(2006) 398–408www.elsevier.com/locate/gr

Gondwana Research 9

Experimental calibration of sapphirine–spinel Fe2+–Mg exchangethermometer: Implication for constraints on P–T condition of

Howard Hills, Napier Complex, East Antarctica

Kei Sato a,b,⁎, Tomoharu Miyamoto c, Toshisuke Kawasaki a

a Department of Earth Sciences, Faculty of Science, Ehime University, Bunkyo-cho 2-5, Matsuyama 790-8577, Japanb Research Center for the Evolving Earth and Planets, Department of Earth and Planetary Sciences, Faculty of Science,

Tokyo Institute of Technology, Ookayama 2-12-1, Meguroku, Tokyo 152-8551, Japanc Department of Earth and Planetary Sciences, Faculty of Science, Kyushu University, Hakozaki 6-10-1, Fukuoka 812-8581, Japan

Received 23 May 2005; accepted 11 January 2006Available online 10 March 2006

Abstract

The Fe2+–Mg distribution coefficients between sapphirine and spinel:

KD ¼ X SplFe X

SprMg

X SplMgX

SprFe

;

were experimentally determined at pressures of 9–13kbar and temperatures of 950–1150°C using a natural ultrahigh-temperature (UHT) granulitewith paragenesis of these minerals from the Napier Complex in East Antarctica [XMg=Mg/ (Fe+Mg); XFe=Fe / (Fe+Mg)]. A new sapphirine–spinel geothermometer has been obtained as:

lnKD ¼ −1:257þ ½2940þ 9:5ðP kbar−11Þ�=T ðKÞ:

We applied the exchange thermometer to UHT or high-grade metamorphic rocks that were reported from various complexes in the world. If theKD values of 2.63–4.34 obtained from low-Cr mineral pairs such as X Cr

Spr <0.016 and X CrSpl <0.047 were substituted into the equation, their

temperature conditions would be estimated as 806–1050°C at 11kbar. The XCr means Cr / (Al+Cr(+Fe3+)). These temperatures are reasonableretrograde or near peak metamorphic condition.© 2006 International Association for Gondwana Research. Published by Elsevier B.V. All rights reserved.

Keywords: Fe2+–Mg partitioning; Sapphirine–spinel exchange thermometer; Ultrahigh-temperature (UHT) metamorphism; Napier Complex; East Antarctica

1. Introduction

Sapphirine is one of major minerals often found inultrahigh-temperature (UHT) metamorphic rocks. In manyisobarically cooled UHT metamorphic rocks, coexistence of

⁎ Corresponding author. Research Center for the Evolving Earth and Planets,Department of Earth and Planetary Sciences, Faculty of Science, Tokyo Instituteof Technology, Ookayama 2-12-1, Meguroku, Tokyo 152-8551, Japan.

E-mail address: ksato@geo.titech.ac.jp (K. Sato).

1342-937X/$ - see front matter © 2006 International Association for Gondwana Rdoi:10.1016/j.gr.2006.01.001

sapphirine and spinel is commonly observed as a mineralassemblage of sapphirine corona that encloses spinel grains.In majority of the cases, robust thermodynamic assemblagesuch as coexistence of garnet and orthopyroxene is rarelypresent. If there is no proper geothermometer, it is difficult toprecisely constrain temperature condition of this early stage ofisobaric cooling history. The Fe2+–Mg partitioning betweensapphirine and spinel has been already reported by severalprevious studies. As the first sapphirine–spinel geotherm-ometer, Owen and Greenough (1991) provided an empirical

esearch. Published by Elsevier B.V. All rights reserved.

Table 1Thermodynamic data of sapphirine (7 :9 :3) and spinel from previous studies

End member H°(cal/mol)

S°(cal/K/mol)

V°(cal/kbar/mol)

Mg3.5Al9Si1.5O20a −2646153.32 106.30 4723.99

Fe3.5Al9Si1.5O20a −2346823.44 131.63 4871.14

MgAl2O4b −549513.58 19.47 950.29

FeAl2O4b −468022.14 25.68 973.46

ΔH°=−4031(cal)

ΔS°=−1.024(cal/K)

ΔV°=−18.87(cal/kbar)

For the Fe2+–Mg exchange reaction between sapphirine (7 :9 :3) and spinel ofthe Eq. (2): 2/7 Fe3.5Al9Si1.5O20+MgAl2O4=2/7 Mg3.5Al9Si1.5O20+FeAl2O4,the ΔH°, ΔS° and ΔV° are −4031 cal, −1.024cal/K and −18.87 cal/kbar,respectively.a Source of data: Ouzegane et al. (2003). These thermodynamic data were

recalculated from those of Holland and Powell (1998).b Soure of data: Holland and Powell (1998).

399K. Sato et al. / Gondwana Research 9 (2006) 398–408

exchange thermometer derived from literature data of theseminerals. For the Fe2+–Mg exchange reactions betweensapphirine (2 :2 :1) and spinel:

1=2 Fe2Al4SiO10Spr

þMgAl2O4Spl

¼ 1=2Mg2Al4SiO10Spr

þ FeAl2O4Spl

;

ð1Þ

or sapphirine (7 :9 :3) and spinel:

2=7 Fe3:5Al9Si1:5O20Spr

þMgAl2O4Spl

¼ 2=7Mg3:5Al9Si1:5O20Spr

þ FeAl2O4Spl

; ð2Þ

the free energy change ΔG° at the standard state referring tothe pure phases at pressure P (kbar) and temperature T (K) ofinterest can be written as:

−DG-P;T ¼ −DG-

1 bar;T−ðP kbar−1 barÞDV -1 bar;T

¼ −DH -1 bar;T þ TDS-1 bar;T−ðP kbar−1 barÞDV -

1 bar;T

¼ RT lnK; ð3Þ

assuming that the volume change ΔV° is independent of P.The ΔH°, ΔS° and R are the enthalpy, entropy changes of thereactions (1) or (2) and gas constant (=8.3144J/mol/K=1.9862cal/mol/K), respectively. The K of Eq. (3) isequilibrium constant (see Thermodynamic treatments in thispaper). If the ΔV° is independent of temperature, Eq. (3) can berewritten as:

RT lnK¼−DH -1 bar;TþTDS-1 bar;T−ðP kbar−1 barÞDV -

1 bar; 298 K;

ð4Þi.e.,

lnK ¼ −DH -1 bar;T−ðP kbar−1 barÞDV -

1 bar; 298 K

h i=RT

þ DS-1 bar;T=R: ð5Þ

Owen and Greenough (1991) expressed the thermometer asrelation between lnK vs. T. However, it contradicts thethermodynamic relation between lnK vs. 1 /T in Eq. (5). Asthe first experimental study of sapphirine–spinel thermometry,Das et al. (2003) developed an exchange thermometer based onthe Fe2+–Mg partitioning data for the systems in which the bulkXMg [=Mg/ (Fe2+ +Mg)] values are 0.81 and 0.72. Theseexperiments were conducted at pressures of 9–12kbar andtemperatures of 850–1100°C. However, their experimental dataand thermodynamic treatments for calibration of the exchangethermometer were not described clearly because this study wasmainly focused on determination of P–T petrogenetic grid inKFMASH system rather than geothermometry. Das et al. (2006)mixed their new experimental data (bulk XMg values=0.78 and0.87; P=8kbar and T=975–1075°C) with previous results, and

revised the sapphirine–spinel thermometer. For the Fe2+–Mgexchange reaction between sapphirine (7 :9 :3) and spinel of Eq.(2), the ΔH° can be calculated as −4031 cal (Table 1) by usingthe thermodynamic data set (Ouzegane et al., 2003, based ondata by Holland and Powell, 1998). In the previous thermo-meters of Das et al. (2003, 2006), however, the ΔH° is >0. Theslope in lnKD vs. 1 /T space is diametrically reversed by thediscrepancy in the ΔH° between the thermodynamic calcu-lated data and these previous thermometers (KD=distributioncoefficient; see Thermodynamic treatments in this paper). Inthe present study, we would determine new experimentaldata on the Fe2+–Mg partitioning between sapphirine andspinel at pressures of 9–13kbar and temperatures of 950–1150°C with 50°C intervals for the granulite (TM981229-03E) system in which the bulk XMg is 0.805. Thisexperimental data would put constraints on the temperatureestimates of the evolutionary history of UHT metamorphicterranes with sapphirine–spinel-bearing rocks.

