Contrib Mineral Petrol (1981) 78 : 99-109 Contributions to [603712]

Contrib Mineral Petrol (1981) 78 : 99-109 Contributions to
Mineralogy and
Petrology 9 Springer-Verlag 1981
The A|20 3 Contents of Enstatite in Equilibrium with Garnet
in the System MgO – Al203 – SiO2
at 15-40 kbar and 900~ ~ C
D. Perkins III*, T.J.B. Holland**, and R.C. Newton
Department of the Geophysical Sciences., University of Chicago, Chicago, Illinois 60637, USA
Abstract. Forty-six reversed determinations of the AI203 content
of enstatite in equilibrium with garnet were made in the PIT
range 15-40 kbar/900-1,600~ C in the MgO-AlzO3-SiO2 sys-
tem. Starting materials were mixtures of synthetic pyrope+A1-
free enstatite and pyrope+enstatite (5-12% AlzOs). AI/O3 con-
tents in reversal run pairs closely approached common values
from both the high- and low-A1 sides. Most experiments were
done in a piston-cylinder device using a NaC1 medium; some
runs at very high temperatures were made in pyrex/NaC1 or
pyrex/talc assemblies.
The measured enstatite compositions, expressed as mole frac-
tions 9 En of Mg2(]V[gA1)(A1S13)Ole(Xopy ) were fitted by a Monte-
Carlo method to the equilibrium condition:
0 0 AH97o-970zJS97 o
P T
+ ~ AV9~ alP- ~ AS~ dT+ RTlnXnpy=O
1 9"70
where the best fit parameters of AH, AS and A V (1 bar, 970 K)
for the reaction pyrope=opy are 2,040 cal/mol, 2.12 eu and
9.55 cc/mol.
In addition to the determination of A120 s contents of ensta-
tite, the univariant reaction pyrope + forsterite = enstatite + spinel
was reversibly located in the range 1,100-1,400 ~ C. A "best-fit"
line passes through 22, 22.5 and 25 kbar at 1,040, 1,255 and
1,415 ~ C, respectively. Our results for the univariant reaction
are in agreement with previous studies of MacGregor (1974)
and Haselton (1979). However, comparison of the experimentally
determined curve with thermochemical calculations suggests that
there may be a small error in the tabulated AH~f(97o,1 ) value
for enstatite. A value of -8.32 rather than -8.81 kcal/mole
(Charlu et al. 1975) is consistent with the present data.
Application of garnet-enstatite-spinel-forsterite equilibria to
natural materials is fraught with difficulties. The effects of non-
ternary components are poorly understood, and the low solubili-
ties of A1203 in enstatite under most geologically reasonable
conditions make barometric or thermometric calculations highly
sensitive. More detailed studies, including reversed determina-
tions in low-friction assemblies, are sorely needed before the
effects of important diluents such as Fe, Ca and Cr can be
fully understood.
* Present address: Geology Department, University of North Dako-
ta, Grand Forks, North Dakota 58202, USA
** Present address: Department of Earth Science, University of Cam-
bridge, Cambridge CB2 3EW, England
Reprint requests to." D. Perkins III Introduction
The solubility of A120 3 in orthopyroxene is of fundamental
importance to petrologists. At high pressures, the saturation
A120 3 content of orthopyroxene (opx) is buffered by
coexisting garnet (gt) according to the reaction (Lane and
Ganguly 1980):
MgsAlzSi3012 (pyrope)=Mg3AlzSi3012 ("opy"). (1)
in gt in opx
This reaction was expressed using different end members by
Wood and Banno (1973):
Mg3A12Si3012 = MgAlzSiO6("MgTs" )
in gt in opx
+ Mg2Si20 6 (enstatite) (1 a)
in opx
Reaction (1) is the basis of the most important geobarometer
for garnet peridotites because of the large pressure depen-
dence of the solubility of A120 3 in enstatite in the presence
of garnet. At lower pressures in peridotite systems,
A120 3 saturation is determined by the presence of olivine
(ol) and spinel (sp) as described by the reaction:
Mg4Si40~z+MgAlzOr
in opx in sp
=MgsA12Si3012+Mg~Si204 (forsterite)
in opx in ol
or its MgTs analogue:
Mg2Si20 6 + MgA120 4 = Mg2A1SiO 6 + Mg2SiO 4
in opx in sp in opx in ol (2)
(2a)
These reactions have small volume change and are thus un-
suitable as geobarometers, but may have considerable po-
tenital for geothermometry (Fujii 1977; Danckwerth and
Newton 1978). Reactions (1) and (2) are related by a third
which represents the transition of garnet peridotite to
spinel peridotite:
Mg3A12Si3012 + Mg2SiO4 = Mg4Si4012 -~ MgAI204
in gt in ol in opx in sp (3)
or:
Mg3A12Si3012 + Mg2SiO, = 2Mg2Si206 + MgA120~
in gt in ol in opx in sp (3a)
0010-7999/81/0078/0099/$02.20

100
All three of the reactions have been investigated experimen-
tally in chemically simple systems approximating natural perido-
tites. The simplest and most-studied model system is MgO-
A1203–SiOz in which reactions (1) and (2) are divariant. Reac-
tion (1) was studied by Boyd and England (1964), MacGregor
(1974), Howells and O'Hara (1978) and Lane and Ganguly
(1980). Reaction (2) was studied by MacGregor (1974), Fujii
(1977) and Danckwerth and Newton (1978), and additional ex-
perimental data were provided by Anastasiou and Seifert (1972)
and Arima and Onuma (1977). Reaction (3) is univariant, and
was studied by MacGregor (1974), Danckwerth and Newton
(1978) and Haselton (1979), and Staudigel and Schreyer (1977)
have provided additional data.
In spite of the numerous studies, inconsistencies and uncer-
tainties remain, for two major reasons: difficulty of pressure
calibration of the solid-medium piston-cylinder high pressure
apparatus, and reliance on synthesis, rather than reversed equi-
librium data in most of the studies. Only Danckwerth and New-
ton (1978) and Lane and Ganguly (1980), among previous inves-
tigators, experimentally demonstrated equilibrium enstatite com-
positions by reaction from Al203-undersaturated and -oversatur-
ated directions at fixed P and T in runs with the low-strength
NaC1 solid pressure medium (Johannes 1973; Mirwald et al.
1975), for which pressure uncertainties are very small. Only three
unambiguous reversals of reaction 2) by Danckwerth and New-
ton (1978) and two of reaction (1) by Lane and Ganguly (1980)
have been achieved heretofore. Numerous determinations of re-
action (3) were achieved by MacGregor (1974), but enstatites
of high and low A1203 content were not used as starting materials
in the experiments, and it is possible that metastable pyroxene
compositions can bias the P- T location of the univariant reac-
tion.
Experimental data on reactions (1) and (2) in the three-com-
ponent system serve as the starting point for geobarometry-
geothermometry of natural peridotites. For this purpose, the
experimental data may be reduced to simple thermodynamic
expressions which relate P, T, and the orthopyroxene A1203
content. A simple but effective model of ideal solution of
MgzSi/O 6 (En) and MgAI2SiO6 (MgTs) components in enstatite
was first used by Wood and Banno (1973) to smooth experimen-
-tal data in the garnet field. This model has been used for garnet-
field A1203 isopleth calculations by Wood (1975) and Obata
(1976) and by Wood (1975), Obata (1976), Stroh (1976), Fujii
(1977), and Danckwerth and Newton (1978) for similar calcula-
tions in the spinel field. Depending on which set of experiments
were used in the modeling, quite different A1203 isopleth curves
and standard reaction enthalpy and entropy changes, AH ~ and
AS ~ for reactions (1 a) and (2a) were derived. Lane and Ganguly
(1980) advocated use of the aluminous orthopyroxene molecule
Mg3A12Si3012 (Opy), recognizing that an aluminous end-
member molecule closer to pure enstatite in composition is likely
to mix more ideally with it.
