 3 9 r o o . 3 2 1 0 0 . 9 8 l o o . 2 3 1 0 0 . 2 9 9 a . 6 4 9 9 - 1 4 1 0 1 . 4 9 9 9 n 9 A - 7 5 9 8 . 9 6 9 8 . 4 5 9 7 . 9 0

CItr Norna

t -49* L9 .64

3 2 . 9 0

1 c

d i I 7 2 4hy 2 14o t 2 3 , 3 2i l 2 . 6 L

ap

0 . 0 6

5 . 0 8 3 . 0 6 1 . 3 6 0 . 4 12 6 . 2 1 1 8 . 8 8 1 2 0 6 1 6 . 1 82 7 . 8 8 2 5 . 5 1 2 7 - 5 6 2 6 . 6 1

o 2 9

1 5 . 9 1 2 2 - A E I 0 . 5 5 1 4 . 0 58 . 8 0 3 4 . 6 0 2 3 . 7 9

2 3 . 6 1 1 8 . 5 8 1 2 . 2 7 L 1 . t O0 . 8 6 2 . O 4 1 . 3 7 1 . 5 4

o . t o 0 1 5 0 . 2 a o - 2 1

3 - 7 3 6 0 - 2 4 4 6 - 2 61 3 . 6 7 0 . 2 2 6 . 4 71 8 . 5 8 ] f , 4 4 t 4 . 5 3

1 1 7 6 7 . 0 3

9 . 4 7 4 . l I 4 . 9 21 5 . 8 73 6 . 6 8 r 0 . 4 6 t 5 . 2 9I . 8 7 1 . 4 9 I - 4 2o . 2 L0 . 3 3 0 . 0 7 0 . 0 7

3 6 4 2 2 9 - 5 9 2 3 - 1 21 0 . 8 1 1 0 . 4 6

7 . 4 2 1 4 . 1 3 1 2 . 9 91 0 , 0 0 8 . 0 7 f 0 . 7 7I I . O I1 2 . 1 3 r 0 . 5 9 r 2 . 0 3

2 r - 2 5 2 r - 4 5 2 4 - 1 41 . 6 t 3 . 4 5 3 . 9 4

o . 9 2 I 2 50 . r 3 0 . 1 1

5 . 8 7 3 . 7 1I 7 - L 9 1 9 . 7 224 51 24.40

5 . 8 6

1 7 . 3 5 1 I 4 81 4 - 3 2

2 5 - 5 4 2 3 6 32 9 3 2 . t 60 . 7 1 0 . 3 70 , 0 6 0 . 2 2

7 4 2 4 . 5 12 1 t r 1 8 4 32 4 1 3 2 5 . 2 0

2 6 A

I 4 . 0 7 I 1 . 6 6t 1 . 7 5

2 6 - 6 L 2 5 . 4 9I 1 4 2 . 2 20 . 6 9 0 . 4 80 . 1 5 0 . 2 6

7 . 2 5 3 - 2 41 4 . o l 1 5 . 5 21 9 . 3 4 2 2 . 3 44 . 8 4

L 3 . 3 4 1 3 . 7 61 6 0

3 7 7 0 3 5 . 0 13 3 9 2 . 0 8

o . 1 6a . L 2 0 . 2 3

4 - O 7 0 . 5 31 6 , 1 3 1 7 . 9 02 0 - 1 A 2 2 . 1 4

7 . 5 2 L 7 , O lL 3 - 7 1 2 - 1 63 4 - 8 7 3 6 . 8 52 . 2 3 2 - 0 60 . 3 50 . 2 8 0 . 1 6

F o * ' 8 5 . 6 8 5 . 3 a 8 . 9

K D * * l . 3 0 3 . 3 2 5 - 3 2 2

8 7 . 6 8 8 . 0 8 5 . 9

, 3 3 8 - 3 2 6 . 3 3 I

8 8 5 8 8 . 0 8 8 . 7 A 9 - 1

- 3 2 6 3 3 5 . 3 4 3 . 3 1 1

8 9 . 9 9 1 . 0 9 0 . 2 8 5 . 8 8 6 . 5 8 6 . 4 8 5 . 8 8 5 . 4

. 2 4 5 . 3 3 5 . 3 0 6 . 3 1 4

a 1 . 6

346

r * | fo rs te r i te mofe Frcent o f coex is t ing o l i v ine t

Dar t i t ion c@f f ic ien t

Table 4. Selected mineral compositions

Run fl 2A

P (bar ) 8

T ( ' C ) 1 2 0 0

Conpo KS

Phase ol opx opx' cpx cpx* sp

3 5

1 0

I215

K 2 1

o1 opx cpx

4 0

1 5

I 3 0 0

K 2 I

01 opx opxr pig cpx

4 3

20

1 3 7 5

KB

oI opx cpx o l

4 7

3 0

r 5 2 0

KP

S i o 2 3 8 . 9 3 5 3 . 3 0 5 2 - 4 5 4 9 - 4 4 4 9 . 2 ! 0 . 5 9T i o 2 0 . 0 7 0 . 5 9 0 . 1 4 l . 3 o 1 . 3 0 0 . 4 5A l 2 o 3 O . I 3 3 . 4 4 5 . 5 4 3 4 1 7 . 3 4 5 5 . 5 7F e o 6 1 5 . a 7 9 . 7 8 9 . 1 7 6 4 2 5 . I s 1 2 . 4 9h o 0 . 1 6 0 . I 5 0 . 2 0 0 . 0 7 0 . 0 7 0 . 1 5

M s o 4 4 3 A 2 9 - 3 2 3 0 . 7 2 1 5 . 8 1 1 4 . 8 4 1 9 . 1 8C a O O . 3 4 2 . 2 4 0 . 7 1 2 1 1 5 1 8 . 1 7 0 . I 0N a 2 O 0 . 0 0 0 . O 0 0 . O O 0 , 2 5 0 . 8 4 0 . 0 0c t 2 i 3 o . o a 0 . 4 6 0 I 7 0 4 0 0 . I 3 1 0 . 7 5N i O 0 . 1 0 O 0 9 0 . 2 0 0 , 0 I 0 . 0 1 0 . 2 1

T o t a l 1 0 0 . 0 5 9 9 . 4 1 9 9 , 4 4 9 a . 3 6 9 7 . A 6 9 9 - 4 9

3 8 4 3 5 4 . 1 9 5 4 , 0 70 . 0 2 0 . 3 5 0 . 3 20 . 0 0 3 - 9 4 2 , 9 6

1 4 . 5 7 A . 2 6 7 - 7 20 . 1 4 0 . 1 7 0 . 1 3

4 7 . 0 t 3 0 3 4 2 4 . a O0 . 3 3 2 , 8 0 1 0 . 1 80 . 0 0 0 . 0 0 0 . 3 00 . 0 6 0 . 4 1 0 . 2 3o . o I 0 . 0 7 0 . 0 5

l o o . 5 7 1 0 0 . 5 3 r 0 0 . 7 5

3 9 . 0 3 5 3 . 0 0 5 2 . 0 5 5 0 . 6 9 5 0 . 6 7o . 0 4 0 . 2 5 0 . 1 4 0 - 3 2 0 . 4 80 . 0 2 3 6 6 4 - 9 6 6 . 6 6 1 . O 4