Napier Complex in Enderby Land, East Antarctica is anArchaean UHT metamorphic terrane (e.g., Sheraton et al.,1987; Harley and Hensen, 1990), where peak temperature hadattained to over 1120°C (Harley and Motoyoshi, 2000).Preliminary experimental study by Sato et al. (2004a) wereconducted at 9–13kbar and 950–1200°C using a phlogopite-bearing orthopyroxene granulite (sample no. TM981229-03E;Miyamoto et al., 2004), which was collected from HowardHills in the Napier Complex. This natural sample containscoexistence of sapphirine and spinel. In this granulite, allspinel grains are enclosed by sapphirine envelopes (Miyamotoet al., 2004; Sato et al., 2004a). From our previousexperimental results, the sapphirine was regarded as a reactionproduct that was generated at temperatures ≤1100°C afterprimary spinel crystallization at higher T condition (Sato et al.,2004a,b). It means that the sapphirine corona surroundingspinel grains formed through retrograde metamorphism. As thesupplement issue in this paper, we also attempt to properlyconstrain the isobaric cooling history of the Howard Hills areaby using the new thermometer.

Fig. 1. Plot of chemical composition of sapphirine in the phlogopite-bearingorthopyroxene granulite (TM981229-03E).

400 K. Sato et al. / Gondwana Research 9 (2006) 398–408

2. Brief description of Howard Hills in Napier Complex,East Antarctica

The Howard Hills is situated east of Amundsen Bay withinthe Napier Complex. Sheraton et al. (1987) reporteddistribution map of spotted outcrops around the AmundsenBay. A geological survey was carried out by the summer partyof the 40th Japanese Antarctic Research Expedition (Yoshi-mura et al., 2000; Miyamoto et al., 2004). According tocompiled data of Harley and Hensen (1990), this area isrepresented as very high-temperature region that is character-ized by the mineral assemblage of sapphirine+quartz (Hensenand Green, 1973; Bertland et al., 1991). The Howard Hills ismainly composed of layered gneisses of various bulkchemistry and origins; and small amounts of metamorphosedultramafic rocks are also present as blocks (Yoshimura et al.,2000). Yoshimura et al. (2000) evaluated metamorphictemperatures of these gneisses as 1000–1200°C using feldsparsolvus geothermometers fitted for 8kbar and 10kbar (Kroll etal., 1993). Yoshimura et al. (2000) also estimated retrogradeP–T conditions as 5.0–9.5kbar and 830–950°C usingcompositional isopleths (Hensen and Harley, 1990) andgarnet–orthopyroxene exchange thermometers (Sen and Bhat-tacharya, 1984; Lee and Ganguly, 1988). Based on these P–Testimates and descriptive data, Yoshimura et al. (2000)

Table 2Chemical compositions of bulk and ferromagnesian minerals in the phlogopite-bearing orthopyroxene granulite (TM981229-03E)

Bulk a,n=15(wt.%)

σ Spr,n=10(wt.%)

σ Spl,n=10(wt.%)

σ Opx,n=10(wt.%)

Phl,n=10(wt.%)

SiO2 47.67 0.36 13.56 0.54 0.03 0.01 50.21 39.22TiO2 1.47 0.07 0.03 0.03 n.d. n.d. 0.09 4.30Al2O3 13.82 0.25 60.37 1.07 60.53 1.08 8.31 13.64Cr2O3 0.28 0.04 1.47 0.09 4.20 0.50 0.22 0.22FeOb 9.51 0.50 5.92 0.26 18.11 0.49 12.83 5.43MnO 0.15 0.04 0.08 0.02 0.14 0.02 0.23 0.03MgO 22.02 0.20 17.17 0.22 14.70 0.21 26.65 20.70NiO 0.10 0.05 0.24 0.05 0.62 0.05 0.07 0.17CaO 0.49 0.04 0.01 0.01 n.d. n.d. 0.11 0.01Na2O 0.46 0.03 0.03 0.02 0.05 0.03 0.02 0.12K2O 2.28 0.08 0.01 0.01 0.02 0.02 0.02 10.43BaO 0.25 0.21 0.06 0.06 0.20 0.20 0.14 0.56ZnO 0.07 0.07 0.17 0.07 1.11 0.10 0.19 0.14F 0.55 0.23 n.d. n.d. n.d. n.d. n.d. 4.05O= −0.23 −1.71H2O 2.22Total 98.89 99.12 99.73 99.11 99.54

XMg 0.805 0.838 0.591 0.787 0.872XCr 0.013 0.016 0.044

Original data are presented in Sato et al. (2004a). Mineral abbreviations are afterKretz (1983): Spr, sapphirine; Spl, spinel; Opx, orthopyroxene; Phl, phlogopite.σ, standard deviations; n, number of point analysis; n.d., not determined. H2O,calculated value based on stoichiometry (after Motoyoshi and Hensen, 2001).XMg, Mg/ (Fe2++Mg). XCr, Cr / (Al+Cr).a Bulk composition was determined by the defocused X-ray microprobe

analysis for glass fused at 10kbar and 1670°C for 2min in a graphite capsuleusing a piston-cylinder apparatus.b Total Fe as FeO.

suggested the possibility of isobaric cooling in the HowardHills. This implication of isobaric cooling agrees withsuggestion that was presented by Harley (1998). Moreover,based on concentrated pattern of Y or Yb in garnet grains inthese gneisses, Yoshimura et al. (2000) also mentionedpossibility of partial melting that was accompanied with suchUHT metamorphic condition. Miyamoto et al. (2004) pre-sented geochemical and geochronological data of UHTmetamorphic rocks from the Howard Hills. In this report,apparent ages of about 2.65Ga within analytical error onisochron diagrams were yielded from the Rb–Sr and Sm–Ndanalytical data. Based on this geochemical data, Miyamoto etal. (2004) also concluded that metamorphic rocks in HowardHills experienced partial melting during the UHT metamor-phism. Possibility of the partial melting of the granulite(TM981229-03E) was solidly supported by our previousexperimental studies (Sato et al., 2004a,b). These experimentalresults suggested that sapphirine corona including spinel grainsformed through retrograde metamorphism and the sapphirinecoexisted with almost H2O-free granitic melt at UHT conditionthat exceeds 1100°C.