For application of the simple system data to complex natural
systems, it is necessary to evaluate the effects of Fe 2+, Ca, Na
and Cr and, to a lesser extent, Fe 3 § and Mn, on the activities
of garnet, pyroxene and spinel components. This was done by
Wood and Banno (1973) as ideal ion-site substitutions for Mg
and A1 in the minerals, an approximation which works best
if the concentrations of the non-ternary cations are relatively
low, as they generally are in garnet and spinel peridotites.
Important information results from comparison of the stan-
dard AH ~ and AS ~ of reaction derived from experimental iso-
pleth sets with calorimetrically measured values. Danckwerth and
Newton (1978) concluded that AH ~ for reaction (2a) based on their three reversals of enstatite composition was in good agree-
ment with high temperature solution calorimetry of Charlu et al.
(1975) on synthetic MgzSi206, MgA1204, MgzSiO4, and an alu-
minous pyroxene of composition Eno.,MgTs0.2. AH ~ for reac-
tion (1 a) derived by Wood (1975) from the MacGregor (1974)
isopleth set is 7.0 kcal, much higher than the 3.1 + 1.9 from solu-
tion calorimetry, but the Boyd and England (1964) garnet-field
isopleths lead to AH ~ (1 a)=4.2 kcal (Wood and Banno 1973).
Thus, the existing isopleth data do not allow definitive testing
of the ideal solution theory against solution calorimetry. Low-
temperature and high-temperature heat capacity measurements
now exist for pyrope (Haselton and Westrum 1980) and enstatite
(Krupka et al. 1979) which permit evaluation of AS ~ for the
reactions and provide another test of the ideal aluminous pyrox-
ene theories, if definitive experimental enstatite compositions
are secured.
Forty-six reversed determinations of the A1203 content of
enstatite in equilibrium with garnet in the system MgO-
A1203-SiO2 were made in the present study. Fluxes were used
in all experiments to produce enstatite crystals large enough
for electron microprobe analysis. The objective was to define
as accurately as possible the enstatite A1203-isopleth curves in
the garnet field, for purposes of testing the ideal aluminous
pyroxene theories and to provide a data base for theoretical
extension to more complex chemical systems emulating natural
peridotites.
Experimental Techniques
High Pressure Apparatus and Calibration. Experiments were car-
ried out in conventional piston-cylinder apparatus with 1/2"-
diameter and 3/4"-diameter chambers. Four different pressure
assemblies with internal graphite heater sleeves were used. Scale
drawings of these assemblies may be obtained from the authors
on request. Brief descriptions follow:
a) At pressures less than 32 kbar and temperatures below
the melting point of NaC1, 3/4" NaC1 cells (assembly A) identical
to those of Holland (1980) were used. This assembly has been
shown to require essentially zero pressure correction in the
"piston-out" mode, even at 600 ~ C, based on Holland's (1980)
comparisons with the albite to jadeite + quartz equilibrium deter-
mined in a gas-pressure apparatus (Hays and Bell 1973).
b) At pressures less than 32 kbar and temperatures close
to or above the melting point of NaC1, the pressure medium
outside the heater sleeve was NaCI, but the inner parts were
pyrex and boron nitride, similar to the assembly used by Newton
(1972), except for the smaller (3/4") diameter. This assembly
is termed B in the run tables. The inner diameter of the heater
sleeve was 0.4". The pyrex assembly is not frictionless, but re-
quires pressure calibration relative to equilibria determined in
frictionless apparatus. Four calibrants were used. The reaction
albite=jadeite+quartz was reversibly bracketed at 1,025 ~
1,050~ C using the same starting materials as Holland (1980).
The reaction ferrosilite–fayalite+quartz was bracketed hyd-
rothermally with synthetic starting materials at 900 ~ and 950 ~ C
with the NaC1 assembly, giving results identical to those of Boh-
len et al. (1980), and then bracketed with the NaCl-pyrex-BN
setup, which showed substantial pressure loss (negative pressure
correction). The reaction pyrope + forsterite = enstatite + spinel
with 7 wt.% Al/O3 in the enstatite was reversed hydrothermally
at 1,085~ ~ C in the NaC1 assembly, giving results identical
to those of Danckwerth and Newton (1978), and then bracketed
with assembly B at 1,110~ ~ C. Finally, the reaction anhy-

101
drous Mg-cordierite = sapphirine § quartz (Newton 1972) was re-
versed at 1,250 ~ and 1,380~ C with synthetic starting materials
and compared to 'a bracked at 1,250~ ~ C of 7.2_+0.2 kbar
done in a gas apparatus (R.C.N. unpublished data). The dP/dT
slope of the univariant equilibrium was computed accurately
to be 0.4 bar/~ from heat of solution measurements (Charlu
et al. 1974). The results of all of these calibration runs, given
in Table 1 (available from the authors), was a negative pressure
correction (pressure loss) of 6-1l %, with the highest corrections
below 1,000 ~ C. As a reasonable overall compromise, we have
used 9 _+ 1% correction for all runs with pressure medium B.
c) At pressures of 32 kbar and above and temperatures of
1,300 ~ C or less, a 1/2" NaC1 assembly was used. The albite=
jadeite+quartz reaction was bracketed at 1,050 ~ and 1,300~ C
with results exactly in agreement with Holland's (1980) calibra-
tion formula, thus indicating zero friction correction.
d) At temperatures of 1,500 ~ C and above, a 1/2"-diameter
assembly with talc and pyrex outer sleeves and pyrex inner parts
was used. This assembly was calibrated at 1,200~ C with the
albite=jadeite+quartz reaction. The bracketing runs are given
in Table 1. A plot of the ratio of the intensities of the 27.8~
(CuKc 0 albite reflection to the 30.6 ~ jadeite reflection versus
pressure for runs of equal duration (3 h) gave -15% correction
on the basis of the same ratio in the starting mix.
In all calibration runs and experimental runs with NaC1,
the" piston-out" technique was used. A nominal pressure several
kbar lower than the final run pressure was applied. Then the
sample was heated to the final temperature. Thermal expansion
of the NaCI increased the pressure by a precalibrated amount,
which brought the sample very near to the final desired pressure.
Sometimes a small amount of pressure bleeding was required
to prevent overshoot. With the pyrex assemblies, the procedure
was as follows. First a "locking" pressure of several kbar was
applied. Then the sample was heated to 700o-800 ~ C to soften
the pyrex and prevent brittle fractures during loading. The assem-
bly was taken to the final desired nominal pressure and heated
to the final temperature. Continual pumping was necessary to
maintain gauge pressure against relaxation of the glass. Thus,
the final condition was advancing piston, or "piston-in". The
pressures reported in the run tables are the ranges of gauge
pressure observed during the runs.
Temperature Measurement. Temperature uncertainty has been
a problem in previous experimental studies of subsolidus equilib-
ria. Thermocouple emf drift can occur, especially with Pt-PtRh
thermocouples, in the long run times often necessary for equili-
bration. The pressure effect on thermocouple calibration is not
well known, and is highly dependent on the specific geometry
of the pressure assembly used. These problems were addressed
in the present study by using side-by-side chromel-P-alumel ther-
mocouples and W-3% Re vs. W-25% Re thermocouples in many
of the runs. Temperatures were controlled automatically on the
W-Re thermocouple. Both thermocouples were monitored at
intervals with a potentiometer. The readings, given in the run
tables, were always within a few degrees of each other, indicating
that the combined effects of poisoning and pressure effect are
small, because it is highly unlikely that they would result in
identical readings of thermocouples of greatly differing emf.
