American Mineralogist, Volume 66, pages 1164-1174, 1981

lnfluence of COz on melting of model granulite facies assemblages:a model for the genesis of charnockites

Rtcn^lno F. Wer.tlrnNor

Lunar and Planetary Institute3303 NASA Roqd l, Houston, Texas 77058

Abstract

Partial melting studies at crustal pressures in SiO2-rich portions of the system KAlSiO4-

Mg2SiOa-H2O-CO2 can be used to model the anatectic origin of charnockites. The

univariant reaction

phlogopite i sanidine + quartz + vapor: enstatite + liquid

produces a SiOz-rich melt (granite analog) at 3 kbar; the vapor composition at the solidus isbuffered to high HzO-contents by the coexistence of phlogopite with its breakdownproducts. At higher pressures, 8 and 15 kbar, the fluid phase is buffered to higher CO2-

contents and the melt composition becomes enriched in K2O and MgO (charnockite

analog). Melting relations are controlled by the expansion of the quartz liquidus field

relative to the enstatite and sanidine fields with increasing pressure. Partial melts generated

at the base of the crust in the presence of a CO2-rich fluid will be of an alkaline nature and

will crystallize enstatite at lower pressures.COz-saturated melting of similar SiO2-rich bulk compositions (phlogopite-absent) by the

reaction

enstatite + sanidine * quartz + CO2 = liquid

occurs at temperatures in excess of 1000'C to about 15 kbar. Liquid compositions showanalogous trends, however, with increasing pressure, to those observed in the 5-compo-

nenr sysrem as a consequence of the expansion of the quartz liquidus surface relative to the

enstatite surface.This partial melting model for charnockite genesis satisfies the constraints of observed

charnockite mineralogies, P and T estimations for charnockite assemblages (5-12 kbar and

750"-1000"C), and reports of high-temperature CO2-rich fluid inclusions that are believed to

approximate solidus vapor compositions (Ormaasen, 1977; KOnnerup-Madsen, 1979).

Also, observations of anatexis preceding granulite facies metamorphism are reconciled:

Influx of CO2 vapor (to a melt-crystal system) in sufficient amount to exhaust the vaporphase buffer will result in H2O extraction from the melt and crystallization of the melt,

although temperatures may continue to rise.

IntroductionCharnockites form a series of ultrabasic to felsic

composition rocks, containing orthopyroxene, clin-opyroxene, mesoperthite (or perthite), orthoclase,plagioclase, and quartz, as well as biotite, garnet,amphibole, olivine, or opaques, which crystallized(or recrystallized) during high-grade metamorphicconditions (Pichamuthu , 1969). They are often tem-porally and spacially associated with massif-typeanorthosites. Models for the genesis of charnock-ites typically fall in 5 general categories:

0m3-004x/8 I /1 l l 2-l 164$02.00

(1) Intrusion of water-undersaturated magma indry crust during, or with subsequent, granulitefacies metamorphism (Martignole, 1979);

(2) in situ dry anatexis during granulite faciesmetamorphism (Martign ole, 1979) ;

(3) charnockites are the residua after removal ofgranitic melt (Pride and Muecke, 1980; Nesbitt,1980);

(4) charnockites are the cumulates of igneousfractionation (Field et al., 1980);

(5) granulite metamorphism involving volatiles

n64

WENDLANDT: INFLT]ENCE OF CO2 ON MELTING oF GRANT]LITE I 165

containing little HzO (primarily CO2), and possiblyoccurring at high crustal levels (Smith et al., 1979:Newton et al.,1980).

In this paper, -experimental data on H2O-deficientmelting of granulite facies assemblages are pre-sented. It is not an objective of this paper to selectbetween the alternative modes of origin, all modelsbeing conceivable, but rather, the data will be usedto constrain all the models.

Physical and chemical constraints on charnockitegenesN

Observed mineralogies, measured and estimatedvolatile compositions, and thermobarometry esti-mates can be used to constrain conditions of char-nockite genesis. The occurrence of orthopyroxeneis an essential feature of the charnockite mineralasemblages (Pichamuthu, 1969). Its presence isreconciled in either of two ways: (a) by conditionsof low afri$ (or low afrl$, if an occurrence ismetamorphic in origin) resulting in orthopyroxenecrystallization, whereas amphibole or biotite crys-tallization is favored by high ay1,s, or (b) orthopy-roxene crystallization may refleCt more significantdifferences in composition between charnockitesand granites than volatile content alone. In otherwords, charnockites may be compositionally dis-

tinct from granites giving rise to different mineralo-gies. Figure I shows SiO2 vs.(Mg,Fe)O vs.(Na,K)zO + AlzO3 (recalculated to 700Vo, wt.) forcharnockites and granites from the Adirondacks(Letteney, 1968) and SW Sweden (Hubbard andWhitley, 1979). If considered as a quartz ys. pyrox-ene-component ys. feldspar-component pseudoter-nary diagram, then charnockites are noticeablyenriched in both ferromagnesian and feldspathiccomponents relative to granites. Whether this dis-tinction relates to a difference in genetic mechanismas well, has not been determined; however, varia-tion of relative proportions of volatiles, especiallyH2O and CO2, in the source regions may be asignificant genetic variable inasmuch as volatilecompositions have been shown to have a strongeffect on melt compositions (e.g., Mysen andBoettcher, 1975) and liquidus phase relations, ingeneral (Eggler, 1974, 1975; Wyllie and Huang,1975; Wendlandt and Eggler, l980a,b).