1 4 . 0 5 8 . 6 9 8 . 7 0 8 . 0 4 7 . 0 60 . I 2 0 , r r o 1 8 0 . I 4 0 . I l

4 6 . s 7 3 0 - 1 3 3 0 - 7 0 2 6 - 2 2 2 1 6 30 . 3 5 2 . 4 L 0 . 8 0 5 . 4 1 1 r . 5 70 . o o o . 0 0 0 . 0 3 0 . 1 8 0 . 3 ro . o ? o l 5 0 . 3 6 0 . t 9 0 . 4 40 . 0 3 0 . 0 5 0 . 0 8 0 . 0 0 0 . 0 3

I 0 0 . 2 8 9 9 . 2 s 9 a . 0 0 9 a . 0 5 9 9 . 3 6

3 9 . 5 4 5 4 . 6 3 5 2 . 7 00 , 0 2 0 . 2 0 0 . 2 0o . 0 2 4 . 1 3 3 . 7 1

t4-r2 8.59 1 -4r0 . 1 9 0 . 1 3 0 , 0 8

4 6 . 9 5 2 9 . 9 5 2 4 . O 5o - 2 2 2 - O 2 7 . 3 00 . 0 0 0 , 1 4 0 . s 30 . 0 7 0 . 3 2 0 . 3 2o . 1 3 0 . 0 0 0 - 1 4

I O t . 2 5 l O O . 7 1 9 8 . 5 1

3 9 . 9 9 5 3 . 3 50 , 0 2 0 . 2 5o - 2 0 5 . 0 4

1 0 . 9 5 6 , 7 0o . 1 4 0 . 0 4

4 1 . 9 4 3 1 . 3 90 . 3 0 L , 7 30 , 0 0 0 . 1 80 - 1 5 0 . 2 50 . 0 4 0 , 0 0

9 9 . 7 3 9 8 . 9 7

4 - 0 0 0 6 . o o o 5 . 0 0 0 6 . 0 0 0 6 . 0 0 0 4 . o o o

0 . 9 8 3 1 . 8 9 6 r . 8 5 1 1 . 8 6 2 1 . 8 4 0 0 . 0 r 5o . o o o o . o t 6 0 . 0 o 3 0 0 3 6 0 . 0 3 6 0 . 0 0 8o . o o 3 0 . 1 4 4 0 . 2 3 6 0 . r 5 0 0 - 3 2 3 I - 1 2 00 3 3 5 0 . 2 9 0 0 . 2 7 0 0 . 2 0 2 0 . 1 6 0 0 . 2 7 4o . o o 3 0 . 0 0 4 0 . 0 0 5 0 . 0 0 2 0 , o 0 2 0 . 0 0 3

1 . 6 7 2 r . 5 5 4 1 . 6 1 6 0 . 8 9 0 0 . A 2 7 0 . 7 5 00 . 0 0 9 0 . 0 8 6 0 . 0 2 6 0 . 8 5 3 0 7 2 1 0 . O O 20 . o o o 0 . 0 0 0 0 . 0 0 0 0 . 0 1 8 0 . 0 6 0 0 . 0 0 00 . 0 0 1 0 . 0 1 2 0 . 0 0 4 0 . 0 1 1 0 0 0 3 0 . 2 2 3o . o o r 0 . 0 0 2 0 . 0 0 5 0 . 0 0 0 0 . 0 0 0 0 0 0 3

TiA18eh

M9

CrNi

I c a t i o n 3 . 0 0 7 4 . 0 0 4 4 . 0 1 6 4 . o 2 4 3 . 9 a 2 2 9 9 8

4 0 0 0 6 . o o o 5 , o 0 0

o 9 6 2 1 a 9 3 1 , 9 2 0o o o o 0 . 0 0 9 0 . 0 0 a0 0 0 0 0 . 1 6 2 0 , 1 2 30 3 0 5 0 . 2 4 1 0 , 2 2 90 . 0 0 3 0 - 0 0 4 0 . 0 0 3

r . 7 5 s 1 . 5 7 9 1 . 3 1 30 . 0 0 9 0 . 1 0 4 0 . 3 8 70 . 0 0 0 0 . 0 0 0 0 . 0 2 00 . 0 0 0 0 . 0 1 0 0 . 0 0 60 . 0 0 1 0 . 0 0 1 0 . 0 0 1

3 . 0 3 s 4 . 0 0 3 4 0 1 0

4 . O O O 5 . O O O 6 . O O O 6 . O 0 0 6 . O 0 0

0 . 9 7 5 I . a 7 9 1 - a 6 r 1 . 8 3 1 f . 8 3 20 . 0 0 0 0 . 0 0 6 0 . 0 0 3 0 . 0 0 8 0 . 0 r 30 . 0 0 0 0 . I 5 3 0 , 2 0 8 0 . 2 4 3 0 . 2 9 9o - 2 9 3 0 - 2 5 7 0 . 2 6 0 0 . 2 4 3 0 . 2 1 3o . o o 2 0 . 0 0 3 0 . 0 0 5 0 . 0 o 4 0 . 0 0 3

\ . 1 3 5 r - 6 2 5 I 6 3 6 1 . 4 1 2 1 . 1 6 50 . 0 0 9 0 0 9 0 0 . 0 3 0 0 . 2 0 8 0 . 4 4 70 - o 0 0 0 . 0 0 0 0 . 0 0 I 0 . 0 I 3 0 . 0 2 10 . 0 0 0 0 0 0 9 0 . 0 0 9 0 0 1 0 0 . 0 1 10 , 0 0 0 o . o o l 0 . 0 0 2 o . o o 0 o . o o 0

3 . 0 1 6 4 . 0 2 3 4 . 0 1 5 4 . 0 1 2 4 . 0 0 4

4 . 0 0 0 6 . 0 0 0 6 . 0 0 0

0 . 9 ? 9 r . 8 9 9 1 . 8 9 40 . 0 0 0 0 . 0 0 5 0 . 0 0 50 . o o o 0 . 1 9 3 0 . 2 4 4o . 2 9 2 0 . 2 4 9 0 . 2 2 2o . 0 0 3 0 . 0 0 3 0 . 0 0 2

1 . 7 3 3 1 . 5 5 r r . 2 8 90 . 0 0 5 0 . 0 7 4 0 . 2 8 00 . 0 0 0 0 . 0 0 9 0 . 0 3 70 , o o o o . o o a 0 . o o a0 . 0 0 2 0 . 0 0 0 0 , 0 0 3

3 . 0 1 4 3 . 9 9 1 3 . 9 8 4

4 . O O O 6 , 0 0 0

0 . 9 8 9 1 . 8 7 5o , o o o 0 . 0 0 6o . 0 0 6 0 . 2 0 9o . 2 2 6 0 . 1 9 60 . 0 0 3 0 , 0 0 2

1 . 7 6 8 I . 5 4 40 . o 0 7 0 . 0 5 50 . 0 0 0 0 . 0 I 2o . o o 3 0 , 0 0 10 . 0 0 I 0 , 0 0 6

3 . O 0 3 4 , 0 1 6

MG 4 5 . 2 8 6 . 8 8 5 - 1 s 5 . 6 4 6 . 3 4 6 . 2 4 5 . 3 8 4 , 5 8 6 . I 8 5 . 3I 5 . 5 8 8 . 7 8 9 . 3

* , reT ic t rd rera ls ; * * , 100M9/ (Mg+Fe)


1 0 0 0

' o

, r " r tu r " 'o ruoor ,

40 50

Fig. 9. Solidus curve of the potassic peridotite (circles andbold line) is considerably lower in temperature than the soliduscurve defined in Figure 8 (light line). The solidus curve of thepotassic peridotite may intersect with the normal oceanicgeotherm of Clark and Ringwood (1964) (broken line) at apressure between 30 and 50 kbar.

peridotite is lowered by about 75'C at 20 kbar and about150"C at 30 kbar (Fig. 9).