After several hours in runs at temperatures above 1,000 ~ C the
thermocouples usually showed a tendency to deverge, with the
Cr-A1 indication falling below that of the W-Re thermocou-
ple. The runs were always terminated at this point. Quenching
by power cut-off brought temperatures below 100 ~ C in about
10 seconds. Table 2. Synthesis conditions of starting materials
Phase P T Comments
(Kb) (~
1 MgSiO 3 20
(Opx)
2 MgSiO395A1203, 20
(Opx)
3 MgSiO39~A12037 20
(Opx)
4 MgSiO3~.~A1203~.~ 20
(Opx)
5 MgSiO39oA12031o 15
(Opx)
6 MgSiO3~sA1203,2 18
(Opx)
7 Mg3A12Si3Olz 30
(Gt)
8 MgAI204 1 atm.
(Sp)
9 MgzSiO r 1 atm.
(Vo) 1,000 hydrothermal
from clinoenstatite
1,000 hydrothermal"
from oxides
1,000 hydrothermal
from oxides
t,000 hydrothermal
from oxides
1,100 hydrothermal
from oxides
1,000 hydrothermal
from oxides, reground
and recycled at 1,600 ~
and 40 Kb
1,100 hydrothermal
from oxides
1,425 dry from a sintered
oxide pellet
1,425 dry from a sintered
oxide pellet
Starting Materials and Encapsulation. The synthetic crystalline
starting materials used in reversal experiments were made hyd-
rothermally in sealed gold capsules from oxides, except as noted
in Table 2. Homogeneous high-A1 enstatites were difficult to
synthesize. A nominal 12% A1203 enstatite was prepared hyd-
rothermally at 1,000~ C and 18 kbar. It is possible to achieve
very high metastable Al/O3 contents in hydrothermal runs under
these conditions (Fawcett and Yoder 1963). The charge was
rerun at 1,600 ~ C and 28 kbar for several hours. The X-ray pat-
tern improved greatly, indicating 11.3 weight percent A1203 by
the reflection difference method of Boyd and England (1960),
and a small amount of garnet. Details of synthesis of the other
materials are given in Danckwerth and Newton (1978) and Per-
kins and Newton (1981).
In the 3/4" apparatus NaC1 runs, reversal pairs of sealed
Pt capsules were run side-by-side to determine the enstatite
A1203 contents. The thermocouple tip was lodged between the
capsules, whose combined width was about 4 ram. One capsule
contained pyrope+orthopyroxene oversaturated in A1203 (5,
10, or 12 wt. %) in nearly equal amounts, and the other contained
pyrope + orthopyroxene undersaturated in A1203 (0 or 5 wt.%).
Only one capsule at a time could be run in the 1/2" apparatus
and the 3/4" NaCl-pyrex assemblies. Single sealed Pt capsules
containing a mixture of pyrope + forsterite + enstatite + spinel in
reacting proportions were used to locate reaction (3). The ensta-
tire had 7 wt.% A1203 for runs near 1,100 ~ C, 8.5 wt.% AlzO3
for runs near 1,250 ~ and 10wt.% AlzO3 for runs near
1,400 ~ C. These were previously-determined equilibrium AlzO3
contents at the different run conditions, selected so as to mini-
mize effects of metastable A1203 content on the position of
the univariant boundary.
In order to grow run products of sufficient size (and to
avoid fine intergrowths of garnet and pyroxene) for accurate
microprobe analysis it was necessary to use a flux. Starting mixes
for runs below 1,500~ C were sealed into 1/16" Pt tubes with
fluxing materials. At temperatures below 1,150 ~ C, 30 wt.% H20

102
Table 3. Reversal runs defining the A120 3 content of enstatite in the presence of garnet
Run # Pressure P TW-Ro Tcr Al Run duration Initial Final
media (Kbar) (~ C) (~ C) (h) wt% A1203 wt% A1203
G1 A 24.8-25.2 1,089-1,094 1,095-1,097 5 0 5.40
G1 A 24.8~5.2 1,089 1,094 1,095-1,097 5 10 4.60
G3 A 29.8-30.2 1,093-1,105 1,090-1,104 1 0 3.45
G3 A 29.8-30.2 1,093-1,105 1,090-1,104 1 10 3.05
G5 A 20.6-21.0 1,073 1,079 1.076–1,080 6 0 6.60
G5 A 20.6 21.0 1,073 10079 1,076-1,080 6 10 6.00
G7 A 25.0-25.4 1,043-1,046 1.048-1,049 12 0 4.90
G7 A 25.0-25.4 1,043-1,046 1,048-1,049 12 10 4.30
G9 A 19.8-20.2 1,047 1,052 1,046-1,051 12 0 7.60
G9 A 19.8-20.2 1,047-1,052 1,046-1,051 12 10 6.20
G13 C 30.4-32.0 1,088-1,091 1,095-1,105 5 0 3.10
G14 C 31.2-32.0 1,093-1,100 1,090-1,100 5 10 2.45
G19 C 39.6-41.2 – 1,092-1,098 4 0 1.80
G20 C 39.641.2 1,096-1,105 1,097-1,101 5 10 1.40
G22 C 34.4-35.2 1,104-1,107 1,092-1,106 7 0 2.70
G23 C 34.8-35.2 1,099 1,105 1,104-1,106 11 10 1.50
P3 A 22.0-22.8 1,038-1,044 1,036-1,049 50 0 6.05
P3 A 22.0-22.8 1,038 1,044 1,036-1,049 50 10 5.65
P8 A 21.1~2.2 898- 918 901- 903 104 0 3.95
P8 A 21.1-22.2 898 918 901- 903 104 5 3.10
P16 A 29.5-30.5 951 966 943- 950 23 0 2.25
P16 A 29.5-30.5 951- 966 943- 950 23 5 1.45
P17 A 29.6-31.1 865- 873 872- 877 143 0 1.60
.P17 A 29.6-31.i 865 873 872- 877 143 5 1.20
P19 A 29.2 30.1 1,000-1,001 985-1,003 33 0 2.65
P19 A 29.2-30.1 1,000-1,001 985-1,003 33 10 2.05
P24 A 26.0-26.5 1,092 1,095 – 3 0 4.80
P24 A 26.0-26.5 1,092-1,095 – 3 10 4.00
P25 A 27.9-28.3 1,09~1,098 1,092-1,107 4 0 3.72
P25 A 27.9-28.3 1,092-1,098 1,092-1,107 4 10 2.77
P26 A 29.7-30.3 1,028-1,050 1,027-1,035 3 0 3.20
P26 A 29.%30.3 1,028-i,050 1,027-1,035 3 10 2.40
P27 A 29.6 30.5 1,070-1,075 1,066-1,077 3 0 3.25
P27 A 29.6 30.5 1,070-1,075 1,066-1,077 3 10 2.45
P28 A 23.7-24.3 1,094-1,096 1,091-1,097 3 0 5.35
P28 A 23.7 24.3 1,094-1,096 1,091-1,097 3 10 4.65
P30 A 21.8 22.6 944- 951 942- 948 67 0 4.10
P30 A 21.8 22.6 944- 951 942 948 67 10 3.20
P32 A 23.8-24.4 899- 901 890- 900 24 0 4.20
P32 A 23.8 24.4 899- 901 890- 900 24 10 2.70
P33 A 25.3~6.5 887 901 887 900 47 0 3.45
P33 A 25.3-26.5 887 901 887- 900 47 10 2.55
P34 A 27.4-28.5 896- 902 892- 901 50 0 2.15
P34 A 27.4-28.5 896- 902 892- 901 50 10 1.65
P36 A 29.6-30.4 897- 902 892 900 8 0 1.70
P36 A 29.6-30.4 897- 902 892 900 8 10 1.35
P37 A 21.8-23.0 1,046 1,054 1,043-1,053 12 0 6.00
P37 A 21.8-23.0 1,046-1,054 1,043 1,053 12 10 5.85
P41 A 22.6 23.5 1,090-1,095 1,080-1,091 5 0 5.55
P41 A 22.6-23.5 1,09(~1,095 1,080-1,091 5 10 4.95