Although some charnockite occurrences are be-lieved to approach H2O-saturation in the terminalstages of crystallization (Martignole, 1979), charnockites are thought to be the products ofrelativelydry processes. Mineral assemblages in both char-nockites and granulites do not contain appreciablequantities of hydrous phases suggesting low as,s. A

t 0 c h a r n o c k i t e s

a o g r a n i t e s

+ * t r a n s i t i o n a l g r a n i t e s

K2o+Na2O + Al2O3

s i02

wt pe rcen t

comoi led l rom Let teTy (1968) , Hubb€rd & Whi i ley (1979)

Fig. l. SiO2 vs. MgO * FeO vs. K2O + Na2O + Al2O3 for charnockites and granites from southwest Sweden (Hubbard andWhitley' 1979) and the Adirondack Mountains (Letteney, 1968). Transitional granites are intermediare in composition betweengranites and charnockites; they are charnockitic in mineralogy, and either spacially transitional into both granites and charnockites(Hubbard and Whitley, 1979) or spatially distinct from both homogeneous charnockite and granite (Letteney, 1968).

( M g , F e ) O

I 166 WENDLANDT: INFLUENCE OF CO2 ON MELTING OF GRANULITE

Table 2: Thermobarometry constraints on charnockite genesisfurther observation, and one that deserves addition-al close scrutiny, is that high-temperature COz-richfluid inclusions occur in both charnockite (Ormaa-sen, 19771' Konnerup-Madsen, 1979) and granulitefacies minerals (Touret, l97lb) suggesting higha6sr. Unfortunately, the interpretation of whatthese CO2-rich inclusions represent and how theyoriginate is ambiguous. The proposed interpreta-tions, summaized in Table 1, include suggestionsthat CO2-rich fluid inclusions may be unrelated tocharnockite genesis, due instead to oxidation ofgraphite, loss of H2 from HzO-COz fluid inclusionsby ditrusion, or decarbonation of carbonate-bearingcountry rocks. A number of investigators contend,however, that the COz-rich inclusions are juvenileand their compositions may approximate peakmetamorphic or magmatic fluid compositions. Forthe purposes of this study, it is assumed that thelatter interpretation is correct at least in someinstances.

Also bearing on the nature of volatiles associatedwith charnockite genesis is the observation ofBlattner (1980) of high Cl and F contents in char-nockites from Kondapalli, India.

Estimates of pressures and temperatures for theformation or crystallization of charnockites by igne-ous processes (Table 2) define a window from about4 to 12 kbar and 750'to 1000'C; those by metamor-phic processes may be formed at somewhat lowertemperatures. The lowest estimated magmatic tem-peratures (ca. 750'C) are 100o-150'C above HzO-saturated granite solidi for this pressure window(Tutt le and Bowen, 1958; Piwinski i , 1968;Boettcher and Wyllie, 196E) implyitg attzo signifi-cantly less than unity.

Experiments were designed to test the hypothesisthat compositional features of felsic charnockitesmay be generated by partial melting of a HzO-

T O a Source o f es t imate

2-47-r0"125-6

800-900800-900-1000700-8001000 i. 100

'600750-800

5 a x e n a ( r v l / . ,t v ta r t igno le (1979)ormaasen ( 1977)Konnerup-l ' ladsen ( 1979)Boh len and Essene (1978)S m i t h e t a l . ( 1 9 7 9 )J a f f e e t a l . ( 1 9 7 8 )

3-59-12

undersaturated source over a P-Zinterval compati-ble with thermobarometry estimates.

Experimental model

Most models for charnockite genesis invoke acrustal derivation from salic gneiss or granulitefacies material of pelitic or greywacke composition(e.g., Hubbard and Whitley, 1979). The SiO2-richportion of the system KAlSiO4-MgzSiO+-SiOz hasbeen selected as a model for the volatile-absentsource composition; a potassic phase in the sourceis assumed. In Figure 2, the starting compositions(solid dots) are shown relative to the l-atm liquidussurface (solid lines) as determined by Schairer andBowen (1955) and Luth (1966)' All four composi-tions have protoenstatite as a liquidus phase at 1-atm. For anhydrous conditions, or when only COzis present at crustal pressures, all compositionsconsist of orthopyroxene, feldspar and quartz, asshown by the 3-phase triangle, sanidine-enstatite-quartz. If any H2O is present in the source composi-tion, then phlogopite is stabilized as a solidus phaseat pressures above about 0.5 kbar (Luth, 1967). Thisalternative assemblage is shown by the other 3-phase region (projected from H2O), phlogopite-sanidine-quartz. Both assemblages are models forsimple ganulite facies assemblages.

Volatile-absent, CO2-saturated, and H2O-under-

Table 1: Possible origins of CO2-rich fluid inclusions in charnockite and granulite minerals

( 1 )

( 2 \

/ ? \

( 4 )

( s )

Sol idus vapor composi t ' ions

Juven i l e ( uppe r man t l e ) o r i g i n

oxidat ion of graphi te or organic mater ia l in country rocks

Decarbonat ion of carbonate-bear ing country rocks

Dehydrat ion of COr-Hr0 f lu id by the react ion

Pyroxene + (Hr0-C02)r . , = Hornblende + (C02)r"1

D i f f us i ona l l oss o f H , f r om m jxed vo la t i l e i nc l us i ons

Evolved f rom proximal anorthosi te or any basic magma bodyemolaced in the lower crust

0rmaasen (1977), Konnerup-Madsen (1979)

Hoe ts ano l ou re ! [ r v l 5 ,

Madsen ( 1977)

Ho l l i s t e r and Bu r russ (1976 )

Ashwa l ( 1978 ) , Sm i th e t a ] . ( 1979 )

( 6 )

( 7 )

fire25iOo

II

\II

WENDI.ANDT: INFLUENCE OF CO2 ON MELTING OF GRANULITE r167

Ks KAl5i2o6 so

Ic

KAlSiO4-Me25iO4-5iO2

5io2

Qz,lt,Cr

Fig. 2. Starting compositions (solid dots) ae shown relative tothe I atm liquidus surface (Schairer and Bowen, 1955; Luth1966). Dashed l ines show possible sol idus assemblages(discussed in text).