3. Experimental procedures

Description of the phlogopite-bearing orthopyroxenegranulite (TM981229-03E) was reported by Miyamoto et al.(2004). The salient points are repeated below. This granulite iscomposed of aluminous orthopyroxene (modal percentage:45.6%), sapphirine (12.1%), spinel (3.4%), feldspar (23.7%;alkali feldspar and plagioclase), fluorphlogopite (14.0%), rutile(1.2%) and small amounts of quartz (less than 1%).Paragenesis of sapphirine and quartz that suggests UHTcondition (Hensen and Green, 1973; Bertland et al., 1991) iscontained in the granulite although its amount is very little.Mean values of XMg of these constituent minerals are asfollows: aluminous orthopyroxene, 0.787; sapphirine, 0.838;spinel, 0.591; and: fluorphlogopite, 0.872 (Table 2). Sapphi-rine and spinel contain small amounts of Cr2O3 as 1.47 wt.%and 4.20 wt.%, respectively. In these two minerals, remarkablecompositional zoning is not found (Table 2). Chemicalcomposition of sapphirine in the granulite is shown in Fig. 1.This sapphirine is intermediate between 2 :2 :1 and 7 :9 :3 in

Table 3Experimental conditions and phases in run products

Run no. P(kbar)

T(°C)

Time(h)

Startingmaterial a

Spr Spl Additional phases(mean values of XMg in the phases) b

030924A 9 1200 163 1 – ○ Opx (0.84), melt (0.69)021022B 13 1150 161 1 ○ ○ Opx (0.80), Phl (0.85), melt (0.54), ±Rt030613B 12 1150 171 1 – ○ Opx (0.81), Phl (0.86), melt (0.59)020920D 11 1150 172 2 – ○ Opx (0.83), Phl (0.89), melt (0.61)020902C 9 1150 176 2 – ○ Opx (0.81), Phl (0.89), melt (0.52)031001C 13 1100 119 1 ○ ○ Opx (0.80), Afs, Phl (0.85), Rt031020B 12 1100 232 1 ○ ○ Opx (0.80), Afs, Phl (0.85), Rt030526B 11 1100 262 1 ○ ○ Opx (0.80), Afs, Phl (0.85), Rt030804A 9 1100 264 1 ○ ○ Opx (0.80), Phl (0.87), Rt, melt (0.58)021029C 13 1050 309 1 ○ ○ Opx (0.80), Afs, Pl, Phl (0.86), Rt, Grt (0.69)020927Ac 11 1050 569 2 ○ ○ Opx (0.80), Afs, Phl (0.85), Rt030402C 10 1050 340 1 ○ ○ Opx (0.81), Afs, Pl, Phl (0.85), Rt021111A 9 1050 336 1 ○ ○ Opx (0.80), Afs, Pl, Phl (0.85), Rt031030B 12 1000 575 1 ○ ○ Opx (0.80), Afs, Pl, Phl (0.85), Rt, Grt (0.68)030421C 11 1000 597 1 ○ ○ Opx (0.79), Afs, Pl, Phl (0.86), Rt030623B 9 1000 598 1 ○ ○ Opx (0.79), Afs, Pl, Phl (0.85), Rt031123B 12 950 770 1 ○ ○ Opx (0.79), Afs, Pl, Phl (0.84), Rt, Grt (0.69)021127A 9 950 584 1 ○ ○ Opx (0.79), Afs, Pl, Phl (0.85), Rt, ±Qtza Starting materials: no. 1, mixture of 90 wt.% glass and 10 wt.% seed mineral aggregate of the pulverized granulite sample (constituent minerals for TM981229-

03E: Opx, Spr, Spl, Afs (=alkali feldspar), Pl (=plagioclase), Phl, Rt (=rutile) and minor amounts of Qtz (=quartz)); no. 2, glass (seed-free).b In this table, the XMg values of rimed orthopyroxene are shown.c This data was omitted from calibration of the present sapphirine–spinel exchange thermometer.

Table 4Chemical compositions of sapphirine, spinel and melt in a run product (run no.030804A; 9kbar and 1100°C)

Spr in rim,n=16 (wt.%)

Spr in core,n=10 (wt.%)

Spl, n=13(wt.%)

Melt a, n=10(wt.%)

σ

SiO2 13.46 13.86 0.13 66.87 1.30TiO2 0.09 0.08 0.31 1.17 0.23Al2O3 61.54 61.35 64.00 16.25 0.64Cr2O3 1.31 1.28 3.02 0.01 0.01FeO 5.51 6.44 13.76 2.51 0.32MnO 0.08 0.08 0.14 0.03 0.03MgO 17.68 17.53 18.01 1.97 0.55NiO 0.19 0.20 0.35 0.04 0.04CaO 0.03 0.01 0.01 1.50 0.21Na2O 0.01 0.02 0.02 1.25 0.11K2O 0.02 0.02 0.01 6.13 0.13BaO 0.06 0.05 0.03 0.42 0.18ZnO 0.06 0.05 0.49 0.04 0.04F 0.91 0.41O= −0.38Total 100.04 100.97 100.27 98.71

XMg 0.851 0.829 0.700 0.583XCr 0.014 0.014 0.031

The significances of symbols are the same with those of Table 2a Original data is presented in Sato et al. (2004a).

401K. Sato et al. / Gondwana Research 9 (2006) 398–408

composition. The H2O content in the granulite was measuredusing the Karl–Fischer Method (after Muroi, 1979) asfollows: H2O (−)=0.12 wt.% and H2O (+)=0.25 wt.%(Miyamoto et al., 2004).

Chemical data of run products obtained by experiments ofSato et al. (2004a) was used to present sapphirine–spinelgeothermometry. Experimental treatments were already de-scribed in Sato et al. (2004a). Salient features of theseexperiments are repeated below. High-pressure and high-temperature experiments were conducted at pressures of 9–13kbar and temperatures of 950–1200°C using a 16.0mmpiston-cylinder apparatus at Ehime University (Japan). Forpreparing starting materials in these experiments, glassedsample was synthesized by fusing the granulite at 10kbar and1670°C for 2min in a graphite capsule using the piston-cylinder apparatus. The defocused X-ray microprobe analysisrevealed that the XMg of glass is 0.805 (Table 2). Mixture of90 wt.% glass and 10 wt.% seed mineral aggregate of thepulverized granulite sample was mainly used as the startingmaterial (Table 3). Glass sample without the rock powder wasalso used as a supplement starting material in a few runs (runnos. 020902C, 020920D and 020927A; Table 3). The mixture(or rock-free glass) was ground in agate mortar (10–50μm ingrain size) and was put into an inner molybdenum-foil capsulewithin an outer platinum tube (Kawasaki and Motoyoshi,2005). In the experimental study by Kawasaki and Motoyoshi(2005), low-contrast thin film of MoO2 with thickness of 2–3μm coating the inner wall of the molybdenum capsule wasfound in their experimental runs. However, Kawasaki andMotoyoshi (2005) mentioned that the fO2 of the charge waskept within the Mo–MoO2 buffer where the intrinsic fO2 is2.9×10−16 bar at 1000°C (O'Neill, 1986). Before theseexperiments, all starting materials within sample container

were dried at 110°C in the oven. Two edges of platinum tubewere welded using carbon arc.

The sectional view of a typical high-pressure cell assemblywas shown in Sato et al. (2004a). Graphite was used as a heater.Temperature was controlled using PtRh thermocouples. Hotpiston-in technique was applied for all runs (e.g., Kawasaki andMotoyoshi, 2005). Pressure was monitored using an oil pressuregauge; It was calibrated by the phase transformations of Bi I–IIat room temperature (25.5kbar; Hall, 1971) and by the quartz–

Table 5The XMg values of sapphirine and spinel in run products

Run no. P(kbar)

T(°C)

XMg data ofMgO-rich pair

XMg data ofFeO-rich pair

Spr Spl KD Spr Spl KD

021022B 13 1150 0.852 0.710 2.35 0.821 0.682 2.13031001C 13 1100 0.848 0.688 2.54 0.829 0.666 2.44031020B 12 1100 0.844 0.695 2.37 0.827 0.682 2.23030526B 11 1100 0.840 0.691 2.35 0.821 0.683 2.13030804A 9 1100 0.861 0.722 2.38 0.821 0.679 2.16021029C 13 1050 0.841 0.684 2.44 0.824 0.676 2.24030402C 10 1050 0.862 0.698 2.70 0.828 0.661 2.46021111A 9 1050 0.859 0.691 2.73 0.823 0.663 2.36031030B 12 1000 0.848 0.671 2.75 0.821 0.643 2.55030421C 11 1000 0.849 0.677 2.69 0.825 0.654 2.49030623B 9 1000 0.853 0.673 2.83 0.827 0.656 2.52031123B 12 950 0.864 0.653 3.36 0.827 0.651 2.57021127A 9 950 0.858 0.653 3.21 0.841 0.631 3.09

Chemical data of MgO-rich pairs were used for calibration of new thermometer.Chemical data of FeO-rich pairs were not used for the thermometry. Total Fe asFeO.