P57 A 20.9-31.9 1,095 1,096 1,088 1,090 4 0 7.45
P63 A 21.~22.1 1,094-1,096 1,086-1,091 4 10 6.50
p65 B 29.4-30.2 ~ 1,298-1,301 – 2 0 6.60
P66 B 29.6-30.1" 1,296-1,300 – 2 10 6.20
P67 A 30.0-30.4 1,195 1,200 – 3 0 3.85
P67 A 30.0-30.4 1,195 1,200 – 3 10 3.55
P68 A 29.9 30.9 1,250-1,252 – 2 0 4.35
P68 A 29.9 30.9 1,250-1,252 – 2 10 4.25
P73 A 25.4-26.6 1,191-1,203 – 3 0 5.80
P73 A 25.4-26.6 1,191 1,203 – 3 10 5.20
P76 A 21.7-22.4 1,000-1,004 995-1,004 29 0 5.05
P76 A 21.7-22.4 i,000-1,004 995-1,004 29 I 0 4.85
P77 A 27.9-28.8 1,185 1,200 – 5 0 5.20
P77 A 27.9 28.8 1,185-1,200 – 5 10 4.70
P78 A 29.8 30.3 1,150-1.151 1,153-1,156 3 0 3.40

Table 3 (continued)
Run # Pressure P Tw- Re Tcr- A~ Run duration Initial Final
media (Kbar) (~ C) (~ C) (h) wt% A1203 wt% A1203 103
P78 A 29.8-30.3 1,150-i,151 1,153-1,156 3 10 3.20
P79 B 24.8-25.3" 1,293-1,298 – 1 0 9.60
P81 B 24.8-25.2" 1,298 1,301 – 2 12 9.50
P82 B 25.0-25.3" 1,195-1,200 – 5 0 8.00
P83 B 24.8-25.3 a 1,199 1,201 – 5 10 7.30
P84 B 27.6-28.1 a 1,398-1,403 – 3 0 9.60
P85 B 27.4-27.8 a 1,299-1,302 – 3 12 7.80
P91 B 29.9-30.3 ~ 1,400 1,401 – 1 12 7.85
P96 B 29.9-30.4 a 1,395-1,405 – 3 5 8.28
P92 B 27.6-28.1 a 1,393-1,407 – 2 12 9.55
P94 B 27.3-27.7 a 1,290-1,298 – 2 5 8.25
P95 A 20.3-21.1 1,088 1,093 1,085 1,088 2 0 7.45
P95 A 20.3-21.1 1,088 1,093 1,085-1,088 2 10 7.00
P97 C 39.6-41.0 1,283-1,308 – 1 5 2.86
P100 C 39.6-40.4 1,296 1,305 – 3 0 3.29
P98 C 36.4-36.9 1,297-1,309 – 2 5 3.68
P101 C 36.6-37.0 1,298-1,305 – 2 0 4.11
P99 C 32.6 33.4 1,295-1,306 – 3 5 4.50
P102 C 32.8-33.4 1,297 1,304 – 1 0 4.95
E4 D 33.0-33.8 a 1,495-1,505 – 5 0 8.85
E5 D 32.8-33.6 a 1,495-1,505 – 8 12 9.65
E6 D 32.8-33.6 a 1,595-1,605 – 2 0 10.80
E7 D 32.8-33.6 a 1,595 1,605 – 3 12 ?b
E8 D 41.2-42.0 a 1,495-1,505 – 3 12 6.05
E9 D 41.242.0" 1,495 1,505 – 3 0 5.75
El0 D 41.642.4" 1,595 1,605 – 2 12 8.32
E11 D 41.6~2.4 a 1,595-1,605 – 3 0 7.78
a The reported pressures are nominal (gauge) pressures. See text for details of friction corrections
b No apparent change
was added. At temperatures of 1,150~ C and above, hydrous
melting occurred, and 10 wt.% oxalic acid was added instead,
to lower H20 activity. At 1,500 ~ C and above, extensive melting
occurred with oxalic acid, and 3 5 wt.% PbO, intimately mixed
with the charge, was used, in graphite rather then Pt containers.
Analysis of the Run Products. For runs to determine A1203 con-
tent of enstatite, the quenched charges were examined optically
and by X-ray diffraction. Charges run with HzO were white
and sugary, with crystals of garnet and pyroxene up to 50 btm
in longest dimension. Charges run with oxalic acid were gray
and coarsely crystalline (30 gin) mixes of garnet and pyroxene,
and, occasionally, a small amount of glass. Charges with PbO
were dark hard pellets consisting of 10-15 btm pyroxene and
garnet crystals separated by thin coatings of PbO-rich glass.
Several hydrothermal samples were examined by 1/8~ per
rain X-ray diffractometer scans with an annealed corundum in-
ternal standard. The degree of equilibration of some of the high-
est pressure hydrothermal runs was sufficient to give an indica-
tion of the enstatite composition from the Boyd and England
(1960) 20-difference method (62-63 ~ peaks). The agreement with
the microprobe analyses was usually good. In all runs, the pres-
ence of garnet and enstatite was verified by fast-scan X-ray
diffraction. The reversal runs for reaction (3) were scanned at
1/2~ per minute in the range 20~176 and the X-ray peak
heights of the phases were compared with those in the diffracto-
gram of the starting mix. Clear indication of the direction of
reaction was usually possible by this method.
Charges for reaction (1) were mounted in epoxy resin, gener-
ally as loose grains or aggregates, for microprobe analysis. The
PbO runs were mounted as cylindrical slugs. The surfaces of the probe mounts were carefully polished to avoid excessive
erosion of the grains. The polished PbO-run slugs were etched
for 30 min in 60% fluoroboric acid to define the crystal shapes
by preferential dissolution of the surrounding glass.
Energy dispersive microprobe analyses were obtained with
an ARL-EMX electron microprobe. All analyses were made
at 15 kv and 0.1 ~tamp beam current with a beam diameter of
about 1 gin. The standards used, synthetic pyrope glass and
several MgSiO3-AlaO3 glasses, were chemically very close to
the unknowns, so that fluorescence and absorption corrections
were negligible. At least 20 grains of orthopyroxene were ana-
lyzed in each sample, and several analyses were made on each
grain. An analysis was rejected if it did not total 100_+ 2%.
Results of Experiments
Enstatite Al203 Contents. All charges contained enstatite plus
garnet, assuring continuous A1203 buffering throughout the
runs, except for one charge, that of run E6 (Table 3). In this
run the pyrope was entirely consumed to make the high-AlzO3
pyroxene from pure MgSiO3. Therefore, a true reversal bracket
was not obtained with this run, but only a lower limit for AlzO3
content.
Most grains of orthopyroxene in the run charges were zoned.
The cores sometimes retained the original starting compositions,
with interiors grading in A1203 content, sometimes to broad,
nearly equilibrated rims. In nearly all hydrothermal and oxalic
acid runs, there was a small overlap of extreme A1203 analyses
coming from opposite directions. Some of this overlap results
from analytic scatter and pressure and temperature variations

104
Si0z P 32
)3/~ /~ ~ 24Kb MgSi03
–,.. / \
0 5 I0
% AI203 A P 99/102
/ k – 33TU-
/ ~k 13) 0~
.- ,oA. ~
/ o,o<
0 5 I0
% AI203
/ \'7 ~
0 5 I0
% Alz03 ~. P 95
/- \ -~i ~-/
/ \,o?c
40,2 Leo Ooo o / o~ 0 \
/ -~ o-o':., t..~ ~.?
