Dear all,
I am a novice user of GULP and this is my first simulation to run with it.
I am trying to use the core-shell model with the Buckingham potential implemented in GULP to optimize the unit cell of BaTiO3. I am facing an “ERROR?!!!” saying
“**** Unit cell is not charge neutral ****
**** Sum of charges = -3.4960000000 ****
**** Check that a special position atom ****
**** coordinate has not been varied ****”.
The reason I am saying it is an error is that I am not seeing the end coordinates of the simulation.
But unfortunately this is not working, a far as i think. Because the screen output looks like this. out.out (10.2 KB)
Is this actually an error? If yes, what might be causing it?
Why are there more atoms in the output than expected? I thought I was simulating only 5 atoms (Ba + Ti + 3 O), which should give 10 particles with shells, but the output shows more.
Am I optimizing the lattice parameters isotropically or anisotropically? Is the current setup allowing the a and c lattice parameters to vary independently (anisotropic)? Or are they being constrained to maintain a fixed c/a ratio (isotropic)?
How can I fix this issue and successfully optimize the unit cell?
Dear all,
I think I solved this issue by deleting the last Oxygen atom definition in the fractional section because it was symmetric to the before last oxygen atom I suppose. The software requires the definition of ONLY the atoms at the asymmetric positions.
I ran into the next error saying:
"
Start of bulk optimisation :
!!!
!! ERROR : Largest core-shell distance exceeds cutoff of cuts
!!!
Largest core-shell distance = 0.9850 Angstroms
Program terminated by processor 0 in cutscheck
"
Any ideas what I should change or if it is related to the edit I used in the input file?
I am guessing the issue comes from not excluding the interactions between the cores and the shells? Could this be the reason?
After increasing the cutoff of the potential using the cuts command, the simulation went on to the end almost. The forces are still high, but it proceeded
As you’ve discovered, GULP expects only the asymmetric unit atoms to be input if you are using symmetry, as demonstrated in example1 for alumina.
The issue with your optimisation is probably not the upper cutoff, but the lower one. You’ve specified in your original bto.gin file that the minimum and maximum cutoffs are both 10 Angstroms. This means that the potentials won’t act at all as there is no distance where they are allowed. Putting rmin to 0 will solve this. Your results will still be very sensitive to the upper cutoff due to the unphysically large C6 term between oxygens.
A final important point is that if you think through the physics of the model, then it’s well known that a shell model can only stabilise cubic or rhombohedral BaTiO3 (depending on the choice of polarisability). There is no stability field for tetragonal BaTiO3 with a standard shell model.
Regards,
Julian
Dear Professor Gale,
Thanks a lot for your suggestions. I will try to implement the idea with the lower cutoff. I think I already did, but i am not sure. When I have certainly done it, I will publish it in case someone might need it.
As for the physics of perovskites, I am aware of your opinion about that. Still a lot of people report using the exact same potential in the literature and show how well it can capture the phase transitions and yield c/a ratios similar to PBEsol and study even finite temperature phenomena. I am not appealing to what they say as I don’t have enough knowledge to pick one side over the other. I will read for sure and wish you a nice evening.
Bets regards,
Mahmoud
Dear Mahmoud,
I’m not sure what other studies you are referring to & so you’ll need to post the references. However, I can tell you that it is physically impossible for the model you are using to reproduce a tetragonal distortion in GULP. You are using a spherically symmetric short-range potential with a dipolar shell model. This means there is nothing of quadrupolar symmetry, which is what a tetragonal distortion requires. You can prove this to yourself by optimising the system and computing the gamma point phonons as a function of spring constant for the shell model. What you should find is a transition between a situation where the cubic phase is phonon stable to one where there are 3 imaginary modes (i.e. rhombohedral distortion) as the spring constants decrease. To get a tetragonal distortion from a standard shell model would require modification in a way that artificially biases the outcome, such as geometrical constraints, making the spring constant depend on the shell direction (GULP doesn’t allow this) or changing atom labels so that different potentials act in different directions (again not a physical mechanism). Anyway, if you can post any references that claim to violate the laws of physics then we’ll see if we can debunk the myth.
Regards,
Julian
Regarding PbTiO₃, I initially thought that its two-phase system might simplify the challenges, but from your explanation, I now understand the fundamental limitation lies in the quadrupolar nature of tetragonality itself. Is this interpretation correct?
I apologize if my initial premise was oversimplified—I’m still in the process of thoroughly reviewing the literature and deeply value your expertise in clarifying these nuances.
Dear Mahmoud,
Thanks for providing the references. I’ve looked at the first paper you mention (Vielma and Schneider) & can largely reproduce the numbers in their Table 2 using their parameters, though given the sizeable C6 term the answers are sensitive to cutoff choice, which isn’t in Table 1.
The issue here is that the calculations at 0 K were symmetry constrained & so you can get different answers for the different structures, if you fix the symmetry. However, if you analyse the gamma point phonons the problem can be spotted. The cubic structure has 3 imaginary modes, while the tetragonal structure has 2 imaginary modes. What this means is that neither structure is dynamically stable and both would spontaneous transform to the more stable rhombohedral structure. As expected, the shell model doesn’t give a tetragonal structure unless you artificially constrain the structure such that it has no choice.
The same thing will be true for any system, including PbTiO3; a dipolar shell model can’t give a quadrupolar distortion as required for stability of the tetragonal structure.
Regards,
Julian