402 K. Sato et al. / Gondwana Research 9 (2006) 398–408

coesite translation at 1000°C (29.7kbar; Bohlen and Boettcher,1982). These calibrations for experimental P–T condition wereafter Kawasaki and Motoyoshi (2005). Run durations weredepending on experimental temperatures (Table 3). However, intwo experiments (run nos. 031001C and 021127A), the PtRhthermocouples were broken after 119h and 584h, respectively(Table 3). During each experiment, the temperature and pressurewere kept constant within about 1% relative.

At the end of all runs, the electric power supply was turnedoff and sample was quenched. Chemical compositions ofproduced sapphirine and spinel were analyzed by the ElectronProbe Micro Analyzer JEOL model JXA-8800 Superprobe atEhime University with the ZAF correction method. Accelerat-ing voltage is 15kV. Electron beam current is 5×10−9 A. Beamdiameter estimated from the size of contamination spots formedby chemical analysis is 1–2μm.

Fig. 2. Histogram of XMg of sapphirine and spinel synthesized at 9kbar and 1100°CMg-rich rim is found in sapphirine grains. It is difficult to distinguish between core agranulite (TM981229-03E). Open triangles: XMg values of constituent minerals forFilled and open squares: XMg values of produced sapphirine and spinel, respectively

4. Experimental results

Table 3 lists experimental conditions and produced phases.Detailed chemical data of these phases and back-scatteredelectron images of synthesized assemblages were presented inSato et al. (2004a). Spinel was found in all run products.Produced spinel crystals were often observed as small as lessthan 2μm to 10μm in grain size. Sapphirine was found in allassemblages synthesized at temperatures of 950–1100°C. It wasalso generated in a run product obtained at 13kbar and 1150°C.At higher temperatures, melt was generated and feldsparsdisappeared from phase assemblages. This melt showed graniticcomposition (Table 4), and it was probably H2O-free as well asbulk composition of the granulite. Garnet crystallized at higherpressures. Original P–T diagram for the granulite system wasshown in Sato et al. (2004a).

In some experiments, the glass sample of rock-free startingmaterial was used. As results, sapphirine was found in only onerun product (no. 020927A; Table 3) that was obtained at 11kbarand 1050°C. Since grain sizes of sapphirine and spinel in thisrun product were too small (<2 μm), we could not collect theirprecise chemical compositions for the determination of the Fe2+–Mg distribution coefficient. Therefore, this chemical data setwas omitted from calibration of the new thermometer.

In the experiments using the mixture of glass+ rock powder,weak compositional zoning with Fe-rich core and Mg-rich rimwas often found in produced sapphirine grains (Tables 4 and 5).In those cases, weak spread of XMg ratio was also observed inspinel grains (Table 4). However, many spinel grains weresmaller in size (Sato et al., 2004a). Thus, it is difficult todistinguish between core and rim of such spinel grains. As anexample, chemical data of phases of sapphirine, spinel and meltin an assemblage that was obtained at 9kbar and 1100°C (runno. 030804A) is given in Table 4. Histogram of XMg of thesesapphirine and spinel is illustrated in Fig. 2. Kawasaki and Sato(2002) conducted phase equilibrium experiments using glass or

for 264h (run no. 030804A). Weak compositional zoning with Fe-rich core andnd rim of spinel grains in composition. Filled triangle: XMg in whole rock for thethe granulite. Numbers in parentheses: modal % of these constituent minerals..

403K. Sato et al. / Gondwana Research 9 (2006) 398–408

glass+ seed minerals as starting materials, and they alsoobtained sapphirine and orthopyroxene grains with composi-tional zoning. Such zoning in synthesized minerals is also quitecommon in previous experiments done by Harley (1984) andSato and Kawasaki (2002). Although origin of the composi-tional change in phases was not clear, we adopted Mg-richmineral pairs that were regarded as equilibrium (or nearequilibrium) for determination of the thermometer. The XMg

of whole rock for the granulite system is 0.805 (Table 2). In thegranulite, the XMg (=0.591) of spinel is the lowest value, and theXMg (=0.838) of sapphirine is higher than the X Mg

Bulk. Thus, if thestarting mixture was kept at high temperature condition, the XMg

of produced spinel should be shifted to Mg-rich side incomposition. If this theory was also applied to synthesizedsapphirine, the X Mg

Spr must be shifted to Fe-rich side incomposition because its previous XMg was higher than theX Mg

Bulk. However, in case of produced sapphirine, compositionalzoning with the Fe-rich core and the Mg-rich rim was found(Table 4 and Fig. 2). Thus, the Mg-rich sapphirine was regardedin equilibrium or near to the equilibrium in this study. For thesereasons, the Fe-rich sapphirine and the Fe-rich spinel were notused for data regress. Analytical data for just phase boundariesand too small crystals were not chosen for the determination ofnew thermometer because we avoided contaminated data.Spinel grains adjoined to sapphirine were measured in manycases. Spinel grains near sapphirine crystals were chosen for thecalibration if they did not adjoin to sapphirine directly.

Distribution coefficients KD between sapphirine and spinelfor the Fe2+–Mg exchange reaction:

KD ¼ X SplFe X

SprMg

X SplMgX

SprFe

; ð6Þ

are given in Table 5. These KD data are plotted versus 1 /T inFig. 3. For comparison, the experimental study of Das et al.

Fig. 3. Relation between the KD and temperature T. The KD is approximated as:lnKD=− (1.257±0.281)+ (2940±370) /T (K). In addition, experimental data ofDas et al. (2006) and thermodynamic data (Holland and Powell, 1998; Ouzeganeet al., 2003; Table 1) are also shown for comparisons with the new thermometer.

(2006) and the thermodynamic data (Holland and Powell, 1998;Ouzegane et al., 2003; Table 1) are also shown in the samefigure.

5. Thermodynamic treatments

The equilibrium constant K in Eq. (4) derived from thereactions (1) and (2) is defined by the quotient of activities a,

K ¼aSplFeAl2O4

aSprMg2Al4SiO10

� �1=2

aSplMgAl2O4aSprFe2Al4SiO10

� �1=2ð7Þ

and

K ¼aSplFeAl2O4

aSprMg3:5Al9Si1:5O20

� �2=7

aSplMgAl2O4aSprFe3:5Al9Si1:5O20

� �2=7; ð8Þ

respectively. If these minerals are not ideal solid solution(a=Xγ, where the γ is defined as the activity coefficient), the Kshould be distinguished from the KD as:

K ¼ X SplFe X

SprMg

X SplMgX

SprFe

dgSplFeAl2O4

g SprMgAl2Si1=2O5

gSplMgAl2O4gSprFeAl2Si1=2O5

¼ KDd Kg ð9Þ

and

K ¼ X SplFe X

SprMg

X SplMgX

SprFe

dgSplFeAl2O4

gSprMgAl18=7Si3=7O40=7

gSplMgAl2O4gSprFeAl18=7Si3=7O40=7

¼ KDd Kg; ð10Þ

respectively. Thus, for both the reactions (1) and (2), thefollowing equation can be derived from Eq. (4):