[ v v v v V v v v v V v \ \
0 5 I0
% AI203
Fig. 1. a Analytical results for experiment P32 projected on the MgO-A1203-SiOz ternary. In this and the following three figures the
open circles represent pyroxenes which started on MgSiO3 composition and the closed circles represent pyroxenes originally oversaturated
in A1203. No apparent crossover of AlaO3 contents as equilibrium was approached from two directions, b Analytical results for experiments
P99 and P102. Note the tight cluster and minor crossover of analyses9 c Analytical results for experiments P84 and P92. Note apparent
excess SiO2 in pyroxene, d Analytical results for experiment P95 showing well equilibrated compositions
40
..0 30
13.
20
10 ' ' 1' ' 2' 3 '4 "5/
/ .1.6 /9 / / / "~
15 t./2.8 -~.~3,3~.%4a / 9 14 ell 9 9 9 9 9 9 9 4.3
9 / 2,3 2/, ~ //~
/ i / / /~176 '~ / .^ / 4.6/9 9 / ~95~"~/ /
/ / 9 1/ "~ 9 07.6..-09.5
PY F.~:i,~':~'-/ / // / /
"~EnSP/~ / / / / / -z-n/P/ / / / / / /
5 6 7 8 9 10 11 12 13
/ .z / /.. /, /, //, /, /, /,
900 1100 1300 1500 1700
T ~ Fig. 2. Experimental results and "best-fit" curves for reaction (1).
Solid circles represent experiments conducted in NaC1 pressure media;
open circles represent experiments with an associated friction correc-
tion. Points are plotted at mean pressures and temperatures. Mean
% AlzO3 determinations are shown. See Table 3 for ranges. The solid
line for the reaction Py+Fo=En+Sp and the isopleths in the spinel
field are calculated from the thermochemical value discussed in text during the runs, but some is undoubtedly because of "path
looping"; that is, overstepping of the equilibrium composition
and reapproach from the opposite direction. In general, cross-
overs were more pronounced at lower temperatures and pressures
(Fig. 1 a) than at higher temperatures and pressures, where tight
clusters of analyses with only slight overlap indicated nearly
complete reequilibration of the charges (Fig. 1 b). The region
of overlap was usually small, and the equilibrium compositions
were taken at the centers of crossover bands. The problem of
overlap due to path looping arose in the hydrothermal runs
on aluminous pyroxenes of Lane and Ganguly (1980) and Perkins
and Newton (1981). The present runs with PbO did not show
crossover; therefore it is presumably a capricious effect of hydro-
thermal nucleation and growth. In some runs, especially at higher
temperatures and pressures, departures from the MgSiO3-
A1103 join in the direction of excess SiO2 were observed
(Fig. 1 c), compared to runs of similar A1203 content at lower
temperatures and pressures (Fig. 1 d). Boyd and England (1960)
also detected a probable excess of Si/Mg ratio over the ideal
value of unity by the X-ray diffraction characteristics of some
of their aluminous enstatites synthesized at very high tempera-
tures and pressures.
Mean values of Al~O3 contents in weight percent are shown
in Fig. 2. The isopleth lines in Fig. 2 were derived from a statisti-

Table 4. Experimental runs defining the reaction anstatite + spineol = pyrope + forsterite
Run r Pressure P T Run duration Starting opx Results
media (Kbar) (+ 5 ~ C) (h) (wt% A1203) 105
PF1 B 23.9~4.0 a 1,315 7 10
PF2 B 21.8-22.1 a 1,260 7 10
PF3 B 28.8-29.3 a 1,465 2 10
PF4 B 28.0-28.3 a 1,435 2 10
PF5 B 26.0 26.1 a 1,375 8 10
PF6 B 26.8 27.2 a 1,405 8 10
PF7 B 24.254.6 a 1,250 4 8a/2
PF8 B 24.8-25.1" 1,270 6 81/z
PF9 B 23.8-24.2" 1,240 5 81/z
PFI0 B 21.4-21.6 a 1,100 4 7
PF11 B 21.8-22.2 a 1,100 5 7
PF12 B 25.0-25.3 ~ 1,260 4 81/2
PF13 B 22.3-22.5 ~ 1,110 4 7
PF14 B 22.7 22.9 a 1,120 3 7
PF15 B 23.0-23.3 a 1,135 3 7
PFI6 A 20.3-20.7 1,100 7 7
PFI7 B 27.4-27.9" 1,425 2 10
PF18 A 20.0-20.0 1,085 5 7
PF19 B 23.6 23.6 a 1,145 3 7 moderate reaction to En + Sp
complete reaction to En + Sp
complete reaction to Py + Fo
complete reaction to Py + Fo
moderate reaction to En + Sp
strong reaction to En + Sp
no apparent reaction
weak reaction to Py + Fo
strong reaction to En + Sp
complete reaction to En+ Sp
complete reaction to En + Sp
complete reaction to Py + Fo
strong reaction to En + Sp
no apparent reaction
no apparent reaction
complete reaction to Py + Fo
strong reaction to Py + Fo
strong reaction to En + Sp
strong reaction to Py + Fo
The reported pressures are nominal (gauge) pressures for pressure media B. See text for details
cal and thermodynamic treatment of all of the experimental
points, as explained in the next section. Complete analytic data
are given in Table 3.
Univariant Reaction. Table 4 presents the experimental observa-
tions bearing on reaction (3) and Fig. 3 shows the reversal runs
and the earlier brackets of Danckwerth and Newton (1978).
Their 1,100 ~ C bracket in the NaC1 medium was repeated in
the present study with identical results. The NaCl-pyrex-BN me-
dium runs just above 1,100 ~ C show that the pressure correction
of -8% to –10% is appropriate, at least in this pressure-
temperature range. The solid curve of Fig. 3 is a statistical and
thermodynamic fit to the nine reversal brackets, as explained
in the following section.
Thermodynamic Treatment of Results
Garnet-Field Al203 Isopleths of Enstatite. Thermodynamic analy-
sis of the data proceeds from the equilibrium relation:
-AG~ P)=RTIn K (4)
where A G O is the Oibbs energy change for reaction (1) or (1 a)
at a fixed temperature and pressure. K, a constant at fixed T
X ~ for the ideal opy theory or and P, is evaluated as Mg3A125i3012
X ~ X ~ for the ideal MgTs theory, where X de- MgAI2SiO6 ' Mg2Si206
notes mole fraction. AG~ P) is evaluated as:
AG~ P)=AH~ 1)-970 AS~ 1)
T
– y AS~ 1)dT+(P-1)AV ~ (5)
970
A reference temperature of 970 K was chosen because the en-
thalpy of solution measurements on pyrope and enstatite to
evaluate AH ~ the standard enthalpy change, were made at that
temperature (Charlu et al. 1975). The approximation of constant
A V ~ and variable-pressure standard state (Lane and Ganguly
1980) are chosen for simplicity. Low-temperature and high-tem-
perature heat capacity data are available for pryope and enstatite 30
e~
0. 20 ~ _ Sp En
15 ~ , ~ , , , ,
800 1000 1200 1400 1600
T ~
Fig. 3. Experimental results and "best-fit" curve for reaction (3). I-
beam brackets are from Danckwerth and Newton (1978); circles are
the present results using NaC1 pressure media and rectangles are the
present results using pyrex pressure media. Solid symbols, half filled
symbols and open symbols represent growth of low-pressure assemblage,
no reaction, and growth of high pressure assemblage, respectively.