saturated (mole fraction of H2O in the vapor phaseless than unity) melting relations have been deter-mined for a pressure range coffesponding to depthsexceeding the thickness of the crust. Examples ofthe latter two types of reactions are shown schemat-ically in Figure 3. This isobaric section depicts thesolidus (heavy line) and subsolidus assemblages forone of the compositions. The CO2-saturated systemmelts at the temperature on the extreme left of thesection. The addition of H2O stabilizes phlogopite,and where both phlogopite and enstatite coexist inthe subsolidus, the phase rule predicts that univar-iant melting will occur. Furthermore, the vaporcomposition is buffered at a unique value for a givenpressure and temperature in the region meaning thatchanging the HzOlCO2 ratio within this regionchanges the proportions of phases but not theassemblage or phase compositions. Two points areof interest:

(1) The interval of univariant melting expands tocover a wider range of volatile contents as the totalamount of volatiles in the system decreases (Eggler,1977b); and

(2) the vapor composition in the subsolidus isbuffered to higher CO2 contents with increasingpressure (Eggler, 1977b; Wendlandt and Eggler,1980b). At temperatures above the solidus. forpressures corresponding approximately to the

crust, the melt preferentially dissolves H2O relativeto COz; hence, if a volatile-rich fluid phase ispresent, then it is likely to be even mor6 CO2-rich.Furthermore, the lower the amount of total vola-tiles, the greater the compositional difference be-tween the dissolved and fluid components (Kadikand Lukanin,1973; Eggler and Rosenhauer, 1978).The two melting reactions I have determined thatare most applicable to charnockite genesis are: (1)The CO2-saturated solidus; and (2) the univariantsolidus characteized by high X6e"relative to H2O.Water-saturated melting occurs at even lower tem-peratures than this univariant portion of the solidusbut has not been determined as most evidencesummarized previously indicates that charnockiteswere not produced at high aH,s conditions.

Existing experimental work applicable to thisstudy include determinations of a COz-saturatedgranite solidus at low pressures (less than 3 kbar) byWyllie and Tuttle (1959) and a granite solidus for arange of CO2[H2O compositions at pressures lessthan 5 kbar by Novgorodov and Shkodzinskiy(r974).

Fig. 3. Schematic T-X section for one of the startingcompositions depicting the solidus configuration (heavy solidline) and subsolidus assemblages. Where phlogopite andenstatite coexist in the subsolidus, the vapor composition isbuffered at a unique value for a given pressure and temperatureand the solidus is isobarically invariant. The hypotheticalpressure for this section is assumed to be greater than 0.5 kbar.

T

S"3, - En . , r - QzUO

P h S a E n Q z V

P h S a Q z V

l t 6 8 WENDLANDT: INFLUENCE OF CO2 ON MELTING OF GRANULITE

Experiment, ", orXiLXll'uoou" 10 kbar were

done with a solid-media, high-pressure apparatus(Boyd and England, 1960) using a "hot piston-out"technique without additional correction for the ef-fect of friction on pressure. Talc-Pyrex furnaceassemblies (1.27 cm diam) were used throughout.Nominal pressures were maintained at +0.1 kbar.

Temperatures were measured with Pt-PteeRhlsthermocouples and automatically controlled to-r2"C. No corrections have been made for the effectof pressure on the emf output of the thermocouple.

Experiments at pressures below 10 kbar weredone with an internally heated, gas-media appara-tus. Temperatures were measured with Pt-PtegRhlg.thermocouples and pressures were measured bystrain gauge. Temperatures are precise to 13'C andpressures are precise to at least -r100 bars.

Starting materials were synthetic crystallinephlogopite (Ph), enstatite (En), sanidine (Sa), mag-nesite (Mag), and quartz (Qz). Anhydrous phlogo-pite component (APh) consisted of crystalline for-sterite, kalsilite, and leucite. Details of phasepreparation have been reported elsewhere (Wend-landt and Eggler, 1980a). Starting mixtures wereof the anhydrous compositions Sa26En1aQz6s,Sa32En13Qz5q, SaasEn22Qz3s, and Sa36F-na2Qz22(wt). Two addit ional mixtures were used,(Phr5+APhr5)3sSa2sQz5s, which is a partially hy-drated equivalent of SaasEn2zQzrs, and Sa3sMag3sQzaq, which is the carbonated equivalent ofSa36Ena2Qzzz (shown in Fig. 2). Carbon dioxide wasadded as Ag2C2Oa or as MgCO:. A Xco, - 0.5 (molfraction) and a total volatile content of less than 10wt.Vo (H2O+COz) was found (by trial and error) tobe optimal for investigating the H2O-undersaturatedmelting reaction; HzO and CO2 were added ascrystalline HzCzO t' zHzO.

Approximately 10 mg of the reactants, encapsu-lated in 3 mm Pt capsules by arc-welding both ends,were run at the pressure and temperature of inter-est. The solidi for all the reactions were determinedby the presence of glass which is assumed to bequenched liquid. The HzO-undersaturated reactionwas located by confirming the presence of bothphlogopite and enstatite in the subsolidus; reactionstoichiometry was determined by the location of thesolidus relative to the low pressure H2O-saturatedunivariant reactions determined by Luth (1967).The present data are not inconsistent with thetopology of Luth' s forsterite-, kalsilite-, and leucite-absent invariant point (labelled (v) in Fig. 4; Luth1967), but do not constrain the location of the

invariant point. Run durations ranged from 2.5hours at 1600"C to 672 hours at 725"C. Whiletemperature reversals were not accomplished, bothwater-undersaturated and C02-saturated solidiwere determined using different crystalline startingcompositions: The former using both Ph + APh +Sa + Qz + H2O + COz and Sa * En-r Qz + H2O +CO2, and the latter using Sa * En * Qz + CO2 andSa * Mag * Qz mixtures. A close approach toequilibrium is inferred from the agreement of differ-ent starting compositions and from the lengthy rundurations. Phases were detected by optical and X-ray diffraction techniques; segregations of glasssufficiently large to be analyzed by electron micro-probe existed only at temperatures 100"-300'Cabove the solidus.