−DH -1 bar;T þ TDS-1 bar;T−ðP kbar−1 barÞDV -

1 bar;298 K

¼ RT lnKD þ RT lnKg: ð11ÞThe thermodynamic parametersΔH°,ΔS° andΔV° are assumedto be constant or nearly constant for variations on the P–Tconditions of the present experiments. The RT lnKγ in Eq. (11) isthe function of T and composition. Kawasaki and Matsui (1983)formulated the strict solution of activities of the components ofdouble-sited FeSiO3–MgSiO3–Al2O3 orthopyroxene. Theeffects of the excess energy terms, arising from non-ideal Mg–Al, Al–Fe and Fe–Mgmixings, can be regarded as negligible orconstant for the system in which bulk chemistry is constant. Leeand Ganguly (1988) referring Kawasaki andMatsui (1983) drewa diagram of KD vs. X Fe

phases, and they argued that the RTlnKγ

term can be ignored for garnet–orthopyroxene pair. In ourexperiments, variations in the X Mg

Spr and Spl are ranged as 0.653–0.722 in spinel and 0.840–0.864 in sapphirine (Table 5).Relationship between KD vs. X Fe

Spr or Spl is given in Fig. 4. Fromthis figure, it can be seen that the KD is independent ofchanges of the X Fe

Spr or Spl. The thermometric formulation is anempirical one, as it does not account for non-ideality in Fe2+–Mg–Cr3+–Fe3+ mixing in sapphirine and spinel. However,

Fig. 4. Relation between Fe2+–Mg distribution coefficient KD and XFe values ofsapphirine and spinel in run products. The XFe data was obtained by “1−XMg

(Table 5)”. Top minerals in this figure stand for constituent minerals for thegranulite. Numbers in parentheses show modal % for these minerals.

404 K. Sato et al. / Gondwana Research 9 (2006) 398–408

since the RTlnKγ in Eq. (11) can be probably assumed asconstant or nearly constant, the RTlnKDwill be approximated as:

−DH -1 bar;T þ TDS⁎1 bar;T−ðP kbar−1 barÞDV -

1 bar;298 K

¼ RT lnKD; ð12Þwhere the ΔS⁎(=ΔS°−RlnKγ) is the effective enthalpy change.For the Fe2+–Mg exchange reaction between sapphirine (7 :9 :3)and spinel of Eq. (2), the ΔV° can be calculated as −18.87cal/kbar (Table 1) by using the thermodynamic data set of Hollandand Powell (1998) and Ouzegane et al. (2003). In our study, theexperimental pressures range from 9kbar to 13kbar (Table 3).Middle value of their pressure conditions is about 11kbar. If weevaluate the pressure effect, Eq. (12) can be rewritten as follows:

−DH -1 bar;T þ TDS⁎1 bar;T þ 18:87ð11 kbar−1 barÞ ¼ RT lnKD:

ð13ÞAs a notice, the ΔV° in Table 1 corresponds to only reaction (2)with 7 :9 :3 sapphirine. If it was assumed as a negligible pressureeffect, Eq. (12) could be modified as:

−DH⁎11 kbar;T þ TDS⁎11 kbar;T ¼ RT lnKD 11 kbar;T ; ð14Þ

where the ΔH⁎ stands for ΔH°+ (P kbar−1bar)ΔV°. Experi-mentally calibrated lnKD values have been plotted against theircorresponding temperatures (in terms of 1 /T) as shown in Fig. 3.The liner fit using the least squares method (Deming, 1943) hasbeen employed, from which the values of ΔH⁎ and ΔS⁎ havebeen calculated as −5839±735cal and −2.497±0.558cal/K,respectively. From these data, the Fe2+–Mg distributioncoefficient KD between sapphirine and spinel can be approxi-mated by the following equation:

lnKD ¼ −ð1:257F0:281Þþ½ð2940F370Þ þ 9:5ðP kbar−11 kbarÞ�=T : ð15Þ

If the pressure effect was neglected, we could rewrite Eq. (15) as:

lnKD ¼ −ð1:257F0:281Þ þ ð2940F370Þ=T : ð16Þ

An approach of this kind has already been carried out with fairsuccess (Sakai and Kawasaki, 1997; Kawasaki and Motoyoshi,2000; Kawasaki and Sato, 2002; Sato and Kawasaki, 2002).

Spinel is a reciprocal solid solution in the Cr-bearing system(e.g., Wood and Nicholls, 1978). Besides, Ozawa (1983)reported that the XCr [=Cr / (Al+Cr(+Fe3+))] ratio of spinelincreases with decreasing ratio of the XMg. Thus, the effect ofthe reciprocal exchange reaction of spinel:

FeCr2O4 þMgAl2O4 ¼ MgCr2O4 þ FeAl2O4; ð17Þshould be considered in order to determine more strict Fe2+–Mgexchange thermometer (Liermann and Ganguly, 2003). How-ever, variations in the XCr of spinel and sapphirine are verysmall in the present experimental conditions [X Cr

Spl =0.044 andX Cr

Spr =0.016 in the granulite (Table 2); X CrSpl =0.025–0.036 and

X CrSpr =0.012–0.022 in synthesized minerals (e.g., Table 4)].

Liermann and Ganguly (2003) estimated the effect of XCr inspinel for Fe–Mg exchange between spinel and orthopyroxene.In our experimental data, the XCr change in produced spinel(ΔX Cr

Spl) is only 0.011. If the XMg values of 0.80 and 0.67 in theproduced orthopyroxene and spinel (as to be KD between theseminerals=about 2.0) were substituted into Liermann andGanguly's empirical equation (Eq. (8) and the third column ofTable 4 in their paper), the effect ofΔX Cr

Spl would be less than 3%at about 1000°C. On the other hand, the ΔX Cr

Spr in thesynthesized sapphirine is as less as 0.010 in our experiments.Although we can not evaluate the effect ofΔX Cr

Spr, it is probablysmall as well as the effect of ΔX Cr

Spl. Thus, in these cases, theeffects of the reciprocal exchange reactions of spinel andsapphirine are assumed to be nearly constant. However, if thepresent thermometer was applied to metamorphic rocksincluding high-Cr spinel, it would be difficult to estimate correcttemperatures (see Discussion and conclusions in this paper).

Because the whole rock system is fixed and the pressurecondition is constrained as only 9–13kbar in the presentpartitioning experiment, it is also difficult to evaluate theeffect of change of Al content by Tschermak substitution insapphirine in produced assemblages. However, the effect ofchange in Al content would probably be small in pressureranges of only 9–13kbar.

6. Discussion and conclusions

6.1. Application to UHT or high-grade metamorphic rocks

In this study, the empirical Eq. (15) [or (16)] is given as thenew sapphirine–spinel Fe2+–Mg exchange thermometer. In Fig.3, slope of our equation does not agree with Das et al. (2006). Forthe Fe2+–Mg exchange reaction between sapphirine (7 :9 :3) andspinel of Eq. (2), the ΔH° is calculated as −4031 cal (Table 1)from the previous thermodynamic data set (Holland and Powell,1998; Ouzegane et al., 2003). This means that the KD betweensapphirine and spinel surely increases with decreasing

Table 6Comparisons of estimated temperatures (°C) of UHT and high-grade metamorphic rocks by various Fe2+–Mg exchange thermometers

Area (sample no.) KD X CrSpr X Cr

Spl P a

(kbar)Spr–SplPresent 1(°C)