Among constituent minerals in dry peridotite, clinopy-roxene is the only phase likely to accommodate potassi-um. According to Erlank and Kushiro (1970), however,the maximum solubility of K2O in clinopyroxene is lessthan 140 ppm. It is likely, therefore, that a trace amountof sanidine is present in the experimental run products ofHK66 (which contains 0.07 wt.%o K2O) at pressuresabove the stability limit of plagioclase under dry subsoli-dus conditions. The solidus of the potassic peridotite(Fie. 9) should, therefore, be the true dry solidus ofHK66.

Throughout the large temperature interval between thetwo solidi, trace amounts of partial melt could be present(e.9., O'Hara, 1968, Fie. 8), but they would be verydifficult to detect. Mysen and Kushiro (1977) found that atrace amount ofliquid is present below 1450"C at 20 kbar(where the degree of partial melting increases abruptly) inthe melting study of a peridotite PHN16l I containing 0.14wt.%b K2O. Although they attributed the observation tothe presence of moisture in the charge, the same observa-tion was made by one of the present authors (E.T.) with avery well dried starting material of the same rock down toat least 1350'C at 20 kbar. It is thus likely that theexistence of such incipient melts below the ordinarilyaccepted solidus temperature is common in the meltingstudy of natural peridotites under dry conditions.

In order to study the nature of the melt compositions inthe temperature interval between the two solidi, a mix-ture of adularia(50wt.%o) and alkali basalt KRB (50wt.%)was sandwiched between layers of peridotite HK66 andrun at 20 and 30 kbar (Runs #ffi-64). Trace amounts ofthe potassium feldspar were also mixed with the perido-tite so as to assure the presence of partial melt in theperidotite matrix. Sanidine was present only in the runclosest to the solidus (Run #62), but reacted with perido-

tite minerals and disappeared to form very potassic meltsin all other runs. Analyses of the melts are listed in Table3. Compared with melts of Runs #26-51, they are rich inK2O and SiO2 and depleted in MgO, FeO and CaO. Themelt formed near the solidus of the potassic peridotite at20 kbar (Run #60) contains normative nepheline, ortho-clase and albite components, whereas the one at 30 kbar(Run #62) contains normative leucite instead of albite(Table 3). This observation is consistent with meltingrelations in the xuns system (Kushiro, 1980; Wendlandtand Eggler, 1980) where the initial melts are very rich inpotassic feldspar component and their normative compo-sitions change from enstatite-bearing to leucite-bearing atabout 20 kbar.

It is important to note that the solidus defined with thepotassic peridotite approaches or might cross the normaloceanic geotherm (e.9., Clark and Ringwood, 1964; Pol-lack and Chapman, 1977) at pressures between 30 and 50kbar (Fig. 9). The presence of volatiles (HzO, CO) hasbeen regarded as necessary to explain the Low VelocityZones in the upper mantle by partial melting (Kushiro etal., 1968; Lambert and Wyllie, 1968; Wyllie, 1973;Eggler, 1976). The present observation, however, sug-gests an alternative possibility, that the earth's uppermantle can be partially molten under dry conditions evenunder a normal geothermal gradient only when it containsa substantial amount of potassium.

If the upper mantle contains 1000 ppm K2O on theaverage (Ringwood, 1975), the degree of partial meltingalong the solidus of the peridotite would be on the orderof I wt.Vo,judging from the K2O content of the liquids inTable 3. Such a small amount of melt will not segregatefrom the peridotite matrix under hydrostatic conditions(Stolper et al., l98l). The presence of I wt.Vo melt in thegrain boundaries of peridotite, however, might accountfor the decrease in seismic wave velocities and theirattenuation observed in the earth's Low Velocity Zones(Anderson and Sammis, 1970; Stocker and Gordon,r97s\.

Potassic magmas represented by unusually high KzOand KA.{a ratio erupted at many localities in the world:oceanic islands, continental rift zones, and continentalshields associated with kimberlites (e.9., see Gupta andYaei, 1980). Despite the isolation of the activity in timeand space, these magmas form a distinct and uniquechemical and petrological province (Turner and Verhoo-gen, 1951; Wendlandt and Eggler, 1980). Some of thepotassic rocks relatively rich in silica (i.e., leucite phono-lite, leucite basalt) might represent the dry partial meltsformed along the solidus shown in Figure 9.

The Fe/Mg partitioning between olivine and liquid

The partitioning of Fe and Mg between olivine andmagmatic liquids has been studied extensively, and it hasbeen demonstrated that Kp(Fe/Mg)or-riq - (Fe/Mg)'ll(Fe/Mg;tie is nearly constant (0.3010.03) at one atmo-

1 5 0 0

aooa

"t6/^rnZq"7

,eo707

1 4 0 0

1 3 0 0

1 2 0 0e

1 1 0 0


b=

I

oOl

oil{

Y

Pressure (kbor )

Fig. 10. The olivine-liquid Fe/Mg partition coefficient Ko : Ge/Mg)"t/(Fe/Mg;t'c as a function of pressure (open squares). Therange at one atmosphere (0.30t0.03) is from }ioeder and Emslie (1970). A slight positive trend with increasing pressure is seen; Kp :0.30 + 0.002 P (kbar). Small solid symbols indicate the Kp values obtained in previous melting studies of peridotite with direct EeMAanalysis ofquenched liquids: circles, Kushiro (1973b), triangles, Mysen and Kushiro (1977).

0.5 0

0 .4 0

0 .3 0

0 .2 0

0 .10

sphere over a large temperature interval and variety ofliquid compositions (e.9., Roeder and Emslie, 1970;Ta-kahashi, 1978). Because of its importance in placingconstraints on igneous petrogenesis, determination of Kpat high pressures is strongly needed. The present resultscan be used to test the pressure efect on the Kp values,inasmuch as compositions of olivine and coexisting liq-uids were measured over a large pressure interval. Chem-ical equilibrium between olivine and liquid is believed tohave been established throughout the basalt melt and theperidotite layers (see Figs. 4 and 7). Ferric iron in the meltcan be neglected because of the low oxygen fugacityimposed by the graphite container.

The Kp values calculated from the analysis of liquidsand coexisting olivine (Table 3) are shown as a function ofpressure in Figure 10. The Kp values show a positivetrend with increasing pressure (Ko : 0.30 + 0.002Pkbar). Even though the Kp value is known to be insensi-tive to temperature and liquid compositions at I atm, it isnot absolutely clear whether the increase in Kp at highpressures is due to the intrinsic pressure effect, becausethe experimental temperatures and liquid compositionsare changing simultaneously with pressure, and the rangeof variation is much wider than those examined at I atm.Nevertheless, the trend is useful, because it gives therange in liquid compositions under pressure-temperature

O--O

aa

A

A

I

100 20 4030

o 8 k b , q 1 5 k b , e 2 0 k b , a 3 0 k b

Fig. ll. Normative liquid compositions obtained with thebasalt/peridotite sandwich technique (see Fig. 3). The sameprojection scheme to Figure 6. All the liquid compositionsobtained at 8, 15, 20 and 30 kbar are plotted (Table 3). Isobaricliquid compositional trends (solid lines) extend toward ol-Q join

so that they agree with the data in the system MgzSiO+-SiOz byChen and Presnall (1975).


conditions relevant to basalt magma genesis in the earth'supper mantle.