Results

Table 3 summarizes the experiments defining themelting reactions

S a * E n + Q z : L ( 1 )

S a * E n + Q z + C O 2 : L ( 2 )

and,

P h + S a + Q z + V : E n * L ( 3 )

In Figure 4, these reactions are shown in P-Tprojection. The CO2-saturated assemblage melts atabout 1000" C over a considerable pressure inter-val. Only at pressures of about 15 kbar and higherdoes the difference between melting points for theCO2-saturated and the volatile-absent reactions be-come detectable. The H2O-undersaturated reactionoccurs at about 750" C in the interval 5-10 kbar;H2O-saturated melting will occur at lower tempera-tures.

The slope, dPldT, of the CO2-saturated solidus(reaction 2) is approximately vertical. Novgorodovand Shkodzinskiy (1974) determined a slightly posi-tive slope for the CO2-saturated solidus of a granite;they estimated a vertical slope for X6o, : 0'95. Thepossibility of generating small amounts of H2O(Xn"o : 0.02) by diffusion of H2 into the experimen-tal charges of this investigation cannot be dismissed(Eggler, Mysen, and Hoering, 1974), and may beaffecting the solidus determination in this study.

Discussion

Melt composit ions and high pressure phaserelations

Figure 5 summarizes the changes in the liquidussur{ace and, thus, in liquid compositions with in-

WENDLANDT: INFLUENCE OF CO, ON MELTING OF GRANULITE

900

TEMPERAIURE, "C

Fig. 4. P-T projection of univariant melting reactions in the silica-rich portion of the system KAlSiOa-MgzSiO4-SiOrH2O-CO2.Open circles are subsolidus experiments; closed circles are runs in the melting intervals. The invariant point at approximately 0.5kbar and 825'C is from Luth (1967). Numerous runs at temperatures above 1200'C are not shown (see Table 3).

n69

e

aovr4fvl|,Itllec

1200

creasing pressure for the CO2-saturated system.The trend of liquid compositions for the H2O-undersaturated melting reaction is approximatelythe same. This is because P66, increases withincreasing Ptotu1 in both reactiona. In reality, thereaction involving small amounts of H2O will have aslightly more siliceous melt composition than willthe higher temperature CO2-saturated reaction at agiven pressure.

With increasing pressure, the liquidus stabilityfield of quartz expands relative to enstatite, and, asa consequence, liquid compositions become en-riched in potash, alumina and magnesia. The shift ofthe enstatite-quartz cotectic with increasing pres-sure (shown in Fig. 5) is constrained in part byobserved liquidus phases for the compositions in-vestigated. Unfortunately, electron microprobeanalyses of quenched melt cannot be obtained nearthe solidus because the melt does not segregate intomasses sufficiently large to analyze (quenched meltoccurred as glass rims on stable crystals). Analysesof melts from runs at much higher temperatures are

possible, however, and have also been used toconstrain the shift of the enstatite-quartz cotecticwith increasing pressure. A glass analysis for a runat 1150' C and 3 kbar plots almost directly on the 1-atm curve; the analysis constraining the 15 kbarcotectic is at 1300" C, and the analysis along the 25kbar cotectic is from a run at 1550" C.

The melt compositions of most interest are thosegenerated at the solidi of the melting reactions.Although melt compositions near the solidus cannotbe determined directly, the trend of evolution ofliquid compositions with increasing Pss, can beestablished: At I atm, the eutectic involving Sa *En + Qz * L is labelled P (Fig. 5), and, it can bedemonstrated that with increasing P66,, the liquidcomposition represented by this point shifts awayfrom the SiO2 apex toward higher alkali, magnesiaand CO2 contents. In the end member system,MgG-SiO2-CO2, as presented by Wyllie and Huang(1976) and Eggler (1974, 1975), the CO2-saturatedquartz liquidus surface is shown to intersect theenstatite-Co2 join at a pressure between 42 and 55

I 170 WENDLANDT: INFLUENCE OF CO" ON MELTING OF GRANULITE

Table 3: Results of experiments

R u n C r y s t a l l i n e s t a r t i n gnumDer composi ti on

P( kua r )

T(oc )

Mol eFract i on

coz

c0, c0r+1.1r9adiled added(wt%) (wt%)

durat ion Final assemblage(h rs )

3 7505 I t 5

3 8003 8003 900

694698682/uo681b

( P h , . A P h , . ) " d s a , n Q Z . n!lt ';; lt i; l#1;4'';"

!Pnt lArnt^s ) 3s5a2sQz5g5a4 o Ln22 qz3 8Sauo En, Qz*

Sauo EnrrQz*Sa3 oMag 3 oQz4 oSaunEnr rQz^^Ja4oEn22qz3SSauo En,Qz*

Sa4 o En2 2Qz3 8Sa .nEn ,

"Qz .n)a26 tn r4qz6Osau o EnrrQz*Sa4 o En2 2Qz3 ISa* o Enr.Qz, ,sau o EnrrQz*( P h , . A P h , . ) " ^ S a ^ ^ Q z . ^5auoEnr rqz?nsa, i Nas, oOii oSa u oEn22Qz3sSa 3 el4a g 3sQza eSa,ol.lag 39Qza 6Saa 6 En2 2Q236Saa e En2 2Qz3s

