![[up]](mini-home.gif){kind=small}
Practical Aspects of Mineral Thermobarometry
Phase diagrams and pseudosections with THERMOCALC
The old pages on this topic are now entirely obsolete, and major revision is needed. I intend to provide here some examples of approaches to pseudosection calculation that do not attempt to duplicate THERMOCALC's own documentation, but rather reflect our own experience here in Oxford.
Pseudosections:
monitoring assemblage and mode changes in a specified bulk
composition
A pseudosection is a diagram that explicitly shows the assemblage in a system of specified bulk composition as a function of two independent variables. The mineral proportions and mineral compositions are also uniquely defined at any point on the pseudosection, so that divariant fields, for example, can be contoured for mineral mode or composition. The pseudosection is the graphical expression of Duhem's Theorem, which says that a system of specified bulk composition has two (and only two) degrees of freedom. It is one of the more important predictive tools, and one of the best practical uses in petrology of a two-dimensional sheet of paper!
Constructing pseudosections
The THERMOCALC documentation provides instructions on how to get started, including practical exercises from THERMOCALC workshops, or you can try this worked example from Julie Baldwin (University of Montana) on the SERC pages.
Some construction tips
Don't be tempted to leap in and calculate any curve you might think useful, hoping to sort out later which ones (or which parts of ones) are stable. Be more systematic. Try to locate any low-variance bands, perhaps by comparison with existing comparable pseudosections, or by trying some free-energy-minimisation traverses. Then work systematically outwards from these.
Things to remember:
- At the corner of an assemblage field on a pseudosection, four lines meet. These lines separate four fields, each differing from its neighbour by one phase, gained or lost. Therefore, if the lowest-variance field has p phases, it is flanked by two fields having p-1 phases, while the fourth (opposite) field has p-2 phases.
- The corners are where the modes of two phases go to zero, so the P-T positions of these points can be precisely calculated in THERMOCALC.
- The approximate locations of corners can be found when you calculate along a low-variance boundary (p against p-1) and the calculation stops when another phase runs out (exit code 7).
- However, if you're calculating along a higher-variance boundary (p-1/p-2) you won't know if an extra phase should appear. These boundaries, then, have metastable parts extending through the intersection.
Tips for dealing with complex cases
If it looks as if a number of boundaries are going to converge on a very small area of P-T space ...
- Sketch the arrangement of boundaries and intersections, at a much-enlarged scale, as you calculate them.
- Work out the variance of the assemblage formed if all the phases involved do occur together. For example in MnCNKFMASH (C=9) you might have a tight tangle involving the full assemblage g st ctd chl pl cz mu q V (P=9), so the minimum variance is 2 and you may expect one (and only one) divariant field where they meet.
- A consequence of each field boundary being where the mode of a phase goes
to zero is that you can view the entire pseudosection as a web of intersecting
curves marking the stability limit in that bulk composition for each phase.
You may have noticed that some limiting curves are often arranged in
sub-parallel pairs, like (in pelites) st-out and ctd-out, while others are
more independent, like the limits for g or pl. Two phases cannot share the
same limiting curve (unless it's a univariant, which we might then prefer to
think of as an infinitely narrow field with two edges).
Therefore, for each phase p, the p-out boundary traces a continuous and unique path through the diagram. So, test possible arrangements by following each limiting curve through the tangle, making sure no sections are shared and all boundaries are accounted for.
This page last modified 18 December 2007