Also shown in Figure 10 are Kp values calculated fromcoexisting olivine and liquid reported in previous meltingexperiments on peridotite under dry conditions. Theprevious data are scattered and systematically lower thanthe trend obtained by the present experiments. As notedpreviously, the major reason for the discrepancy is thatthe liquid compositions reported in the previous studieshave been modified considerably by quench crystalliza-tion (Fig. l). In the present experiments we observed thatthe glasses in the peridotite matrix are heterogeneous inchemical composition regardless of run duration and aremuch poorer in MgO, lower in MgO/FeO and enriched inAl2O3 and SiO2 than the liquids quenched in the glass diskin the same charge. These lines of evidence support theconclusion of Jaques and Green (1979) who claimed thatchemical compositions of the peridotite partial melts areunquenchable in ordinary quenching experiments andthat most previously reported compositions of the meltsby direct erua analysis are greatly in error except thoseobtained near the center of large glass pods.

Based on peridotite melting experiments under hydrousconditions, Mysen (1975) reported that the Kp changesfrom 0.12 at l0 kbar to 0.30 at 30 kbar, at ll00"C andunder oxygen fugacities close to the NNO buffer. Asdescribed above, however, his liquid compositions deter-mined by direct EIMA analysis of quenched glass aresubject to much ,doubt. Furthermore, it can also beargued that his unbuffered experiments with Ag-Pd con-tainers gave oxygen fugacities which increased with runduration due to the alloying reaction of Fe in the contain-er as noted previously. His liquids may, therefore, con-

tain much greater amounts of ferric iron than he expectedafter long runs. The amount of change in the oxygenfugacity is also dependent on the modal proportion ofthemelt to the container and on various physical properties(e.9., cationic diffusion coemcients in melt); hence/o2 isa complex function of pressure and temperature. Furtherexperiments are necessary in order to clarify these pointsand to know the exact Kp values under hydrous condi-tions.

Isobaric liquid compositional trends andcomparison with previous work

Compositions of melts obtained in this study are plot-ted on a ternary projection of the normative basalttetrahedron (Fig. ll). Liquids within and close to theolivine tholeiite field of the basalt tetrahedron (Yoder andTilley, 1962) are also plotted in the Walker's ternaryprojections (Fig. 12A,B) which have been used for discus-sion of mid-ocean ridge basalt (rvrons) petrogenesis(Walker et al., 1979; Stolper, 1980). In rhose ternaryprojections, liquids obtained at each pressure lie along asingle line.

In a four component system such as crr.ras (CaO-MgO-Al2O3-SiO2), liquids coexisting with three solid phases ata given pressure are univariant, hence they must lie on asingle line in ternary projections similar to Figures I I and12. Liquids coexisting only with two solid phases, howev-er, need not lie along such isobaric liquidus boundaries inthe ternary projections. In the case of natural peridotite,melts formed after extensive partial melting contain onlysmall amounts of clinopyroxene and feldspar compo-nents. In other words, melts leaving a harzburgite residuemay differ only slightly in chemical composition at a given

OPI

.5 tb , oEkb, r t0kb , o tskb , ̂ 20kb, e25kb. o30kb, o35kb

Fig. 12. Normative liquid compositions and the isobaric liquid compositional trends in ternary projections by Walker et al. (1979).A, projection from plagioclase onto the plane olivine-diopside-quartz; B, projection from diopside to olivine-plagioclase-quartz.Liquids projected far away from the normative triangle were omirted (cf.,Fie. ll with Fig. l2B).

qzol


pressure-temperature condition even ifthey started fromdifferent lherzolite sources. The trends defined in FiguresI I and 12, which refer to liquids coexisting with the threesolid phases near the solidus and the liquids coexistingonly with olivine and orthopyroxene at higher tempera-tures, thus are considered to be a good approximation ofisobaric liquid trends on which most of the peridotitepartial melts are located.

In all the ternary projections, the location of theisobaric liquid. trends shifts systematically toward theolivine apex as pressure increases (Figs. ll and 12). Inother words, a peridotite partial melt formed at higherpressure is invariably more enriched in normative olivinecompared to the melt formed at a lower pressure and at asimilar AT from the solidus.

In his classic reviews, O'Hara (1965, 1968) estimatedthe initial melt composition of natural peridotite as afunction of pressure. The present observations showgeneral agreement with his estimation in that the liquidsobtained at higher pressures are more olivine rich thanthose at lower pressures. They support O'Hara's viewthat the initial melt formed at pressures above 15 kbar arealkali basalts. O'Hara expected that the initial melt re-turns to the olivine tholeiite field (the ol-opx-di and pl-ol-opx triangles in Fig. 12) at some pressure below 30kbar (O'Hara, 1965, Fig. l4). In the present study,however, the initial melt at 30 kbar turns out to be a verysilica undersaturated potassic melt (leucite phonolite, see

Fig. 1l). Based on the systematics in our experimentalobservations and their good consistency with experimen-tal results in simple oxide systems such as runs (Ku-shiro, 1980; Wendlandt and Eggler, 1980) and Nvas(Kushiro, 1968), it is concluded that the initial peridotitemelt becomes increasingly alkalic and silica undersaturat-ed as pressure increases at least up to 30 kbar.

In the cuns system simulating peridotite, Presnall et al.(1979) estimated initial melt compositions as a function ofpressure. Their estimated liquid compositions were plot-ted in Figure 13. Compared with the solidus melts in thecMAS system, the compositions of the melts formed alongthe solidus of HK66 are systematically silica deficient.This must be due to the presence of alkalis in the naturalrocks used in this study. Hoover and Presnall (1982) havecarried out melting experiments in the Nclraes system(Na2O-CaO-MgO-Al2O3-SiOz). According to them thecompositions of the solidus melts in the Ncues systemare more silica deficient than in cuas, and the locationsof the isobaric liquid trends virtually coincide with thoseofthe present study (Hoover and Presnall, 1982; and alsoD. C. Presnall, pers. comm., 1982).

Using the same basalt/peridotite sandwich technique,Fujii and Scarfe (1981) have studied the effect of bulkchemical composition of the source peridotite on thecomposition of the partial melts at l0 kbar. According tothem, melts coexisting with olivine + orthopyroxene +clinopyroxene of more magnesian compositions than

. s o l i d u s m e l t s i n C M A S s y s t e m

est imoted por t io t me l l comPos i t ions o lp y r o l i l c o t . 5 , . 1 0 , t r l 5 k b

opx qz ol opx

Fig. 13. Comparison of our isobaric liquid compositional trends with previous work. Projection is the same as in Fig. 12. Small solidcircles with lighrsolid lines show compositions of isobaric invariant points in the cruls system at I atm, 7 , 9 .3, ll , | 4, and 20 kbarfrom qz-rich side toward ol-rich side (Presnall et al., 1979). Solid symbols with broken lines denote isobaric liquid compositionaltrends of a pyrolite estimated by mass balance calculation by Jaques and Green (1980). Open symbols show compositions of naturalbasaltic rocks which are experimentally demonstrated to coexist with olivine + orthopyroxene * clinopyroxene at followingconditions: (circle) nronn,7.5 kbar, 1210'C, Kushiro (1971a); (triangle) Mone,8 kbar, 1225'C, Fujii and Kushiro (1977); (square)Mons, I I kbar, 1290'C, Bender et al. (1978); (diamond) alkali-olivine basalt, 14 kbar,1275"C, Takahashi (1980c); (hexagon) picrite, 25kbar, 1470'C, Elthon and Scarfe (1980).