Sa,r 6 En22Qz36Saa s En22Qz3sSaa6 En22Qz 3 sSa 32En1 6Qz5 6Sa 32 En 1 sQz56

Sa 32En 1 s Qz5 ISa 32En 1 s Qz5 eSaa s En2 2Q236Sa 32En 1 sQ z5 eSa25 En I aQz5 e

n q *

0. 5*1 ,0** 20 .580 . 51 .0 t9 .27

1 . 0 1 5 . 6 6

1 . 0 7 5 . 7 71 . 0 1 9 . 3 5

1 . 0 28.221 . 0 u . 2 8

1 . 0 1 4 . 8 1

u . 9

0 . 5 *

1 . 0 1 5 . 6 6

1 . 0 1 5 . 6 61 'o 15 .66

1 . 0 1 3 . 5 76 q

u . 50 . 51 .0 30 .841.0 39 .061 . 0 6 0 . 5 61.0 49 .L21 ' o 9 . 0 71 .0 70 .021 . 0 2 9 . 9 5

1 . 0 1 5 . 6 61 . 0 1 5 . 6 6r . u l 3 . o o1 , 0 1 5 . 6 61 . 0 1 5 . 6 6

1 . 0 2 7 . 5 41 . 0 3 7 . 3 41 . 0 1 5 . 6 6

10 .55 1649.45 167

zoz8.85 168

I Z

- r b 6

168- 9 5

95116

- v b9696

- 9 8 . 5o a q

7.53 672b . J o 1 6 0

t9.77 186- 2 4 . t 5

2 4 . 7 5- 2 4

2424.5

- 24.525

s . 6 9 2 2 . s9 .23 26 .758.08 23 .5

107 . 3

Q z , S a , E n , P h , VQ z , S a , E n , P h , L ( t r ) , VQ z , S a , E n , VQ z , S a ( q ) , E n , L , VQ z , S a , E n , V

Qz ,Sa ,EnQ z , S a , E n , VQ z , S a , E n , LQ z , S a , E n , L , VQ z , E n , L , V

Q z ( t r ) , E n , LQ z , E n , L , VQ z , E n , L , VQ z , S a , E n , L ( t r )Q z , S a , E n , L ( t r ) , V

Q z , S a , E n , P h , VQ z , S a , E n , P h , L ( t r ) , VQ z , S a , E n ( t r ) , L , VQ z , S a , E nQ z , S a , E n , V

Q z , S a , E n , L ( ? )Q z , E n , L , VQ z ( t r ) , E n , L , \ /Qz ,Sa ,En ,L ( t r )Q z , E n , L , V

Q z , S a , E n , P h , VQ z , s a ( ? ) , E n ( t r ) , L , v ( ? )Q z , s a ( ? ) , E n ( ? ) , P h ( q ) , L , v ( ? )Q z , E n , L , V ( ? )Q z , E n , L , V

Q z , E n , L , VQ z , E n , L , VE n , L , VE n , L , v

Q z ' L ' V

Q z , S a , E n , VQ z , S a ( ? ) , E n , VQ z , E n , L , VQ z , S a , E n ( t r ) , M a g , L ( ? )Qz , sa ,En , l ' l ag ,L (? )

Q z , E n , L , Vq z ( E r J , L n , L , v

Qz ,Sa ,Mag

726a7266696a696b688

723a723b723c703aI U J O

712700a700b/ t J a

713b

71 la711b708a708b715a

704702707722725

72872973173?730

? o 7 q

3 10003 10003 1025

I 150I 150115010001000

t z a

750750A I J

97510001000102510251050

6 . 36 . 3

Id

812.5L Z . J

72.5L Z . 31 ' a

72.572.5

t515l5f 3

tf,

l 5

t 3

7507 7 E

80012001300

14001475150015251550

718 Sa36Mag3sQzas7l6a Sa3oMagroQzae724 Sa3oMag3gQz,{s705 SarsMag,rQzaq701 Sa ,oMag ,oQzue

738 Saa6En22Qz36739 SaasEn22Qz3s699 Sa,oMag3oQzae

77.5 95017.5 97577.5 100025 87525 900

25 155025 160026 850

6f,

53 . f ,

o J , 5

45. 52 72618

( q

2 . 520.75

Abbt'euiatLons: Sa, sanidine; En, enstatLte; Qz, quartz; Mag, nagnesite; Ph, phlogopite; APh, anhydrous phlogopite;L, Li.quid; V, uapor; (q), phase intezpreted as quaneh; (tr), traee anountl; (?), plnee belieued tobe preeant.

r XCO^ ie a bit Less tllqn 0.5 &)e to snoll qnolmts of Hr) in the phlogopite.** XrO- ie sLightLy Lese than 1 due to HrO in the phlogopite.

kbar, and intersect the carbonate stability surface ata pressure surmised to be about 60 kbar. That is,Wyllie and Huang propose that the melting reactionEn * CO2 : Qz * L intersects the decarbonationreaction En + CO2 : Qz + Mag at about 60 kbarand about 1600'C. In the system investigated in

this work, containing the additional component ofK2O + AlzO:, the quartz stability volume will alsoexpand with increasing P6s, relative to the enstatitestability volume. It can be reasonably inferred thatthe liquid composition in equilibrium with Sa * En* Qz + CO2 will intersect the Sa-En-CO2 plane at

ile25i04

Pco2=Prorol

25 kbcr .r ,/

WENDLANDT: INFLUENCE OF CO2 ON MELTING OF GRANULITE ltTl

Kr t. rltsiro, or

So

Fig. 5. Evolution of the liquidus surface with increasingpressure for CO2-saturated melting. Solid circles show thestarting compositions. Open circles denote liquid compositions(analysed by electron microprobe) in equilibrium with quartz andenstatite at high temperatures and 3, 15, and 25 kbar. P indicatesthe I atm eutectic involving quartz, sanidine, enstatite andliquid. The cotectics bounding the enstatite liquidus surface atlow silica contents are taken from Schairer and Bowen (1955) forthe I atm data, and from Wendlandt and Eggler (1980a) for the 15and 25 kbar data.