The fine dotted line is the curve of MacGregor (1974); dashed line
is from Haselton (1979)
(Haselton and Westrum 1980; Krupka etal. 1979; Haselton
1979) to evaluate the standard entropy change, AS~ The
heat capacity ofopy was estimated as the sum of the heat capaci-
ties of 3MgSiO3+A1203 (corundum), the latter heat capacity
taken from Robie et al. (1978), and the heat capacity of MgTs
was similarly evaluated. The heat capacity difference, ACp, is
nearly constant in the range 1,000-1,500 K, and was assumed
constant for extrapolation above 1,500 K to get AS~ from
the formula :
AS~176 + i A Cp dT. (6)
970 T

106
Table 5
Reaction AH~gvo (cal/mol) AS~ (cal/mol-deg) AV ~ (cc/mol)
experimental thermochemical experimental thermochernical experimental volumetric
(1) Mg3AlzSi3012 (Gt) 2,040-+170 3,070-+1,900 2.12_+0.08
~ MgaA12Si3012 (Opx)
(la) Mg3A12Si3012 (Gt) 1,640+150 3,070_+1,900 -0.15_+0.i4
= MgA12SiO6 (Opx)
+Mg2Si/O6 (Opx)
(3) Mg3A12Si3012 (Gt) -3,200_+250 -5,510_+460 -0.5+_0.5
+ MggSiO4 (Fo)
= Mg~Si4012 (Opx)
+ MgAlaO4 (Sp)
(3a) Mg3A12Si3012 (Gt) -3,300+400 -5,510_+460 -0.5-+1.5
+ MggSiO4 (Fo)
=2 Mg2Si206 (Opx)
+ MgA1204 (Sp) 9.55_+0.24
8.65 -+ 0.29 9.42
-0.55 7.10_+2 7.73
-0.55 10 _+5 7.73
Equations (4), (5) and (6) were combined to derive best-fit
values of AI~97o, AS~ and A V ~ (Table 5) for both the ideal
opy and MgTs theories. A Monte Carlo method was used, where-
in many combinations of AH~ ' AS~ and A V ~ each varying
independently by closely-spaced increments, were tried. Each
combination resulted in an isopleth set. The sum of the squared
deviations of the experimental A1203 values from the corre-
sponding isopleths were tallied for each combination. The best
fit was the combination with the smallest residual. The opy
and MgTs theories gave equally good fits to the experimental
data. With the exception of 8 experimental points the "best fit"
models are completely consistent with the experimental determi-
nations. For both models, the maximum deviation of the experi-
mental determinations from the smoothed isopleths was about
150 cal/mole (experiments # P98/101 and P67). The standard
deviation of the points from the best-fit isopleths, expressed
as a Gibbs energy difference, is 50 cal for the opy theory and
42 cal for the MgTs theory. Table 5 gives the best fit parameters
AH ~ AS ~ and A V ~ for both models, compared to the results
of calorimetric and volume measurements. Note that the errors
given for these values (Table 5) are two standard deviations from
the mean and are not independent.
The thermochemical AH~97o for reaction (1) of Table 5 de-
pends on a linear extrapolation of the enthalpy of solution of
a (MgSiO3)0.9(A1203)o.x enstatite (Charlu et al. 1975) and has
large uncertainty. The experimental best-fit A//~t is much more
precise, and defines the standard enthalpy of formation at 970 K
from the oxides of the Mg3A12Si3012 pyroxene (opy) as
-18.17 _+ 0.42 kcal, using the value for pyrope of -20.21_+
0.38 kcal (Charlu et al. 1975). Using the entropy of pyrope at
970 K of 182.20 cal/K (Haselton, 1979) the standard entropy
ofopy at the same temperature is 184.32 cal/K. This is 4.59 cal/K
greater then 3Sugsio3+ S~u2o3 and is presumably due to disorder-
ing of A1 on the octahedral and tetrahedral sites. The statistical
mechanical calculations of Ganguly and Ghose (1979) lead to
a maximum disorder entropy of 7.0 cal/K, and to a minimum
disorder entropy of 4.3 cal/K corresponding to the "Al-avoid-
ance" possibility, in which adjacent corner-linked tetrahedra do
not both contain A1. The present findings support the Al-avoid-
ance model. The standard Gibbs energy of formation from the
oxides of opy at 970 K is thus -20.54 kcal.
The experimental volume change of reaction (1) may be com-
pared with X-ray diffraction measurements on aluminous ensta-
tire. A molar volume versus composition curve was constructed from volume data of three synthetic aluminous enstatites of
Danckwerth and Newton (1978) and of five aluminous enstatites
synthesized by Chatterjee and Schreyer (1972). These authors
estimated the compositions of their pyroxenes from the b-axis
plot of Skinner and Boyd (1964). The resulting molar volume
relation is:
V(cm 3) = 125.26 – 4.130X opt + 1.418X zopy. (7)
The molar volume of opy is 122.55 cm 3, and AV ~ of reaction
(1) is 9.42 cm 3, which is in good agreement with the best-fit
volume parameter of 9.55 _+0.24 cm 3.
The MgTs theory gives as good a fit to the isopleth data
as the opy theory in most respects, except that the molar volume
of MgTs cannot be estimated accurately by extrapolation of
the aluminous enstatite volume data. The best-fit volume is
8.65 cm 3, and there is no independent way to tell if this is a
physically meaningful value or simply a data-fitting parameter.
Univariant Curve. A Monte Carlo fit for AH~97o, AS~ and
A V ~ was made to the five experimental brackets of Danckwerth
and Newton (1978) and the four of the present study. The result-
ing parameters from the opy and MgTs theories are given in
Table 5. The uncertainties given in Table 5 are the total range
of H, S and V values consistent with the experiments and are
highly correlated.
The best fit value of AH~970 for reaction (3) is not in agreement
with the enthalpy of solution data of Charlu et al. (1975). More-
over, the solution data are not consistent with the location of
the univariant curve of reaction (3). Using the accurately mea-
sured entropy data, giving AS~ cal/K, the enthalpy
change from calorimetry of – 5,510_+ 460 cal, and the volume
change of 7.73 cm 3, one gets a pressure of 27.0_+2.5 kbar at
1,000 K, compared to the 17.5 kbar of Fig. 3. Therefore, it is
certain that some of the heat of solution values are incorrect.
Since the MgSiO3 heat of solution is multiplied four times in
the calculation, an error of 500 calories in this quantity is suffi-
cient to account for the discrepancy. An enthalpy of formation
of MgSiO3 of -8.32 kcal, rather than the -8.81 measured by
Charlu et al. (1975) would give a AH ~ (3) in agreement with
the experimental value and a pressure for the univariant equilibri-
um with Al-free pyroxene of 15.5 kbar at 970 K. The additional
opy-theory stabilization of enstatite by 3.9 wt.% A1203 brings
the pressure to 17.5 kbar, in agreement with the experimental
curve. The value of -8.32kcal compares well with that of

107
-8.39 kcal which results from the AG o of Zen and Chernosky
(1976) deduced from phase equilibria and the AS ~ from the
recent heat capacity work.
A large range of A V allows agreement with the experimental
brackets in both theories. This parameter is much more loosely
constrained than for reaction (1).