Spr–SplPresent 2PΔV ° effect-free

Spr–SplThermodynamic dataPΔV ° effect-free

Spr–SplDas 2006

Spr–OpxK&S 2002

Source of data

<Antarctica>Napier Complex,

Howard Hills (TM981229-03E) 3.58 0.016 0.044 7–13 873–895 888 860 947 912 Sato et al. (2004a)Howard Hills (9812280101-1) 3.18 0.009 0.043 7–13 930–953 945 941 919 – Yoshimura et al. (2000)Howard Hills (9812280101-2) 4.09 0.012 0.041 7–13 816–837 830 782 979 – Yoshimura et al. (2000)Howard Hills (9812280101-3) 3.03 0.011 0.047 7–13 954–978 970 977 909 – Yoshimura et al. (2000)Tonagh Island (A98021102H) 3.29 0.008 0.024 7–13 913–936 928 917 927 946 Hokada et al. (1999)Bunt Island (Spr–Qtz gneiss) 3.23 0.003 0.032 7–13 921–945 937 929 923 906 Osanai et al. (2001)

Lützow–Holm ComplexRundvågshetta (92011102A) 4.19 0.004 0.055 10–12 817–824 820 769 986 839 Kawasaki et al. (1993)

Rayner Complex,Forefinger Point (4652) 3.26 – – 7–12 917–936 933 923 925 1083 Harley et al. (1990)

<India>Madurai block

Ganguvarpatti (Crd–Spl or Spr–Opx) 2.63 – – 9–11 1041–1050 1050 1096 878 1049 Sajeev et al. (2004)

<Algeria>In Ouzzal Terrane (lnh 131) 4.83 0.046 0.153 5–10 745–762 765 698 1022 640 Ouzegane et al. (2003)In Ouzzal Terrane (lnh 928) 3.12 – – 5–10 931–950 954 955 915 – Ouzegane et al. (2003)

<Japan>Higo metamorphic terrane (Assemblage 1) 4.34 – 0.003 7–8 792–795 806 750 995 – Osanai et al. (1998)

<Scotland>Lewisian Complex, South Harris (95913-7) 2.83 0.005 0.015 9–14 999–1020 1007 1032 894 1075 Baba (2003)Lewisian Complex, South Harris (95919-18) 3.02 0.005 0.025 9–14 964–984 972 979 908 – Baba (2003)

Present equation (Spr–Spl thermometer), lnKD=−1.257+[2940+9.5(P−11)] /T; Other equations, thermodynamic data are after previous literatures (Holland and Powell, 1998; Ouzegane et al., 2003); Spr–Spl Das 2006means Das et al. (2006); Spr–Opx K&S 2002 is Kawasaki and Sato (2002). XCr, Cr / (Al+Cr(+Fe

3+)).a Pressure ranges in Napier Complex and Lützow–Holm Complex are after Harley (1998) and Motoyoshi and Ishikawa (1997), respectively.

405K.Sato

etal.

/Gondw

anaResearch

9(2006)

398–408

Fig. 6. Comparison of estimated temperatures between three Fe2+–Mg exchangethermometers. New sapphirine–spinel exchange thermometer (P=11kbar) wascompared with previous sapphirine–spinel thermometer of Das et al. (2006) andsapphirine–orthopyroxene thermometer of Kawasaki and Sato (2002). Datasources of mineral pairs for the new thermometer vs. the sapphirine–orthopyroxene thermometer are given in Table 6.

406 K. Sato et al. / Gondwana Research 9 (2006) 398–408

temperatures. In the thermometers that were derived by ourstudy and Das et al. (2006), theΔH⁎ in Eq. (14) (orΔH°) valueswere <0 and >0, respectively.

The new thermometer was applied to the UHT phlogopite-bearing orthopyroxene granulite (TM981229-03E) from theHoward Hills. Estimated temperatures are 873–895°C, ifpressures of 7–13kbar are substituted into the thermometer(Table 6). These temperatures can be probably identified withmetamorphic condition (=912°C; Table 6) that is evaluatedusing orthopyroxene–sapphirine exchange thermometer ofKawasaki and Sato (2002). The new thermometer was alsoapplied to UHT or high-grade metamorphic rocks from someterranes in the world. Consequently, their temperatures condi-tions were evaluated as 792–1050°C (Table 6). However, if thepairs of sapphirine and spinel containing high-Cr contents suchas X Cr

Spr =0.046 and X CrSpl =0.153 (lnh 131 from Ouzzal Terrane;

Ouzegane et al., 2003; Table 6) are substituted into the newthermometer, it will be estimated as low temperature condition(=745–765°C). The effect of pressure is shown in Fig. 5. It is assmall as 3°C or 4°C/1kbar. For comparisons with the newthermometer, the thermodynamic data (Holland and Powell,1998; Ouzegane et al., 2003; Table 1) and the previoussapphirine–spinel thermometer (Das et al., 2006) were alsoapplied to these UHTor high-grade metamorphic rocks (Table 6and Fig. 6).

We think that the new thermometer is available to estimate P–Tcondition for UHT or high-grade metamorphic rocks. Howev-er, this thermometer has several problems. The effects of X Cr

Spl,X Cr

Spr and change of Al content by Tschermak substitution insapphirine were ignored in this geothermometry. If thesefactors caused considerable effects for estimation of temper-ature, it would be difficult to evaluate correct temperatures.Besides, in the determined equation for the thermometer, the

Fig. 5. Pressure effect for present sapphirine–spinel exchange thermometer.Solid lines: pressure effect was counted using previous thermodynamic data(Holland and Powell, 1998; Ouzegane et al., 2003). Broken lines: this effect wasignored.

ΔH⁎ and ΔS⁎ in Eq. (14) have error as ±735cal and±0.558cal/K, respectively. User must apply this new ther-mometer to natural rocks carefully.

6.2. Constraints on P–T condition of the UHT phlogopite-bearing orthopyroxene granulite from Howard Hills in NapierComplex

The P–T condition of the UHT granulite (TM981229-03E)that was collected from the Howard Hills in Napier Complexwas constrained using the present exchange thermometer andprevious data set (Newton, 1972; Yoshimura et al., 2000;Kawasaki and Sato, 2002; Sato et al., 2004a) compiled in Fig. 7.Using feldspar solvus thermometer fitted for 10kbar (Kroll etal., 1993), Yoshimura et al. (2000) estimated peak metamorphictemperatures of various UHT gneisses from the Howard Hills as1000–1150°C (T condition of symbol C in Fig. 7). On the otherhand, Sato et al. (2004a) experimentally mentioned that peaktemperature of the granulite is >1100°C at 9–13kbar (P–T fieldof symbol A in Fig. 7). However, it is impossible to reveal upperlimit of the UHT metamorphism from only these two data.

The new sapphirine–spinel exchange thermometer wasapplied to the granulite, and the estimated temperatures were873–895°C when pressures were 7–13kbar (Table 6). Untilnow, only thermometer that was applicable to mineral paircontaining sapphirine was probably the orthopyroxene–sapphirine exchange thermometer of Kawasaki and Sato(2002). This thermometer without calibration for the pressureeffect was also used for temperature estimation for thegranulite. As a result, the temperature was evaluated as912°C (Table 6). These estimations by the present sapphirine–

Fig. 7. Constrained P–T condition of the UHT granulite (TM981229-03E) andcompiled P–T conditions of other UHT metamorphic rocks from Howard Hills.Original data of P–T phase diagram in the granulite (TM981229-03E) systemwas reported by Sato et al. (2004a). Broken allows stand for expected P–T path.(A) Peak metamorphic condition of the granulite was suggested by Sato et al.(2004a). (B) This T estimation of the granulite was estimated using the presentnew sapphirine–spinel and the previous sapphirine–orthopyroxene (Kawasakiand Sato, 2002) exchange thermometers. Pressure condition in the P–T fields ofsymbols A and B was constrained as “garnet-in>P>8.2kbar” by previousexperimental data of Newton (1972) and Sato et al. (2004a). (C) Peaktemperature condition of UHT gneisses that were collected from the HowardHills area was estimated by Yoshimura et al. (2000) using feldspar solvusthermometer (Kroll et al., 1993). (D) and (E) Retrograde metamorphicconditions of these UHT gneisses were evaluated by Yoshimura et al. (2000)using compositional isopleths (D: Hensen and Harley, 1990) and garnet–orthopyroxene exchange thermometers (E: Sen and Bhattacharya, 1984; Lee andGanguly, 1988).