a pressure near 25 kbar because the melting reac-tion involving that assemblage (reaction 2) mustintersect the same decarbonation reaction (shownin Fig. 4) as did the assemblage En + Qz + CO2.Three experiments were made at 25-26 kbar and850'-900"C to locate this decarbonation reaction(see Fig. 4); in the two higher temperature runsquenched liquid (mixed silicate and carbonate) wasobserved suggesting that reaction 2 may intersectthe decarbonation reaction at pressures a bit below25 kbar. This lower pressure of intersection, con-trasted with 60 kbar in the system MgO-SiO2-CO2,is a consequence of the much lower melting tem-peratures (600'-700) of the assemblage Sa * En *Qz + CO2 as opposed to the assemblage En + Qz +CO2. Because the H2O-undersaturated reaction isalso characterized by increasing Pco, (relative toP11re) with increasing total pressure, similar evolu-tion of liquid compositions can be inferred (towardhigher feldspar and pyroxene components and low-er SiO2 contents). Analogous phase relations in-volving both CO2-saturated and H2O-undersaturat-ed melting reactions in the silica-undersaturatedportion of the system KAlSiO4-Mg2SiOa-SiO2-

H2O-CO2 have been discussed in greater detailelsewhere (Wendlandt and Eggler, 1980a,b).

Conditions of charnockite genesis

The formation of silicic partial melts enriched infeldspar and orthopyroxene components (whichmay be defined as protocharnockitic) is enhancedby low as,6. The physical conditions of partialmelting to produce such melts are in concert withthermobarometry estimates for the genesis of natu-ral charnockite occurrences (Table 2). Melting maybe initiated in the pressure interval 4-12 kbar attemperatures near 750"C by reaction 3, or, in thepresence of pure COz at temperatures of approxi-mately 1000"C for the same pressure interval byreaction 2. Melting by reaction 3 at pressures lessthan 4 kbar will produce melts with high SiOzcontents (granite analols) because the compositionof the volatile-rich fluid phase at the solidus will bebuffered at high Xs,6. Similarly, melts generated atlow pressures by reaction 2 will have higher SiO2contents than melts generated at higher pressures.At a given pressure, the melt generated by reaction3 will have higher SiOz content than the meltgenerated by reaction 2 at the same pressure.

A possible scenario for which the data of thisstudy might be applicable is as follows: Emplace-ment of an anorthosite or mafic magma in the lowercrust preferentially releases COz relative to H2Oduring crystallization (Kadik and Lukanin, 1973;Eggler and Rosenhauer, 1978), or promotes decar-bonation of carbonate-bearing country rocks, flux-ing partial melting of the country rocks. This partialmelt solidifies at high grade metamorphic condi-tions as the charnockite rock suite. A partial meltgenerated in the presence of CO2-rich fluids at somepressure, perhaps 8-10 kbar, which crystallizes atsome lower pressure, 5-8 kbar, will initially sepa-rate orthopyroxene and feldspar, followed byquartz. Differences in bulk composition of observedcharnockites may have arisen from variations indepth of genesis (melting and crystallization), de-gree of partial melting, or the composition of thesource material.

This model is analogous to that presented byEmslie (1978) for derivation of essentially bimodalanorthosite-granite suites. Emslie stressed the alka-line nature of the granitic rocks, including char-nockites and rapakivi granites, associated withGrenville anorthosite massifs, and the relation ofthe suite to anorogenic magmatic processes. Thereactions presented in this study reinforce his ob-

n72

servations by predicting the alkaline, as opposed tocalc-alkaline, nature of the granitic fractions. Thecompositions of partial melts generated near thebase of the crust, in the presence of a COz-richvapor phase, will be controlled by the expansion ofthe primary quartz field and characterized by en-richment of alkalis in the melt fractions.

A further application of the data of this studyregards the common association of migmatites andfelsic melts with charnockites (e.9., Quensel, 1950;Touret, I97la; Sen, 1974; Weaver, 1980). Weaver(1980) and Phillips (1980) suggested that partialfusion may be a logical precursor to granulite faciesmetamorphism. Phillips argued that the reduceddH,s necesSary for granulite metamorphism may bedue to partial melting (probably by biotite break-down) with as,6 decreasing with increasing tem-perature as H2O transfers to an increasing fractionof melt. Weaver, on the other hand, suggested thatfluctuations in the composition of the metamorphicfluid, from H2O-rich to CO2-rich, correspond withinitial production of anatectic melt followed byhigh-grade metamorphism at low as,6. Elaborationon these models is possible in light of meltingreaction 3 that has been presented here. Anatexis inthe presence of an incoming COz-rich vapor phasewill commence at approximately 750"C. The assem-blage of liquid + crystals (including biotite) willpersist with continuing COz influx until the biotite isexhausted by HzO extraction, at which point thevapor phase composition will be no longer buffered.Extraction of H2O from the silicate melt will thenoccur, promoting melt crystallization despite tem-peratures which may continue to increase (Fig. 3).The resultant mineralogy of the quench melt islikely to be devoid of volatile-bearing minerals andmore felsic (perhaps aplitic) than the surroundingterrain.

Effect of Fe on the phase relations

It is likely that the inclusion of additional compo-nents to the system investigated will result in de-pression of the melting temperatures. The mostsignificant component is likely to be iron; charnock-ites have Fe/(Fe+Mg) (oxide wt) ratios on the orderof 5-15. In this study, the objective was to presentthe phase relations of CO2-saturated and HzO-undersaturated melting in the most simple waypossible, consequently, the added complexity of(Fe,Mg)O solid solutions was neglected. Some gen-eral observations may be in order, however. Themelt fraction separating from a metapelite or meta-

WENDLANDT: INFLUENCE OF CO2 ON MELTING OF GRANULITE

greywacke source will be enriched in Fe relative toMg (MacRae and Nesbitt, 1980), and thus it is likelythat melting temperatures will occur at lower tem-peratures than those determined in this study. Sax-ena (1969), however, analyzed coexisting biotiteand orthopyroxene in charnockites and observedthat the distribution coefficient for Fe/Mg betweenthe two phases was approximately unity. He pro-jected that a substantial shift of melting equilibria insystems involving (Fe,Mg)O solid solutions in bothmica and orthopyroxene was unlikely. Accordingly,it is expected that the equilibrium temperature ofreaction 2 will be shifted to lower temperatures bythe addition of Fe, while the temperature for reac-tion 3 will decrease to a lesser degree.