Spinel-Field Al203 Isopleths of Enstatite. Elimination of garnet
between Eq. (1) and (3) yields Eq. (2), which governs the spinel-
field isopleths, and the opy-theory parameters of AH~g70 =
5,240 cal, AS~ cal/K, and AV~ cm 3. The slope of
an isopleth Xopy=Constant is given by:
A S~ (2) -R In X~
o A V ~ (2) (8)
The isopleths are shown in Fig. 2. It is to be emphasized that
they are determined by the very oblique intersections of the
garnet-field isoplaths with the univariant curve. Slight shifts in
the positions of either could cause large shifts in the spinel-field
isopleths. Also, the volume change is only loosely constrained
by the best-fit volumes for reactions (1) and (3). Therefore, the
spinel-field isopleths of Fig. 2 have greater uncertainties than
those of the garnet field.
Comparison with Previous Work
Garnet-Field Isopleths. Figure 4 shows the present smoothed
isopleth set and some of the experimental data of previous
workers. All points claiming to be reversals are shown except
that of Howetls and O'Hara (1978) at 30 kbar and 1,500 ~ C.
Also shown are the unreversed data of MacGregor (1974).
The three reversals of Boyd and England (1964) show much
more A1203 than indicated by our isolpleths. This may be due
to the fact that they depended on the exsolution of pyrope from
A1203 saturated pyroxene, which may be hindered by the diffi-
cult nucleation of garnet. The two hydrothermal reversals of
Lane and Ganguly (1980) in the range 1,000~ ~ C and 25-
30 kbar are in excellent agreement with our isopleths. This is
to be expected inasmuch as their runs were carried out with
procedures identical to ours (hydrothermal reversals in the NaC1
medium). Agreement with their highest temperature runs is not
as good. These runs were unfluxed and, in view of the difficulties
encountered in this study of attaining equilibrium, it is suggested
that equilibrium may not have been attained in these three runs
or that the finely sintered nature of the run products led to
inadvertent inclusion of garnet in the spots analyzed. The one
hydrothermal reversal of Wood (1978) agrees quite well with
our curves. The greatest discrepancy is with Howells and O'Hara
(1978), who found a value of about 6 wt.% A1203 at 30 kbar
and 1,500 ~ C. The present data indicate about 9% at the same
conditions. The MacGregor data points are not in bad agreement
with the present ones but diverge to higher A1203 values at
the higher temperatures.
The Wood (1975) and Obata (1976) isopleths were generated
by modelling the MacGregor (1974) data with the ideal MgTs
theory. These sets show somewhat greater A1203 at fixed T
and P over most of the field than do the present set.
Univariant Reaction. Figure 3 shows the experimental curve of
MacGregor, to which a -10% pressure correction has been
applied, and the experimental curve of Haselton (1979). The
latter reversed study was done by much the same methods as 1000 1200 1400
T ~
Fig. 4. Comparison of the present A1203 isopleths for enstatite in
the garnet field with other data. Bold symbols represent reversed
determinations. Squares, circles, triangles and hexagons are the results
of MacGregor (1974), Lane and Ganguly (1980), Wood (1974) and
Boyd and England (1964)
the present. There is good agreement of the three studies. A
single reversal bracket of Staudigel and Schreyer (1977) is two
kbar higher than our curve at ~950 ~ C. The calculated curve
of Obata (1976) is generally in good agreement with the experi-
mental ones.
Spinel-Field Isopleths. The calculated isopleths agree very well
with the 20 kbar reversed points of Danckwerth and Newton:
7.2 wt.% A1203 at 1,080 ~ C, 6.2 wt.% at 1,000 ~ C, and 5.6 wt.%
at 950 ~ C. The latter two points were metastable reversals a
short distance into the garnet field, which was possible because
of the difficulty of spontaneous nucleation of garnet near its
stability limit (Boyd and England 1960). These points may be
compared to the present spinel-field isopleths by short extrapola-
tions past the univariant curve in Fig. 2. At 1,200 ~ C the unre-
versed syntheses of Fujii agree well with our deduced isopleths.
At 1,300~ C the Fujii A1203 contents are slightly greater, and
at 1,400 ~ C the discrepancy is about 1 wt.%.
Among calculated spinel-field isopleth sets, the Wood (1975),
Obata (1976), and Danckwerth and Newton (1978) sets are all
similar to the present ones. The latter set diverges to slightly
higher A1203 contents in the high temperature range. All of
these show steep positive dP/dT slopes, which agree with the
present study. The calculated curves of Stroh (1976), which mo-
deled MacGregor's (1974) spinel-field data, have flatter dP/dT
slopes and do not seem to be in accord with the measured
small volume change of the isopleth reaction (reaction 2).
Potential for Geobarometry-Geothermometry
The three-component garnet-field isopleths of MacGregor (1974)
have been the basis of most recent geobarometers for garnet
peridotites, following the demonstration of Boyd (1973) of the
power of this method. The A1203 contents of peridotite pyrox-
enes are usually very low, so that small analytical error and
differences among model isopleth sets make large differences
in calculated pressures. An additional serious problem is the
combined effect of other garnet and pyroxene components, main-

108
ly Ca and Fe. Akella (1976) and Perkins and Newton (1981)
showed that Ca in the garnet lowers the A1203 content of coex-
isting orthopyroxene over that at a given T and P in the three-
component system. Wood (1974) studied garnet-pyroxene equi-
libria in the FeO-MgO-AlzO 3-SIO2 system and concluded
that Fe lowers the A1 content of orthopyroxene over the three-
component system. No reversals of the A1203 contents were
made. Quantitative garnet-pyroxene geobarometry awaits accu-
rate reversal experimental evaluation of the effects of Ca and
Fe.
Several authors (e.g. Ferguson et al. 1977) have concluded
from the coexistence of garnet and spinel in exotic peridotite
nodules in certain volcanic pipes that the assemblage equilibrated
near the garnet-spinel peridotite boundary, assumed to be near-
univariant. However, Jenkins and Newton (1979) showed that
the combined effect of Ca and Fe on this transition amount
to a pressure lowering of at least 6 kbar for the coexistence
of garnet and olivine at 1,000 ~ C. It is probable that spinel
can coexist with garnet in peridotites over many kbar, especially
when stabilized by high Cr content.
Most workers now agree that the spinel-field AlzO3 isopleths
of enstatite in peridotite have little potential as a geobarometer
but may be useful as a geothermometer for spinel lherzolites.
Obata (1980) used this thermometer for the Ronda, Spain, peri-
dotite body, basing temperatures on his own (1976) spinel-field
isopleth set. The present isopleths, those of Wood (1975) and
Obata (1976) all give uncorrected temperatures within about
80~ of each other. The role of non-ternary ions is crucial
however, especially the large and probably non-ideal effect of
Cr in spinel (Evans and Frost 1975), which results in higher
temperature indications for spinel lherzolites than given by the
uncorrected ternary isopleths. The experimental finding that ad-
dition of Ca results in relatively flat dP/dT slopes of the orthopy-
roxene AI isopleths in the spinel peridotite field (Dixon and
Presnall 1980), which isopleths may therefore be useful for geo-
barometry, has no obvious explanation in terms of simple volume
and entropy considerations and needs additional experimental
verification.
Acknowledgements. This research was supported by a National Science
Foundation grant, # EAR 78-15939 (RCN). Some of the equipment
and materials used were provided by funds from the Materials Re-
search Laboratory (NSF) at Chicago. W. Moloznik kept the high
pressure apparatus in good repair. Cassandra Spooner typed the several
versions of this manuscript.
A gas apparatus determination of the pressure of the reaction
of Mg-cordierite to sapphirine plus quartz at 1,250 ~ and 1,280~
was made possible through the courtesy of G. Lofgren and O. Mullins
at the NASA (Houston) experimental petrology laboratory and of
A. Koster Van Groos at the University of Illinois, Chicago Circle
campus. This information was valuable in solid-medium pressure cali-
bration.
