CFMASH +Qtz,An,H20
3.4 T'esting the calculated phase relations
Although they are internally consistent, the uncertainties
in
the thermodynamiccalcuiations outlined above imply that caution is needed in the application of the phase relations determined from equilibrium thermodynamics calculations to amphibolites. The relevance
of
the calculated phase diagrams can be determined from comparison of the predicted phase relations
with
those determined in natural assemblages. The comparison here comprises three parts: identifying thelikely
stable CFMASH invariant points a¡d univariant reactionsfrom
natural low-variance assemblages; consÍaining the location of the invariants and univariantsin p-T
space; and comparing the calculated phase diagram to those determined by previous authors.3.4.1 CFMASH invariant points and univariant reactions
There are several reported instances of low-variance amphibolites and although these correspond more closely to the
NCFMASH
system, they are closely related to CFMASH andFMASH
sub-systemequilibria.
These assemblages may be used to infer, independent of the calculations, which of the six CFMASH invaria¡rt points arelikely
to be relevant to real rocks i.e. which CFMASH invariant points are stable.(A fuller
account of the relevant occulrences is givenin
Chapter 6.)As shown in the previous chapter, rocks
with low
Ca-,K-
and Na-contents can be approximated by theFMASH
model sysrem. Exampiesof
these reported from the literature include the assemblages (with co-existing quartz):sillimanite/kyanite-staurolite-garnet-orthoamphibole(F{bl,Pl,Crd)
Robinson
&
Jaffe (1969b), sillimanite/kyanite-staurolite-<ordierite--orthoamphibole (Hbl, Pl, Grt)Schumacher
&
Robinson (1987)' Chapter3-CFMASII- 34sillimani te-staurol ite-cordieri
te-garnet
(Hbl, Pl, O am)Sharma
&
MacRae (1981), sillimanite-cordierite-game t--orthoamphibole (Hbl, Pl, S ÐSharma
&
MacRae (1981).These univariant assemblages intersect
in
anFMASH
invariant assemblage involving aluminosilicate, staurolite, cordierite, garnet and orthoamphibolewith
quartz, which suggests that the[tlbl, Pl]
inva¡iantpoint
is stable in theFMASH
system.In
Ca-rich rocks hornblende and plagioclase a¡e commonly equilibrium phases.Examples of four-phase assemblages
involving
hornblende include (with plagioclase and quartz):staurolite-g arnet-orth o amphib ole-hornb
lende
(Crd, Al s)(Spear, 1971
;
1982; ChaPter 4) kyanite-staurolite-garnet-hornblende (Crd,Oam)(Purtscheller
&
Mogessie, 1984; Selverstone et ai., 1984; Chapters 4 &.5), kyanite-staurolite--ortho amphib ole-hornblende (Crd, Grt)(Chapter 4),
sillimanite-staurolite-cordierite-hornblende (Grt, Oam)
(Schumacher
&
Robinson, 1987), kyanite--cordierite--orthoamphibole-hornblende (Grt, S t)(Chapter 4) and cordierite-garnet--orthoamp hibole-hornblen
de (Als'
S t)(Humphreys, 1993).
Apart from the last, all of these four phase assemblages are consistent
with
the stabilityof
theCFMASH
invariant points[Crd]
andtcrtl.
Thus the phirse relations observed in real rocks suggest that the three invariant points,[Hbl], [Crd]
and[Grt]
are potentially stable (cf.Fig. 3.4).
However, the occurrence of cordierite-garnet--orthoamphibole-hornblende- plagioclase-quartz (Als, St) is inconsistentwith
theCFMASH
grid and seems to indicate that either[Als] or
[St] is stable.3.4.2 P-T constraints
Information from natural amphibolite assemblages and the rocks
with
which they are associated can also providetn
independent constraint on the pressures and temperatures of theCFMASH equilibria.
In particular, the aluminosilicate polymo¡phs associated with individual assemblages might be expected to provide broad constraints on the location of the stabilityChapter3-CFMASII- 35
fields of the relevant aluminosilicate-bea¡ing assemblages and here
I
briefly describe examplesof
these.The calculated phase diagram predics that the assemblages aluminosilicate-homblende and aluminosilicate-orthoamphibole
will
be restricted to elevated pressures in the kyanite and kyanite and sillimanite stability fields, respectivelyFig. 3.5).
This isin
accordwith
mostrecorded assemblages. For example, hornblende is generally observed in equilibrium
with
kyanite rather than sillimanite or andalusite, whereas orthoamphibole istypically
obsewed with either sillimanite or kyanite but only rarelywith
andalusite.As discussed
in
Chapter2,themre
examples of andalusite--orthoamphibole rocks reportedin
theliteratue
(e.g. Seki&
Yamasaki, 1957) probably reflect problems associatedwith very
small scale equilibrium volumes, rather than equitibrium associations, however the discrepancies involving aluminosilicateswith
hornblende a¡e more problematic. There are two instairces of hornblende reported in associationwith
sillimanite, neither of which can be easily dismissed. Schumacher and Robinson (1987) report corroded sillimanite in the coresof
aluminous enclaves separating sillimanite from the hornblende--orthoamphibole-plagioclase
matrix.