Conclusions

A model has been presented for the melting of asimple granulite assemblage in the presence of COz-rich fluid phases. Melting can occur between 750'-1000"C at crustal pressures, and these temperaturesare well within the range estimated for regionalmetamorphism in, for example, the Adirondacks(Bohlen and Essene, 1977;Jaffe et al.,1978; Valleyand Essene, 1980). For melting occurring at about750'C in the presence of both HzO and CO2, pres-sures corresponding to the deep crust are requiredto generate a melt enriched in pyroxene and feld-spar components. Melting in the presence of pureCO2at about 1000"C generates analogous melt com-positions at lower pressures. The experimentalmelting reactions are in agreement with a number ofobservations constraining charnockite occurrences :Pressure and temperature constraints, mineralogi-cal constraints, and constraints on the compositionsof volatiles associated with peak conditions of char-nockite formation.

Acknowledgments

Thorough reviews of this paper by Drs. W. I. Harrison and R.C. Newton are greatly appreciated. Low pressure runs weredone by O. Mullins; his assistance was indispensable. Theresearch reported in this paper was done while the author was aStaff Scientist at the Lunar and Planetary Institute which isoperated by the Universities Space Research Association underContract No. NASW-3389 with the National Aeronautics andSpace Administration. The use of the experimental petrology labat the Johnson Space Center is acknowledged. This paper isLunar and Planetary Institute Contribution No. 439.

References

Ashwal, L. D. (1978) Perogenesis of Massif-type Anorthosites:Crystallization History and Liquid Line of Descent of the

Adirondack and Morin Complexes. Ph.D. Thesis, PrincetonUniversity.

Blattner, P. (1980) Chlorine and fluorine in apatite, biotite andhornblende of basic charnockites from Kondapalli, India.Neues Jahrbuch fuhr Mineralogie, Monatshefte, 6, 283-288.

Boettcher, A. L. and Wyllie, P. J. (196E) Melting of granite withexcess water to 30 kilobars pressure. Journal of Geology, 76,235244.

Bohlen, S. R. and Essene, E. J. (1977) Feldspar and oxidethermometry of granulites in the Adirondack Highlands. Con-tributions to Mineralogy and Petrology, 62, 153-169.

Bohlen, S. R. and Essene, E. J. (1978) Igneous pyroxenes frommetamorphosed anorthosite massifs. Contributions to Miner-alogy and Petrology, 65,433442.

Boyd, F. R. and England, J. L. (1960) Apparatus for phase-equilibrium measurements at pressures up to 50 kilobars andtemperatures up to 1750"C. Journal of Geophysical Research,65,741-748.

Eggler, D. H. (1974) Effect of CO2 on the melting of peridotite.Carnegie Institution of Washington Year Book, 73,215-224.

Eggler, D. H. (1975) CO2 as a volatile component of the mantle:The system MgrSiOa-SiO2-H2O-CO2. Physics and Chemistryof the Earth, 9, 869-881.

Eggler, D. H. (1977a) Calibration of a Pyrex solid-media assem-bly. Carnegie Institution of Washington Year Book, '76, 656-658.

Eggler, D. H. (1977b) The principle of the zone of invariantvapor composition: An example in the system CaO-MgO-SiO2-CO2-H2O and implications for the mantle solidus. Car-negie Institution of Washington Year Book, 76, 428-435.

Eggler, D. H., Mysen, B. O. and Hoering, T. C. (1974) Gasspecies in sealed capsules in solid-media, high-pressure appa-ratus. Carnegie Institution of Washington Year Book,73,228-232.

Eggler, D. H. and Rosenhauer, M. (1978) Carbon dioxide insilicate melts: II. Solubilities of CO2 and H2O in CaMgSi2O6(diopside) liquids and vapors at pressures to 40 kb. AmericanJournal of Science, 278, U-94.

Emslie, R. F. (197E) Anorthosite massifs, rapakivi granites, andlate proterozoic rifting of North America. Precambrian Re-search, 7, 6l-98.

Field, D., Drury, S. A. and Cooper, D. C. (1980) Rare-earth andLIL element fractionation in high-grade chamockitic gneisses,south Norway. Lithos, 13,281-289.

Hoefs, J. and Touret, J. (1975) Fluid inclusion and carbonisotope study from Bamble granulites (South Norway). Contri-butions to Mineralogy and Petrology, 52, 165-174.

Hollister, L. A. and Bumrss, R. C. (1976) Phase equilibria influid inclusions from the Khtada Lake metamorphic complex.Geochimica et Cosmochimica Acta, 40, 163-175.

Hubbard, F. H. and Whitley, J. E. (1979) REE in charnockiteand associated rocks, southwest Sweden. Lithos. 12. l-11.

Jatre, H. W., Robinson, P. and Tracy, R. J. (1978) Orthoferrosi-lite and other iron-rich pyroxenes in microperthite gneiss ofthe Mount Marcy area, Adirondack Mountains. AmericanMineralogist, 63, I I 16-1 136.

Kadik, A. A. and Lukanin, O. A. (1973) The solubility-depen-dent behavior of water and carbon dioxide in magmaticprocesses. Geokhimiya, 2, 163-179 (transl. Geochemistry In-ternational, 10, 115-120, 1973).

Konnerup-Madsen, J. (1977) Composition and microthermo-metry of fluid inclusions in the Kleivan granite, south Norway.

lt73

American Journal of Science, 277,673-696.Konnerup-Madsen, J. (1979) Fluid inclusions in quartz from

deep-seated granitic intrusions, south Norway. Lithos, 12, l3-23.