407K. Sato et al. / Gondwana Research 9 (2006) 398–408

spinel and orthopyroxene–sapphirine exchange thermometersare cited within P–T field of symbol B in Fig. 7. Yoshimura etal. (2000) evaluated retrograde P–T condition as 8.0–9.5kbarand 850–950°C using compositional isopleths (Hensen andHarley, 1990). They also reported the P–T condition of 5.0–8.0kbar and 830–900°C using garnet–orthopyroxene ex-change thermometers (Sen and Bhattacharya, 1984; Lee andGanguly, 1988). These data are shown as P–T fields ofsymbols D and E in Fig. 7. Although pressure conditions ofthese symbols B, D and E in Fig. 7 are different from otherfields, temperature conditions of them are well fitted withothers. Needless to say, the temperatures of 873–895°C by thenew sapphirine–spinel exchange thermometer is not regardedas the peak metamorphic condition for the granulite systembecause the peak temperature has been constrained as>1100°C by the phase equilibrium experiment of Sato et al.(2004a) (the P–T field of symbol A in Fig. 7).

Any garnet is not found in this UHT granulite. Stability fieldof garnet in the granulite system is present at higher pressureside (Sato et al., 2004a; Table 3 and Fig. 7). Absence of garnet inthe granulite restricts upper limit of pressure condition from thepeak to retrograde metamorphism. Cordierite is also not

contained in the same granulite. Newton (1972) reportedreversal experimental data on an anhydrous reaction ofcordierite to sapphirine (+enstatite?)+quartz for the Fe-freebulk system. Maximum pressure of stability of anhydrouscordierite has been determined as 8.2kbar by Newton (1972).Motoyoshi and Ishikawa (1997) mentioned that cordierite-bearing corona and symplectites, which were found in UHTmetamorphic rocks from Rundvågshetta in Lützow–HolmComplex (East Antarctica), formed at about 7.5–8.0kbarunder decompressional condition. If pressure of 8.2kbar issubstituted into our empirical equation, temperature conditionof the granulite (TM981229-03E) is estimated as 887°C. Thisproves that the sapphirine crystal surely existed until 887°C atP>8.2kbar (Fig. 7). As results, the isobaric cooling history ofthe Howard Hills area is revealed as shown in Fig. 7. This figurealso means that the present sapphirine–spinel exchangethermometer is applicable to temperature estimations forgarnet-free UHT or high-grade metamorphic rocks.

Acknowledgements

The authors received invaluable discussion and sugges-tions from Prof. S.L. Harley, Drs. D.J. Dunkley, T. Hokada,T. Tsunogae and Y. Yoshimura. We also obtained invaluablecomments from Prof. E. Takahashi, Drs. T. Kogiso, Y. Nishiharaand S. Omori. Constructive criticisms from two anonymousreviewers are much appreciated. We thank Dr. P. Bhalla forhis help in improving the English of this paper. Main part ofthis study was derived from the Ph.D. thesis of Kei Sato at theGraduate School of Science and Engineering, Ehime Univer-sity. The research was financially supported by the SasakawaScientific Research Grant from The Japan Science Society toKei Sato (no. 15-144) and by the Grant-in-Aid for ScientificResearch from the Ministry of Education, Culture, Sports,Science and Technology (MEXT) of the Japanese Governmentto Toshisuke Kawasaki (no. 14654093). It was partly supportedby the 21st Century COE Program “How to build habitableplanets”, Tokyo Institute of Technology, sponsored by theMEXT, Japan.

References

Baba, S., 2003. Two stages of sapphirine formation during prograde andretrograde metamorphism in the Palaeoproterozoic Lewisian Complex inSouth Harris, NW Scotland. J. Petrol. 44, 329–354.

Bertland, P., Ellis, D.J., Green, D.H., 1991. The stability of sapphirine–quartzand hyperthene–sillimanite–quartz assemblages: an experimental investi-gation in the system FeO–MgO–Al2O3–SiO2 under H2O and CO2

condition. Contrib. Mineral. Petrol. 108, 55–71.Bohlen, S., Boettcher, A.L., 1982. The quartz↔coesite transformation: a precise

determination and the effect of other components. J. Geophys. Res. 87,7073–7078.

Das, K., Dasgupta, S., Miura, H., 2003. An experimentally constrainedpetrogenetic grid in the silica-saturated portion of the system KFMASH athigh temperatures and pressures. J. Petrol. 44, 1055–1075.

Das, K., Fujino, K., Tomioka, N., Miura, H., 2006. Experimental data on Fe andMg partitioning between coexisting sapphirine and spinel: an empiricalgeothermometer and its application. Eur. J. Mineral. 18, 49–58.

Deming, W.E., 1943. Statistical Adjustment of Data. John Wiley Sons, NewYork. 261 pp.

408 K. Sato et al. / Gondwana Research 9 (2006) 398–408

Hall, H.T., 1971. Fixed points near room temperature. In: Lloyd, E.C. (Ed.),Proceedings of Symposium of Accurate Characterization of High PressureEnvironment. NBS Spec. Pub., vol. 326, pp. 313–314.

Harley, S.L., 1984. An experimental study of the partitioning of the Fe andMg between garnet and orthopyroxene. Contrib. Mineral. Petrol. 86,359–373.

Harley, S.L., 1998. On the occurrence and characterization of ultrahigh-temperature crustal metamorphism. In: Treloar, P.J., O'Brien, P.J. (Eds.),What Drives Metamorphism and Metamorphic Reaction? Special Publica-tion-Geological Society of London, vol. 138, pp. 81–107.

Harley, S.L., Hensen, B.J., 1990. Archaean and Proterozoic high-grade terranesof East Antarctica (40–80°E): a case study of diversity in granulite faciesmetamorphism. In: Ashworth, J.R., Brown, M. (Eds.), High-TemperatureMetamorphism and Crustal Anatexis. Unwin Hyman, London, pp. 320–370.

Harley, S.L., Motoyoshi, Y., 2000. Al zoning in orthopyroxene in a sapphirinequartzite: evidence for >1120°C UHT metamorphism in the NapierComplex, Antarctica, and implications for the entropy of sapphirine.Contrib. Mineral. Petrol. 138, 293–307.

Harley, S.L., Hensen, B.J., Sheraton, J.W., 1990. Two-stage decompression inAntarctica: orthopyroxene–sillimanite granulites from Forefinger Point,Enderby Land, implications for the evolution of the Archaean NapierComplex. J. Metamorph. Geol. 8, 591–613.

Hensen, B.J., Green, D.H., 1973. Experimental study of the stability ofcordierite and garnet in pelitic compositions at high pressure andtemperatures: III. Synthesis of experimental data and geological applica-tions. Contrib. Mineral. Petrol. 38, 151–166.

Hensen, B.J., Harley, S.L., 1990. Graphical analysis of P–T–X relations ingranulite facies metapelites. In: Ashworth, J.R., Brown, M. (Eds.), High-Temperature Metamorphism and Crustal Anatexis. Unwin Hyman, London,pp. 19–56.

Hokada, T., Osanai, Y., Toyoshima, T., Owada, M., Tsunogae, T., Crowe, W.A.,1999. Petrology and metamorphism of sapphirine-bearing aluminousgneisses from Tonagh Island in the Napier Complex, East Antarctica.Polar Geosci. 12, 49–70.