qzol


/--\( \ MoRBs c tus te r

d i qyeroge l,lORBs aworld. OMid Atlontic, D I

oPx opx

Fig. 14. A schematic diagram illustrating a model for generation of MoRBs from a peridotite partial melt formed at l0 kbar and1275"C (point A). The areas ofthe Naonss' main cluster from Walker et al. (1979) and Stolper (1980). The average vonas compositionsfrom Melson et al. (1976). Other symbols; (normal triangle) the most magnesian glass from Leg.45, Fujii and Kushiro (1977); (opensquare) the most magnesian glass from the FAMous area, Bender e t al. (1978); (inverted triangle) average olivine-phyric tholeiite fromLeg.37,Irvine (1979); (solid square) estimated primary magma for tronrs by an olivine fractionation model, Irvine (1979); (solidhexagon) estimated primary picrite magma for naonns, Elthon and Scarfe (1980).

qzqz

those of the present study (Mg/(MB+Fe) : 0.88-0.93) areslightly more enriched in both MgO and CaO (10-13wt.Vo, and ll-13 wt.Vo, respectively) than those listed inTable 3. Their liquid compositions in the crpw normativeprojections, however, are very close to the l0 kbar liquidtrends shown in Figure 12. Based on the concordancewith the experiments by Fujii and Scarfe (1981) andHoover and Presnall (1982), it is anticipated that theisobaric liquid trends in Figure 12 are valid for peridotiteswith Mg/(Mg*Fe) from 0.85 through 1.0.

Jaques and Green (1980) estimated liquid compositionsin anhydrous melting studies of synthetic peridotites.Their liquid compositions are compared with our resultsin Figure 13. In the ol-pl-Q projection (Fig. 13B), agree-ment of the two results is relatively good but the liquids ofJaques and Green are considerably richer in normativediopside relative to ours in the ol-di-Q diagram (Fig.l3A). The liquids of Jaques and Green (1980) are alsosystematically lower in Al2O3 than ours (by about 2 to 3wt.Vo) when compared at the same AZ from the solidus ata given pressure. A possibility that might account for thediscrepancy is in the uncertainty of the estimation of theliquid compositions by Jaques and Green (1980) based onchemical compositions and modes of solid residues. Cli-nopyroxenes in their run products revealed compositionalzoning (Jaques and Green, 1980, p. 298), and they tenta-tively assumed the most calcic and low alumina clinopy-roxene as equilibrium compositions. It is likely, however,that their pyroxene compositions are metastable becauseof short run durations (see Figs. 2,4 and Table 4).

Stolper (1980) has made melting experiments with anabyssal tholeiite using harzburgite as a container. With

liquid compositions which are coexisting with olivine +orthopyroxene or olivine * orthopyroxene + clinopyrox-ene, and assuming that the composition of the liquidcoexisting with the latter three minerals is pseudoinvar-iant at a given pressure, he constructed polybaric phasediagrams for uonns. However, it is clear from thepresent study and also that ofFujii and Scarfe (1981) thatthe liquids coexisting with olivine + orthopyroxene +clinopyroxene can not be treated as pseudoinvariantpoints and the compositions of those liquids range towardmore silica deficient portions of the diagram than heconsidered (c/. Fig. 12 with Stolper's Figs. 2 and 3).

Compositions of basalts which are experimentally dem-onstrated to coexist with olivine * orthopyroxene +clinopyroxene at certain pressure-temperature conditionsare plotted in Figure 13. The experimentally bracketedpressures for those basalts are in excellent agreementwith the isobaric liquid compositional trends defined bythe present study. Based on the above observations, itmay be argued that we have shown a general set ofisobaric liquid trends in Figure 12 and the dry basalticmelts formed by partial melting of mantle peridotite areplotted on the trend at their depth of last equilibrationwith the upper mantle. And it follows that we may be ableto estimate the depth of segregation of a basalt magmafrom its chemical composition by fitting the isobaricliquid trends in Figure 12. An example of such anapplication is illustrated below.

Genesis of MoRBs from dry peridotite

The origin of mid-ocean ridge basalts (uonns) is asubject of much debate (e.9., O'Hara 1968; Green, l97l;


fncteasing Temperature (degree of rel t ing)

( be low 5 kb )

(between 5 and 15 kb)

(between 15 and 25 kb)

(above 25 kb)(K rO r i ch

Fig. 15. A petrogenetic grid showing the compositional diversity of basaltic magmas formed by partial melting of dry peridotite atvarious pressure and temperature conditions.

E75

o

a

o

b,d

60

o

Kushiro, 1973a; Presnall et al., 1979; Stolper, 1980;Elthon and Scarfe, 1980). An important consensus ex-sists, however, among petrologists and geochemists thatmost fresh MoRBs contain very little water and othervolatile constituents (Moore, 1970; Delaney et al., 1978).Experimentally determined phase relations at high pres-sures under dry conditions, therefore, are directly appli-cable for petrogenesis of Monss. It has been demonstrat-ed experimentally that some magnesian N{ones coexistsimultaneously with olivine * orthopyroxene * clinopy-roxene at pressures of7 to ll kbar (Kushiro, 1973a; Fujiiand Kushiro, 1977; Bender et al.,1978; see Figure 13 fornormative compositions of the basalts), whereas otherauthors have failed to find such a relation using similarMoRB compositions (e.9., Green et al., 1979).

Based on experimental results in the cues system,Presnall et al. (1979) predicted that initial melts ofnaturalperidotite at a pressure of about 9 kbar are close to thecomposition field of the least fractionated Naonss (see Fig.l3). Hoover and Presnall (1982) supported the aboveprediction based on their experimental results in theNcMAS system. On the other hand, Jaques and Green(1980) and Stolper (1980) concluded that peridotite partialmelts formed at about l0 kbar are notably diferent fromthe typical MoRBs based on their experiments discussedpreviously. They proposed high pressure origins for theMORBS.

The present experiments have shown that initial meltcompositions formed near the solidus of the peridotiteHK66 at 8 kbar are rich in Al2O3 and CaO (about 16 andll wt.Vo, respectively), moderate in SiO2 (49-50 wt.%),olivine normative (about l0 mole Vo) and poor in K2O.These liquids are virtually indistinguishable from themost typical MoRBs that cluster in the center of theMoRB's variation (Fig. la). At l0 kbar, the initial meltcomposition becomes more silica deficient, olivine richand away from the main Monss cluster. Compositions ofthe magnesian variety of the rvronss, however, fall close

to our liquid trend at l0 kbar (open symbols in Fig.l4A,B). Furthermore, it is important to note that overallcompositional variations of the Lronss could be producedfrom the melts formed at l0 kbar in the following way.

If a magma produced at l0 kbar and 1275"C (point A inFig. 14) rose and underwent fractional crystallization in anear surface magma chamber, the first phase to crystal-lize would be olivine (O'Hara, 1968). At point B, clino-pyroxene and plagioclase begin to crystallize togetherwith olivine, and the composition of the liquid wouldfollow the path B to C and then to D (Walker et al.,1979).In the same way, the variety of uonns could be producedby fractional crystallization in a shallow magma chamberstarting from peridotite partial melts formed at any pres-sure above 10 kbar. It is emphasized, however, that themagmas produced at 8 and l0 kbar by partial melting needonly a very small amount of crystallization to becometypical MoRBS.