Letteney, C. D. (1968) The anorthosite-norite-charnockite seriesof the Thirteenth Lake Dome, south-central Adirondacks. InY. W. Isachsen, Ed., Origin of Anorthosite and RelatedRocks, p. 328-342. New York State Museum and ScienceService, Memoir 18, Albany.

Luth, W. C. (1966) The system KAlSiO4-KAlSizOe-MgzSiOr.Journal of the American Ceramic Society, 50, 174-176.

Luth, W. C. (1967) Studies in the system KAISiO+-MgzSiO+-SiOrHzO: I, Inferred phase relations and petrologic applica-tions. Journal of Petrology, 8,372416.

MacRae, N. D. and Nesbitt, H. W. (19E0) Partial melting ofcommon metasedimentary rocks: A mass balance approach.Contributions to Mineralogy and Petrology, 7 5, 2l:26.

Martignole, I. (1979) Chamockite genesis and the proterozoiccrust. Precambrian Research, 9. 303-310.

Mysen, B. O. and Boettcher, A. L. (1975) Melting of a hydrousmantle: II. Geochemistry of crystals and liquids formed byanatexis of mantle peridotite at high pressures and hightemperatures as a function of controlled activities of water,hydrogen, and carbon dioxide. Journal ofPetrology, 16,549-593.

Nesbitt, H. W. (1980) Genesis of the New Quebec and Adiron-dack granulites: Evidence for their production by partialmelting. Contributions to Mineralogy and Petrology, 72,3O3-310.

Newton, R. C., Smith, J. V. and Windley, B. F. (1980) Carbonicmetamorphism, granulites and crustal growth. Nature, 2EE,45-50.

Novgorodov, P. G. and Shkodzinskiy, V. S. (1974) Experimentson melting of granite in H2O-CO2 mixtures and some problemsof granite formation. Geokhimiya, 5,741-:150 (transl. Geo-chemistry International, ll, 522-531, 1974).

Ormaasen, D. E. (1977) Petrology of the Hopen mangerite-charnockite intrusion, Lofoten, north Norway. Lithos, 10,291-310.

Phillips, G. N. (1980) Water activity changes across an amphibo-lite-granulite facies transition, Broken Hill, Australia. Contri-butions to Mineralogy and Petrology, 75,377-3E6.

Pichamuthu, C. S. (1969) Nomenclature of chamockites. TheIndian Mrncraiogist, 10, 23-35.

Piwinskii, A. J. (196E) Experimental studies of igneous rockseries: central Sierra Nevada Batholith. Califomia. Joumal ofGeology, 76, 548-570.

Pride, C. and Muecke, G. K. (1980) Rare earth element geo-chemistry of the Scourian Complex N.W. Scotland--evidencefor the granite-granulite link. Contributions to Mineralogy andPetrofogy, 73,403412.

Quensel, P. (1950) The charnockite series of the Varberg districton the south-western coast of Sweden. Archiv for Mineralogioch Geologi, l, 227-332.

Saxena, S. K. (1969) Chemical study of phase equilibria incharnockites, Varberg, Sweden. American Mineralogist, 53,1674-r69s.

Schairer, J. F. and Bowen, N. L. (1955) The system KzO-AlzOr-SiO2. American Journal of Science, 253,6E1:746.

Sen, S. K. (1974) A review of some geochemical characters ofthe type area (Pallavaram, India) charnockites. Joumal of theGeological Society of India, 15,413420.

WENDLANDT: INFLUENCE OF CO2 ON MELTING OF GRANULITE

t t14

Smith, J. V., Newtori, R. C. and Janardhan, A. S. (1979)Significance of granulite metamorphism for stabilization ofplanetary crust: Charnockite formation at Kabbaldurga, S.India: Roles of CO2 and HzO; Speculations on Venus andMars. (abstr.) Lunar and Planetary Science X, l134-1136.

Touret, J. (1971a) Le facies granulite en Norvege meridionale. I.Les associations mineralogiques. Lithos 4, 239-249.

Touret, J. (1971b) Le facies granulite en Norvege meridionale.II. Les inclusions fluides. Lithos,4, 423-436.

Tuttle, O. F. and Bowen, N. L. (1958) Origin of granite in thelight of experimental studies in the system NaAlSi3Os-KAl-Si3OFSiO2-H2O. Geological Society of America Memoir, 74.

Valley, J. W. and Essene, E. J. (19E0) Akermanite in theCascade Slide xenolith and its significance for regional meta-morphism in the Adirondacks. Contributions to Mineralogyand Petrology, 7 4, 143-152.

Weaver, B. L. (1980) Rare-earth element geochemistry of Ma-dras granulites. Contributions to Mineralogy and Petrology,7r.27r-279.

Wendlandt, R. F. and Eggler, D. H. (1980a) The origins of

WENDI,ANDT: INFLUENCE OF CO, ON MELTING OF GRANULITE

potassic magmas: l. Melting relations in the systems KAI-SiO+-MgzSiOcSiOz and KAlSioa-MgO-SrO24o2 to 30 kilo-bars. American Journal of Science, 280, 3E5-420.

Wendlandt, R. F. and Eggler, D. H. (l9E0b) The origins ofpotassic magmas: 2. Stability of phlogopite in natural spinellherzolite and in the system KAISiO+-MgO-SiOrH2O1O2athigh pressures and high temperatures. American Journal ofScience, 280,421458.

Wyllie, P. J. and Huang, W. L. (1975) Peridotite, kimberlite, andcarbonatite explained in the system CaO-MgO-SiOrCOz.Geology, 3, 621-624.

Wyllie, P. J. and Huang, W. L. (1976) High COz solubilities inmantle magmas. Geology, 4,21-24.

Wyllie, P. J. and Tuttle, O. F. (1959) Effect of carbon dioxide onthe melting of granite and feldspars. American Journal ofScience, 257, 6/,8-655.

Manuscript received, April 27, 1981accepted for publication, July 20, 1981.