Holland, T.J.B., Powell, R., 1998. An internally consistent thermodynamic dataset for phases of petrological interest. J. Metamorph. Geol. 16, 309–343.

Kawasaki, T., Matsui, Y., 1983. Thermodynamic analyses of equilibriainvolving olivine, orthopyroxene and garnet. Geochim. Cosmochim. Acta47, 1661–1679.

Kawasaki, T., Motoyoshi, Y., 2000. High-pressure and high-temperature phaserelations of an orthopyroxene granulite from McIntyre Island, EnderbyLand, East Antarctica. Polar Geosci. 13, 114–134.

Kawasaki, T., Sato, K., 2002. Experimental study of Fe–Mg exchange reactionbetween orthopyroxene and sapphirine and its calibration as a geotherm-ometer. Gondwana Res. 5, 741–747.

Kawasaki, T., Motoyoshi, Y., 2005. Experimental constraints on the decom-pressional P–T paths of Rundvågshetta granulites, Lützow–Holm Complex,East Antarctica. In: Fütterer, D.K., Damaske, D., Kleinschmidt, G., Miller,H., Tessensohn, F. (Eds.), Antarctica: Contributions to Global EarthSciences. Springer-Verlag, Berlin, pp. 23–36.

Kawasaki, T., Ishikawa, M., Motoyoshi, Y., 1993. A preliminary report oncordierite-bearing assemblages from Rundvågshetta, Lützow–Holm Bay,East Antarctica: evidence for a decompressional P–T path? Proc. NIPRSymp. Antarct. Geosci. 6, 47–56.

Kretz, R., 1983. Symbols for rock-forming minerals. Am. Mineral. 68,277–279.

Kroll, H., Evangelakakis, C., Voll, G., 1993. Two-feldspar geothermometry: areview and revision for slowly cooled rocks. Contrib. Mineral. Petrol. 144,510–518.

Lee, H.Y., Ganguly, J., 1988. Equilibrium compositions of coexisting garnet andorthopyroxene: experimental determinations in the system FeO–MgO–Al2O3–SiO2, and applications. J. Petrol. 29, 93–113.

Liermann, H.P., Ganguly, J., 2003. Fe2+–Mg fractionation between orthopyr-oxene and spinel: experimental calibration in the system FeO–MgO–Al2O3–Cr2O3–SiO2, and applications. Contrib. Mineral. Petrol. 145,217–227.

Miyamoto, T., Yoshimura, Y., Sato, K., Motoyoshi, Y., Dunkley, D.J., Carson,C.J., 2004. Occurrences of metamorphosed ultramafic rock and associating

rocks in Howard Hills, Enderby Land, East Antarctica: evidence of partialmelting from geochemical and isotopic characteristics. Polar Geosci. 17,88–111.

Motoyoshi, Y., Hensen, B.J., 2001. F-rich phlogopite stability in ultra-high-temperature metapelites from the Napier Complex, East Antarctica. Am.Mineral. 86, 1404–1413.

Motoyoshi, Y., Ishikawa, M., 1997. Metamorphic and structural evolution ofgranulites from Rundvågshetta, Lützow-Holm Bay, East Antarctica. TheAntarctic Region: Geological Evolution and Processes, pp. 65–72.

Muroi, K., 1979. Water in analytical chemistry; determination of water Karl–Fischer Method. Bunseki 50, 74–82 (in Japanese).

Newton, R.C., 1972. An experimental determination of the high-pressurestability limits of magnesian cordierite under wet and dry conditions. J.Geol. 80, 398–420.

O'Neill, H.St.C., 1986. No–MoO2 (MOM) oxygen buffer and the free energy offormation of MnO2. Am. Mineral. 71, 1007–1010.

Osanai, Y., Hamamoto, T., Maishima, O., Kagami, H., 1998. Sapphirine-bearinggranulites and related high-temperature metamorphic rocks from the Higometamorphic terrane, west-central Kyushu, Japan. J. Metamorph. Geol. 16,53–66.

Osanai, Y., Toyoshima, T., Owada, M., Tsunogae, T., Hokada, T., Crowe, W.A.,Kusachi, I., 2001. Ultrahigh temperature sapphirine–osumilite and sapphi-rine–quartz granulites from Bunt Island in the Napier Complex, EastAntarctica— reconnaissance estimation of P–T evolution. Polar Geosci. 14,1–24.

Ouzegane, K., Guiraud, M., Kienast, J.R., 2003. Prograde and retrogradeevolution in high-temperature corundum granulites (FMAS and KFMASHsystems) from In Ouzzal Terrane (NW Hoggar Algeria). J. Petrol. 44,517–545.

Owen, J.V., Greenough, J.D., 1991. An empirical sapphirine–spinel Mg–Feexchange thermometer and its application to high grade xenoliths in thePopes Harbour dyke, Nova Scotia, Canada. Lithos 26, 317–332.

Ozawa, K., 1983. Evolution of olivine–spinel geothermometry as an indicator ofthermal history for peridotites. Contrib. Mineral. Petrol. 82, 52–65.

Sajeev, K., Osanai, Y., Santosh, M., 2004. Ultrahigh-temperature metamorphismfollowed by two-stage decompression of garnet–orthopyroxene–sillimanitegranulites from Ganguvarpatti, Madurai block, southern India. Contrib.Mineral. Petrol. 148, 29–46.

Sakai, S., Kawasaki, T., 1997. An experimental study of Fe–Mg partitioningsbetween orthopyroxene and cordierite in the Mg-rich portion of theMg3Al2Si3O12–Fe3Al2Si3O12 system at atmospheric pressure: calibration ofits geothermometry for high-temperature granulites and igneous rocks. Proc.NIPR Symp. Antarct. Geosci. 10, 165–177.

Sato, K., Kawasaki, T., 2002. High-pressure and high-temperature experimentson the phase relations in the system of Mg-rich garnet composition(Prp75Alm25): implication for the Fe–Mg partitioning between garnet andorthopyroxene. Polar Geosci. 15, 66–79.

Sato, K., Miyamoto, T., Kawasaki, T., 2004a. Experimental constraints ofmetamorphic pressure and temperature, and phase relations of a phlogopite-bearing orthopyroxene granulite from Howard Hills, Napier Complex, EastAntarctica. J. Mineral. Petrol. Sci. 99, 191–201.

Sato, K., Miyamoto, T., Kawasaki, T., 2004b. High-pressure and high-temperature annealing experiment of a phlogopite-bearing orthopyroxenegranulite from the Howard Hills, Napier Complex, East Antarctica:synthesis of the spinel–sapphirine corona. The 24th Symposium onAntarctic Geosciences, Program and Abstracts, 14–15 October 2004. Natl.Inst. Polar Res, Tokyo, p. 87 (in Japanese).

Sen, S.K., Bhattacharya, A., 1984. An orthopyroxene–garnet thermometer andits application to the Madras charnockites. Contrib. Mineral. Petrol. 88,64–71.

Sheraton, J.W., Tingey, R.J., Black, L.P., Offe, L.A., Ellis, D.J., 1987. Geologyof Enderby Land and Western Kemp Land, Antarctica. Bur. Miner. Resour.,Geol. Geophys., Bull. 223 (51 pp.).

Wood, B.J., Nicholls, J., 1978. The thermodynamic properties of reciprocal solidsolutions. Contrib. Mineral. Petrol. 66, 389–400.

Yoshimura, Y., Motoyoshi, Y., Grew, E.S., Miyamoto, T., Carson, C.J., Dunkley,D.J., 2000. Ultrahigh-temperature metamorphic rocks from Howard Hills inthe Napier Complex, East Antarctica. Polar Geosci. 13, 60–85.