At 09:34 AM 24/09/2001 +0100, Bruce Yardley wrote:
>...for almost any rock with this assemblage, biotite
>and garnet will be major repositories of Fe2 and so this assemblage will
>itself be the buffer for the oxygen fugacity. As a result it will
>potentially be stable over a very large P-T range. ... So a useless
>thermobarometer but a valuable fO2 buffer!

Agreed except for a semantical point. An fO2 buffer is able to hold oxygen
fugacity constant at specified P and T in respect to reversible loss or
gain of oxygen. The assemblage Grt+Bt+Ms+Mt+Qtz does not meet this
condition because its variance is too large. It will _define_ fO2 (by means
of the equilibrium among the endmembers alm, ann, mu, mt, qtz), but it is
not able to buffer fO2 in the same sense as Ni+NiO or QFM buffer fO2.

>Any additional
>presence of graphite at lower-T's will tend to drive the reaction as
>written to the left, destroying magnetite on heating.

Yes. Using the HP98 database we can put limits on this process too. The
pertinent equilibrium would be
 Mu + 3 Qtz + 2 Mt + Grf  =  Alm + Ann + CO2 .

In the presence of pure CO2 the curve passes through 1 kbar 440 deg C and
10 kbar 490 deg C. If CO2 is diluted down to 10% of the fluid phase, the
corresponding T's are 370 and 400. Above the curve, either Mt or Grf (but
not both) can coexist with Mu and Qtz. Changing Alm to Grt and Ann to Bt
will displace the curve to lower T and expand it to a zone of finite P-T

Cheers, Dugald