John Buckeridge

Thanks to Alex Squires for finding a bug when the total concentration of a defect is frozen. There was a problem with the neutral charge state concentration, when one charge state dominated (over all other non-zero charge states) the total concentration. The neutral concentration ended up being zero due to the omission of a term in the denominator (essentially of Eq. 20 in the Comput. Phys. Commun. paper). This has now been rectified.

Thanks to Ben Morgan for pointing out a bug in the DOS integrations - an additional 'slice' was included erroneously when summing up to the Fermi level. The differences are very minor but the bug has now been fixed.

Corrected bug which made the routine get stuck in some cases where large exponentials were encountered. A hard-coded large number is used to avoid overflow, but sometimes there were problems. Now the routine is better able to deal with large numbers and avoids those regions in the solution space. Also - corrected a couple of lines so that the code can be compiled with ifort and gfortran.

Found a series of bugs associated with calculated concentrations at very low T. The corrections are very minor, and low T will not be of much interest to anyone studying defect formation anyway. Also, the input has changed so that the lattice vectors are read from a file called 'unitcell.dat' (if that file is not present, but a VASP POSCAR is, the POSCAR can still be read).

Fixed another minor bug - in some cases where the exponential terms are large, the routine could fail to find the solution as step sizes were too large. This problem has been fixed by checking that, when the exponentials are large, if the charge neutrality condition is not changing as the Fermi level changes, then a more refined search is performed around the point where changes are observed (the old routine would continue scanning across the full energy range then report that no solution could be...

Found a bug in the new version of 'frozen-sc-fermi', where fixed total concentrations could lead to negative neutral defect concentrations. The error occured due to an incorrect term in a summation. This error has been fixed as far as I can tell (please let me know if you find any bugs!)

Found a bug in the part of the calculation that determines the lowest energy charge...