The Mg-value (l00Mei(Mg+Fe2+)) of typical MoRBSare in the range between 50 and 70 and their magnesianvariety contains 200-250 ppm Ni (e.g., Schilling et al., inpress). Using the Kp partition coefficients between oli-vine and liquid, 0.32 for Fe-Mg at l0 kbar (Fig. 10, thisstudy) and 2.3 for Ni-Mg at l0 kbar (Takahashi, 1980a),compositions of olivines in equilibrium with the MoRBs at10 kbar can be calculated. According to the above calcu-Iation, a uonn with Mg-value 70, l0 wt.Vo MgO and 250ppm Ni would coexist with olivine of Fos and0.35 wt.%oNiO on its liquidus at l0 kbar. The above olivine compo-sition is within the compositional range of olivines inmantle peridotites (e.9., see Fig. 2 of Sato , 1977). In fact,such magnesian olivine phenocrysts or xenocrysts(Fos8-so) often have been reported from normal magne-sian Monss (e.g., Frey et al., 1974). Whether the magne-sian Monss actually are direct partial melt products orhave undergone a small amount oflow pressure fraction-ation needs further clarification. But it is less likely thatthey have evolved from picritic primary magmas after

. a l k a l i p i c r i t e , p j c r i t e ,Dasa I ts I

hyp - r i ch p ' i c r i t e , pe r i do t i t i c koma t i i t e

hyp - r i ch p i c r i t e , pe r i do t i t i c koma t i i t e

hyp - r iCh p ' i c r i t e , pe r i do t i t i c koma t i i t e

hyp - r i ch p i c r i t e , pe r i do t i t i c koma t i i t e

qz thole i i te

ol thole i i te

a l ka l i - o l basa l t , o l t ho l e i i t e ,


extensive olivine fractionation (O'Hara, 1968: Irvine.1979; Green et al., 1979; Elthon and Scarfe, 1980),because such olivine fractionation would have eliminatednickel from the magmas (Sato, 1977; Hart and Davis,1978).

Compositional diversity of magmas formed bypartial melting of dry peridotite

Based on the present experimental results a petrogenet-ic grid was constructed illustrating the compositionaldiversity of magmas formed by partiat melting of dryperidotite at pressures of5 to 30 kbar (Fig. 15). Along thesolidus of dry peridotite, a variety of basalts such asquartz tholeiite, olivine tholeiite, alkali-olivine basalt andalkali picrite are produced. Very potassic magmas equiv-alent to leucite basalt and leucite phonolite can be pro-duced at relatively low temperatures (1300-1400'C) in thepressure range above about 20 kbar. The amount of suchpotassic melt, however, is strongly dependent on the K2Ocontent ofthe peridotite. In the case ofperidotite deplet-ed in alkalis, only picritic tholeiite melt could be producedin appreciable amount at pressures above 25 kbar.

A very silica deficient variety of alkali-mafic magmas(basanite, melilitite, nephelinite), however, can not bederived from dry peridotite by partial melting alone. Thelowest SiO2 value of the liquid obtained in our experi-ments is 46.5 wt.Vo (Run #47). Thus, the highly silica-deficient magmas may be partial melt products of perido-tite with CO2 (Eggler, 1978; Wendlandt and Mysen,1980).

High-alumina basalts (occcasionally with 18 wt.%Al2O3 or more) which occur in island-arcs and continentalmargins have been identified as a primary basalt magmatype (Kuno, 1960). Compared to the liquids of the presentstudy, they are closest to the solidus melt at 8 to 15 kbarexcept that Kuno's high-alumina basalts are more en-riched in normative plagioclase components. The high-alumina basalts in the oceanic belts subduction zonesmay be partial melt products of peridotite under slightlywet conditions (Takahashi et al., l98l).

Conclusions

(1) The solidus curve of a dry peridotite HK66 hascusps at points where the subsolidus phase assemblagechanges, and it is lower in temperature by about 100'Cthan previously determined.

(2) Along the solidus the composition of partial melt ofthe peridotite changes; quartz tholeiite at 5 kbar, olivinetholeiite at 8 and l0 kbar, alkali-olivine basalt at 15 and 20kbar, and alkali picrite at25 and 30 kbar.

(3) Addition of potassium feldspar component to rheperidotite, lowers the dry solidus significantly, and itsextension might intersect the normal oceanic geotherm atpressures between 30 and 50 kbar. The solidus curve ofthe potassic peridotite is considered to be the true drysolidus of the earth's upper mantle.

(4) At temperatures between the two solidi, a variety ofpotassic magmas such as leucite phonolite and leucitebasalt are produced as incipient melt of dry mantleperidotite.

(5) The olivine/liquid Fe-Mg partition coemcient showsa slight positive trend with increasing pressure Ko = 0.30+ 0.002 P (kbar). Judging from the Kp values, mostpreviously reported glass compositions in peridotite melt-ing experiments by direct ErMA analysis are considerablyin error.

(6) In normative projections, isobaric liquid composi-tional trends have been defined based on the presentexperiments. The isobaric trends shift systematicallytoward the olivine apex as pressure increases and theinitial melt compositions become increasingly silica-defi-cient. These trends are consistent with previous meltingstudies in synthetic systems and those with natural ba-salts.

(7) The partial melting model of mantle peridotite atpressures of 8 to l0 kbar for origin of uonns is supported,both because compositions of the melts formed near thesolidus at above pressures are very close to typicalmagnesian MoRBs and because the observed diversity ofthe vonns can be produced from the partial melts byfractional crystallization at shallower depths.

Acknowledgments

We are grateful to Dr. H. S. Yoder, Jr., director of theGeophysical Laboratory, Carnegie Institution of Washington,where most of the present experiments were made. We arethankful to Drs. F. R. Boyd and B. O. Mysen of the GeophysicalLaboratory for providing their experimental facilities. Specialthanks are due to Drs. T. N. Irvine, A. Navrotsky, and D. C.Presnall who contributed greatly to improve the manuscript bycareful review. We also appreciate criticisms and comments onthe manuscript by Drs. F. Chayes, P. A. Danckwerth, T. Fujii,M. J. Holdaway, S. Jakobsson, A. L. Jaques, Y. Matsui, andC. M. Scarfe at various stages of its preparation. Thanks are alsodue to Mr. C. G. Hadidiacos who assisted the electron micro-probe work.

References

Anderson, D. L. and Sammis, C. (1970) Partial melting in theupper mantle. Physics ofthe Earth and Planetary Interiors, 3,41-50.

Bender, J. F, Hodges, F.N. and Bence, A. E. (1978) Petrogene-sis of basalts from the project nevous area: experimentalstudy from 0 to 15 kbars. Earth and Planetary Science Letters,41,277-302.

Bowen, N. L. (1913) Melting phenomenon of plagioclase feld-spar. American Journal of Science (4th series), 35,577-599.

Bowen, N. L. (1928) The Evolution of the Igneous Rocks.Princeton University Press, New Jersey.

Boyd, F. R. and England, J. L. (1960) Apparatus for phase-equilibrium measurements at pressures to 50 kilobars andtemperatures up to 1750'C. Journal of Geophysical Research,65,74t-:748.

Chen, C. H. and Presnall, D. C. (1975) The system Mg2SiO4-


SiO2 at pressures up to 25 kilobars. American Mineralogist,60,398-406.

Clark, S. P. and Ringwood, A. E. (1964) Density disrributionand constitution of the mantle. Review in Geophysics, 2, 35-88 .

Crank, J. (1956) The Mathematics of Diffusion. Oxford Universi-ty Press, London.

Delaney, J. R., Muenow, D. W. and Graham, D. G. (1978)Abundance and distribution of water, carbon and sulfur in theglassy rims of submarine pillow basalts. Geochimica et Cos-mochimica Acta. 42. 581-594.

Eggler, D. H. (1976) Does CO2 cause partial melting in the lowvelocity layer of the mantle? Geology, 4, 69-72.

