When I see a problem with a bunch of jagged lines like I see in your diagram, my first thought is to check that I have added enough ferric iron to the bulk composition. The system composition basis set for that file use the system components iron (FE) and oxygen (O), and does not use an explicit ferric iron system component that is different than ferrous iron. Thus, to control the ferric/ferrous iron in your bulk composition case, you use oxygen. Think of controlling iron valence using FeO and FeO1.5. The O(?) tells Theriak-Domino to add the proper amount of oxygen using the default valence of the elements given in the database file header section, so it adds O to the FE to give all FeO, no FeO1.5. If you use the O(?) feature, you can then explicity add some 'extra' O to your bulk composition by adding a second O(...) entry.
What is the consequence of doing a calculation with a bulk system composition with just FeO when using that (or some other) database file? The program will try to find solutions using the ferric-bearing species in the database file when they should not be stable. The problem is that, numerically, the ferric-bearing species in some of the ax models are predicted to have a very small amount stable, and the program ends up going in wrong directions trying to find a solution, as in this case. If you run theriak in the problematic space, the program sometimes recognizes something is wrong and it usually tells you so in the output (look for double stars in the output section listing the solid solution phase compositions).
Thus, I always try to add a bit more ferric iron to the bulk system composition to see if this resolves the issue, and it usually does (but not always, the jagged lines could be due to other problems). In your case, you have 0 ferric iron. If you want to work on a ferric-free calculation using that database file, then you should go through the file and deactivate/comment the ferric bearing members. This problem is discussed on this website
under the "Ferric iron problems" section.
The lack of ferric iron in your bulk composition while keeping ferric database species active in your file does appear to be the cause of the problem in your case.
If you wanted FeO1.5/(FeO+FeO1.5) = 0.06, your bulk compo to 3 dec digits would be entered:
0 SI(67.870)AL(15.772)FE(3.303)MG(2.750)MN(0.025)CA(0.948)NA(5.725)K(3.043)TI(0.441)H(100.000)O(?)O(0.099) * js TD Forum