References
Akella J (1976) Garnet pyroxene equilibria in the system CaSiOs-
MgSiO3-A1203 and in a natural mineral mixture. Am Mineral
61:589-598
Anastasiou P, Seifert F (1972) Solid solubility of A1203 in enstatite
at high temperatures and 1-5 kb water pressure. Contrib Mineral
Petrol 34: 272-287
Arima M, Onuma K (1977) The solubility of alumina in enstatite
and the phase equilibria in the join MgSiO3-MgAlzSiO6 at 10-
25 kbar. Contrib Mineral Petrol 61:219-230
Bohlen SR, Essene EJ, Boettcher AL (1980) Reinvestigation and appli- cation of olivine-quartz-orthopyroxene barometry. Earth Planet Sci
Lett47:l 10
Boyd FR (1973) A pyroxene geotherm. Geochim Cosmochim Acta
37 : 2533-2546
Boyd FR, England JL (1959) Pyrope. Ann Rept of the Dir of the
Geophys Lab, Carnegie Inst Washington Yearb 58:83-87
Boyd FR, England JL (1960) Minerals of the mantle. Carnegie Inst
Washington Yearb 59:47 52
Boyd FR, England JL (1964) The system enstatite-pyrope. Carnegie
Inst Washington Yearb 63:157-161
Charlu TV, Newton RC, Kleppa OJ (1975) Enthalpies of solution
at 970 K of compounds in the system MgO-A1203-SiO2 by high
temperature solution calorimetry. Geochim Cosmochim Acta
39:1487-1497
Chatterjee ND, Schreyer W (1972) The reaction eutstatite+silliman-
ite=sapphirine+quartz in the system MgO-AlzOa-SiOz. Con-
trib Mineral Petrol 36:49-62
Danckwerth PA, Newton RC (1978) Experimental determination of
the spinel peridotite to garnet peridotite reaction in the system
MgO-AlzO3-SiO2 in the range 900~176 C and A1203 iso-
pleths of enstatite in the spinel field. Contrib Mineral Petrol
66 : 18%201
Dixon JR, Presnall DC (1980) A1203 content of enstatite: a spinel
lherzolite barometer. Geol Soc Amer Prog W Abstr, 1980 Ann
Meeting in Atlanta, 414
Evans BW, Frost BR (1975) Chrome-spinel in progressive metamor-
phism – a preliminary analysis. Geochim Cosmochim Acta 39 : 959-
972
Fawcett JJ, Yoder HS (i963) The liquidus region at 10 kilobars PH2o.
Carnegie Inst Washington Yearb 62:143-145
Ferguson J, Ellis DJ, England RN (1977) Unique spinel-garnet lherzo-
lite inclusion in kimberlite from Australia. Geology 5:278-280
Fujii T (1977) Pyroxene equilibria in spinel lherzolite. Carnegie Inst
Washington Yearb 76:569-572
Ganguly J, Ghose S (1979) Aluminous orthopyroxene: order-disorder,
thermodynamic properties, and petrologic implications. Contrib
Mineral Petrol 69:375 385
Haselton HT (1979) Calorimetry of synthetic pyrope-grossular garnets
and calculated stability relations. Ph D thesis, Dept of Geophy
Sci, Univ of Chicago, 98 pp
Haselton HT, Westrum EF (1980) Low-temperature heat capacities
of synthetic pyrope, grossular, and pyrope60grossular40. Geochim
Cosmochim Acta 44: 701-709
Hays JF, Bell PM (1973) Albite-jadeite-quartz equilibrium : a hydrosta-
tic determination. Carnegie Inst Washington Yearb 72:706 708
Holland TJB (1980) The reaction albite=jadeite+quartz determined
experimentally in the range 600-1,200 ~ C. Am Mineral 65:129-134
Howells S, O'Hara MJ (1978) Low solubility of alumina in enstatite
and uncertainties in estimated paleogeotherms. Philos Trans Roy
Soc London 288 (Ser A), 471-486
Jenkins DM, Newton RC (1979) Experimental determination of the
spinel peridotite to garnet peridotite inversion at 900~ and
1,000 ~ C in the system CaO-MgO-A120 3-SIO2, and at 900 ~ C
with natural garnet and olivine. Contrib Mineral Petrol 68:40%419
Johannes W (1973) A simplified piston-cylinder apparatus of high
precision. N Jb Mineral Monats 8:337-351
Krupka KM, Kerrick DM, Robie RA (1979) Heat capacities of syn-
thetic orthoenstatite and natural anthopyllite from 5 to 1,000 K.
Eos 60:305
Lane DL, Ganguly J (1980) A1203 solubility in orthopyroxene in
the system MgO-A1303-SiO2: a reevaluation and mantle geo-
therm. J Geophys Res 85:6963~5972
MacGregor ID (1974) The system MgO-A1203-SiO2: Solubility of
A1203 in enstatite for spinel and garnet peridotite compositions.
Am Mineral 59:110 119
Mirwald PW, Getting IC, Kennedy GC (1975) Low-friction cell for
piston-cylinder high-pressure apparatus. J Geophys Res 80:1519-
1525
Newton RC (1972) An experimental determination of the high pressure
stability limits of magnesian cordierite under wet and dry condi-
tions. J Geol 80:398420

109
Newton RC, C]harlu TV, Kleppa OJ (1974) A calorimetric investigation
of the stability of anhydrous magnesium cordierite with application
to granulite facies metmorphism. Contrib Mineral Petrol 44:295
311
Obata M (1976) The solubility of A1203 in orthopyroxenes in spinel
and plagioclase peridotites and spinel pyroxenite. Am Mineral
61:804-816
Obata M (1980) The Ronda peridotite: garnet-, spinel-, and plagioclase
lherzolite facies and the ~T trajectories of a high-temperature
mantle intrusion. J Petrol 21 : 533 572
Perkins D, Newton RC (1981) The compositions of coexisting pyrox-
enes and garnet in the system CaO-MgO-A1203- SiO2 at 900 ~
1,100 ~ C and high pressures. Contrib Mineral Petrol 75:291-300
Robie RA, Hemingway BS, Fisher JR (1978) Thermodynamic proper-
ties of minerals and related substances at 298.15 ~ K and 1 bar
(105 pascals) pressure and at higher temperatures. US Geol Surv
Bull 1452:456 pp
Skinner B J, Boyd FR (1964) Aluminous enstatites. Carnegie Inst Wash-
ington Yearb 63:163-165
Staudigel H, Schreyer W (1977) The upper thermal stability of clino- chlore, MgsAI[A1Si30~0](OH)8, at 10-35 kb PH~o. Contrib Mineral
Petrol 61 : 187-198
Stroh JM (1976) Solubility of alumina in orthopyroxene plus spinel
as a geobarometer in complex systems : applications to spinel-bear-
ing alpine-type peridotites. Contrib Mineral Petrol 54:173 188
Wood BJ (1974) The solubility of alumina in orthopyroxene coexisting
with garnet. Contrib Mineral Petrol 46:1-15
Wood BJ (1975) The application of thermodynamics to some subsoli-
dus equilibria involving solid solutions. Fortschr Mineral 52:2145
Wood BJ (1978) The influence of Cr203 on the relationships between
spinel- and garnet-peridotites. Prog Exptl Petrol (NERC, Gt Bri-
tain) 4: 78-80
Wood B J, Banno S (1973) Garnet-orthopyroxene and orthopyroxene-
clinopyroxene relationships in simple and complex systems. Contrib
Mineral Petrol 42:109 124
Zen E-A, Chernosky JV (1976) Correlated free energy values of antho-
phyllite, brucite, clinocrysotile, enstatite, forsterite, quartz, and talc.
Am Mineral 6I:1156-1166
Received March 31, 1981 ; Accepted in revised form July i6, 1981