Eggler, D. H. (1978) The effect of CO2 upon partial melting ofperidotite in the system Na2O-CaO-Al2O3-MgO-SiOr-CO, to35 kb, with an analysis of melting in a peridotite-H2o-Co2system. American Journal of Science, 278. 3,05-34j.

Elthon, D. and Scarfe, C. M. (1980) High-pressure phase equi-libria of a high-magnesia basalt: implications for the origin ofmid-ocean ridge basalts. Carnegie Institution of WashingtonYear Book. 79.277-281.

Erlank, A. J. and Kushiro, I. (1970) Potassium contents ofsynthetic pyroxenes at high temperatures and pressures. Car-negie Institution of Washington Year Book, 68,233-236.

Frey, F. A., Bryan, W. B. and Thompson, G. (1974) Atlanticocean floor: Geochemistry and petrology ofbasalts from Leg 2and 3 of the Deep-Sea Drilling Project. Journal of GeophysicalResearch. 79. 5507-5527 .

Fujii, T. and Kushiro, l. (1977) Melting relations and viscosity ofan abyssal tholeiite. Carnegie Institution of Washington yearBook,76, 461-465.

Fujii, T. and Scarfe, C. M. (l9El) Partial melting of spinellherzolite and its bearing on the origin of uonrs. (abstr.)Geological Society of America Abstracts with programs, 13,456.

Green, D. H. (1971) Compositions of basaltic magmas as indica-tors of conditions of origin: application to oceanic volcanism.Philosophical Transaction of the Royal Society of London,268a,707-:725.

Green, D. H. (1973) Experimental melting studies on a modelupper mantle composition at high pressure under water-saturated and water-undersaturated conditions. Earth andPlanetary Science Letters, 19, 37-53.

Green, D. H. and Ringwood, A. E. (1967a) The genesis ofbasaltic magmas. Contributions to Mineralogy and Petrology,15, 103-190.

Green, D. H. and Ringwood, A. E. (1967b) The stabiliry ofaluminous pyroxene peridotite and garnet peridotite and theirrelevance in upper mantle structure. Earth and planetaryScience Letters. 3. 15l-160.

Green, D. H., Hibberson, W. O. and Jaques, A.L. (19j9)Petrogenesis of mid-ocean ridge basalts. In M. W. McElhinny,Ed., The Earth: Its Origin, Structure and Evolution, p. 265-29. Academic Press, London.

Gupta, A. K. and Yagi, K. (1980) Petrology and Genesis ofLeucite-bearing Rocks. Springer-Verlag.

Hart, S. R. and Davis, K. E. (1978) Nickel partitioning betweenolivine and silicate melt. Earth and Planetary Science Letters,40,203-219.

Hofmann, A. W. (1980) Diffusion in natural silicate melts: acritical review. In R. B. Hargraves, Ed., Physics of MagmaticProcesses, p. 385-418. Princeton University press, New Jer-sey.

Hoover, J. D. and Presnall, D. C. (1982) Melting relations ofsimplified peridotite in the SiOz-CaO-Al2O3-MgO-Na2O(scerr,tN) system from I atm to 20 kb-II. Results and applica-tion to basalt generation. (abstr.) Geological Society of Ameri-ca Abstracts with Programs, 14, 517.

Irvine, T. N. (1977) Chromite crystallization in the joinMg2SiOa-CaMgSizOe-CaAlzSi2Os-MgCrrOa-SiO2. CarnegieInstitution of Washington Year Book, 76, 465-472.

Irvine, T. N. (1979) Rocks whose composition is determined bycrystal accumulation and sorting. In H. S. Yoder Jr., Ed., TheEvolution of the Igneous Rocks, p. 245-306. Princeton Uni-versity Press, New Jersey.

Irvine, T. N. and Baragar, W. R. A. (1971) A guide to thechemical classification of the common volcanic rocks. Canadi-an Journal of Earth Sciences. 8. 523-548.

Ito, K. and Kennedy, G. C. (1968) Melting and phase relations inthe plane tholeiite-lherzolite-nepheline basanite to 40 kilobarswith geological implications. Contributions to Mineralogy andPetrology, 19, 177-211.

Jaques, A. L. and Green, D. H. (1979) Determination of liquidcompositions in experimental, high pressure melting of perido-tite. American Mineralogist, 64, 1312-1321.

Jaques, A. L. and Green, D. H. (1980) Anhydrous melting ofperidotite at 0-15 kb pressure and the genesis of tholeiiticbasalts. Contributions to Mineralogy and Petrology, 73, 287 -3 1 0 .

Kuno, H. (1960) High-alumina basalt. Journal of Petrology, 1,t2l-145.

Kushiro, I. (1968) Composition of magmas formed by partialzone melting of the earth's upper mantle. Journal of Geophysi-caf Research. 73. 619-634.

Kushiro, l. (1973a) Origin of some magmas in oceanic andcircum-oceanic regions. Tectonophysics, 17, 2ll-222.

Kushiro, I. (1973b) Partial melting of garnet lherzolites fromkimberlite at high pressures, In P. H. Nixon, Ed., LesothoKimberlites, p. 294-29. National Development Corporation,Lesotho.

Kushiro, l. (1974) Melting of hydrous mantle and possiblegeneration of andesitic magma: an approach from syntheticsystems. Earth and Planetary Science Letters, 22,294-299.

Kushiro, I. (1975) On the nature of silicate melt and its signifi-cance in magma genesis: regularities in the shift of the liquidusboundaries involving olivine, pyroxene and silica minerals.American Journal of Science, 275, 4ll-431.

Kushiro, L (1976) A new furnace assembly with a small tempera-ture gradient in solid-media, high-pressure apparatus. Carne-gie Institution of Washington Year Book, 75, 832-833.

Kushiro, I. (1980) Changes with pressure of degree of partialmelting and K2O content of liquids in the system MgzSiO+-KAlSiO4-SiOr. Carnegie Institution of Washington YearBook.79. 267-271.

Kushiro, I. and Kuno, H. (1963) Origin of primary basaltmagmas and classification of basaltic rocks. Journal of Petrol-ogy , 4, 7 5-89 .

Kushiro, I., Syono, Y. and Akimoto, S. (1968) Melting of aperidotite nodule at high pressures and high water pressures.Journal of Geophysical Research, 7 3, 6023-6029.

Kushiro, I., Shimizu, N., Nakamura, Y. and Akimoto, S. (1972)Compositions of coexisting liquid and solid phases formedupon melting of natural garnet and spinel lherzolites at highpressures: A preliminary report. Earth and Planetary ScienceLetters. 14. 19-25.


Lambert, L B. and Wyllie, P. J. (1968) Stability of hornblendeand a model for the low velocity zone. Nature ,219, 1240-1241 .

Lindsley, D. H. (1980) Phase equilibria of pyroxenes at pres-sures )l atmosphere. In C. T. Prewitt, Ed., Reviews inMineralogy, 7, Pyroxenes, p. 289-308. Mineralogical Societyof America, Washington D.C.

McCallister, R. H. (1980) Determinations of major cationic diffu-sion constants in pyroxenes. (abstr.) Geological Society ofAmerica Abstracts with Programs , 13 , 479.

Melson, W. G., Val l ier, T. L., Wright, T. L., Byerly, G. andNelen, J. (1976) Chemical diversity of abyssal volcanic glasserupted along Pacific, Atlantic, and Indian ocean spreadingcenters. In G. H. Sutton et a/., Eds., Geophysical Monograph,19, The Geophysics of the Pacific Ocean Basin and its Margin,p.351-367. American Geophysical Union, Washington D.C.

