Multiferroic tunnel junctions

[1ight0ne]

Hello! I have a problem with build up that kind of junction. I try to build Fe-BaTiO₃-Fe junction by using various publications. But when i try to optimize this structure it breaks. I think that problem in lattice constant of supercell but I can be wrong.

In attachment you can see my structure and forces and structure from publication.

Can you help me build this junction?

In addition, some times if I do just fit lattice and then begun to calculate, it say error: 

Calculating Eigenvalues    :
** Back Engine Exception : Diagonalization error, overlap matrix not positive definite. This may be caused by atoms that are too close to each other or situated in equivalent positions, or (in the Extended Huckel model) a too low value of interaction_max_range
** Location of Exception : mathutils.cpp:748



This application has requested the Runtime to terminate it in an unusual way.
Please contact the application's support team for more information.

[Anders Blom]

The system is not correctly set up which you can verify by repeating it 2x in A and B. You have vacuum in the cell, and one layer of atoms is repeated.

[1ight0ne]

The system is not correctly set up which you can verify by repeating it 2x in A and B. You have vacuum in the cell, and one layer of atoms is repeated.

I understand that my system is not correct, but how i can build the sytem from publication ?
If I repeat structure in A or B direction it will be wrong structure, but if I delete one atomic layer and fit cell for remove vacuum it will be structure like from pub. But optimization shows very big forces which break structure.
Help me solve this problem.

[Anders Blom]

I doubt it will break since it's highly symmetry-constrained - at most some atoms move a bit on the Z direction. Maybe if you attach the system (what you consider the correct version) I can tell if it actually is correct or not.

[1ight0ne]

I doubt it will break since it's highly symmetry-constrained - at most some atoms move a bit on the Z direction. Maybe if you attach the system (what you consider the correct version) I can tell if it actually is correct or not.

Here is my system. This structure seems to be more stable and no more large forces in various directions, only Z. But I do not quite understand how to properly set the supercell and more exactly the boundaries of the supercell. Because if I set it on the atoms I have one result and if on the length of the atomic bound it will differ.

The main question is how to build up the same structure like in publication?
Does it correct if i repeat my structure in B direction 2x times and then delete one layer and fit Cell in B direction ?

[Anders Blom]

The structure you sent is the one in the publication.

[1ight0ne]

The structure you sent is the one in the publication.

Thanks, but does it correct if i repeat my structure in B direction 2x times and then delete one layer and fit Cell in B direction ?

[Anders Blom]

Certainly not! As I said, the current structure is correct. 2x repeat in B would also be correct, just larger. But if you then delete part of the cell, it becomes a very different structure. If you are in doubt, try centering the structure in A and B (which doesn't change the structure, since the origin is arbitrary in an infinite periodic structure).

[1ight0ne]

Certainly not! As I said, the current structure is correct. 2x repeat in B would also be correct, just larger. But if you then delete part of the cell, it becomes a very different structure. If you are in doubt, try centering the structure in A and B (which doesn't change the structure, since the origin is arbitrary in an infinite periodic structure).

I see... Can you tell me does the repeating is necessary for best results in transport characteristics or i can calculate my structure with one layer and get the same result as in many layers (3 is min ) ?
And one more question.. Why in publication shows two-layers structure (They are performed All calculations in the same ATK) ?

[Anders Blom]

The publication does not show 2 layers - it simply includes equivalent atoms at the opposite edge of the unit cell (for instance with fractional coordinate a=1) to make it easier to see the symmetries. Those atoms can't be included in the calculation, as they are on top of the existing atoms (e.g. the one at a=0).

Compare the two pictures for the conventional cell of Si below. The first one can be used in a calculation, the second can not.

[1ight0ne]

Some times ago we have discussion about how to set polarization (FE/NFE state) in MFTJ. My calculations based on results of that pub with doi: 10.1021/nl803318d
In result, I spent many time to set FE state in ATK software, but I dont got needed result. For NFE state I have some result but not sure in reability, one reason is that i have different results for GGA and LDA (see attachment).
To set FE state i try to displace atoms and set small bias but geometry optimization gave me constantly different wrong results.

Help me solve this problem with setting FE state!
See my structure in attachments, I build it with parametrs like in the publication, make interface in ATK builder with bothstrain option.

[1ight0ne]

Hello! Can anyone answer my question or help me with that problem ?

The main question is how to set ferroelectric state with ATK ? Is this possible at all ?
All my calculations lead to para phase in structure.

Maybe it connected with LCAO technique that do not allow to optimize super-cell ?

[Umberto Martinez]

Hello,

To set FE state i try to displace atoms and set small bias but geometry optimization gave me constantly different wrong results.

can you please clarify what do you mean here with "wrong results"? zero displacement?
Also how do you displace the atoms and add the small bias? (entire script maybe?)

to answer to your last question, how to setup ferroelectronic state, you can apply a small voltage and optimize as suggested in the old discussion.
In ATK you can optimize the geometry under an applied voltage. It can be a time consuming calculation though.