They suggest rhar sillimanite-hornblende-gedrite-plagioclase dehned a relatively early equilibrium assemblage which later became metastable with respect to cordierite-plagioclase*
other aluminous phases. Similarly, Humphreys (1993) reports equilibrium assemblages including sillimanite-staurolite-cordierite-hornblende-plagioclase-qua¡tz. Since, there is no indication that either of these authors are in error this discrepancy be¡,veen the calculated phase relations and those obsewed in natural amphibolites probably reflects the relatively large uncertainties attached to the thermodynamic data from which the phase diagram was calculated.
The calculated uncertainties on the reactions bounding the hornblende-aluminosilicate held (Grt, St), (Grt, Oam) are
of
the orderof
150'C and thus allow the possibility that the aluminosilicate-hornblende stability field may extend as far as 100oC into the sillimanite stabiliry held.Further constraints on the relevance of the calculated CFMASH phase diagram to natural amphibolites might be expected from the evolving compatibility relations
of
amphibolites in a prograde metamorphic sequence. The only well-documented example of this type of
field
setting involves four separate outcrops of the Post Pond and AmmonoosucVolcanics (Spear, 1978). However, the compatibiliry diagrams constructed
for
each outcrop do not detail thefull
sequence of reactions and the assemblages are thought to have equilibratedwith fluids
of variable composition (Spear, 1982). Thus even these relativelywell
described amphibolites are unlikely to provide a complete enough record to enable comparison with the calculated phase relations.Similarly, reaction textures might be expected to yield information about the
applicability of the calculated grid to the phase relations in natural
rocks.
Later in this thesisI
discuss a set of low-variance rerction textures
from
amphibolites in the Harts Range areaof cen¡al
Australia (Chapter4).
These rocks provide evidence of the reactions:kyanite + hornblende
=
staurolite + anorthite,Chapter3-CFMASII- 36
and
þanite
+ orthoamphibole=
cordierite + plagiocl ase' kyanite + hornblende=
cordierite + plagioclasestaurolite + hornblende
=
garnet + orthoamphibole'However, the multi-phase metamorphic history and the complexiry of the structural relations between the compositional units in the area mean that the interpretation of the pressure- temperature history of these rocks is by no means
trivial.
Later in this thesis the calculatedCFMASH
gridwill
be used to interpret these reaction textules and thus the P-T history of the area (Chapter 4).3.4.3 Phase diagrams developed by other workers
Anappealingaspectofthecalculatedgridisthatitisessentiallytopologicallyequivalent
ro rhat of Froese andHall
(1983) which was developedfor
aluminous amphibolites' The Froese andHall
(1983) grid was constnrctedfrom
a combination of published pressure-temperature information
for
stable invariant points in potassium-poor pelites and in mafic rocks which were then linked by GFMASH reactions and invariant points using Schreinemakersanaiysis.
Stight inco;tsistencies between the two grids stem from Froese andHall's
(1983) assumptions that the phases involved have fixed compositions, that garnet is more Fe-rich than co_existing staurolite and that orthoamphibore is more Fe-rich than the staurolite-hornblendetie-line.
These assumptions aleconfary
to the phase relaúonsin
some observed assembiages (e.g. Purtscheller and Mogessie, 1984; Schreyer etal''
1984)'Spear and Rumble (1986; see also Spear, 197S) constructed an
NCFMASH
P-T Schreinemakers net for the phases observedin
the Post Pond and Ammonoosuc Volcanics' They used fixed composition phases and the AV and AS of the reactions to determine theslopes
of
the discontinuous reactions. The invariant points were positioned by comparisonwith
peak metamorphic conditions determined from natural assemblages (Spear&
Rumble'1936). Because the grid calculated here and the
NCFMASH
gdd of Spear and Rumble (1986) involve different compositional systems, they are not directly comparable (however' seesection 3.6
for
a qualitative extension intoNCFMASH)'
A
problem encountered in the comparisonof
the calculated phase relations and those determined from natural amphiborites is the continued assertion that staurolite-hornblende assemblages occur under high pressure conditions of)
6 kba¡ (e'g' Selverstone etal''
1984;Grew
&
Sandiford, 1gg5). The calculated phase diagram (Fig. 3.a) suggests that staurolite- hornblende may be stable under highPII
ratios, but only at limited pressures' up toapproximately 7
kbar.
The P-T grids of Froese andHall
(1983) and Spear and Rumble (1986) also predict that hornblende-staurolite assembrageswill
be limited to a highPft
stability freld which is closed to higher pressure by a reactioninvolving
staurolite, gafnet' hornblende'aluminosilicate(+chloriteororthoamphiboleforSpear&Rumble,1986)withanorthite,
quütz
and aqueous vapour in excess. However, there is a large discrepancy in the maximum ChaPter3-CFMASII- 37pressure at which staurolite-hornblende is stable. spear and Rumble (1986) predict that staurolite-hornblende
will
be a stable assemblage at plessures up to-14
kbar, whereas the grid calculated here predicts a maximum pressure of-
7 kbar (Fig'3'a)'
This problem is add¡essed in later sections.3.4.4 SummarY
The comparison between calculated phase equilibria and those observed in real rocks or predicted by other methods is largely encouraging, suggesting that, topologically' the
calculated grid provides a reasonable approximation of natural rock systems' However' Some problems are