Misener, D. J. (1974\ Cationic diffusion in olivine to 1400'C and35 kilobars. In A. W. Hofmann, Ed., Geochemical Transportand Kinetics, p. ll7-129. Carnegie Institution of Washington,Washington D.C.

Moore, J. G. (1970) Water content of basalt erupted on the oceanfloor. Contributions to Mineralogy and Petrology, 28,272-279.

Morioka, M. (1981) Cation diffusion in olivine-Il, Ni-Mg, Mn-Mg, Mg and Ca. Geochimica et Cosmochimica Acta, 45, 1573-1580.

Mysen, B. O. (1975) Partitioning of iron and magnesium betweencrystal and melts in peridotite upper mantle. Contributions toMineralogy and Petrology, 52, 69J6.

Mysen, B. O. and Boettcher, A. L. (1975) Melting of a hydrousupper mantle: II, Geochemistry of crystal and liquids formedby anatexsis of mantle peridotite at high pressure and hightemperatures as a function of controlled activities of water,hydrogen, and carbon dioxide. Journal of Petrology, 16,549-593.

Mysen, B. O. and Kushiro, l. (197'7) Compositional variations ofcoexisting phases with degree of melting of peridotite in theupper mantle. American Mineralogist, 62, 843-865.

O'Hara, M. J. (1965) Primary magmas and the origin of basalts.Scottish Journal of Geology, l, 19-40.

O'Hara, M. J. (1968) The bearing of phase equilibria studies insynthetic and natural systems on the origin and evolution ofbasic and ultrabasic rocks. Earth-Science Review, 4, 69-134.

Pollack, H. N. and Chapman, D. S. (1977) On the regionalvariation of heat flow, geotherms, and lithospheric thickness.Tectonophysic s, 38, 27 9 -296.

Presnall, D. C. and Brenner, N. L. (1974) A method for studyingiron silicate liquids under reducing conditions with negligibleiron loss. Geochimica et Cosmochimica Acta. 38. 1785-1788.

Presnall, D. C. and Hoover, J. D. (1982) Melting relations ofsimplified peridotite in the SiOrCaO-AlzOr-MgO-NazO(sc^r.ruN) system from I atm to 20 kb-I. Methodology. (abstr.)Geological Society of America Abstracts with Programs, 14,593.

Presnall, D. C., Dixon, J. R., O'Donell, T. H. and Dixon, S. A.(1979) Generation of mid-ocean ridge tholeiites. Journal ofPetrology, 20,3-35.

Ringwood, A. E. (1975) Composition and Petrology of theEarth's Mantle. McGraw-Hill, New York.

Roeder, P. L. and Emslie, R. F. (1970) Olivine-liquid equilibri-um. Contributions to Mineralogy and Petrology, 29,275-289.

Sato, H. (1977) Nickel content of basaltic magma: identificationof primary magmas and measure of the degree of olivinefractionation. Lithos, 10, ll3-120.

Schilling, J.-G., Zajac, M., Evans, R., Johnston, T., White, W.,Devine, J. D. and Kingsley, R. (in press) Petrologic andgeochemical variations along the Mid-Atlantic Ridge from29'N to 73'N. American Journal of Science.

Shiraki, K. and Kuroda, N. (1977) The boninite revisited.Joumal of Geography, Tokyo Geographical Society, 86, 174-190 (in Japanese).

Stocker, R. L. and Gordon, R. B. (1975) Velocity and internalfriction in partial melts. Journal ofGeophysical Research, 80,4828-4836.

Stolper, E. (1980) A phase diagram for mid-ocean ridge basalts:preliminary results and implications for petrogenesis. Contri-butions to Mineralogy and Petrology, 74, 13-27.

Stolper, E., Walker, D., Harger, B. H. and Hays, J. F. (1981)Melt segregation from partially molten source regions: theimportance of melt density and source region size. Journal ofGeophysical Research, 86, 6261-627 l.

Takahashi, E. (1978) Partitioning of Ni2*, Co2+, Fe2*, Mn2* andMg2* between olivine and silicate melts: compositional depen-dence of partition coefficient. Geochimica et CosmochimicaActa, 42, 1829-1844.

Takahashi, E. (1980a) Olivine/liquid nickel partitioning at highpressures: experiments with an olivine capsule (abstr.) EOS,Transaction of the American Geophysical Union, 61, 397.

Takahashi, E. (1980b) Thermal history of lherzolite xenoliths-I,Petrology oflherzolite xenoliths from the Ichinomegata crater,Oga peninsula, Northeast Japan. Geochimica et Cosmochi-mica Acta. 44. l6y''3-1658.

Takahashi, E. (1980c) Melting relations of an alkali-olivine basaltto 30 kbar and their bearing on the origin of alkali basalticmagmas. Carnegie Institution of Washington Year Book, 79,271-276.

Takahashi, E., Kushiro, I. and Tatsumi, Y. (1981) Melting of aperidotite at high pressures and its bearing on island arcmagmas. (abstr.) International Association of Volcanologyand Chemistry of the Earth's Interior, p. 365-366, Tokyo.

Takahashi, E., Yamada, H. and Ito, E. (1982) A new furnaceassembly to 100 kbar and 1500'C with minimum temperatureuncertainty. Geophysical Research Letters, 9, 805-807.

Thompsom, R. N. and Kushiro, L (1972) The oxygen fugacitywithin graphite capsules in piston-cylinder apparatus at highpressures. Carnegie Institution of Washington Year Book, 71,615-616.

Tsuchiyama, A. and Takahashi, E. (in press) Melting kinetics ofa plagioclase feldspar. Contribution to Mineralogy and Petrol-ogy.

Tumer, F. J. and Verhoogen, J. (1951) Igneous Petrology.McGraw-Hill, New York.

Walker, D., Shibata, T. and Delong, S. E. (1979) Abyssaltholeiites from the Oceanographer fracture zone. II, Phaseequilibria and mixing. Contributions to Mineralogy and Petrol-ogy,70, l l t -125.

Watson, E. B. and Bender, J. F. (1980) Diffusion of cesium,samarium, strontium and chlorine in molten silicate at hightemperatures and pressures. (abstr.) Geological Society ofAmerica Abstracts with Programs , 13, 545.

Wendlandt, R. F. and Eggler, D. H. (1980) The origins ofpotassic magmas: I, Melting relations in the systems KAlSiO4-MgzSiOq-SiOz and KAISiO+-MgO-SiO2{O2 to 30 kilobars.American Journal of Science. 280. 385-420.

Wendlandt, R. F. and Mysen, B. O. (1980) Melting phase rela-tions of natural peridotite + CO2 as a function of degree ofmelting at l5 and 30 kbar. American Mineralogist, 65, t7-44.

TAKAHASHI AND KUS.HIRO: PERIDOTITE MELNNG E79

Wyllie, P. J. Qn, Experimental petrology arid global tecton- magmas: an experimental study of rtatural and syntheticics-A preview. Tectonophysics, 17, 189-209. systems. Journal of Petrology, 3,342-532.

Yoder, IL S. Jr. (1976) Generation of Basaltic Magma. NationalAcademy of Science, Washington D,C. Manuscript recei'ved, June 17, 1982;

Yqder, H. S. Jr. and Tilley; C. E- (1962) Origiir of basalt acceptedfot public.atioi, February 28, 1983.

Documents

Melting of a dry peridotite at high pressures and basalt ... · Melting of a dry peridotite at high pressures and basalt magma genesis ErIcst Tera.uesul Institute for Thermal Spring