Chapter 1: Chapter 1: Introduction
L. L Introduction
L.3 Calculated phase relations in amphibolites
2.5 Uncertainties in the calculated P-T projection
The internally consistent nature of the data used to calculate the phase diag¡am (Fig.
2.3) allows some confidence in the predictions of these diagrams (especially in the relative positions of the stability fields and the slopes of the univariant reactions which separate them).
However, there are
still
substantial uncertaintiesin
the data themselves and these may have asignificant effect on rhe loci of the equilibria in P-T space. For example, the slopes of most
of
the reactions involved in the mid-pressure, mid-temperature invariant points,
[Grt]
and[Chl]'
are
fairly
similar and their sub-parallelnatue
means that the positionsof [Grt]
and [Chl], marked by the intersection of these univariants,will
not bewell
defined. However, bothof
these bundles of reactions also involve the near-perpendicular reaction;
orthoamphibole + sillimanite
=
staurolite +cordierite (Grt'
Chl)'The high angle at which this reaction intersects the other reactions means that
it
effectively controls the positionof
both of the inva¡iantpoints. It follows
that any error in the positionof
the reaction (Grt, Chl) may have a considerable effect on the absolute position of the critical assemblages
in
P-T space.Although many of the data required for equilibrium thermodynamics calculations are relatively
well
characterised from experimentalwork,
standa¡d molal enthalpies of formation are notoriouslydifficult
to determine to an acceptable level of accuracy (llelgeson et al., 1978;Powell
&
Holland. 1985). The ur¡certainties on standard molal enthalpies of formation determined from both molten salt calorimetry andlow
temperature heat by solution inIIF
arehigh and may cause errors
in
the calculated positionof
a reaction of as much as 100"C(Helgeson er al., 1973). For this reason
it
is assumed that the enthalpy data of the mineral end memberswill
be the major contributors to uncertaintiesin
the P-T conditions of the calculated mineral equilibria.The uncertainties on the thermodynamic data used in the calculation of Fig' 2,3 a¡e listed in Appendix
A2.
Although the aluminosilicates, quartz and water vapour all havefairly
insignificant uncertainties on AH¡, and those for the end members clinochlore, aÍnesite,almandineandpyropearealsoquitelow (3.21,2.98,2.78,2.33kJ mol'lrespectively),the
uncertaintiesfor
the enthalpies of formation of the other end-members are comparatively high' The most uncertain dataue
those for Mg-staurolite, Fe-staurolite , gedrite,ferro-anthophylliteChaptcr2-FMASII-
16and anthophytlíte (8.31, 8.17, 6.55, 6.44, 5.58 kJ
mol-l respectively). It
is unfortunate that the reactionconrolling
the position of both[chl]
and [Grt] involves phaseswith
signif,rcant uncertainties. Figure 2.4 illustrates the uncertainryin
the P-T position of the invariant point[chl]
due to altering theñI
of gedrite and bothFe-
and Mg-staurolir¿ wirhin the limits of their uncertainties.Another source of potential problems in the calculated equilibria is in the relative Fe-Mg partitioning exhibited by staurolite and garnet which, in natural rocks, may have overlapping Xps (see section
2.2).
Small changes in the enthalpy of oneof
the staurolite end members (orin
the thermodynamic data or a-X (activiry-composition) relations of garnet) may cause an exaggeration (or, lesslikely,
a reversal) of the differences in the calculated compositionsof
staurolite or garnet.
2.6
^comparison
between obserYed and calculated phase relations in aluminous schists
The positions
of
the invariant points in the calculated P-T projection(Fig'
2'3)correspond quite
well with
the relative positions of the real assemblages in P-T space and wittt theestimatedtemperaturesofequilibrationof
theserocks (seesections2'3&'2'4)' T\e
calculated pressures of the inva¡iant points in Fig. 2.3 also seem to correspond reasonably well
with
the information obtained from the natural rocks, however the presence of additional components islikely
to have a significant effect on this conformity'Most natural aluminous schists contain at least a small proportion of plagioclase' a titanium phase and sometimes a small amount of biotite (e'g'
Harley'
1985;Arnold &
sandiford,
1990). Whilerutile
(or ilmenite), biotite and anorthite are each stabilised by an additional component which is not easily incorporated into other phases, albite is stabilised by the presenceof
sodium which may also be partitioned into the strucn¡re of orthoamphiboles.This means that
for
nearly a// natural examples, the phase relations in aluminous schists are mofe properly describedin
theNFMASH
rather than theFMASH
system'If
the phase diagram is projected from plagioclase (asit
is effectiveiy for the other additional components'K
andTi)
then the orthoamphibole-bearing assemblageswill
be stabilised with respect to the orher phases and those stability fieidswill
be expanded along (down) the [Oam] reactions' Becauseit
isthe
modelsysremFMASH
which is of primary interest here (a system which can later be used to explain the effects of minor additional components such as Na)' the calculatedequilibria
in Fig.2.3 are considered to occur ai excessivelylow
pfessures'of
mostsignificance is the likelihood that
with
the additionof
Na to the phase diagram inFig' 2'3'the
invariant point Ip1 islikely
to move to lower pressures, thus allowing orthoamphiboie to bestable
with
andalusite and thus contradicting the observed phase relations'As mentioned previously, the uncertainties
in
the thermodynamic data used to calculate the P-T projection in Fig.2.3
¿requite considerable, especiallyfor
the mst' fst' ged'fath
andanth
endmembers. Therefore it is possible to adjust theaH
data for these endmember phases without invalidating the results of the calculations. Becauseof
the old and inaccurate natureof
ChaPter2-FMASll-
177
6
(ú
-o l¿
o-
5
600
2.4. The
to 20 kJmol-l
650
T ("C)
700 750
Figure from 0
[Chl] or Ip1, with irrcreasing ÂHr"a 5 kJmol-l (squares).
-15
-14 -12
-10 10
I
-8 6 -64 -4 Ky
2 siil -2
0
20 18
16 14 12
Chapter2-t;MASII-
I8the thermodynamic data
for
rhe gedriteend.member which was generated from the experimenral dataof
Schreyer and Seifert (1969b; see alsoHowell,
1991), altering the enthalpy of gedrite is considered to be the most appropriate adjusunent to make to the dataset'
In order to bener represent the phase relations in theFMASH
model system, the enthalpy o1 gedrite (^HguÐshould be altered by a factor which places Ip1 high up
in
the sillimanite stabilityfield
(Howell, 1991;Xu
et al., 1993). For rocks containing sodium, Ip1 may then move down pressure' along the[oam] FMASH
univariantreaction, without leaving the sillimanite stability freld'2.7FMr1isHgridwithadjustedthermodynamicdata
AsecondP-Tprojection(Fig2.5)hasbeencalculatedfortheFMAsHsystem(with
qJarïzand aqueous vapour in excess)